Datasets:

---

TomTBT

/

pmc_open_access_xml

like 5

[ "<title>Background</title>", "<p>Musculoskeletal conditions are commonly seen in health care practices and have been identified as the leading cause of chronic health problems, long term disability and consultations with health professionals in Canada [##REF##2944216##1##, ####REF##17124551##2##, ##REF##7791173##3####7791173##3##]. Given the fact that many of these conditions are associated with aging (i.e. osteoporosis, osteoarthritis), this burden on society is estimated to increase in the future as the proportion of elderly individuals increases [##REF##1444632##4##]. A 2007 Statistics Canada report revealed that 13.7% of Canadians are 65 years of age or older with the fastest growing group being those between 55–64 years of age [##UREF##0##5##]. In 1998, it was estimated that musculoskeletal issues accounted for approximately 20% of the daily care administered by family practitioners [##REF##12730505##6##]. However, it is also known that signs and symptoms of musculoskeletal conditions may be under diagnosed by primary care physicians as it has been suggested that the examination of this system is often omitted from routine patient assessments [##REF##12730505##6##,##REF##8311541##7##]. This may be related to the notion that some primary care physicians may not feel that, for the most part, they can alter the clinical course by early detection, such as may be the case in osteoarthritis. Studies of medical students and practicing physicians have cited a lack of interest and/or a lack of perceived importance of the musculoskeletal system, time constraints and an overall lack of confidence in providing assessment as reasons for the discomfort with managing patient musculoskeletal conditions [##REF##12730505##6##,##REF##8882045##8##, ####REF##12421993##9##, ##REF##1586238##10##, ##REF##2352196##11##, ##REF##9632059##12####9632059##12##].</p>", "<p>Given the gap between the incidence and diagnosis of musculoskeletal diseases in the primary care population, a simple screening exam may enable practitioners to accurately identify abnormalities of this system. The GALS locomotor screening exam, an acronym that stands for Gait, Arms, Legs and Spine, has been developed by Doherty and colleagues in response to this need [##REF##1444632##4##]. This 3-minute examination consists of three questions about pain, difficulty dressing and difficulty with stairs, followed by assessment of the appearance and movement of the four regions. Although it cannot be considered a substitute for a more detailed locomotor exam, it may be useful as a diagnostic tool for the identification of musculoskeletal abnormalities and possible subsequent early intervention.</p>", "<p>To date, the GALS exam is primarily carried out by rheumatologists or musculoskeletal specialists as a teaching tool [##REF##12421993##9##,##REF##10976078##13##]. However, its introduction to the medical school curriculum in Britain and recently in Canada has proven to be beneficial. Medical students taught the GALS examination reported feeling more confident when assessing the locomotor system and, when evaluated by rheumatology consultants in an examination, performed the screen with the same degree of skill as other clinical areas (i.e. chest examination, blood pressure) [##REF##10976078##13##,##UREF##1##14##]. The GALS screening exam is ideally suited for the family physician who has the opportunity, as a first contact in the health care system, to identify musculoskeletal disorders since it can be easily incorporated into a routine physical exam. However, while the examination is proven to be both valid and reliable when conducted by specialists in rheumatology, its effectiveness in the primary care setting has not yet been determined [##REF##8311541##7##,##REF##17013854##15##]. The aims of this pilot study were 1) to evaluate the accuracy of the GALS examination by primary care physicians as compared to rheumatologists; 2) to test a sampling frame method for participant recruitment and 3) to determine if the instructional DVD is sufficient in teaching the GALS exam.</p>" ]

[ "<title>Methods</title>", "<title>Participants</title>", "<p>Potential study participants from a local academic family practice clinic were selected through a database generated from the electronic medical record, Open Source Clinical Applications Resource (OSCAR) [##UREF##2##16##]. This is a large family practice centre which is representative of the general population. All patients 65 years of age and older who were capable of giving informed consent were considered eligible to participate, regardless of their medical history. International Statistical Classification of Diseases and Related Health Problems (ICD-9) codes were known for each potential study participant in the OSCAR database. The intention was to select approximately half of the subject population with no record of ICD-9 code between 710 and 739 (Diseases of the Musculoskeletal System and Connective Tissue) with the other half having such a record. A list of approximately 261 eligible patients was generated by the database manager and given to the study co-coordinator who was responsible for mailing information letters to all eligible participants. The first 50 patients with no known musculoskeletal disorders and the first 50 patients with known musculoskeletal disorders who positively responded to a follow-up telephone call inviting them to participate were to comprise the study population. Those willing to take part in the study were then scheduled for a one hour appointment on one of three study days.</p>", "<p>To estimate sample size we attempted to estimate the reliability coefficient with as much accuracy as possible to be certain that the true reliability coefficient was reasonably close to the estimate. We hypothesized that an intraclass correlation coefficient (ICC) between the two groups of physicians (family physicians and rheumatologists) was approximately 0.7. Based on the equation derived from Bonett, we determined that 100 subjects would be needed for an estimated ICC of 0.7 and precision ± 0.10 [##REF##12111881##17##]. Thus, recruitment was closed once the target convenience sample of 100 subjects was achieved. This study was approved by the Research Ethics Board at Hamilton Health Sciences and McMaster University.</p>", "<title>Physician Participation</title>", "<p>Four rheumatologists (AC, RB, PB, WK) agreed to participate in this validation study, all of whom had previous experience using the GALS exam in routine clinical practice. Three family physicians (IB, DC, IS), all members of the Canadian College of Family Physicians (CCFP), from the Stonechurch Family Health Centre also volunteered to participate in the study. Participating family physicians had never been previously exposed to or received training on the GALS examination. An instructional DVD of the GALS exam, endorsed by the Canadian Rheumatology Association, was used as the primary teaching method and was distributed to each family physician 2 months prior to conducting the study. This DVD, which takes approximately five minutes to review, demonstrates a rheumatologist performing the exam, as well as 3 case studies on adult patients with specific abnormalities. All of the family physicians received a call one week prior to the scheduled physical exam date to clarify any questions regarding the exam. No additional training was provided.</p>", "<title>Study Procedures</title>", "<p>Each study participant was assessed by one family physician and one rheumatologist, both of whom were blinded to the medical history of the patient. Family physicians and rheumatologists examined the patients immediately following one another on the same exam day. During the examination, physicians posed three questions to each participant and then proceeded to score each of the 7 components of the GALS examination as being either abnormal or normal as shown in Table ##TAB##0##1##. Details about what features of each of the GALS components were examined and assessed to yield a normal or abnormal appearance or movement are shown in Table ##TAB##1##2##. All examiners were blinded as to the assessments of the other physicians. Family physicians were asked to check the appropriate box in the record form (abnormality: yes or no) and document the observed abnormalities. They were not required to make a diagnosis. Rheumatologists, on the other hand, were also asked to state the presence of absence of abnormalities and, in the case where an abnormality was identified, perform a focused exam to assess the abnormality and make a diagnosis, if possible.</p>", "<p>Analyses were performed to assess the degree of overall agreement (P_observed), as well as the degree of agreement on traits considered to be abnormal (P_positive) and normal (P_negative) between the family physicians' and rheumatologists' scores on the GALS assessment. Kappa statistics and 95% confidence intervals (95% CI) were calculated as a composite of the overall GALS examination as well as for each component of the exam (Gait, Arms, Legs, and Spine).</p>", "<p>In order to assess the usefulness of the GALS examination in identifying abnormalities not previously detected in routine family practice, electronic charts (e-charts) including family physician and radiographic reports and referrals to physical therapy and tertiary care (i.e. rheumatologists), were retrospectively reviewed from the family practice clinic. Each chart was evaluated by two independent individuals. Physical abnormalities identified by the rheumatologists were sought in the chart history during the preceding 2 years; those that had not been previously documented were considered \"new\" abnormalities/diagnoses. Due to the difficulty in differentiating between acute and chronic pain during the examination, pain noted by rheumatologists on the GALS assessment form but not in the patient chart was not considered to be a new abnormality. In addition, all abnormalities recorded during the GALS examinations were reviewed by two rheumatologists (AC, RB). Those that could be investigated in further detail or treated, but not found in the patient charts, were considered to be previously undetected. All newly detected abnormalities were then grouped by trait, and further subdivided into affected region including: gait (antalgic, abnormal stride length), arms (fingers and hand, wrist, elbow, shoulder), legs (toes and feet, ankle, knee, hip, other), and spine (scoliosis, lordosis, kyphosis, decreased range of motion, other). Analyses were performed to assess the total number of abnormalities in each region were considered to be new as compared to those that were not, as well as the relative contribution of new abnormalities from each of gait, arms, legs and spine to the total. Data were analyzed using SPSS 12.0 for Windows XP Professional (SPSS Inc; Chicago, IL).</p>" ]

[ "<title>Results</title>", "<p>Of 261 individuals who were mailed a letter informing them about the study, 221 were reached by phone and invited to participate. Unfortunately however, the database-generated list of eligible patients included very few potential participants with no previous musculoskeletal diagnosis (ICD-9 codes 710–739) thus supporting the evidence of high prevalence of MSK disorders in those over 65 years of age. Regardless of ICD-9 code selection, those who responded positively (N = 103) were scheduled to be seen on one of three study days. Of those 103 scheduled for examination, 99 individuals were seen in the clinic, 92 of whom were identified as having a previously diagnosed musculoskeletal condition. Two individuals who were scheduled but not seen were sick on the exam day while two others did not show for their appointments. Those who did not agree to participate gave various reasons for declining, including illness and unavailability on study days. Of those who consented to participate, 61 (62%) were women and 38 were men with a group mean age of 75.2 (SD = 6.1) years (min. to max.: 65 to 89 years). Of the 99 participants, 7 were assessed by their own family physician (7%) by chance alone. Because the names of study participants' family physicians were not known at the time the study was designed and conducted, this occurred purely by chance. However, it should be noted that family physicians had no advance knowledge of the names of the patients they would be assessing and had no access to the patients' medical charts at any time before, during or after the study.</p>", "<p>Overall, the observed agreement (P_obs) of the GALS examination was 0.698 with a P_posof 0.614 and a P_negof 0.752. The composite GALS score had a coefficient of agreement (estimated Kappa) of 0.3675 (95% CI: 0.3009, 0.4342). Agreement was further subdivided into each component of the GALS exam as outlined in Table ##TAB##0##1##. The number of normal and abnormal features graded by each of the family practitioners and rheumatologists are displayed in Table ##TAB##2##3##. Comparisons between family physicians and rheumatologists are presented in Table ##TAB##3##4##. As shown in the table, agreement between physicians was highest in response to the three questions posed. It is intuitive that, when asked the same question by two different physicians, the patient would answer the same way the vast majority of the time since there is no bias or interpretation introduced by the physician.</p>", "<p>Electronic charts for retrospective review were available for 92 of the 99 participants. Ten (10%) participants were identified with a new gait abnormality, nine requiring further investigation/referral. One-third of participants (N = 30) were identified with ≥1 new arm abnormality, the majority (N = 84%) being in the fingers/hand. Of 31 arm abnormalities, 7 would require further investigation or referral and 23 would be treated if symptomatic. In the legs, 35 (38%) participants had 46 new abnormalities, 18 requiring further investigation and 24 requiring treatment if symptomatic. Of these, 52% (N = 24) were in the toes/foot, 7% (N = 3) in the ankle, 24% (N = 11) in the knee, 15% (N = 7) in the hip and 2% (N = 1) \"other\". The prevalence of new spinal abnormalities was 29% (N = 27), with a total of 40 identified, 14 of which would require further investigation/referral. Scoliosis accounted for 15% of abnormalities (N = 6), kyphosis for 30% (N = 12), loss of lordosis for 20% (N = 8), decreased cervical ROM for 23% (N = 9), DDD for 10% (N = 4) and \"other\" for 3% (N = 1). These results are shown in Table ##TAB##4##5##.</p>" ]

[ "<title>Discussion</title>", "<p>While a few studies have investigated the reliability, sensitivity, and specificity of the GALS examination [##REF##8311541##7##,##REF##17013854##15##,##REF##11723316##18##,##REF##2066919##19##], we believe this to be the first study to investigate its use in primary care by comparing the results of the GALS exam between family physicians and rheumatologists. Results of this pilot study revealed a reasonable level of agreement between rheumatologists and family physicians recently taught to perform the GALS examination via an instructional DVD (estimated Kappa = 0.3675; 95% CI: 0.3009, 0.4342, P_obs= 0.698). Upon further analysis of the individual components of the exam, assessments of gait and arm movement were found to have the greatest level of agreement, while the appearance of the legs and spine were identified as the sources of greatest disagreement. Gait is an extremely important component of the GALS exam since its assessment often contributes information with respect to a patients' propensity to falling [##REF##3205267##20##,##REF##15341555##21##,##REF##15341555##21##].</p>", "<p>To more accurately assess the source of disagreement, positive and negative agreement of all components of GALS were determined. Results revealed that family physicians were more likely to agree with rheumatologists when the trait being assessed was considered normal as opposed to abnormal, as shown in both Tables ##TAB##2##3## and ##TAB##3##4##. Similarly, Hood and colleagues reported greater negative predictive values in the assessment of 200 patients suffering from acute or chronic musculoskeletal conditions, also suggesting that negative or normal traits are more easily identified [##REF##11723316##18##].</p>", "<p>Despite the fact that few studies have examined the use of GALS by different health care professionals, a pattern that has previously emerged, and one that was also noted in the current study, is the variation in the assessment of the appearance of the spine. Plant et al. investigated the reliability of the GALS examination when conducted by senior house officers and registrars in rheumatology (N = 30) and reported the greatest disagreement when scoring of the appearance of the spine [##REF##8311541##7##]. Jones et al. also reported difficulties in the identification of other spinal abnormalities, particularly for lateral cervical flexion [##REF##2066919##19##]. It has been suggested that age-related changes affecting flexibility of the neck and back are the likely source of the difficulties encountered in differentiating normal from abnormal spinal appearance [##REF##8311541##7##,##REF##2066919##19##]. Thus, it is plausible that these and other age-related changes may also contribute to difficulties distinguishing normal from mildly abnormal traits in other components of the GALS examination.</p>", "<p>Further comparisons of our results with those of Plant et al. revealed a similar level of observed agreement; however, the reported reliability (estimated kappa) differed significantly. A well-known and frequently observed trend is that of the relation between reliability and degree of scale complexity (i.e. dichotomous scales, Likert scales etc.) where an increase in the number of possible outcomes (i.e. none, mild, moderate, or severe) results in increased reliability [##UREF##3##22##]. The decreased level of reliability as assessed by the kappa statistics (min = 0.13, max = 0.49) in this study may, in part, be attributable to the dichotomous nature of the scale employed where appearance and movement could be labeled only as normal or abnormal. In contrast, Plant et al. replaced the traditional dichotomous scale with one that allowed examiners to rate features as normal, mildly, moderately, or severely abnormal and subsequently reported higher kappa statistics varying from 0.49 to 0.74 [##REF##8311541##7##].</p>", "<p>Although the results of the current study appear to suggest that difficulties persist in the recognition of musculoskeletal abnormalities, one should be cautioned about making definitive conclusions without acknowledging factors which may have contributed to or limited the observed level of agreement. For instance, a review of patient scoring sheets completed by family physicians and rheumatologists revealed that while both examiners recognized similar patient characteristics, there was discrepancy between the comments recorded and the identification of these features as normal or abnormal. In a given patient, for example, some physicians recorded gait to be abnormal due to an observed limp, while others also noted the presence of a limp but incorrectly labeled this as normal. This observation helps to explain the trend observed in other features of the GALS exam where rheumatologists consistently labeled more features as abnormal than family practitioners as seen in Table ##TAB##2##3##. This may be another example of what some physicians may deem normal, age-related changes, thus assessing the feature as normal, while others would assess the feature as being abnormal relative to a healthy standard. This discrepancy may be linked to differences in the perception of abnormalities between family physicians and rheumatologists. However, given the fact that the recorded observations could not be objectively quantified or assessed as being mildly or moderately abnormal as in the study by Plant et al., these differences ultimately resulted in a decreased level of agreement. It is believed that agreement would have improved significantly had the newly trained family physicians been given an opportunity to directly observe the GALS examination as conducted by a rheumatologist and to meet with rheumatologists prior to the study to discuss characteristics that differentiate normal features from those that are abnormal. By coming to a consensus as to how to score certain features (i.e. the limp), it is anticipated that agreement would have been higher. In addition, variation in scoring between family physicians and between rheumatologists was not assessed. Characteristics of the cohort can also influence the measures of agreement, particularly the kappa statistic. For instance, the lack of symmetry in the study population (i.e. the majority have a musculoskeletal condition) will tend to produce lower kappa values [##REF##2189948##23##,##REF##15733050##24##]. One of the major limitations to this study was the asymmetry in the study population which consisted of only 7 participants who had never been identified with any musculoskeletal conditions by the ICD codes.</p>", "<p>The prevalence of MSK abnormalities in this ambulatory study population was also estimated for each anatomical region. These were further subdivided by the joints that were involved. It was apparent that the most common features assessed as being abnormal by the rheumatologists were those in the joints of the fingers/hands (20.8% of patients) and the toes/feet (19.2% of patients). The vast majority of these abnormalities were cases of osteoarthritis in the peripheral joints, none of which had previously been documented in the patients' family practice charts. Decreased cervical range of motion (9.6% of patients) and abnormal knees (8.8%) were also prevalent in this population and were regions that had not been documented as abnormal by the patients' family physicians. There may be a couple of reasons for these \"new\" abnormalities; a) lack of documentation by the family physician, b) the patient has experienced these problems but not expressed/reported them to his/her family physician. The majority of these newly identified abnormalities would require further investigation (i.e. kyphotic posture being assessed for osteoporosis) or treatment (swollen joint treated with medication). Only one other study has used the GALS exam to investigate the prevalence of MSK abnormalities. This study was conducted in acute and chronic medical in-patients [##REF##11723316##18##]. Here the GALS screening tool was positive (abnormality identified) in 53% of acute patients and 94% of chronic patients where osteoarthritis accounted for the majority of rheumatological conditions identified in the both study populations.</p>", "<p>A future study will include a wider variation in subject ages so as to obtain a sample population without any previous musculoskeletal diagnoses allowing the sensitivity and specificity of the GALS exam to be investigated. This study will also involve the analyses of subgroup of patients who are assessed by all family physicians and all rheumatologists to assess the inter-observer variation. In addition, these results also suggest that an instructional DVD alone may not be the most effective and consistent method of teaching the GALS exam but that the DVD should be accompanied by oral instruction/interaction. This may be more important when instructing physicians who have already developed their skill set or routine as compared to medical students who have little to no background in this area.</p>", "<p>Although the ability of family physicians to assess the MSK system prior to the introduction of the GALS examination was not assessed, our results suggest that family physicians can efficiently use the GALS examination to assess the MSK system, by integrating it into their routine physical exam. Nevertheless, previous studies of medical professionals whose ability to assess the MSK system before versus after learning the GALS examination was evaluated have revealed that physicians' confidence and efficiency in examining the system had increased significantly [##REF##10976078##13##,##UREF##1##14##]. These results suggest that the same may be true for the family physicians of the current study.</p>" ]

[ "<title>Conclusion</title>", "<p>A report from the Summit on Standards for Arthritis Prevention and Care, a large multidisciplinary group including health care professionals and patients, clearly identified a need for a screening exam that could facilitate the identification of musculoskeletal diseases [##UREF##4##25##]. The results of this study support the notion that the GALS examination is a simple, reliable, and valid screening tool that may improve recognition of musculoskeletal abnormalities when used by specialists and family physicians alike. However, it also appears as though further validation needs should be considered by investigating different methods of training and scoring in a more diverse study population. This will be the focus of a future larger study. Although it cannot serve as a replacement for a focused rheumatological exam, it is believed that the use of the GALS exam in primary care settings may lead to increased detection of previously unrecognized abnormalities and early intervention in hopes of preventing further deterioration. For example, the early diagnosis of rheumatoid arthritis is associated with better long term outcome [##REF##8546524##26##]. While detection of musculoskeletal conditions has significantly improved over the last decade, additional emphasis must be placed on educating primary care physicians to differentiate between normal age-related changes of the musculoskeletal system and early signs of deterioration that are mildly abnormal in nature as these are typically overlooked.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>As the proportion of the Canadian population ≥65 grows, so too does the prevalence of musculoskeletal (MSK) conditions. Approximately 20% of visits to family physicians occur as a result of MSK complaints. The GALS (Gait, Arms, Legs, and Spine) screening examination was developed to assist in the detection of MSK abnormalities. Although MSK exams are primarily performed by rheumatologists or other MSK specialists, expanding their use in primary health care may improve the detection of MSK conditions allowing for earlier treatment. The primary goal of this study was to evaluate the use of the GALS locomotor screen in primary care by comparing the results of assessments of family physicians with those of rheumatologists. The secondary goal was to examine the incidence of MSK disorders and assess the frequency with which new diagnoses not previously documented in patients' charts were identified.</p>", "<title>Methods</title>", "<p>Patients ≥65 years old recruited from an academic family health centre were examined by a rheumatologist and a family physician who recorded the appearance of each participant's gait and the appearance and movement of the arms, legs and spine by deeming them normal or abnormal. GALS scores were compared between physicians with the proportion of observed (P_obs), positive (P_pos) and negative (P_neg) agreement being the primary outcomes. Kappa statistics were also calculated. Descriptive statistics were used to describe the number of \"new\" diagnoses by comparing rheumatologists' findings with each patient's family practice chart.</p>", "<title>Results</title>", "<p>A total of 99 patients consented to participate (92 with previously diagnosed MSK conditions). Results showed reasonable agreement between family physicians and rheumatologists; P_obs= 0.698, P_pos= 0.614 and P_neg= 0.752. The coefficient of agreement (estimated Kappa) was 0.3675 for the composite GALS score. For individual components of the GALS exam, the highest agreement between family physicians and rheumatologists was in the assessment of gait and arm movement.</p>", "<title>Conclusion</title>", "<p>Previously reported increases in undiagnosed signs and symptoms of musculoskeletal conditions have highlighted the need for a simple yet sensitive screening exam for the identification of musculoskeletal abnormalities. Results of this study suggest that family physicians can efficiently use the GALS examination in the assessment of populations with a high proportion of musculoskeletal issues.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>KAB, AC, AP, DC and RB were responsible for the development and design of the study. AC, RB, PB and KW were the rheumatologists involved in the assessment of all patients. IB, DC, and IS were the family physicians involved in the assessment of all patients. GI was charged with all statistical analyses. KAB, JO and RMcC were involved in participant recruitment and organization of the study. All authors were involved in the writing of the manuscript. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2474/9/115/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>The authors would like to thank the funding source for this project, the Institutional Capacity Enhancement (ICE) Pilot Project Grant: Injury Prevention Across the Lifespan (IPALS) (originally funded by CIHR), ICH 63069, and its Principal Investigator, Dr. Parminder Raina. We would also like to thank the Stonechurch Family Health Centre for their collaboration.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>GALS Recording Sheet Completed by Physicians</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Yes</bold></td><td align=\"center\"><bold>No</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Do you have any pain or stiffness in your muscles, joints or back?</bold></td><td/><td/></tr><tr><td align=\"left\"><bold>Do you have any difficulty dressing yourself completely?</bold></td><td/><td/></tr><tr><td align=\"left\"><bold>Do you have difficulty walking up or down stairs?</bold></td><td/><td/></tr><tr><td align=\"left\"><bold>Gait</bold></td><td align=\"center\" colspan=\"2\">Abnormal or Normal</td></tr><tr><td/><td align=\"center\"><bold>Appearance (✔ or ✘)</bold></td><td align=\"center\"><bold>Movement (✔ or ✘)</bold></td></tr><tr><td align=\"left\"><bold>Arms</bold></td><td/><td/></tr><tr><td align=\"left\"><bold>Legs</bold></td><td/><td/></tr><tr><td align=\"left\"><bold>Spine</bold></td><td/><td/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Individual features of the GALS exam which were examined (6)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>GAIT</bold></td></tr><tr><td align=\"left\">• Symmetry &amp; smoothness of movement</td></tr><tr><td align=\"left\">• Stride length &amp; mechanics</td></tr><tr><td align=\"left\">• Ability to turn normally &amp; quickly</td></tr></thead><tbody><tr><td align=\"left\"><bold>ARMS (Hands)</bold></td></tr><tr><td align=\"left\">• Wrist/finger swelling/deformity</td></tr><tr><td align=\"left\">• Squeeze across 2^ndto 5^thmetacarpals for tenderness (indicates synovitis)</td></tr><tr><td align=\"left\">• Turn hands over, inspect muscle wasting &amp; forearm pronation/supination</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\"><bold>ARMS (Grip Strength)</bold></td></tr><tr><td align=\"left\">• Power grip (tight fist)</td></tr><tr><td align=\"left\">• Precision grip (oppose each finger to thumb)</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\"><bold>ARMS (Elbows)</bold></td></tr><tr><td align=\"left\">• Full extension</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\"><bold>ARMS (Shoulders)</bold></td></tr><tr><td align=\"left\">• Abduction &amp; external rotation of shoulders</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\"><bold>LEGS (Feet)</bold></td></tr><tr><td align=\"left\">• Squeeze across metatarsals for tenderness (indicates synovitis)</td></tr><tr><td align=\"left\">• Calluses</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\"><bold>LEGS (Knees)</bold></td></tr><tr><td align=\"left\">• Knee swelling/deformity, effusion</td></tr><tr><td align=\"left\">• Quadriceps muscle bulk</td></tr><tr><td align=\"left\">• Crepitus during passive knee flexion</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\"><bold>LEGS (Hips)</bold></td></tr><tr><td align=\"left\">• Check internal rotation of hips</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\"><bold>SPINE (Inspection from behind)</bold></td></tr><tr><td align=\"left\">• Shoulders &amp; iliac crest height symmetry</td></tr><tr><td align=\"left\">• Scoliosis</td></tr><tr><td align=\"left\">• Paraspinal, shoulder, buttocks, thighs &amp; calves muscles normal</td></tr><tr><td align=\"left\">• Popliteal or hind foot swelling or deformity</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\"><bold>SPINE (Inspection from front)</bold></td></tr><tr><td align=\"left\">• Quadriceps normal in bulk &amp; symmetry</td></tr><tr><td align=\"left\">• Swelling or at Varus or valgus deformity at knee</td></tr><tr><td align=\"left\">• Forefoot of midfoot deformity, action normal</td></tr><tr><td align=\"left\">• Ear against shoulder on either side to check lateral cervical spine flexion</td></tr><tr><td align=\"left\">• Hands behind head with elbows back (check rotator cuff muscles, acromioclavicular, sternoclavicular &amp; elbow joints)</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\"><bold>SPINE (Inspection from side)</bold></td></tr><tr><td align=\"left\">• Normal thoracic &amp; lumbar lordosis</td></tr><tr><td align=\"left\">• Normal cervical kyphosis</td></tr><tr><td align=\"left\">• Normal flexion (lumbosacral rhythm from lumbar lordosis to kyphosis) while touching toes</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\"><bold>SPINE (Trigger point tenderness)</bold></td></tr><tr><td align=\"left\">• Supraspinatus muscle tenderness (exaggerated response)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Number of features graded normal and abnormal for each patient</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\"><bold>Primary Care Physician</bold></td><td align=\"center\" colspan=\"2\"><bold>Rheumatologist</bold></td></tr><tr><td/><td colspan=\"4\"><hr/></td></tr><tr><td/><td align=\"center\"><bold>Normal (N)</bold></td><td align=\"center\"><bold>Abnormal (N)</bold></td><td align=\"center\"><bold>Normal (N)</bold></td><td align=\"center\"><bold>Abnormal (N)</bold></td></tr></thead><tbody><tr><td align=\"left\">Arms – Appearance</td><td align=\"center\">66</td><td align=\"center\">33</td><td align=\"center\">62</td><td align=\"center\">36*</td></tr><tr><td align=\"left\">Arms – Movement</td><td align=\"center\">62</td><td align=\"center\">36*</td><td align=\"center\">67</td><td align=\"center\">30*</td></tr><tr><td align=\"left\">Legs – Appearance</td><td align=\"center\">53</td><td align=\"center\">43*</td><td align=\"center\">45</td><td align=\"center\">53*</td></tr><tr><td align=\"left\">Legs – Movement</td><td align=\"center\">43</td><td align=\"center\">54*</td><td align=\"center\">47</td><td align=\"center\">52</td></tr><tr><td align=\"left\">Spine – Appearance</td><td align=\"center\">73</td><td align=\"center\">25*</td><td align=\"center\">57</td><td align=\"center\">41*</td></tr><tr><td align=\"left\">Spine – Movement</td><td align=\"center\">62</td><td align=\"center\">36*</td><td align=\"center\">52</td><td align=\"center\">45*</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Agreement between Family Physician &amp; Rheumatologist GALS Scores</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>P_POS</bold></td><td align=\"center\"><bold>P_NEG</bold></td><td align=\"center\"><bold>P_OBS</bold></td><td align=\"center\"><bold>Estimated Kappa (95% CI)</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Pain/Stiffness</bold></td><td align=\"center\">0.944</td><td align=\"center\">0.758</td><td align=\"center\">0.910</td><td align=\"center\">0.704 (0.500, 0.908)</td></tr><tr><td align=\"left\"><bold>Difficulty Dressing</bold></td><td align=\"center\">0.731</td><td align=\"center\">0.904</td><td align=\"center\">0.858</td><td align=\"center\">0.636 (0.444, 0.832)</td></tr><tr><td align=\"left\"><bold>Difficulty on Stairs</bold></td><td align=\"center\">0.909</td><td align=\"center\">0.911</td><td align=\"center\">0.910</td><td align=\"center\">0.821 (0.694, 0.947)</td></tr><tr><td align=\"left\"><bold>Gait</bold></td><td align=\"center\">0.676</td><td align=\"center\">0.813</td><td align=\"center\">0.784</td><td align=\"center\">0.490 (0.310, 0.670)</td></tr><tr><td align=\"left\">Arms – Appearance</td><td align=\"center\">0.617</td><td align=\"center\">0.793</td><td align=\"center\">0.742</td><td align=\"center\">0.412 (0.222, 0.601)</td></tr><tr><td align=\"left\">Arms – Movement</td><td align=\"center\">0.634</td><td align=\"center\">0.821</td><td align=\"center\">0.760</td><td align=\"center\">0.458 (0.271, 0.646)</td></tr><tr><td align=\"left\">Legs – Appearance</td><td align=\"center\">0.574</td><td align=\"center\">0.583</td><td align=\"center\">0.578</td><td align=\"center\">0.164 (0.000, 0.359)</td></tr><tr><td align=\"left\">Legs – Movement</td><td align=\"center\">0.711</td><td align=\"center\">0.666</td><td align=\"center\">0.690</td><td align=\"center\">0.379 (0.196, 0.563)</td></tr><tr><td align=\"left\">Spine – Appearance</td><td align=\"center\">0.400</td><td align=\"center\">0.697</td><td align=\"center\">0.597</td><td align=\"center\">0.128 (0.000, 0.314)</td></tr><tr><td align=\"left\">Spine – Movement</td><td align=\"center\">0.632</td><td align=\"center\">0.743</td><td align=\"center\">0.697</td><td align=\"center\">0.385 (0.204, 0.566)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Prevalence of newly detected abnormalities by subcategory</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Subcategory</td><td align=\"center\">Prevalence <break/>[% in category (% overall)]</td></tr></thead><tbody><tr><td align=\"left\"><bold>Gait</bold></td><td/></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\">Stride Length</td><td align=\"center\">90.0 (7.2)</td></tr><tr><td align=\"left\">Antalgic</td><td align=\"center\">10.0 (0.8)</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>Arms</bold></td><td/></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\">Fingers/hand</td><td align=\"center\">83.9 (20.8)</td></tr><tr><td align=\"left\">Wrist</td><td align=\"center\">3.2 (0.8)</td></tr><tr><td align=\"left\">Elbow</td><td align=\"center\">3.2 (0.8)</td></tr><tr><td align=\"left\">Shoulder</td><td align=\"center\">9.7 (2.4)</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>Legs</bold></td><td/></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\">Toes/foot</td><td align=\"center\">52.2 (19.2)</td></tr><tr><td align=\"left\">Ankle</td><td align=\"center\">6.5 (2.4)</td></tr><tr><td align=\"left\">Knee</td><td align=\"center\">23.9 (8.8)</td></tr><tr><td align=\"left\">Hip</td><td align=\"center\">15.2 (5.6)</td></tr><tr><td align=\"left\">Other</td><td align=\"center\">2.2 (0.8)</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>Spine</bold></td><td/></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\">Scoliosis</td><td align=\"center\">15.8 (4.8)</td></tr><tr><td align=\"left\">Kyphosis</td><td align=\"center\">26.3 (8.0)</td></tr><tr><td align=\"left\">Loss of lumbar lordosis</td><td align=\"center\">21.1 (6.4)</td></tr><tr><td align=\"left\">Decreased cervical ROM</td><td align=\"center\">31.6 (9.6)</td></tr><tr><td align=\"left\">Degenerative disk disease</td><td align=\"center\">2.6 (0.8)</td></tr><tr><td align=\"left\">Other</td><td align=\"center\">2.6 (0.8)</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>✔ = normal, ✘ = abnormal</p></table-wrap-foot>", "<table-wrap-foot><p>*Please note that these features contain missing data (i.e. N(normal) + N(abnormal) ≠ 99).</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Over the past decade, the treatment of cervical cancer has evolved registering a gradual abandonment of radical surgery in favor of more conservative approaches: this becomes even more relevant considering that approximately 15% of all cervical cancers, and 45% of surgically treated stage IB cervical cancers occur in women &lt; 40 years of age [##REF##14551007##1##]. These figures are expected to increase due to the widespread use of cervical cancer screening which results in overall younger age and an earlier stage of disease at diagnosis. In addition, more and more frequently women defer childbearing, so that an increasing number of women would be diagnosed cervical cancer before having started or completed their reproductive program. Among the uterus preserving techniques, radical vaginal trachelectomy (RVT) with laparoscopic pelvic lymphadenectomy [##UREF##0##2##] has gained acceptance over the years by the gynecologic oncology community due to the favorable results in terms of oncological and obstetrical outcome [##REF##16739746##3##].</p>", "<p>Among the strict criteria employed in the selection of cases who can potentially be offered uterus preserving approaches, tumor histology <italic>per se </italic>seems not to be a relevant factor [##REF##12548192##4##], with the exception of rare histological types such as adenosquamous, neuroendocrine tumors or glassy cell carcinomas which have been generally associated with a higher risk of recurrence [##REF##14675676##5##,##REF##12831923##6##], and considered a contraindication to conservative treatment [##REF##17291273##7##,##REF##17534391##8##]. In particular, glassy cell carcinomas first described by Glücksmann and Cherry [##REF##13364884##9##] in the uterine cervix, are typically composed of malignant cells showing a moderate amount of cytoplasm with \"ground glass\" appearance, distinct cell membranes stained with eosin or periodic acid-Schiff, and large nuclei with prominent nucleoli. These tumors have been considered since the beginning as an uncommon variant of poorly differentiated adenosquamous carcinoma [##REF##13364884##9##], endowed with resistance to radiation therapy and unfavorable prognosis [##REF##1541437##10##].</p>", "<p>To our knowledge, only three cases of glassy cell carcinomas undergoing conservative treatment by laparoscopic pelvic lymphadenectomy and radical vaginal trachelectomy have been reported [##REF##15350349##11##].</p>", "<p>Here, we report the case of a stage IB1 cervical glassy cell carcinoma patient, who was safely treated with cold knife conization plus laparoscopic pelvic lymphadenectomy.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<p>We report a case of early stage glassy cell cancer in a patient, who was conservatively treated by conization and laparoscopic pelvic lymphadenectomy. Indeed, among the fertility preservation approaches to early stage cervical carcinoma, RVT has gained much attention because of the recognized oncologic efficacy and safety. Intra- and postoperative complications have been reported to be approximately 4% and 12% of cases, respectively [##REF##17534391##8##], and even less radical procedures such as conization plus laparoscopic pelvic lymphadenectomy have been investigated in selected cases of stage IB1 squamous cell carcinoma &lt; 2 cm diameter [##REF##17291273##7##]. While the fertility preserving procedures are widely accepted for tumors with squamous histological type, and also adenocarcinomas, which <italic>per se </italic>should not be considered a contraindication to conservative treatment, some concerns have been raised for rare histological types such as adenosquamous, neuroendocrine or glassy cell carcinomas. In particular, conservatively treated neuroendocrine and adenosquamous tumors have been reported to carry out a very unfavorable prognosis [##REF##14675676##5##,##REF##12831923##6##]. On the other hand, very few data about early stage glassy cell cervical cancer have been reported: of 3 cases treated with laparoscopic pelvic lymphadenectomy and RVT, all were reported as having no evidence of disease at time of publication [##REF##15350349##11##]. No case of early stage glassy cell carcinoma treated with conization plus laparoscopic pelvic lymphadenectomy has been reported until now.</p>", "<p>Despite the extensive counseling about the possibility to perform trachelectomy or adjuvant treatment after final diagnosis, our patient decided only to undergo strict follow-up procedures, and is currently without evidence of disease after 38 months since initial diagnosis.</p>" ]

[ "<title>Conclusion</title>", "<p>We report a case of an early stage glassy cell cervical carcinoma patient, who was successfully treated with conization and laparoscopic pelvic lymphadenectomy. Given the rarity of this tumor histological type, and the paucity of data about its natural history, which has been reported to be similar to other histological types only with the employment of multimodal treatment strategies [##REF##11972387##12##], caution should be taken to i) carefully evaluate the patients' fertility potential; ii) extensively counsel the patients about the risk/benefit of a conservative treatment; iii) investigate the patients' compliance to undergo strict follow-up procedures.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Very little data about the conservative treatment of early stage glassy cell cervical cancer have been reported.</p>", "<title>Case presentation</title>", "<p>A 30-year old patient, nulligravida was admitted to the Gynecologic Oncology Unit of the Catholic University of Campobasso for irregular post-coital vaginal bleeding. The patients was staged as having FIGO stage IB1 (tumor diameter = 2 cm) squamous cervical cancer. After extensive counseling of the patient and her family, laparoscopic pelvic lymphadenectomy and cold knife conization were performed. The final diagnosis was FIGO Stage IB1 glassy cell carcinoma. Currently, after a follow-up of 38 months, she has no evidence of disease.</p>", "<title>Conclusion</title>", "<p>We reported a case of early stage glassy cell cancer patient, who was conservatively treated by conization and laparoscopic pelvic lymphadenectomy.</p>" ]

[ "<title>Case presentation</title>", "<p>A 30-year old patient, nulligravida was admitted in March 2005, to the Gynecologic Oncology Unit of the Catholic University of Campobasso, for irregular post-coital vaginal bleeding. Her medical history was unremarkable. Her gynecological history was negative with menarche at the age of 12 years, and regular menses until 6 months before the occurrence of the symptoms.</p>", "<p>Gynaecological examination revealed a normal size uterus, and no adnexal masses. A circumscribed, ulcerated lesion (maximum diameter = 2 cm) was documented in the posterior esocervix. Parametria and vagina appeared uninvolved. Colposcopy-guided biopsy and curettage of endocervical canal were performed revealing an invasive squamous cell cervical carcinoma with areas of poor differentiation. Transabdominal and transvaginal ultrasound examination documented the presence of a normal size uterus showing normal echogenicity with the exception of a vascularized hypoechogenic area (18 × 14 × 11 mm) located in the cervix.</p>", "<p>Staging evaluation including chest X-ray, total body CT scan, and pelvic magnetic resonance imaging (MRI) documented the presence of a tumor mass (maximum diameter = 2 cm) located in the uterine cervix, and no enlarged lymph nodes. Examination under anesthesia revealed an ulcerated lesion of maximum diameter of 2 cm, without vaginal and parametrial involvement. Squamous cell carcinoma antigen levels were negative. The patient was staged as having FIGO stage IB1 cervical cancer.</p>", "<p>After extensive counseling of the patient and her family, she opted for a conservative approach. Open laparoscopy was carried out: peritoneal washing and a careful inspection of the adnexae and intra abdominal organs was performed. Systematic pelvic lymphadenectomy was performed up to internal iliac lymph nodes, and they returned as negative at frozen section examination. Several biopsies of the vaginal walls were obtained; these were negative for disease on frozen section. A cold knife conization was performed, and frozen section analysis showed that the lateral and deep margins of the tissue specimen were uninvolved. The biopsy of the endocervical canal also resulted negative at frozen section.</p>", "<p>At definitive pathological examination, a nodular lesion of maximum diameter of 2.0 cm (width extension) located in the cone (height = 2 cm, width = 3 cm), was detected. Microscopic examination revealed a tumor composed of nests of large cells with large eosinophilic cytoplasm presenting a ground-glass appearance (Figure ##FIG##0##1##). Cell membranes were easily recognizable, and tumor nuclei appeared large, presenting prominent nucleoli, and also areas of abundant eosinophil infiltration were present. The tumor showed a stromal invasion of 8 mm out of 1.7 stromal thickness. The lateral and deep margins of the cone were uninvolved for at least 9 mm. All peritoneal biopsies, as well as pelvic lymph nodes (n = 18) were negative. No lymphovascular space involvement was observed. The final diagnosis was FIGO Stage IB1 poorly differentiated carcinoma with &gt; 90% of the tumor represented by neoplastic cells with glassy cell features. A second pathologist, blinded to the first's impression confirmed the diagnosis. Given the rarity of this histological type and its prognostic features, therapeutic options including radical trachelectomy, hysterectomy, or adjuvant treatment were carefully discussed with the patient, who nevertheless decided to undergo only strict follow-up procedures. The patient was then followed with gynecological examination, pap smear, and colposcopy every 3 months for the first 2 years, and every 6 months thereafter, and was also requested to perform chest x-ray and pelvic MRI every year. Currently, after a follow up of 38 months, she has no evidence of disease.</p>", "<p>Cervical stenosis was documented after 21 months since surgery, and was easily managed by cannulation of the cervical canal under anesthesia.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>GF conceived of the study, participated in its design and drafting. VS participated in the design of the study and collected the clinical data. MP participated in the design of the study and collected the clinical data. AC carried out the histopathological evaluation. GS conceived of the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.</p>", "<title>Consent</title>", "<p>Written informed consent was obtained from the patient for publication of this case report and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.</p>" ]

[]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Glassy cell carcinoma of the cervix: the undifferentiated, glassy cells display large nuclei with prominent nucleoli and granular cytoplasm.</bold> Areas of abundant eosinophils infiltration are present. (Hematoxylin &amp; Eosin, magnification: 200×).</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1477-7819-6-92-1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Total mesorectal excision (TME) is the standard surgical treatment used for patients with primary rectal cancer. TME involves removal of a distinct anatomic compartment, the mesorectum, containing the rectal tumor, all local draining nodes and the mesorectal fat, by means of sharp dissection along the mesorectal fascia [##REF##1365683##1##, ####REF##7489148##2##, ##REF##6751457##3####6751457##3##]. There is substantial evidence for efficacy of neoadjuvant therapy in combination with TME as being important to reduce local tumor recurrence rates [##REF##11884047##4##, ####REF##11547717##5##, ##REF##11550163##6##, ##REF##8094488##7####8094488##7##]. When performing TME, knowledge of the relationship of the tumor to the circumferential resection margin (CRM) is of importance. When CRM is involved by the tumor, the risk of local recurrence is high [##REF##11872058##8##, ####REF##11859207##9##, ##REF##11923599##10##, ##REF##9715152##11##, ##REF##8696739##12##, ##REF##7915774##13##, ##REF##8443747##14##, ##REF##1691810##15##, ##REF##2430152##16####2430152##16##]. The local prognostic factors assessed at preoperative magnetic resonance imaging (MRI) of rectal cancer include the extent of extramural tumor spread, involvement of the lateral resection margin, involvement of neighboring organs in the pelvis, presence of local lymph node metastases, extramural lymphovascular infiltration and peritoneal involvement [##REF##1691810##15##,##REF##7995147##17##]. This information helps select patients who should receive neoadjuvant treatment. This applies especially to cases with locally advanced rectal cancer, in order to maximize the chances of a complete resection and survival [##REF##8665179##18##,##REF##7489174##19##], and at the same time, to minimize morbidity and loss of quality of life. It is therefore of paramount interest to provide detailed anatomic knowledge of tumor and tumor invasion toward neighboring organs before treatment.</p>", "<p>Although evaluated in several studies during the past two decades, it is only during recent years that MRI gained wide acceptance as a valuable method for assessment in patients with rectal cancer [##REF##15865021##20##, ####REF##14736677##21##, ##REF##12904995##22##, ##REF##15286305##23##, ##REF##11229667##24##, ##REF##10189474##25##, ##REF##10525811##26##, ##REF##9112909##27##, ##REF##9191437##28##, ##REF##8915293##29##, ##REF##8448394##30##, ##REF##2911656##31##, ##REF##3720456##32##, ##REF##3486559##33####3486559##33##].</p>", "<p>As a tertiary referral center responsible for patients with advanced rectal cancer, we assess magnetic resonance (MR) examinations from other institutions and hospitals at multidisciplinary team (MDT) meetings. When demonstrating these examinations at MDT meetings, variations in imaging sequences among different centers are noted. These differences may be related to both different equipments and level of dedicated experience in pelvic MRI.</p>", "<p>To our knowledge, no study has reported the importance of the imaging protocol for assessment of tumor involvement of neighboring organs in locally advanced rectal cancer. The aim of the present study was to compare the equivalence between MRI and histopathology in patients with locally advanced rectal cancer based on the effects of using different MRI protocols.</p>" ]

[ "<title>Patients and methods</title>", "<p>Forty-one patients assessed as clinically suspicious for locally advanced primary rectal cancer by surgeons from 2000 to 2005, were included. 37 patients, 27 male and 10 female, with a mean age of 60.1 ± 9.8 (mean ± SD, range 28–79) who had available MRI of the pelvis were studied further. The surgeon's decision that a cancer might be advanced was based on findings at diagnostic laparotomy and/or by means of digital rectal examination.</p>", "<title>Radiological assessment</title>", "<p>All examinations were provided from ten different hospitals or institutions (two of which were university hospitals). Each MR examination (all done on 1.5 T) was assessed by two or three radiologists (C.T., M.R.T. and L.B.) in consensus without knowledge of the clinical and histopathological results prior to this study according to a standard evaluation looking specifically at which organs and/or structures had been involved. However, the radiologists were aware of the high suspicion for locally advanced tumors by the clinicians. Radiologists had evaluated the morphological characteristics of the primary tumor, local prognostic factors including threatening or involvement of the mesorectal fascia, and adjacent organs in each patient.</p>", "<p>For the part of this study, anterior organs were defined as those positioned ventral to the rectum and included the seminal vesicles, the prostate gland, the perineal body, uterus, vagina, ovaries, the small and large intestines, and the urinary bladder. Inferior and posterior organs had been defined as those that were located inferior and dorsal to the rectum, respectively, and included the levator ani muscles, obturator muscles, piriformis muscles and the sacral bone. Involvement of the abovementioned organs was defined as T4-tumor stage.</p>", "<p>The imaging protocol of each MR-examination was recorded by one author (C.T.). Those examinations that showed the following prerequisites were defined as compliant rectal imaging protocol vs. those that did not demonstrate the same sequences (called henceforth noncompliant):</p>", "<p>1. Sagittal and axial T2-weighted images of the pelvis performed,</p>", "<p>2. T2-weighted images with equal to or less than 3 mm slice thickness perpendicular to the rectal length at the level of the tumor with a 16–20 cm field of view and at least a 256 × 256 matrix, otherwise called 'high resolution imaging' [##REF##15865021##20##,##REF##14736677##21##,##REF##10189474##25##,##REF##15730990##34##].</p>", "<p>3. For low rectal tumors, coronal imaging obtained.</p>", "<p>If the patients underwent MR examinations twice but at two different institutions, with different protocols, one compliant and the other non compliant; these were noted separately as combination protocol but categorized with the compliant group regarding some aspects. The number of other sequences and different types of artifacts (if distinguishable) were also noted.</p>", "<p>The common denominators of all MR examinations, whether compliant or otherwise, were that they had to be performed on the request of a surgeon or oncologist for assessment of local extension of the rectal tumor preoperatively, and that the radiologist at the primary institution had not called the examination incomplete.</p>", "<title>Histopathological examination</title>", "<p>All evaluations were performed according to the protocol of Quirke, et al [##REF##2430152##16##,##REF##3045231##35##], by one pathologist (J.L.) with more than 10 years of experience in gastrointestinal pathology. The pathologist was blinded to the MRI study protocol. The tumor site was sliced transversely at 0.5–1.0-cm intervals. The extent of tumor spread into mesorectal fascia and other structures or organs was assessed both macroscopically and with high magnification. Tumor extension into the surrounding structures and organs at microscopical examination were used as the standard of reference against which MRI findings were compared. The extension of tumor cells into mesorectal fascia and other structures or organs was assessed from inspection of the histological macrosection by light microscopy at 20× – 200× magnification.</p>", "<title>Statistical analysis</title>", "<p>All MRI findings including the size of tumor, the name and number of involved fascia(e) and organ(s), the pattern of tumor involvement according to MRI and histopathology as well as the MR imaging protocol were recorded using Microsoft Excel 2003 and Microsoft Access 2000. Sensitivity and specificity of MRI between different groups were compared and 95% confidence interval (CI) was calculated with P-value &lt; 0.05 considered significant using Stat View J-5.0 (SAS Institute. Inc., Cary, NC).</p>", "<title>Ethical considerations</title>", "<p>The study was approved by the local ethical committee. No separate informed consent was obtained for this retrospective study.</p>" ]

[ "<title>Results</title>", "<title>Tumor staging according to MRI</title>", "<p>Nineteen patients were evaluated as T4 rectal tumors based on MRI. The remaining 18 were evaluated as T3 tumors without obvious invasion of neighboring organ.</p>", "<title>Assessment of imaging quality</title>", "<p>Eleven patients were assessed as having compliant (D) protocols and 13 patients as combination protocols (C) and 13 patients a noncompliant imaging (N).</p>", "<p>Regarding imaging parameters, compliant imaging protocols were used with smaller field of view (FOV) (D, 201.7 ± 77.0 mm; N, 263.5 ± 129.8 mm; mean ± SD, p = 0.03), thinner slice thickness (D, 3.8 ± 1.4 mm; N, 5.3 ± 1.9 mm; mean ± SD, p &lt; 0.01), smaller slice gap (D, 0.2 ± 0.9 mm; N 2.0 ± 2.4 mm; mean ± SD, p &lt; 0.01) and smaller voxel size (D, 1.3 ± 1.5 mm³; N, 6.7 ± 6.0 mm³; mean ± SD, p &lt; 0.01). The total number of MR sequences performed in each patient was also larger in the N group (N, 9.2 ± 3.2 sequences vs. D, 5.2 ± 0.7 sequences; mean ± SD, p &lt; 0.01 (table ##TAB##0##1##). One patient from the noncompliant group had some motion artifacts.</p>", "<title>Involvement of the anterior organs</title>", "<p>In the group with compliant protocols and the group with combination protocol, preoperative MRI indicated tumor involvement of anterior pelvic organs in seven out of the 24 patients. Compared to pathological examination, six cases were true positives and one was false positive. Among the remaining 17 patients without organ involvement on MRI, pathological examination revealed one false negative case and 16 true negatives (table ##TAB##1##2##). Figure ##FIG##0##1## demonstrates the false-negative case. In this case, there appears to be no continuity between the tumor and the uterus. However, histopathological examination showed tumor invasion along the fascia, reaching the posterior wall of the uterus and the left adnexa. The radiologist failed to ascertain the anterior extension of the tumor correctly.</p>", "<p>In the noncompliant imaging group, preoperative MRI was indicative of organ involvement in eight cases. Pathological examination revealed two as true positives and six as false positives (Figure ##FIG##1##2##). Among the remaining five patients without organ involvement, pathological examination revealed two false negatives and three true negatives.</p>", "<p>Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) in the compliant and combination protocol group were 85.7%, 94.1%, 85.7%, and 94.1%, respectively. On the other hand, in the group with non-compliant protocol, the sensitivity, specificity, PPV and NPV were 50.0%, 33.3%, 25.0%, and 60%, respectively. Statistically significant difference (p &lt; 0.05) was observed regarding measured specificity (95% CI; 7–70 for group N vs. 95% CI; 71–99 for the other two groups, D and C). The difference in sensitivity in the two groups did not reach statistical significance levels (Table ##TAB##1##2##).</p>", "<title>Posterior or inferior organ involvement</title>", "<p>Only three out of the present 19 patients with locally advanced tumor, showed involvement of an inferior organ (levator ani muscle, piriformis muscle) or a posterior organ (Os sacrum) by the tumor, without simultaneous involvement of any anterior organ. Two of these patients used compliant imaging, and pathological examination revealed both to be true positives. In one patient with noncompliant imaging an inferior organ involvement was suspected but pathological examination proved no obvious tumor infiltration or fibrosis in that organ (false-positive). The number of cases was too few to make any meaningful statistical analysis.</p>" ]

[ "<title>Discussion</title>", "<p>The results of this study indicate considerable differences in correlation between preoperative imaging and histopathology depending on the imaging protocol. Using compliant imaging, despite fewer imaging sequences, a considerably better prediction of tumor invasion towards anterior pelvic organs is seen. On the contrary, this study also indicates that MRI performed with noncompliant imaging protocol does not allow accurate prediction. One other observation is that the radiologist tends to over-stage when the imaging protocol is not optimal. This could be due to the fear of positive resection margins caused by a false negative assessment and partial volume effect observed with thick slices not obtained in the appropriate planes. This could of course be due to nature of the study as well. The radiologists assessing the MR exams were aware of the selection criteria and might have felt compelled to over-stage.</p>", "<p>The lack of compliant imaging, and as we suspect the lack of high resolution T2-weighted imaging, probably forced the radiologists to rely on images with considerable volume averaging. Compared to the compliant imaging, both slice thickness including gap and voxel size were significantly larger in the noncompliant imaging group (<italic>P </italic>&lt; 0.05). Larger slice thickness and gap yield more partial volume effect, thus leading the radiologists to make overestimation of tumor extent. In areas of the pelvis where there are small interfaces between tissues, such as in the anterior and low part of the rectum, this is probably of particular importance. In the compliant and combination groups, there was one false positive and one false negative finding of anterior organ involvement out of 24 cases.</p>", "<p>In the noncompliant imaging group, there were six false positive and two false negative cases out of 13 cases. This means that one patient out of 24 from D and C groups and six patients out of 13 from the N group might receive unnecessary extensive surgery and prolonged, preoperative chemoradiotherapy. Anterior pelvic organs are closely related to urinary and sexual function, and anterior organ surgery has great impact on the patient's quality of life after surgery. By contrast at least partially because of false negative assessments by radiologists, one out of 24 cases from D and C groups, and two out of 13 cases from the N group had involved resection margins.</p>", "<p>Although the low number of cases prohibits any meaningful analysis to be done regarding accuracy of MRI for assessment of organs inferior or dorsal to rectum, our findings suggest that compliant imaging might be superior to noncompliant imaging also for these patients. This low frequency could be due to less likelihood of involvement of posterior organs compared to anterior organs due to more distance between rectum and these neighboring organs [##REF##15906032##36##].</p>", "<p>The number of MR sequences was different between various groups with larger numbers observed in the noncompliant imaging group. It seems that whenever the compliant sequences were not employed, there was a tendency to conduct several other sequences. One of the most widely used sequences in the N group was the one with usage of gadolinium intravenous contrast. Recently, Vliegen and others have shown that gadolinium-enhanced MRI does not improve the diagnostic accuracy in local staging of rectal cancer [##REF##15550372##37##]. Unnecessary use of contrast agents might only lead to increased rate of adverse events and increased costs and time needed for examination, without any proven benefit for the patients.</p>", "<p>There are a number of other limitations in this study. First, we did not compare the same patients using different imaging protocols.</p>", "<p>Second, there was a difference in the sensitivity of MR examinations using different protocols when assessing detection of anterior organ involvement, however, the difference did not reach statistical significance which is probably due to the low power of the study and perhaps the nature of the study (i.e. the radiologists knew that these cases were more likely to be advanced cases).</p>", "<p>However, even with these limitations, the compliant imaging improves accuracy, especially in advanced and complicated cases. It is therefore of utmost importance that radiologists are made aware of pitfalls and the problems, and that radiologist are made up-to-date about recent developments in imaging. This current study reveals that there is a need for continued education in this field.</p>" ]

[ "<title>Conclusion</title>", "<p>For local staging of locally advanced rectal cancer, the correlation between MRI and histopathology was better when a predefined compliant rectal imaging protocol was used. It is possible that this also holds true for all patients assessed with rectal cancer and not only for anterior structures in the pelvis. However, this has to be assessed in further studies. Furthermore, this study indicates that continuous training of radiologists and radiology technicians, including work-shops and seminars seems to be an appropriate way to improve accuracy of MRI in patients with rectal cancer.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Magnetic resonance imaging (MRI) is used for preoperative local staging in patients with rectal cancer. Our aim was to retrospectively study the effects of the imaging protocol on the staging accuracy.</p>", "<title>Patients and methods</title>", "<p>MR-examinations of 37 patients with locally advanced disease were divided into two groups; compliant and noncompliant, based on the imaging protocol, without knowledge of the histopathological results. A compliant rectal cancer imaging protocol was defined as including T2-weighted imaging in the sagittal and axial planes with supplementary coronal in low rectal tumors, alongside a high-resolution plane perpendicular to the rectum at the level of the primary tumor. Protocols not complying with these criteria were defined as noncompliant. Histopathological results were used as gold standard.</p>", "<title>Results</title>", "<p>Compliant rectal imaging protocols showed significantly better correlation with histopathological results regarding assessment of anterior organ involvement (sensitivity and specificity rates in compliant group were 86% and 94%, respectively vs. 50% and 33% in the noncompliant group). Compliant imaging protocols also used statistically significantly smaller voxel sizes and fewer number of MR sequences than the noncompliant protocols</p>", "<title>Conclusion</title>", "<p>Appropriate MR imaging protocols enable more accurate local staging of locally advanced rectal tumors with less number of sequences and without intravenous gadolinium contrast agents.</p>" ]

[ "<title>Abbreviations</title>", "<p>MR(I): Magnetic resonance (imaging); TME: Total mesorectal excision; CRM: Circumferential resection margin; T2-w (image): T2 weighted (image); FOV: Field of view; MDT: Multidisciplinary team; PPV: Positive predictive value; NPV: Negative predictive value; TR: Repetition Time; TE: Echo Time; NEX: number of excitations.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>CS idea, data collection, radiological assessment, manuscript preparation. MT idea, data collection, radiological assessment, manuscript preparation. ST idea, data collection, surgical and clinical assessment, histopathological evaluation, manuscript preparation. GP idea, data collection, surgical and clinical assessment, manuscript preparation. TH idea, data collection, surgical and clinical assessment, histopathological evaluation, manuscript preparation. JL idea, data collection, histopathological evaluation, manuscript preparation. LB idea, supervision, manuscript preparation. All authors read and approved the final version</p>" ]

[ "<title>Acknowledgements</title>", "<p>The authors wish to thank Roberto Vargas, R.T. for his outstanding technical support and knowledge of MRI and all the colleagues the in Department of the Diagnostic Radiology, Karolinska University Hospital, Solna.</p>", "<p>This study is supported partially by ALF project funding, Stockholm County Council and Karolinska Institute.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>MR images of the 'false negative' case in the group with a compliant protocol</bold>. A-63-year-old female with rectal cancer involving the mesorectal fascia, peritoneal reflection and the parietal pelvic fascia. Imaging parameters: TR; 4056, TE; 130, NEX; 2, Thickness; 5 mm, Gap; 0 mm, FOV; 240 mm. (a) Sagittal T2-w image of the pelvis. Primary lesion is located at the rectosigmoid junction with an extramural component, extending dorsally toward the presacral fascia (arrowhead). The tumor seems to be very distant from the inner genitalia (arrow). b-e) Axial T2-w images demonstrated in a craniocaudal direction with b being the uppermost image. In b, the extramural component reaches and thickens the peritoneal fold (arrow), and more inferiorly even the pelvic side wall fascia (arrowheads in c). This fascial thickening continues (arrowheads in d, 15 mm below b), until it sweeps forward (arrow in e, 25 mm below b) and at this point the inner genitalia were involved. At the first glance, there appears to be no continuity between the tumor and the mesorectal fascia, however, histopathological examination proved tumor cells inside the fibrotic tissue and infiltrating the uterine parenchyma and the left adenxa (arrowhead in e).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>MRI of the false positive case in the group with a noncompliant protocol</bold>. A 76-year-old male with rectal cancer suspected of invasion to the urinary bladder. Imaging parameters: TR 7000; TE 132; NEX 2; thickness 5 mm; gap 1.5 mm; FOV 400 mm. (a) Sagittal T2-WI of the pelvis. The large primary lesion (asterisk) originating from the upper part of rectum with accompanying desmoplastic and edematous changes seems to be invading the muscular wall of the bladder dorsally (white arrows). The tumor appears to penetrate into the muscular layer of the urinary bladder which shows higher signal intensity compared to the normal part. (b) Sagittal contrast-enhanced T1-WI of the pelvis with fat-suppression. The posterior bladder wall is not distinguishable, yet the tumor is seen enriching ventrally (white arrowheads) and therefore, it is suspicious for penetrating into the bladder wall. (c-f) Corresponding axial images. c, e, and f are T2-WI and d is T1WI with contrast-enhancement and fat-suppression. T1-w images after Gadolinium contrast enhancement with fat saturation give the impression of the tumor (asterisk) growing into the dorsal wall of the urinary bladder (arrowheads). However, histopathological examination revealed no tumor involvement of the urinary bladder.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Comparison of various MR imaging parameters, average number of sequences in each group and imaging protocols.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"left\"><bold>Compliant protocol (D)</bold></td><td align=\"left\"><bold>Noncompliant protocol (N)</bold></td><td align=\"left\"><bold><italic>P</italic>-value</bold></td></tr><tr><td align=\"left\"><bold>Parameters on T2-WI*</bold></td><td/><td/><td/></tr></thead><tbody><tr><td align=\"left\">Field of view</td><td/><td/><td/></tr><tr><td align=\"left\"> Mean ± SD (mm)</td><td align=\"left\">201.7 ± 77.0</td><td align=\"left\">263.5 ± 129.8</td><td align=\"left\">0.03</td></tr><tr><td align=\"left\">Slice thickness</td><td/><td/><td/></tr><tr><td align=\"left\"> Mean ± SD (mm)</td><td align=\"left\">3.8 ± 1.4</td><td align=\"left\">5.3 ± 1.9</td><td align=\"left\">&lt; 0.01</td></tr><tr><td align=\"left\">Gap</td><td/><td/><td/></tr><tr><td align=\"left\"> Mean ± SD (mm)</td><td align=\"left\">0.2 ± 0.9</td><td align=\"left\">2.0 ± 2.4</td><td align=\"left\">&lt; 0.01</td></tr><tr><td align=\"left\">Matrix size</td><td/><td/><td/></tr><tr><td align=\"left\"> Mean (mm × mm)</td><td align=\"left\">0.5 × 0.5</td><td align=\"left\">0.9 × 1.1</td><td align=\"left\">0.02</td></tr><tr><td align=\"left\">Voxel size</td><td/><td/><td/></tr><tr><td align=\"left\"> Mean ± SD (mm³)</td><td align=\"left\">1.3 ± 1.5</td><td align=\"left\">6.7 ± 6.0</td><td align=\"left\">&lt; 0.01</td></tr><tr><td align=\"left\">No. of sequence</td><td/><td/><td/></tr><tr><td align=\"left\"> Mean ± SD (mm)</td><td align=\"left\">5.2 ± 0.7</td><td align=\"left\">9.2 ± 3.2</td><td align=\"left\">&lt; 0.01</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Comparison of various MR protocols in terms of diagnostic accuracies regarding involvement anterior to rectum.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Compliant and</bold><break/><bold>combination</bold><break/><bold> protocol (D and C)</bold></td><td align=\"center\"><bold>Noncompliant protocol (N)</bold></td></tr><tr><td align=\"left\"><bold>Imaging accuracies</bold></td><td/><td/></tr></thead><tbody><tr><td align=\"left\">True positive</td><td align=\"center\">6</td><td align=\"center\">2</td></tr><tr><td align=\"left\">True negative</td><td align=\"center\">16</td><td align=\"center\">3</td></tr><tr><td align=\"left\">False positive</td><td align=\"center\">1</td><td align=\"center\">6</td></tr><tr><td align=\"left\">False negative</td><td align=\"center\">1</td><td align=\"center\">2</td></tr><tr><td align=\"left\">Sensitivity (%) (95% CI)</td><td align=\"center\">85.7 (42–99)</td><td align=\"center\">50.0 (6–93)</td></tr><tr><td align=\"left\">Specificity (%) (95% CI)</td><td align=\"center\">94.1 (71–99)</td><td align=\"center\">33.3 (7–70)</td></tr><tr><td align=\"left\">Positive Predictive Value (%) (95% CI)</td><td align=\"center\">85.7 (42–99)</td><td align=\"center\">25.0 (3–65)</td></tr><tr><td align=\"left\">Negative Predictive Value (%) (95% CI)</td><td align=\"center\">94.1 (71–99)</td><td align=\"center\">60.0 (14–94)</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>*T2 weighted image;</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1477-7819-6-89-1\"/>", "<graphic xlink:href=\"1477-7819-6-89-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The occurrence of skin metastases are rare events in the course of a follicular thyroid carcinoma (FTC) and usually indicate advanced tumor stages. The scalp is the most affected area of these metastases [##REF##16121050##1##, ####REF##11072371##2##, ##REF##9848721##3##, ##REF##9092737##4##, ##REF##8935380##5##, ##REF##8257865##6####8257865##6##]. Operations are mostly performed with palliative intention. We present a case with extensive and symptomatic scalp metastases in a female patient. The tumors were resected under general anaesthesia. Mesh graft was successfully used to cover the skin defects.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<p>Follicular thyroid carcinomas (FTC) often spread to bones and lung [##REF##16684830##7##]. The occurrence of cutaneous metastases is a rare event. Many different locations have been decribed as abdomen, back and front thigh [##REF##16208262##8##], but predominantely the skin of head and neck is affected [##REF##16121050##1##, ####REF##11072371##2##, ##REF##9848721##3##, ##REF##9092737##4##, ##REF##8935380##5##, ##REF##8257865##6####8257865##6##]. In a review of the literature Quinn and coworkers [##REF##16121050##1##] found scalp metastases in 9 of 14 patients with cutaneous metastases of FTC. In a study of Erickson and coworkers [##REF##17387744##9##] none of 5 FTC metastatic to the skin showed BRAF(V600E) mutation (T1799A).</p>" ]

[ "<title>Conclusion</title>", "<p>We demonstrate another case with multicentric form. Because of its location and size a primary wound closure was not possible. A healing could be reached using vacuum therapy and mesh graft transplantation. The palliative long term cosmetic and functional result was excellent.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>The occurrence of skin metastases are rare events in the course of a follicular thyroid carcinoma (FTC) and usually indicate advanced tumor stages. The scalp is the most affected area of these metastases.</p>", "<title>Case presentation</title>", "<p>We present a case of a 76 year old Woman with multiple giant scalp metastases of a follicular carcinoma. These metastases had been resected and wounds had been closed with mesh graft. The 14-months follow up is presented.</p>", "<title>Conclusion</title>", "<p>We demonstrate another case with multicentric form. Because of its location and size a primary wound closure was not possible. A healing could be reached using vacuum therapy and mesh graft transplantation.</p>" ]

[ "<title>Case presentation</title>", "<p>A 76-year old female patient had the initial diagnosis of FTC 18 years ago. She had total thyroidectomy with bilateral neck dissection and multiple reoperations for recurrent tumor. Because of an irresectable local recurrence with tracheal infiltration a tracheotomy was performed two years ago. Five sets of internal radiation therapy, had been performed one year ago with a cumulative activity of 55.400 MBq¹³¹I. She was admitted to our hospital because of four intensively vascularized scalp tumors, two of them of hen's egg size (Fig. ##FIG##0##1a, b##, and ##FIG##1##2##) which showed recurrent episodes of contact bleeding during hair dressing. Computed tomography revealed multiple pulmonary, hepatic and bone metastases. Thyreoglobulin level was highly elevated (6750 ng/ml) Nevertheless the patient was in a good general condition. We performed a resection of the scalp tumors under general anesthesia. Histopathology confirmed cutaneous metastases of FTC (Fig. ##FIG##2##3##). The places of resection were primary left for granulation. After achievement of a clean granulation area using vacuum therapy (<italic>V.A.C.^®, KCI International, Amsterdam, The Netherlands</italic>) we performed a mesh graft skin transplant (Fig. ##FIG##3##4a, b##).</p>", "<p>A follow up examination fourteen months later showed a very good cosmetic result with nearly complete healing of the mesh graft transplant (Fig. ##FIG##4##5a, b##). Because the local neck tumor had continued to grow the patient was now convinced to accept external radiation therapy and was admitted to our department of radiation oncology.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>KC had idea to publish the case report and drafted the manuscript, UR was our pathologist and performed the immunohistochemistry, AR and MK helped to to search and analyse thoroughly the literature, AR and WTK performed the initial operation, mesh graft transplantation and follow up examination of the patient. They also initiated the temporary vaccum therapy. All authors read and approved the final manuscript.</p>", "<title>Consent</title>", "<p>Written informed consent was obtained from the patient for publication of this case report and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.</p>" ]

[]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>a) and b) Two scalp tumors at admission of the patient, lateral view</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Giant parietal scalp tumor and to additional smaller tumors, intraoperative view</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Operative specimen, Haematoxylin-Eosin and Thyreoglobulin staining.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Mesh Graft transplants</bold>. a) Retroauricular mesh graft transplant; b) Parietal mesh graft transplant.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>14 months follow up</bold>. a) view from on high; b) lateral view.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1477-7819-6-82-1\"/>", "<graphic xlink:href=\"1477-7819-6-82-2\"/>", "<graphic xlink:href=\"1477-7819-6-82-3\"/>", "<graphic xlink:href=\"1477-7819-6-82-4\"/>", "<graphic xlink:href=\"1477-7819-6-82-5\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Type 2 diabetes mellitus (DM) is one of the most common chronic diseases the world over, and the number of people with DM has risen sharply in recent years [##REF##15111519##1##]. In the United Kingdom (UK), DM affects almost 1.8 million people, representing 3% of the population. There are also up to a million people with undiagnosed (asymptomatic) DM [##UREF##0##2##]. Other European countries report similarly high figures [##REF##12502659##3##]. Individuals with DM are at high risk for cardiovascular disease (CVD). Adequate treatment of DM and associated risk factors such as hypertension and dyslipidemia greatly reduces the risk of complications.</p>", "<p>While it is argued that there is no justification for universal screening for diabetes, there is strong support for screening and early treatment among population subgroups where DM is common and CVD risk is high [##REF##11312236##4##]. Glucose levels are likely to be elevated for 10 years before DM is diagnosed [##UREF##1##5##]. This has led to recommendations for selective screening for DM by the American Diabetes Association (ADA) and Diabetes UK [##UREF##2##6##,##UREF##3##7##]. Current guidelines include questionnaires based on risk factors (e.g. age ≥ 45 years, BMI &gt; 25) or the use of more complex risk scores which require invasive specimen collection [##REF##12032103##8##, ####UREF##4##9##, ##REF##10333936##10##, ##REF##9096967##11##, ##REF##7555482##12####7555482##12##].</p>", "<p>These guidelines have been developed and tested particularly among populations of white European origin. This raises the question of whether they are valid for other ethnic groups as well, in particular populations of South Asian and African origin in Europe. As the prevalence of DM is higher in these groups, it may affect the efficiency of screening. For instance, while type 2 DM in white Europeans usually appears over the age of 40, it often appears before the age of 40 among South Asians and African origin populations [##UREF##0##2##,##REF##10097925##13##, ####REF##10417082##14##, ##REF##9470870##15##, ##REF##2492840##16####2492840##16##]. In addition, the association between DM and its determinants might vary between ethnic groups.</p>", "<p>The limited evidence indeed indicates that a risk score for DM developed for the white population is less efficient among South Asians and Africans [##REF##14693976##17##]. Because of this, some guidelines recognize the necessity of adapting screening programs to ethnicity. The ADA questionnaire includes a question about ethnicity, and suggests screening before the age of 45 among specific groups [##REF##15618112##18##]. However, to our knowledge, no tailored risk score for DM has been developed specifically for populations of South Asian or African origin. Moreover, it is not yet known at what age screening for DM should start, and which selection criteria are most efficient for these ethnic groups. Thus, the aim of our study was to provide information needed to optimize screening criteria for DM among different ethnic groups, e.g. Hindustani and African Surinamese migrants. Given the similarities in geographic and ethnic origin, it is expected that Hindustani Surinamese have much in common with South Asian migrants, and that African Surinamese are very similar to migrants of Afro-Caribbean ancestry in the United Kingdom (UK).</p>", "<p>First, we determined the prevalence of known and newly detected DM in Hindustani Surinamese, African Surinamese, and ethnic Dutch (Dutch) in the Netherlands in two age groups: 35 to 44 and 45 to 60 years. Second, we developed a new risk score, based on ethnicity and biomedical risk factors that do not require invasive specimen collection in clinical practice. Finally, we evaluated the performance of that risk score and compared it with the current criteria derived from guidelines.</p>" ]

[ "<title>Methods</title>", "<p>The study population consisted of participants in the SUNSET study (Surinamese in the Netherlands: Study on health and Ethnicity) [##REF##16208137##19##]. In 1975, almost half the population of the former Dutch colony Surinam migrated to the Netherlands. Approximately 80% of these Surinamese immigrants in the Netherlands are Hindustani ('South Asian', originally from the Indian sub-continent) or African (mixed African, Indian and European, but predominantly of African origin). The SUNSET study is based on a random sample of 2975 individuals, aged 35 to 60 years of age, drawn from the approximately 389000 ethnic Dutch (Dutch) and 72000 Surinamese listed in the Amsterdam population register (figure ##FIG##0##1##). For the sampling procedure, persons who were born in the Netherlands and whose parents were both born in the Netherlands were presumed to be Dutch. Persons of whom both parents were born in Surinam and persons who were born in Surinam and who had at least one parent who was born in Surinam were presumed to be Surinamese participants.</p>", "<p>Between 2001 and 2003, all persons in the sample were approached for face-to-face, structured interviews by trained interviewers who had been matched by sex and presumed ethnicity. The interview included questions on self-identified ethnicity, migration history, demographic variables, lifestyle, and health status. If information on the self-identified ethnicity of the individual was lacking, the origin of the mother, the father and the mother's ancestors were used to classify participants.</p>", "<p>The overall response to the interview was 60% (figure ##FIG##0##1##). Participation rates were higher among women than among men. In addition, participants in the interview were more likely to be married and living with a partner and/or children, and to live in a less urban area (address density of 1500–2500 addresses/km2 vs. ≥ 2500) as compared to non-participants (both non-response and not eligible). However, the absolute and relative differences between participants and non-participants for these characteristics were small and reported trends were similar across ethnic groups (data not shown).</p>", "<p>Participants of Hindustani Surinamese, African Surinamese or Dutch origin were also invited for a physical examination at a local health care centre. During the examination, trained physicians recorded the following characteristics: weight in light clothing on a SECA mechanical scale to the nearest 200 grams and height to the nearest 0.01 meter by wall tape measure. Waist circumference midway between the lower rib margin and the iliac crest and hip circumference at the maximum point over the greater trochanters were determined to the nearest 0.01 meter by tape measure. After the subjects had emptied their bladder and had been seated for at least 5 minutes, blood pressure and resting heart rate measurements were obtained from each subject's arm at heart level using an OMRON-M4 semi-automatic sphygmomanometer with an appropriate-sized cuff. All anthropometric measurements were obtained twice and the means (rounded off to the nearest integer) were used for analysis. Fasting glucose (HK/Glucose-6-P dehydrogenase test; Roche Diagnostics, In), high density lipoprotein cholesterol (HDL, Homogenous enzymatic colorimetric test; Roche Diagnostics, In) and triglyceride (GPO-PAP Enzymatic test; Roche Diagnostics, In) levels were determined in serum samples obtained at the time of the physical examination. DM was defined as fasting glucose ≥ 7.0 mmol/l and/or self-reported DM, excluding the self-reported diagnoses of gestational diabetes.</p>", "<p>The SUNSET-study was approved by the Institutional Review Board of the Academic Medical Centre of the University of Amsterdam, and carried out in compliance with the Helsinki Declaration. All participants provided a written informed consent.</p>", "<p>In the present analysis, we included participants who had participated in both the interview and the physical examination. Of all participants in the interview, 71 were excluded due to missing information on self-identified ethnicity, 182 persons were excluded because they had not undergone the physical exam and 10 persons because fasting glucose measurements were not available (non-response for blood sample). As compared to those who were left in the study, those excluded were similar with regard to gender, self-reported DM and self-rated health (data not shown).</p>", "<p>In total, 1434 participants remained in the study, divided into 339 Hindustani Surinamese, 605 African Surinamese, and 490 Dutch (figure ##FIG##0##1##). Of the Hindustani participants 98.8% were born in Surinam, 99.4 had two parents who were born in Surinam and 92.1% had two parents who were of Hindustani origin. Of the African Surinamese participants, 99.2% were born in Surinam and 99.5% had two parents who were born in Surinam. Moreover, 79.3% of the African Surinamese had two parents who were of African origin.</p>", "<p>Characteristics of the ethnic groups were described using means or proportions. In addition, the prevalence of determinants of DM was calculated, directly standardised to the age distribution of the total population. The association of determinants with DM was studied using univariate logistic regression analysis. All variables showing an association of p ≤ 0.25 for the Wald test were selected for the multivariate analyses. Stepwise multiple logistic regression was then performed to construct the optimal risk score for the occurrence of DM among Hindustani Surinamese, African Surinamese and Dutch. Criteria for entry into and exclusion from the model were a p-value for the likelihood ratio test of 0.05 and 0.10, respectively. A new risk score was developed to determine the probability of having DM with a logistic regression model using data that would be routinely available in general practice. Variables considered were ethnicity, biomedical parameters and disease history, e.g. age, BMI, waist circumference, resting heart rate, first-degree relative with DM, hypertension, history of CVD. The risk score is based on the sum of the score of the variables included in the full model (see additional file ##SUPPL##0##1##: Risk score for DM SUNSET.pdf).</p>", "<p>Subsequently, we evaluated the performance of the risk score and compared it to the sets of screening criteria derived from current guidelines by calculating the area under the Receiver-Operator Characteristic curve (AUC) as a measure of diagnostic accuracy. Before analyzing the data, it was decided to consider an AUC of less than 0.60 to be poor, 0.60–0.75 to be moderate, and higher than 0.75 to be good.</p>", "<p>The cut-off for the risk scores at which fasting plasma glucose screening was indicated, was chosen such that the sensitivity was approximately 80%, but not over. Additionally, we determined the specificity, the total population selected for screening, the prevalence in the screened population (the positive predictive value), and the number needed to screen to detect a case of DM (NNS).</p>", "<p>Finally, we estimated the diagnostic accuracy of the risk scores for the detection of a new case of DM, to simulate a situation where persons with known DM are excluded from screening. This was done because, ideally, only previously unknown cases would have been used in the derivation of the risk score for screening for unknown DM. However, given issues of power to enable statistical modelling, it was decided to base the score on all cases and to then also estimate the diagnostic accuracy of the risk scores for newly detected DM.</p>", "<p>We assessed the ability of a simplified version of our risk score, with points corresponding to the calculated odds ratio, to detect new cases of DM by calculating the number needed to screen to identify a new case of newly detected DM (NNS_new).</p>", "<p>To assess the validity of the risk scores, we used bootstrapping techniques to estimate a 'confidence interval' (bCI) around the estimated AUC of all risk scores in our population. We took 1000 random samples, with replacement, from the study population. At each step the parameters in the predictive models were calculated. We subsequently estimated the AUC for each of the models and calculated the 2.5 and 97.5th percentile to indicate the 'confidence interval'[##REF##11470385##20##].</p>", "<p>All analyses were performed using the the SAS package, version 9.1 (SAS Institute Inc., Cary, USA.).</p>" ]

[ "<title>Results</title>", "<title>Characteristics of the population</title>", "<p>Table ##TAB##0##1## shows the characteristics of the study population by ethnic group. Compared to the Dutch, the Surinamese migrants tended to be congregated at the lower end of the socio-economic structure. People of Hindustani Surinamese origin had more abdominal obesity than the African Surinamese and Dutch.</p>", "<title>Prevalence and determinants of DM</title>", "<p>Overall, Hindustani Surinamese had the highest prevalence of DM 25.6%, followed by African Surinamese 12.7% and Dutch 6.8%. The age-standardised prevalence among these groups was 26,7, 14,2 and 5,5, respectively. The prevalence of known and newly detected DM by age-group is shown in table ##TAB##0##1##. In the age-group 35 to 44 years, the sex-adjusted odds ratio (OR) for DM was 4.6 [2.0–10.9] for Hindustani Surinamese and 1.9 [0.8–4.6] for African Surinamese as compared to the Dutch. In the age group 45 to 60 years, the OR was 6.1 [3.7–10.3] for Hindustani Surinamese and 2.7 [1.6–4.6] for African Surinamese. Hindustani Surinamese and African Surinamese with DM had a higher odds of being detected, i.e. having known diabetes, than the Dutch (Hindustani Surinamese OR: 2.6 [1.1–6.2], African Surinamese OR: 2.7 [1.1–6.7]).</p>", "<p>The determinants of DM in the risk score are shown in Table ##TAB##1##2##. History of CVD (OR 5.4 [2.9–10.3]) and waist circumference (OR 5.3 [3.2–8.6]) showed the strongest association with DM in the univariate analysis. In multivariate analysis, the strongest determinants were history of CVD (OR 3.0 [1.5–6.3]), a first-degree relative with DM (OR 2.7 [1.8–4.2]) and Hindustani Surinamese origin (OR 2.7 [1.7–4.5]).</p>", "<title>Performance of criteria for screening</title>", "<p>Table ##TAB##2##3## shows the performance of four sets of screening criteria in the three population subgroups. Selection based on age alone (set 1) showed the lowest diagnostic accuracy. Application of the risk score (set 4) resulted in a moderate to good diagnostic accuracy: the AUC was 0.74 (0.70–0.79) for the Hindustani Surinamese, 0.80 (0.75–0.85) for the African Surinamese, and 0.78 (0.73–0.85) for the Dutch, with a NNS of 3 among the Hindustani Surinamese, 5 among the African Surinamese, and 7 among the Dutch. Trends were similar when the analysis was restricted to persons with normoglycemia and newly detected DM (Table ##TAB##3##4##).</p>", "<p>In Table ##TAB##4##5##, we listed the performance of the simplified version of the score (as specified in additional file ##SUPPL##0##1##: Risk score for DM SUNSET.pdf), among the population without a prior diagnosis of DM (known DM). The AUC varied between 0.58 among the Hindustani Surinamese to 0.79 among the African Surinamese. At a cut off of 8 points, 13 Hindustani Surinamese, 22 African Surinamese or 13 ethnic Dutch would have to be screened to detect a new case of DM.</p>" ]

[ "<title>Discussion</title>", "<p>To our knowledge, this is the first European study outside the UK, to report on an evaluation of a risk score as a screening test for DM across different ethnic groups. The findings of our study confirm that DM varies strongly across ethnic groups; Hindustani Surinamese had the highest prevalence of DM, followed by African Surinamese and Dutch. Although the Hindustani Surinamese population seemed to have a higher proportion of known DM, the absolute prevalence of newly detected DM was still higher as compared to the other ethnic groups in both age categories. These results indicate that Hindustani Surinamese in particular, but also the African Surinamese, could benefit from screening starting before the age of 45 (i.e. the currently advised threshold).</p>", "<p>The high prevalence of DM among the Hindustani Surinamese and African Surinamese is consistent with studies among South Asian and Afro-Caribbean populations in the UK [##REF##10097925##13##, ####REF##10417082##14##, ##REF##9470870##15##, ##REF##2492840##16####2492840##16##]. Research in the 1990s suggested that for every known case there was another undiagnosed case of DM [##REF##11312236##4##,##REF##7712700##21##,##REF##8612442##22##]. The proportion of newly detected DM in our study was lower. The proportion among the Dutch in our sample (nearly one-third) is in agreement with the UK and the recent data in the USA [##REF##16433718##23##,##UREF##5##24##]. The proportion of newly detected DM among the Surinamese was lower. A higher awareness of the risk of diabetes among Hindustani Surinamese in Dutch clinical practice may contribute (at least in part) to this low proportion, particularly among those aged 45 years and older. The relatively low proportion of newly detected diabetes among African Surinamese has also been found in Afro-Caribbeans in the UK [##REF##9470870##25##]. This may be linked to the high prevalence of hypertension among African origin populations, as clinicians may be triggered to test for elevated (fasting) glucose levels during check-ups for hypertension.</p>", "<p>Despite the relatively small proportion of newly detected DM in Hindustani Surinamese persons, the absolute prevalence of newly detected DM is relatively high in Hindustani Surinamese, particularly among those aged 35 to 45 years. This emphasizes the importance and potential benefits of (selective) screening among young people of Hindustani Surinamese origin. More research is needed to find out whether screening below 35 years of age can be useful for Hindustani and African Surinamese and other South Asian and African origin people.</p>", "<p>The performance of the new risk score, that was developed in this study, appears to be at least as accurate or even more accurate as a screening test for DM than other sets of screening criteria derived from current guidelines. Compared to the criteria based on age and BMI alone, inclusion of a number of additional parameters significantly improved the performance for both the South Asian and African ethnic groups. The parameters included waist circumference and resting heart rate. In our analyses, we found that waist circumference appeared to be superior for predicting the risk of DM when compared to BMI, even if we used a lower cut-off (BMI ≥ 23) in South Asians as recommended by the WHO [##UREF##6##26##]. Moreover, an elevated heart rate is known to be a risk marker for CVD and associated with an increased risk of DM [##REF##15451917##27##,##REF##9369286##28##]. However, despite a broad range of specific determinants, South Asian ethnic origin in itself remained one of the most important predictors.</p>", "<p>The advantage of our risk score as compared to the extended ADA criteria is that it does not require invasive specimen collection. Previous prospective studies have also looked at diagnostic criteria based on non-invasive parameters and shown that the diagnostic accuracy was potentially good [##REF##12610029##29##, ####REF##16043747##30##, ##REF##17327313##31####17327313##31##]. However, only one of these studies incorporated ethnicity, only 'black' (African American), but did not specifically assess the performance of the score with only clinical parameters (i.e. non-invasive) by ethnic group [##REF##16043747##30##]. The two other studies did not include ethnicity as part of the risk score and were therefore not able to design and evaluate the criteria in a multi-ethnic population.</p>", "<p>Before drawing a conclusion, this study has some limitations which should be discussed. First, the performance of screening criteria in a study population in which the model is developed, is known to often be too optimistic. However, the sub-group analyses and the results of the bootstrapping procedure indicate that our estimate of the performance of our risk score was valid.</p>", "<p>Second, as in many surveys, the diagnosis of DM was based on a single fasting plasma glucose, which might have underestimated the true prevalence rates for DM as compared to a situation where an oral glucose tolerance test had been employed [##REF##9773731##32##]. Studies show that over 30% of persons with DM are missed if the diagnosis is based on fasting glucose alone, suggesting that the total prevalence of DM may be higher than reported in our analyses [##REF##10382583##33##, ####REF##10097924##34##, ##REF##11151755##35####11151755##35##].</p>", "<p>Moreover, these studies showed that the agreement between the two methods is dependent on determinants such as on age and BMI [##REF##10382583##33##,##REF##11151755##35##]. This may affect the validity of our risk score. The performance of the score for the identification of persons with DM diagnosed by means of the oral glucose tolerance test may be insufficient, as our score was solely based on determinants associated with DM persons identified by the fasting plasma glucose measurement. Further studies will have to explore the validity of the risk score for the identification of DM diagnosed by means of the oral glucose tolerance test.</p>", "<p>Third, our study is based on cross-sectional self-reported data. This could have biased the results of behavioural factors if, as part of treatment, persons with DM changed their lifestyle. This would imply an underestimation of how lifestyle-related determinants contribute to DM. In addition, the cross-sectional nature of the study prevented us from assessing the risk of incident DM.</p>", "<p>Fourth, the participation rate among those invited for the study was 60%. Although this is reasonable for this type of study, selective non-response may influence the representativeness of the results, i.e. the generalisability to the original sample. In our study, only small differences were found in participation in the interview by gender, marital status, household composition and urbanisation. Unfortunately, no data were available on determinants of DM in this population. Therefore, we could not determine whether participants and non-participants were comparable with regard to the determinants and risk of DM. However, further comparison of participants and non-participants in the physical examination revealed no differences with regard to self-reported DM and self-rated health between these groups. This suggests that the study population may be largely representative for the entire sample.</p>" ]

[ "<title>Conclusion</title>", "<p>Management of DM – especially in ethnic groups at high risk – deserves a great deal of attention in the form of early detection and prompt treatment. In our study, the prevalence of DM was so high among Hindustani Surinamese and to a lesser extent among African Surinamese that universal rather than selective screening may be indicated. In any case, detection and treatment among these ethnic groups should not just focus on persons 45 years or older, as is advised in most guidelines, but also include persons under the age of 45 years. If a choice for selective screening is made, an ethnicity-specific approach is required. Our risk score, which includes ethnicity, may be relatively easy to use in clinical practice. We have shown that it is as accurate as or more accurate than screening criteria derived from current guidelines. However, we do recommend further validation of this new risk score in practice.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>While the prevalence of type 2 diabetes mellitus (DM) is high, tailored risk scores for screening among South Asian and African origin populations are lacking. The aim of this study was, first, to compare the prevalence of (known and newly detected) DM among Hindustani Surinamese, African Surinamese and ethnic Dutch (Dutch). Second, to develop a new risk score for DM. Third, to evaluate the performance of the risk score and to compare it to criteria derived from current guidelines.</p>", "<title>Methods</title>", "<p>We conducted a cross-sectional population based study among 336 Hindustani Surinamese, 593 African Surinamese and 486 Dutch, aged 35–60 years, in Amsterdam. Logistic regressing analyses were used to derive a risk score based on non-invasively determined characteristics. The diagnostic accuracy was assessed by the area under the Receiver-Operator Characteristic curve (AUC).</p>", "<title>Results</title>", "<p>Hindustani Surinamese had the highest prevalence of DM, followed by African Surinamese and Dutch: 16.7, 8.1, 4.2% (age 35–44) and 35.0, 19.0, 8.2% (age 45–60), respectively. The risk score included ethnicity, body mass index, waist circumference, resting heart rate, first-degree relative with DM, hypertension and history of cardiovascular disease. Selection based on age alone showed the lowest AUC: between 0.57–0.62. The AUC of our score (0.74–0.80) was higher than that of criteria from guidelines based solely on age and BMI and as high as criteria that required invasive specimen collection.</p>", "<title>Conclusion</title>", "<p>In Hindustani Surinamese and African Surinamese populations, screening for DM should not be limited to those over 45 years, as is advocated in several guidelines. If selective screening is indicated, our ethnicity based risk score performs well as a screening test for DM among these groups, particularly compared to the criteria based on age and/or body mass index derived from current guidelines.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>NRB assisted in the design of the SUNSET study, carried out measurements, participated in the statistical analysis and drafted the manuscript. IGMV carried out the statistical analysis, drafted the manuscript and carried out the revisions. GM, carried out measurements and commented on the interpretation of the results and the drafts of the manuscript. RPK conceived and designed the SUNSET study and commented on the interpretation of the results and the drafts of the manuscript. FH, JBLH and RPJM contributed to the interpretation of the results and the drafts of the manuscript. KS conceived and designed the SUNSET study, contributed to the statistical analyses and participated in the writing and revisions of the manuscript. All authors read and approved the final version of the manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2458/8/271/prepub\"/></p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>We are very grateful to Prof. Niek Klazinga for his comments on earlier drafts of this manuscript. Furthermore, we are endebted to Dr. Michael Tanck, statistician, for his assistance with the analysis, in particular the bootstrapping procedure.</p>", "<p>This study was funded by The Netherlands Organisation for Health Research and Development (ZonMw) and the Academic Medical Centre (AMC).</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Flow chart of the inclusion into the study</bold>. ^aPersons who had moved, were deceased or could not be reached at the registered address were not considered as potential participants. ^bOnly persons of Hindustani Surinamese, African Surinamese or ethnic Dutch origin were invited for the physical examination (n = 1626). Javanese or Chinese Surinamese persons and persons with missing ethnicity were excluded (n = 71). ^cPersons without a fasting plasma sample (n = 10) or were excluded.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Characteristics of 35–60 year old Hindustani Surinamese, African Surinamese and ethnic Dutch participants</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td/><td align=\"left\"><bold>Hindustani Surinamese</bold><break/><bold>(n = 336)</bold><break/>median (IQR) or n (%)</td><td align=\"left\"><bold>African Surinamese</bold><break/><bold>(n = 593)</bold><break/>median (IQR) or n (%)</td><td align=\"left\"><bold>Ethnic Dutch</bold><break/><bold>(n = 486)</bold><break/>median (IQR) or n (%)</td></tr></thead><tbody><tr><td align=\"left\"><bold>Age</bold></td><td align=\"left\" colspan=\"2\">(years)</td><td align=\"left\">44 (39–50)</td><td align=\"left\">43 (39–48)</td><td align=\"left\">48 (42–54)</td></tr><tr><td align=\"left\"><bold>Sex</bold></td><td align=\"left\" colspan=\"2\">(female)</td><td align=\"left\">187 (55.6)</td><td align=\"left\">400 (67.5)</td><td align=\"left\">242 (49.8)</td></tr><tr><td align=\"left\"><bold>Level of education</bold></td><td align=\"left\" colspan=\"2\">(high^a)</td><td align=\"left\">34 (10.3)</td><td align=\"left\">116 (19.8)</td><td align=\"left\">175 (36.5)</td></tr><tr><td align=\"left\"><bold>Level of profession</bold></td><td align=\"left\" colspan=\"2\">(high^b)</td><td align=\"left\">98 (32.9)</td><td align=\"left\">257 (48.9)</td><td align=\"left\">279 (58.3)</td></tr><tr><td align=\"left\"><bold>Physical activity</bold></td><td align=\"left\" colspan=\"2\">(at least 30 minutes 5× per week^c)</td><td align=\"left\">166 (52.2)</td><td align=\"left\">315 (56.2)</td><td align=\"left\">306 (63.2)</td></tr><tr><td align=\"left\"><bold>History of CVD</bold></td><td align=\"left\" colspan=\"2\">(myocardial infarction and/or stroke)</td><td align=\"left\">21 (6.3)</td><td align=\"left\">13 (2.2)</td><td align=\"left\">10 (2.1)</td></tr><tr><td align=\"left\"><bold>First degree relative with DM</bold></td><td/><td/><td align=\"left\">273 (81.3)</td><td align=\"left\">371 (62.7)</td><td align=\"left\">202 (41.7)</td></tr><tr><td align=\"left\"><bold>BMI</bold></td><td align=\"left\" colspan=\"2\">(kg/m2)</td><td align=\"left\">26.7 (23.8–29.6)</td><td align=\"left\">27.6 (24.5–31.5)</td><td align=\"left\">25.4 (22.9–28.3)</td></tr><tr><td/><td align=\"left\" colspan=\"2\">(&gt; 25 kg/m2, Hindustani Surinamese: &gt; 23 kg/m2)</td><td align=\"left\">276 (82.1)</td><td align=\"left\">420 (71.0)</td><td align=\"left\">253 (52.1)</td></tr><tr><td align=\"left\"><bold>Waist circumference</bold></td><td align=\"left\" colspan=\"2\">(cm)</td><td align=\"left\">94.2 (85.9–101.0)</td><td align=\"left\">92.9 (83.1–102.1)</td><td align=\"left\">90.3 (81.5–100.3)</td></tr><tr><td/><td align=\"left\" colspan=\"2\">(increased waist circumference^d)</td><td align=\"left\">261 (77.7)</td><td align=\"left\">398 (67.2)</td><td align=\"left\">280 (57.6)</td></tr><tr><td align=\"left\"><bold>Resting heart rate</bold></td><td align=\"left\" colspan=\"2\">(resting heart rate in bpm)</td><td align=\"left\">72.5 (66.5–80.5)</td><td align=\"left\">72.0 (65.0–80.0)</td><td align=\"left\">68.5 (61.0–76.0)</td></tr><tr><td/><td align=\"left\" colspan=\"2\">(resting heart rate ≥ 90 bpm)</td><td align=\"left\">29 (8.8)</td><td align=\"left\">46 (7.9)</td><td align=\"left\">21 (4.4)</td></tr><tr><td align=\"left\"><bold>Dyslipidemia</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">HDL-cholesterol</td><td align=\"left\" colspan=\"2\">(mmol/l)</td><td align=\"left\">1.2 (1.0–1.5)</td><td align=\"left\">1.4 (1.2–1.7)</td><td align=\"left\">1.4 (1.2–1.7)</td></tr><tr><td/><td align=\"left\" colspan=\"2\">(&lt; 0.9 mmol/l (35 mg/dl) and/or treated)</td><td align=\"left\">63 (18.6)</td><td align=\"left\">21 (3.5)</td><td align=\"left\">31 (6.4)</td></tr><tr><td align=\"left\">Triglyceride</td><td align=\"left\" colspan=\"2\">(mmol/l)</td><td align=\"left\">1.2 (0.9–1.8)</td><td align=\"left\">0.8 (0.6–1.2)</td><td align=\"left\">1.1 (0.8–1.7)</td></tr><tr><td/><td align=\"left\" colspan=\"2\">(&gt; 2.8 mmol/l (250 mg/dl) and/or treated)</td><td align=\"left\">51 (15.2)</td><td align=\"left\">15 (2.5)</td><td align=\"left\">41 (8.4)</td></tr><tr><td align=\"left\"><bold>Blood pressure</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">SBP</td><td align=\"left\" colspan=\"2\">(mmHg)</td><td align=\"left\">124.0 (112.0–137.5)</td><td align=\"left\">125.0 (114.5–140.5)</td><td align=\"left\">121.5 (111.0–134.5)</td></tr><tr><td align=\"left\">DBP</td><td align=\"left\" colspan=\"2\">(mmHg)</td><td align=\"left\">81.5 (75.0–89.0)</td><td align=\"left\">82.0 (75.5–91.0)</td><td align=\"left\">78.5 (71.0–86.0)</td></tr><tr><td align=\"left\">Hypertension</td><td align=\"left\" colspan=\"2\">(&gt; 140/90 mmHg and/or on anti-hypertensive therapy)</td><td align=\"left\">115 (35.9)</td><td align=\"left\">223 (38.0)</td><td align=\"left\">125 (25.7)</td></tr><tr><td align=\"left\"><bold>Diabetes mellitus</bold></td><td align=\"left\" colspan=\"2\">(fasting plasma glucose ≥ 7.0 mmol/l and/or self-reported)</td><td/><td/><td/></tr><tr><td/><td align=\"left\">in 35–44 years</td><td align=\"left\">known</td><td align=\"left\">17 (9.8)</td><td align=\"left\">21 (6.1)</td><td align=\"left\">5 (3.0)</td></tr><tr><td/><td/><td align=\"left\">newly detected</td><td align=\"left\">12 (6.9)</td><td align=\"left\">7 (2.0)</td><td align=\"left\">2 (1.2)</td></tr><tr><td/><td align=\"left\">in 45–60 years</td><td align=\"left\">known</td><td align=\"left\">50 (30.7)</td><td align=\"left\">38 (15.4)</td><td align=\"left\">14 (4.4)</td></tr><tr><td/><td/><td align=\"left\">newly detected</td><td align=\"left\">7 (4.3)</td><td align=\"left\">9 (3.6)</td><td align=\"left\">12 (3.8)</td></tr><tr><td/><td align=\"left\">Age standardised^e</td><td align=\"left\">known</td><td align=\"left\">298 (21.1)</td><td align=\"left\">164 (11.6)</td><td align=\"left\">46 (3.2)</td></tr><tr><td/><td/><td align=\"left\">newly detected</td><td align=\"left\">80 (5.7)</td><td align=\"left\">37 (2.6)</td><td align=\"left\">32 (2.2)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Determinants of diabetes mellitus among Hindustani Surinamese, African Surinamese and ethnic Dutch aged 35–60 years^a</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold><italic>Determinant</italic></bold></td><td align=\"left\"><bold><italic>DM</italic></bold><break/><bold><italic>(n = 194)</italic></bold><break/><italic>Prevalence (%)</italic></td><td align=\"left\"><bold><italic>no DM</italic></bold><break/><bold><italic>(n = 1221)</italic></bold><break/><italic>Prevalence (%)</italic></td><td align=\"left\"><bold><italic>Univariate </italic></bold><break/><italic>OR [95% CI]</italic></td><td align=\"left\"><bold><italic>Multivariate</italic>^b</bold><break/><italic>OR [95% CI]</italic></td></tr></thead><tbody><tr><td align=\"left\">Age ≥ 45 years</td><td align=\"left\">67.0</td><td align=\"left\">49.0</td><td align=\"left\">2.2 [1.6–3.0]</td><td align=\"left\">1.8 [1.2–2.6]</td></tr><tr><td align=\"left\">BMI &gt; 25 kg/m2^c</td><td align=\"left\">89.1</td><td align=\"left\">63.6</td><td align=\"left\">4.8 [3.0–7.7]</td><td align=\"left\">1.9 [1.0–3.4]</td></tr><tr><td align=\"left\">Increased waist circumference^d</td><td align=\"left\">89.6</td><td align=\"left\">62.7</td><td align=\"left\">5.3 [3.2–8.6]</td><td align=\"left\">2.3 [1.3–4.1]</td></tr><tr><td align=\"left\">Resting heart rate ≥ 90 bpm</td><td align=\"left\">15.4</td><td align=\"left\">5.6</td><td align=\"left\">3.1 [1.9–4.9]</td><td align=\"left\">2.4 [1.4–4.0]</td></tr><tr><td align=\"left\">First-degree relative with DM</td><td align=\"left\">82.9</td><td align=\"left\">56.2</td><td align=\"left\">3.8 [2.5–5.6]</td><td align=\"left\">2.7 [1.8–4.2]</td></tr><tr><td align=\"left\">Hypertension^e</td><td align=\"left\">59.0</td><td align=\"left\">28.9</td><td align=\"left\">3.6 [2.6–4.9]</td><td align=\"left\">2.4 [1.7–3.4]</td></tr><tr><td align=\"left\">History of CVD</td><td align=\"left\">9.8</td><td align=\"left\">2.1</td><td align=\"left\">5.4 [2.9–10.3]</td><td align=\"left\">3.0 [1.5–6.3]</td></tr><tr><td align=\"left\">Ethnic group:</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Hindustani Surinamese</td><td align=\"left\">44.3</td><td align=\"left\">20.5</td><td align=\"left\">4.8 [3.1–7.4]</td><td align=\"left\">2.7 [1.7–4.5]</td></tr><tr><td align=\"left\"> African Surinamese</td><td align=\"left\">38.7</td><td align=\"left\">42.4</td><td align=\"left\">2.0 [1.3–3.1]</td><td align=\"left\">1.5 [0.9–2.4]</td></tr><tr><td align=\"left\"> Ethnic Dutch</td><td align=\"left\">17.0</td><td align=\"left\">37.1</td><td align=\"left\">1</td><td align=\"left\">1</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Performance of sets of selective screening criteria for diabetes mellitus among participants in SUNSET</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td align=\"left\"><bold><italic>Source</italic></bold></td><td align=\"left\"><bold><italic>AUC (bCI)</italic>^c</bold></td><td align=\"left\"><bold><italic>Sens (%, CI)</italic></bold></td><td align=\"left\"><bold><italic>Spec (%, CI)</italic></bold></td><td align=\"left\"><bold><italic>Screen (%)</italic></bold></td><td align=\"left\"><bold><italic>Prev (%)</italic></bold></td><td align=\"left\"><bold><italic>NNS</italic></bold></td></tr></thead><tbody><tr><td align=\"left\" colspan=\"9\"><bold>Set criteria for Hindustani Surinamese</bold></td></tr><tr><td colspan=\"9\"><hr/></td></tr><tr><td align=\"center\">1</td><td align=\"left\">age ≥ 45 years</td><td align=\"left\">DCGP</td><td align=\"left\">0.62<break/>(0.58–0.67)</td><td align=\"left\">66.3<break/>(55.2–75.9)</td><td align=\"left\">57.6<break/>(51.2–63.8)</td><td align=\"left\">48.5</td><td align=\"left\">35.0</td><td align=\"left\">3</td></tr><tr><td align=\"center\">2</td><td align=\"left\">age ≥ 45 and BMI &gt; 25 kg/m2</td><td align=\"left\">ADA</td><td align=\"left\">0.65<break/>(0.61–0.71)</td><td align=\"left\">66.3<break/>(55.2–75.9)</td><td align=\"left\">57.6<break/>(51.2–63.8)</td><td align=\"left\">48.5</td><td align=\"left\">35.0</td><td align=\"left\">3</td></tr><tr><td align=\"center\">3</td><td align=\"left\">criteria 2, and have additional risk factors, as follows^a</td><td align=\"left\">ADA extended</td><td align=\"left\">0.74<break/>(0.70–0.79)</td><td align=\"left\">79.5<break/>(69.0–87.3)</td><td align=\"left\">47.4<break/>(41.0–53.8)</td><td align=\"left\">56.4</td><td align=\"left\">35.5</td><td align=\"left\">3</td></tr><tr><td align=\"center\">4</td><td align=\"left\">tailored risk score^b</td><td align=\"left\">present study</td><td align=\"left\">0.74<break/>(0.70–0.79)</td><td align=\"left\">75.9<break/>(65,.0–84.3)</td><td align=\"left\">54.1<break/>(47.6–60.4)</td><td align=\"left\">53.5</td><td align=\"left\">35.8</td><td align=\"left\">3</td></tr><tr><td colspan=\"9\"><hr/></td></tr><tr><td align=\"left\" colspan=\"9\"><bold>Set criteria for African Surinamese</bold></td></tr><tr><td colspan=\"9\"><hr/></td></tr><tr><td align=\"center\">1</td><td align=\"left\">age ≥ 45 years</td><td align=\"left\">DCGP</td><td align=\"left\">0.62<break/>(0.58–0.67)</td><td align=\"left\">62.7<break/>(50.7–73.3)</td><td align=\"left\">61.4<break/>(57.0–65.6)</td><td align=\"left\">41.7</td><td align=\"left\">19.0</td><td align=\"left\">6</td></tr><tr><td align=\"center\">2</td><td align=\"left\">age ≥ 45 and BMI &gt; 25 kg/m2</td><td align=\"left\">ADA</td><td align=\"left\">0.68<break/>(0.65–0.75)</td><td align=\"left\">59.5<break/>(47.4–70.5)</td><td align=\"left\">72.6<break/>(68.5–76.3)</td><td align=\"left\">81.1</td><td align=\"left\">14.2</td><td align=\"left\">8</td></tr><tr><td align=\"center\">3</td><td align=\"left\">criteria 2, and have additional risk factors, as follows^a</td><td align=\"left\">ADA extended</td><td align=\"left\">0.79<break/>(0.76–0.85)</td><td align=\"left\">78.1<break/>(66.6–86.6)</td><td align=\"left\">60.7<break/>(56.3–65.0)</td><td align=\"left\">44.1</td><td align=\"left\">22.1</td><td align=\"left\">5</td></tr><tr><td align=\"center\">4</td><td align=\"left\">tailored risk score^b</td><td align=\"left\">present study</td><td align=\"left\">0.80<break/>(0.75–0.85)</td><td align=\"left\">79.2<break/>(67.7–87.5)</td><td align=\"left\">62.3<break/>(57.9–66.5)</td><td align=\"left\">42.8</td><td align=\"left\">23.0</td><td align=\"left\">5</td></tr><tr><td colspan=\"9\"><hr/></td></tr><tr><td align=\"left\" colspan=\"9\"><bold>Set criteria for ethnic Dutch</bold></td></tr><tr><td colspan=\"9\"><hr/></td></tr><tr><td align=\"center\">1</td><td align=\"left\">age ≥ 45 years</td><td align=\"left\">DCGP</td><td align=\"left\">0.57<break/>(0.51–0.63)</td><td align=\"left\">78.8<break/>(60.6–90.4)</td><td align=\"left\">35.6<break/>(31.2–40.2)</td><td align=\"left\">65.4</td><td align=\"left\">8.2</td><td align=\"left\">13</td></tr><tr><td align=\"center\">2</td><td align=\"left\">age ≥ 45 and BMI &gt; 25 kg/m2</td><td align=\"left\">ADA</td><td align=\"left\">0.72<break/>(0.65–0.77)</td><td align=\"left\">72.7<break/>(54.2–86.1)</td><td align=\"left\">72.7<break/>(60.3–69.3)</td><td align=\"left\">79.8</td><td align=\"left\">8.0</td><td align=\"left\">13</td></tr><tr><td align=\"center\">3</td><td align=\"left\">criteria 2, and have additional risk factors, as follows^a</td><td align=\"left\">ADA extended</td><td align=\"left\">0.79<break/>(0.75–0.87)</td><td align=\"left\">71.9<break/>(53.0–85.6)</td><td align=\"left\">71.2<break/>(66.7–75.3)</td><td align=\"left\">31.7</td><td align=\"left\">15.0</td><td align=\"left\">7</td></tr><tr><td align=\"center\"><bold>4</bold></td><td align=\"left\">tailored risk score^b</td><td align=\"left\">present study</td><td align=\"left\">0.78<break/>(0.73–0.85)</td><td align=\"left\">77.4<break/>(58.5–89.7)</td><td align=\"left\">68.8<break/>(64.2–73.1)</td><td align=\"left\">34.3</td><td align=\"left\">14.9</td><td align=\"left\">7</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Performance of sets of selective screening criteria for the identification of newly detected diabetes mellitus among participants in SUNSET^c</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td align=\"left\"><bold><italic>Source</italic></bold></td><td align=\"left\"><bold>Hindustani Surinamese</bold><break/><bold><italic>N = 269</italic></bold><break/><bold>AUC (bCI)</bold></td><td align=\"left\"><bold>African Surinamese</bold><break/><bold><italic>N = 534</italic></bold><break/><bold>AUC (bCI)</bold></td><td align=\"left\"><bold>ethnic Dutch</bold><break/><bold>n = 467</bold><break/><bold>AUC (bCI)</bold></td></tr></thead><tbody><tr><td align=\"left\">1</td><td align=\"left\">age ≥ 45 years</td><td align=\"left\">DCGP</td><td align=\"left\">0.53 (0.50–0.61)</td><td align=\"left\">0.59 (0.51–0.69)</td><td align=\"left\">0.61 (0.54–0.68)</td></tr><tr><td align=\"left\">2</td><td align=\"left\">age ≥ 45 and BMI &gt; 25 kg/m2</td><td align=\"left\">ADA</td><td align=\"left\">0.61 (0.54–0.70)</td><td align=\"left\">0.69 (0.62–0.78)</td><td align=\"left\">0.72 (0.65–0.82)</td></tr><tr><td align=\"left\">3</td><td align=\"left\">criteria 2, and have additional risk factors, as follows^a</td><td align=\"left\">ADA extended</td><td align=\"left\">0.69 (0.64–0.83)</td><td align=\"left\">0.87 (0.83–0.95)</td><td align=\"left\">0.80 (0.74–0.91)</td></tr><tr><td align=\"left\">4</td><td align=\"left\">tailored risk score^b</td><td align=\"left\">present study</td><td align=\"left\">0.70 (0.66–0.83)</td><td align=\"left\">0.87 (0.83–0.93)</td><td align=\"left\">0.78 (0.73–0.90)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Performance of the simplified risk score for the identification of newly detected diabetes mellitus^a</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td align=\"left\"><bold><italic>AUC (bCI)</italic></bold></td><td align=\"left\"><bold><italic>Sens (%)</italic></bold></td><td align=\"left\"><bold><italic>Screen (%)</italic></bold></td><td align=\"left\"><bold><italic>Prev (%)</italic></bold></td><td align=\"left\"><bold><italic>NNS</italic>_new</bold></td></tr></thead><tbody><tr><td align=\"left\" colspan=\"2\"><bold>Hindustani Surinamese</bold></td><td align=\"left\" colspan=\"5\">0.58 (0.49–0.70)</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td/><td align=\"left\">Total population</td><td/><td/><td/><td align=\"left\">6.0</td><td/></tr><tr><td/><td align=\"left\">≥ 5 points</td><td/><td align=\"left\">94.4</td><td align=\"left\">97.3</td><td align=\"left\">6.6</td><td align=\"left\">16</td></tr><tr><td/><td align=\"left\">≥ 8 points</td><td/><td align=\"left\">94.4</td><td align=\"left\">80.7</td><td align=\"left\">8.0</td><td align=\"left\">13</td></tr><tr><td/><td align=\"left\">≥ 11 points</td><td/><td align=\"left\">38.9</td><td align=\"left\">42.0</td><td align=\"left\">6.3</td><td align=\"left\">16</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\" colspan=\"2\"><bold>African Surinamese</bold></td><td align=\"left\" colspan=\"5\">0.79 (0.70–0.89)</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td/><td align=\"left\">Total population</td><td/><td/><td/><td align=\"left\">3.1</td><td/></tr><tr><td/><td align=\"left\">≥ 5 points</td><td/><td align=\"left\">93.8</td><td align=\"left\">86.8</td><td align=\"left\">3.3</td><td align=\"left\">31</td></tr><tr><td/><td align=\"left\">≥ 8 points</td><td/><td align=\"left\">87.5</td><td align=\"left\">58.9</td><td align=\"left\">4.5</td><td align=\"left\">22</td></tr><tr><td/><td align=\"left\">≥ 11 points</td><td/><td align=\"left\">81.3</td><td align=\"left\">25.8</td><td align=\"left\">9.6</td><td align=\"left\">11</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\" colspan=\"2\"><bold>Ethnic Dutch</bold></td><td align=\"left\" colspan=\"5\">0.77 (0.68–0.85)</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td/><td align=\"left\">Total population</td><td/><td/><td/><td align=\"left\">3.1</td><td/></tr><tr><td/><td align=\"left\">≥ 5 points</td><td/><td align=\"left\">100</td><td align=\"left\">56.2</td><td align=\"left\">5.5</td><td align=\"left\">19</td></tr><tr><td/><td align=\"left\">≥ 8 points</td><td/><td align=\"left\">64.3</td><td align=\"left\">24.9</td><td align=\"left\">8.0</td><td align=\"left\">13</td></tr><tr><td/><td align=\"left\">≥ 11 points</td><td/><td align=\"left\">14.3</td><td align=\"left\">6.4</td><td align=\"left\">6.9</td><td align=\"left\">15</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional File 1</title><p>'Appendix: The risk score for DM from the population based SUNSET cohort'. Instructions on how to calculate the simplified risk score for diabetes mellitus from the population based SUNSET cohort.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>^adefinition: higher vocational or more (i.e. primary, secondary or lower vocational versus higher vocational or more) ^bdefinition: grade of employment (as classified by the EGP scheme (routine non-manual employees or higher (high) vs. low) ^cDutch physical activity guideline ^dwaist circumference &gt; 80 cm for women and &gt; 94 cm for men ^eDirectly standardised to the age distribution of the total population (in 5-year categories). Data represent expected number of cases (prevalence). BMI = body mass index, bpm = beats per minute, SBP = systolic blood pressure, DBP = diastolic blood pressure, CVD = cardiovascular disease, HDL = high density lipoprotein, DM= diabetes mellitus, IQR = interquartile range</p></table-wrap-foot>", "<table-wrap-foot><p>^aonly determinants selected for the multivariate analysis, based on a Waldtest for univariate association of p &lt; 0.25 are shown ^bmutually adjusted ^c&gt; 23 kg/m2 for Hindustani Surinamese ^dwaist circumference &gt; 80 cm for all women, &gt; 94 cm for African Surinamese and ethnic Dutch men and &gt; 90 cm for Hindustani Surinamese men ^eblood pressure &gt; 140/90 mm Hg and/or being on anti-hypertensive therapy</p><p>DM = diabetes mellitus (fasting plasma glucose ≥ 7.0 mmol/l and/or self-reported), OR = odds ratio, CI = confidence interval, BMI = body mass index, bpm = beats per minute, CVD = cardiovascular disease (myocardial infarction and/or stroke)</p></table-wrap-foot>", "<table-wrap-foot><p>^aphysically inactive or 1st-degree relative with DM or high risk ethnic group (Hindustani Surinamese, i.e. South Asian) or hypertensive or reduced HDL and/or elevated triglyceride or history of cardiovascular disease.</p><p>^boptimal screening criteria identified from multivariate analysis (ethnic groups, age, BMI, waist circumference, resting heart rate, first-degree relative with DM, hypertension, history of cardiovascular disease)</p><p>^cbCI: The 'Confidence interval' around the score was estimated via a bootstrapping procedure [##REF##11470385##20##], the 2.5 and 97.5 percentiles are listed between brackets. The cut-off value for the SUNSET score was 0.200 for Hindustani Surinamese, 0.077 for African Surinamese and 0.055 for ethnic Dutch. DM = diabetes mellitus (fasting plasma glucose ≥ 7.0 mmol/l and/or self-reported), CI = 95%-confidence interval, DCPG = Dutch College of General Practitioners, ADA = American Diabetes Association, AUC = area under the curve, a measure of diagnostic accuracy, sens = sensitivity of the criteria (value closest to, but not over 80%), spec = specificity corresponding to the listed sensitivity, screen = percentage of the total population in this study that is screened, prev = prevalence in the screened population (the predictive value positive), NNS = number needed to screen to detect a case of DM in this study, BMI = body mass index</p></table-wrap-foot>", "<table-wrap-foot><p>^aphysically inactive or 1st-degree relative with DM or high risk ethnic group (Hindustani Surinamese, i.e. South Asian) or hypertensive or reduced high density lipoprotein and/or elevated triglyceride or history of cardiovascular disease.</p><p>^boptimal screening criteria identified from multivariate analysis (ethnic groups, age, BMI, waist circumference, resting heart rate, first-degree relative with DM, hypertension, history of cardiovascular disease)</p><p>^cPersons with known DM were excluded from the analysis. Data are area under the curve (AUC) and a 'confidence interval' (bCI), based on the 2.5 and 97.5 percentiles estimated via a bootstrapping procedure [##REF##11470385##20##].</p><p>DM = diabetes mellitus (fasting plasma glucose ≥ 7.0 mmol/l and/or self-reported), DCPG = Dutch College of General Practitioners, ADA = American Diabetes Association</p></table-wrap-foot>", "<table-wrap-foot><p>^aPersons with known DM were excluded from the analysis. Simplified score is presented in additional file ##SUPPL##0##1##: Risk score for DM SUNSET.pdf. Cut off values (number of points) chosen based on the 20–50–75 percentiles in the total population.</p><p>DM = diabetes mellitus (fasting plasma glucose ≥ 7.0 mmol/l and/or self-reported), sens = sensitivity of the criteria at the given cut-off, screen = percentage of the total population that is screened at the given cut-off, prev = prevalence in the screened population (the predictive value positive), NNS_new= number needed to identify a case of newly detected DM, i.e. a <underline>previously undiagnosed</underline> case of DM in this study, AUC = area under the curve, a measure of diagnostic accuracy, bCI = 'confidence interval' estimated via a bootstrapping procedure [##REF##11470385##20##], the 2.5 and 97.5 percentiles are listed between brackets.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2458-8-271-1\"/>" ]

[ "<media xlink:href=\"1471-2458-8-271-S1.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The north-east Indian states of Manipur and Nagaland are characterised by political instability, unemployment, and easy availability of heroin from across the Myanmar border. They are classified by the National AIDS Control Society [##UREF##0##1##] as high prevalence states for HIV, and intravenous drug use is an important route of HIV transmission [##UREF##0##1##,##UREF##1##2##]. The injecting drug user (IDU) population in these states constitutes 1.9–2.7% of the adult population [##UREF##1##2##]. In 2005, the HIV prevalence among IDUs in Manipur and Nagaland was reported to be 24% and 5% respectively, representing an increase in both states from the previous year [##UREF##0##1##]. However, much higher rates have also been reported: in a sample of IDUs from north-east India, 75% were found to be HIV positive [##UREF##1##2##]. Most IDUs in north-east India are men, an estimated 40% are married [##UREF##2##3##], and death rates have been high in the last five years, consequently the number of widows of IDUs has increased.</p>", "<p>Widows in India are socially and economically disadvantaged, and the situation for widows of IDUs is often worse. They are frequently stigmatised on three levels – for being a woman, being a widow, and being HIV positive [##UREF##3##4##]. In 2004, a situation assessment of widows of IDUs conducted in Manipur found that they were faced with a range of psycho-social, economic and health problems. Many IDU widows and their children were HIV-infected and experiencing poor health, social isolation, grief, loneliness, discrimination, and poverty; all factors likely to be compromising their mental health [##UREF##4##5##]. Some widows reported engaging in HIV risk behaviours including alcohol and drug misuse, sex work and unprotected sex. Accessing HIV prevention services was not a priority for these women who were predominantly concerned about livelihood and their children's future [##UREF##4##5##].</p>", "<p>Mental health is more than simply the absence of mental illness; it is the foundation for well-being and effective functioning for individuals and communities [##UREF##5##6##]. Mental illness is associated with indicators of poverty including low levels of education, poor housing and low income [##REF##14576893##7##], and with other illnesses such as HIV infection [##REF##15994563##8##]. Substance misuse, violence and health problems such as HIV and depression are more prevalent and more difficult to cope with in conditions of low income, limited education and unemployment [##UREF##6##9##].</p>", "<p>The nexus between gender and mental health is well recognised. Socio-cultural beliefs about gender roles often diminish women's control over their own lives and restrict their access to economic resources and power within the society, and this in turn has a significant effect on their mental health status and the risk of mental illness especially depression [##UREF##7##10##, ####UREF##8##11##, ##UREF##9##12####9##12##].</p>", "<p>The influence of gender on women's mental health is most evident in relation to depression. Globally, including in India, women are approximately twice as likely as men to experience depression and widows in India are more likely to suffer from mental illness than single or married women [##UREF##7##10##, ####UREF##8##11##, ##UREF##9##12####9##12##]. There is a need to promote the mental health of women in general and of widows in particular, both in India and further a field [##UREF##9##12##].</p>", "<p>Mental health is associated with HIV in a range of ways. In relation to engagement in HIV risk behaviours, individuals with poor mental health as a group and including those with untreated mental illness and substance misuse problems, have a greater chance of exposure to HIV related risk behaviours. Many have less control over their lives than other populations, are more likely to find themselves in situations of risk, and have diminished ability to negotiate safe behaviours [##REF##15817956##13##,##UREF##10##14##]. Interactions between drug and alcohol use and depression are common, and studies in India indicate that the former is associated with engagement in HIV risk behaviours, especially among those with mental health problems [##UREF##11##15##,##REF##12935383##16##]. People living with HIV and AIDS have an increased risk of developing mental health problems including depression and substance misuse [##REF##15817956##13##, ####UREF##10##14##, ##UREF##11##15####11##15##]. These conditions adversely affect HIV and AIDS treatment adherence, contribute to risk behaviours and exacerbate social difficulties associated with stigma and discrimination. These considerations raise the possibility that promoting the mental health of vulnerable groups may reduce the risk of engagement in HIV risk behaviours and thereby contribute to HIV prevention.</p>", "<p>Emerging evidence indicates that mental health can be promoted by public health actions with vulnerable groups [##UREF##5##6##]. Just as physical health can be promoted, so too can mental health. A recent WHO report draws on a public health framework proposed initially by the Victorian Health Promotion Foundation [##UREF##12##17##] that identifies three key social and economic determinants of community and individual mental health: (1) social inclusion; (2) freedom from discrimination and violence; and (3) access to economic resources. This framework recognises that psychosocial and economic factors influence (protect or negate) a number of health-related behaviours such as substance misuse and risky sexual behaviours, that in turn affect all areas of health, including mental health [##REF##12935383##16##,##UREF##12##17##].</p>", "<p>While the health benefits of community participation are well understood in development work, health policy does not always reflect this, partly because the published evidence related to this approach is limited. Our intervention drew on participatory action research (PAR) approaches to health development that seek to empower target communities to actively identify problems and develop solutions in relation to particular research questions. This enhances their self-confidence and leadership skills, and assists them to address their own health and social needs [##REF##11468849##18##,##REF##15988448##19##]. For example, a study in Nepal demonstrated that community based participatory action had a significant positive impact on maternal and infant mortality [##UREF##13##20##]. Studies such as these and the one described in this paper help to narrow the evidence gap on the effectiveness of community participation to contribute to changes in health status.</p>", "<p>This pilot intervention study began with the hypothesis that the implementation of structured and peer-facilitated participatory action groups (PAGs) among widows of IDUs in Manipur and Nagaland, with a focus on promoting mental health and well-being and informed by a strengths-based approach [##UREF##14##21##], would be associated with: (1) improved mental health; and (2) a reduced likelihood of engagement in HIV risk behaviours.</p>", "<p>The objectives of the study were to: (1) learn about the women's perspectives on mental health and well-being and the links between mental health and HIV; (2) assess changes in the women's quality of life and mental health during the course of the intervention; (3) assess changes in engagement in HIV risk behaviours; (4) describe the process and outcome of the intervention from the perspective of the women; (5) document the process of establishing and conducting the intervention so it can be repeated or adapted in the future. This paper reports on the findings in relation to objectives 2, 3 and 4. A full description of the background, the intervention and the methods for this study has been published elsewhere [##REF##17442121##22##], so the following methods section is brief.</p>" ]

[ "<title>Methods</title>", "<p>Six groups of IDU widows were established (three in each state) in mid 2006, with 9–16 widows in each group. The women were recruited through partnerships with local non-government organisations (NGOs) working in the field of HIV prevention. The NGOs contacted IDU widow's known to them, and through these women's networks, contacted other widows. All interested widows attended a meeting where the nature of the study and intervention were explained, and those women interested in participating were recruited. The districts covered by the participating NGOs were Imphal and Churachandpur in Manipur and Dimapur and Kohima in Nagaland. The intervention is described elsewhere [##REF##17442121##22##] and more information is available from the authors on request. In brief, the intervention was based on the framework for mental health promotion [##UREF##12##17##], and consisted of ten peer-facilitated PAG meetings that were held every fortnight for half a day over a twenty week period. All meetings were participatory, strengths-based and comprised of a combination of structured activities and open discussion (Table ##TAB##0##1##). Two peer facilitators were trained and supported for each group, and were provided with flexible written guidelines for each session. The women's travel and childcare costs were covered, refreshments were provided, and the activities deliberately engendered fun and enjoyment for the women. Each group participated in an action planning process to develop strategies for promoting mental health and the sustainability of the groups. A range of quantitative and qualitative data was collected to assess the impact of the intervention on the lives of the women.</p>", "<title>Quality of life, mental health, somatic symptoms and HIV risk behaviours</title>", "<p>Three brief questionnaires were completed by the women during the first and last PAG meetings:</p>", "<p>1. The short version of the WHO Quality of Life questionnaire (WHOQOL-BREF): Quality of life is a broad ranging concept and is assessed on a person's perceptions of various factors divided into four domains: (i) Physical health domain including pain and discomfort, energy and fatigue, mobility, sleep and work capacity; (ii) Psychological domain including spirituality, body image and appearance, thinking and learning and self-esteem; (iii) Social domain including personal relationships, sexual activity and social support; (iv) Environment domain including physical safety and security, physical environment (pollution, climate), financial resources, participation in recreation, home environment, transportation, and access and quality of health and social care [##UREF##15##23##,##REF##9626712##24##]. Cronbach alpha values for each of the domain scores in the WHOQOL range from 0.71 to 0.86, demonstrating good internal consistency [##UREF##16##25##].</p>", "<p>2. The General Health Questionnaire (GHQ12): The GHQ12 is a screening instrument for common mental disorders suitable for use in community, primary care and medical settings. It is used in many different countries and although the original version consisted of 60 items; shorter versions have been subsequently developed and validated [##REF##9122299##26##]. The Cronbach's alpha for the GHQ12 is estimated to be between 0.82 – 0.86 [##REF##9122299##26##]. Analysis of the GHQ12 data used a cut-off of three points, greater than three indicating potential presence of a common mental disorder such as depression or anxiety. The selection of this cut-off was based on findings from an earlier study using the GHQ12 in an Indian setting [##REF##9483689##27##].</p>", "<p>3. A Health and Well-being Questionnaire; Women in India who experience mental health problems such as depression often express their distress as somatic symptoms [##REF##7894865##28##, ####REF##15104094##29##, ##REF##16585469##30####16585469##30##], so this questionnaire was adapted from an existing somatic symptom scale [##UREF##17##31##]. It asked participants to select how often they experience pain, various bodily sensations (e.g. weakness, trembling), disturbances of body functions (e.g. sleep, appetite), and reproductive symptoms (e.g. menstrual disturbance, vaginal discharge), as an indication of mental health. The second part of the questionnaire consisted of a small number of questions about sexual partners and drug and alcohol use adapted from a widely used Behavioural Surveillance Survey [##UREF##18##32##].</p>", "<p>The WHOQOL-BREF and the GHQ12 are usually self-administered but assisted administration is possible for people with low literacy. All questionnaires were translated into the local languages, back translated and piloted with literate and non-literate women. The research teams and peer facilitators assessed the sensitivity and appropriateness of all questions before they were included in the study. Special attention was given to the more sensitive questions seeking information about sexual and substance use behaviours.</p>", "<p>Data were analysed using SPSS 10. Individual assessment scores were summarised using frequency distributions and mean and standard deviations. Comparisons between baseline and follow up scores were done using paired t-test for continuous scores (WHOQOL-BREF and Health and Well-being Questionnaire) and McNamara Chi-square test for categorical variables (GHQ12). Data were analysed with the two states combined and separately.</p>", "<title>Value of the intervention from the participant and NGO perspectives</title>", "<p>The Most Significant Change (MSC) method was used to evaluate the intervention from the perspectives of the participants and the partnering NGOs. MSC is a qualitative, participatory approach to monitoring and evaluation used in development projects [##UREF##19##33##]. In this study, we collected 'stories of change' from participants during the eighth PAG meeting, and involved different stakeholders in systematic selection of the stories that best captured what they considered to be the 'most significant change'. Reasons for selecting particular stories were also documented. The point of MSC is that both the content of the selected stories and the reasons for choosing them make the values of the different stakeholders explicit, and this can be used to foster dialogue between potentially competing perspectives, in this case that of the IDU widows and the NGOs. As the MSC approach tends to elicit positive stories (in relation to the intervention), this method is not used as an evaluation tool in isolation.</p>", "<p>Initially, stories of change were collected from all participants who were willing and able to provide them. The stories were told to the state-based research officers (BD, TG) in private and were recorded and subsequently transcribed and translated. Two panels were convened in each state: one participant panel consisting of two peer facilitators, two widows, and an NGO liaison worker from each group (three groups in each state), and one NGO panel consisting of two senior staff from each of four NGOs connected to the project. The story selection process involved reading all of the stories from the relevant state to the panel members who were given time to reflect on each story and then, through a collective process, select the four stories that, in their view, best represented the 'most significant change'. The two panels met and made their selections independent of each other (Figure ##FIG##0##1##). Panel members were encouraged to discuss the reasons why they selected these particular stories and this discussion was recorded, transcribed and translated. We were interested in the similarities and the differences between the stories selected by the participant and NGO panels, as well as similarities and differences in their reasons for selection. Additionally all of the stories were thematically analysed. This is an inductive approach that involves systematically coding and recoding the data in order to identify and organise both explicit and implicit patterns embedded in the data.</p>", "<title>Ethics and funding</title>", "<p>This intervention study was funded by the United Kingdom's Department for International Development (DFID) through the Research and Learning Fund. Ethics approval was obtained from the University of Melbourne Human Research Ethics Committee (Australia) and the Emmanuel Hospital Association Institutional Review Board (India) in early 2006. Participation in the study was entirely voluntary, all participants provided informed consent and confidentiality was assured.</p>" ]

[ "<title>Results</title>", "<title>Participation and demographics</title>", "<p>Seventy-four women participated in the first PAG meeting, and 59 women completed the intervention (80% retention). The level of participation varied between NGOs (p = 0.04) and states (p = 0.002) (Table ##TAB##1##2##). The women were relatively young (mean age 32.5 years) and came from diverse ethnic and religious backgrounds (Table ##TAB##2##3##). The average length of widowhood was 4.3 years and the average number of children was 2.4. The majority of participants (81%) were not currently employed. Sources of income for those not formally employed included small businesses, support from family, sex work, and making handicrafts. Slightly more than half of the participants (53%) had completed secondary school and 30% had undergone some form of tertiary education. Nine percent of participants reported receiving no education. Although we did not systematically collect information regarding the participants' HIV status and drug use history, many of the women revealed their HIV positive status in the course of the intervention, and a small proportion had a history of injecting drug use.</p>", "<title>Changes in quality of life, mental health and somatic symptoms</title>", "<p>In both states, all four Quality of Life domain scores increased across the course of the intervention indicating a general trend towards improvement in their quality of life (Table ##TAB##2##3##). The women experienced significant improvements in their physical and psychological health, and their interaction with the environment was more positive after participating in the intervention (p ≤ 0.05). While there was an increase in the social domain score, this did not reach statistical significance. The Quality of Life results in Manipur mirrored those for the group as a whole, but those in Nagaland were somewhat different as even though they all improved, none reached statistical significance (Table ##TAB##3##4##).</p>", "<p>The results from the GHQ12 indicate that almost three-quarters of the group (70%, 49/70) were possibly experiencing a common mental disorder such as depression or anxiety at baseline compared with 42% (24/57) at the end of the intervention (Table ##TAB##3##4##). This represents a significant decrease in the proportion likely to be experiencing a common mental disorder (p &lt; 0.01). Similar improvements were noted in both states. Those who dropped out of the intervention compared to those who remained were no different at baseline in relation to their GHQ12 scores and a range of demographic measures (ethnicity, religion, education, employment and marital status).</p>", "<p>In both states, the women experienced fewer somatic symptoms across the course of the intervention (as indicated by lower scores). However, pain was the only symptom that significantly decreased for the two states combined (p &lt; 0.01).</p>", "<p>In summary, these results indicate an overall improvement in several quality of life and mental health parameters across the course of the intervention, although the patterning of these improvements varied by state. Unfortunately, data were insufficient to conduct a complete analysis of the effect of the intervention on participants' engagement in HIV risk behaviours as too many of the respondents left the relevant questions blank. Only a handful of participants openly engaging in sex work responded adequately to these questions. This is discussed further in the Discussion section of the paper.</p>", "<title>Thematic analysis of the Most Significant Change stories</title>", "<p>Stories of change were collected from 33 participants in Manipur and 22 in Nagaland. Analysis of the stories uncovered a range of themes, most of which aligned with the socio-economic determinants of mental health i.e. social inclusion, freedom from discrimination, and economic participation. This is not surprising as the intervention was somewhat structured around these themes. Two additional themes were physical health and future orientation. A sample story from each state can be found in Figure ##FIG##1##2##.</p>", "<title>Social inclusion</title>", "<p>Many participants described how joining the group helped them overcome social isolation and provided them with a sense of belonging that was previously absent. The benefits of meeting and building relationships with other women in the same situation were frequently highlighted and the group became a source of nurturance and support for many of the women, and a stepping stone into the wider society for some.</p>", "<p>I felt that others discriminated against me as my husband died from drugs. This made me ashamed and I never wanted to attend any social gatherings. I didn't want to interact with others and this resulted in depression and stress and a feeling that life had no value. Since I attended this meeting, it has encouraged me to interact socially. I have even begun to attend the church and other social gatherings. It has been good meeting other widows like me, who I never knew before, but now they have become my friends with whom I can share my problems. (Manipur 11)</p>", "<p>A lot of participants gained confidence in their own abilities as a result of having to interact in a group setting.</p>", "<p>Before I attended this PAG, I could never stand in a group and speak but during the PAG sessions I had many opportunities to share my opinion and speak in the group, which has helped me to realise some of my potentials. (Manipur 33)</p>", "<p>Family conflict (mainly with in-laws) was one of the major sources of stress for the women. Many felt that the intervention had provided them with skills to better manage anger, thus enhancing their relationships with family.</p>", "<p>My relationship with my family and in-laws has improved. Earlier I used to fight with them because I have very negative thoughts about them. Now I understand that this will only increase our worries and affect my mental health. I have been able to handle my anger. After I stopped shouting, I have observed that even my children have stopped fighting amongst themselves. This has helped me to strengthen my relationship with my family and friends as well. (Nagaland 1)</p>", "<p>The PAG meetings deliberately incorporated activities that were fun such as games and singing, and this provided many of the women with a rare opportunity to experience joy and pleasure.</p>", "<p>The fun and laughter during the PAG has also helped me a lot. It is only here that I experience fun and laughter. (Nagaland 20)</p>", "<p>Finally, several of the participants said that they were inspired to help others.</p>", "<p>I have been able to meet other people like me and have started thinking of helping other people like me as well. I have been able to learn how to access resources and this is helping me to manage my plans and finances better. I have already started sharing my experiences and benefits with other widows. Now I have <bold>some peace of mind</bold>. (Nagaland 12)</p>", "<title>Discrimination</title>", "<p>The theme of discrimination against people with HIV and AIDS generally, and IDU widows in particular, was evident in many of the women's stories. They described intense feelings of anger in relation to perceived discrimination from both family and society. Several of the women reported that learning to manage their anger allowed them to better cope with this source of adversity.</p>", "<p>Prior to joining the PAG, I had a strong resentment fermenting inside me for a very long time. I would not be able to eat, sleep or do anything... nor take care of my children... People do not respect us because we were are widowed very young. They think that we do not have any other means of earning except by selling our body... Thinking about these things made my resentment grow from bad to worse. It has been about four months since I began attending this PAG meeting. I have realised that these problems will not go away... I can not hope nor expect any of them to change, but I can adjust with them. I need not brood over their attitude and let it affect my life and my relationship with my children. I can control my anger and seek my own course in life. Now I keep my priority concern on my children's welfare, so I've started attending the parents meetings at my children's school. (Manipur 20)</p>", "<title>Economic participation</title>", "<p>This theme was connected to both an appreciation of the allowance received as compensation for their travel and childcare costs, and a growing awareness of their collective ability to form self help groups for income generation.</p>", "<p>Our allowances are very helpful because most of us are income-less at home. It is no longer necessary to demand money from the family as we can meet some of our needs now. With this allowance I buy Raja [tobacco] and sell it to get some profit... and I spend that on my other essential needs. With this allowance we have a sense of ownership in the family. (Manipur 1)</p>", "<p>There are changes in my work also. I go to the agricultural field and forest to earn my livelihood. I was very shy and could not even think of doing something in the town. But now I have the confidence and I have started selling vegetables in the market and also go house to house. (Nagaland 9)</p>", "<p>We do not have a voice, so we can become strong if we come together... We can start income generation and share our problems. If we are able to generate income it will help us. (Nagaland 13)</p>", "<title>Improved physical health</title>", "<p>Many women reported feeling much better physically and several had initiated contact with service providers as a result of participation in the group. Additionally, the relationship between physical and mental health was recognised by some women.</p>", "<p>After coming to the meeting I realised that I am not alone and that has encouraged me to become more productive. I have also started accessing free medical services and medicines, which I never did even if I knew there were facilities available. I attended a free health camp for the first time last month. (Nagaland 11)</p>", "<p>Even my physical health has improved in the last two to three months. I used to have severe gastric problems but now that has reduced. Being able to share and discuss my problems with friends has lightened my mental worries. (Nagaland 9)</p>", "<title>Future orientation</title>", "<p>Many of the women commented that the intervention contributed to renewed feelings of hope for the future and several appreciated the opportunity to create personal goals.</p>", "<p>Before all my hope was gone but this meeting has given me new hope. I was like a dead soul, but now with this new hope I can carry on my life and my responsibility. This way it improves my mental health. (Manipur 11)</p>", "<title>Reasons for selecting stories</title>", "<p>All stories were considered by the participant and NGO panels in each state, and each panel selected the four stories that best represented the 'most significant change'. The participant and NGO panels selected stories that were similar in thematic content, although the emphasis given to the reasons for the significance of the stories varied. The reasons the participant and NGO panel members give for selecting particular stories make explicit the type of changes valued by the different stakeholders. The emphasis given to different types of changes was reflected by the number of times they were mentioned in the course of the discussion. From the perspective of the storytellers, changes took place within themselves, and in relation to their interactions with the group, their families, and the broader community. The changes in these four spheres are summarised in Figure ##FIG##2##3##.</p>", "<p>The participant panels particularly valued changes in the intra-personal, group and familial spheres, while the NGO panels tended to focus a lot more on the importance of changes located in the societal sphere.</p>" ]

[ "<title>Discussion</title>", "<p>This participatory intervention study to promote the mental health of widows of IDUs apparently had a positive impact on the widows' quality of life and their mental health. However, the sustainability of the improvement and the links between this and their engagement in HIV risk behaviours remain to be demonstrated.</p>", "<p>While the physical, social and environmental domain scores of the quality of life measure improved significantly, the change in the social domain score did not. Qualitative findings, however, demonstrated improvements in social connectedness for the women. Had the sample sizes been larger, the change in the social domain scores may also have achieved statistical significance. For the women in Nagaland, the change in the scores for all four domains was not significant. The Quality of Life scores at baseline were higher among participants from Nagaland, and this together with the smaller sample size may account for the non-significant change observed in that state.</p>", "<p>While the intervention's positive impact on mental health as assessed by the GHQ12 is encouraging, identifying a high proportion of the group (70%) as potentially having a common mental disorder (anxiety or depression) is a matter for concern. An overestimation of this proportion is possible for technical reasons. The GHQ12 includes questions about the presence of physical symptoms; as some of the women were HIV positive and experiencing physical illness, their resulting higher scores may have resulted in some instances of (mis)classification as 'cases'. On the other hand, the level of distress among this vulnerable group is evident to those familiar with their situation, and this finding is not surprising.</p>", "<p>Given that women in India often express psychological distress as somatic symptoms [##REF##7894865##28##,##REF##15104094##29##], the observed reduction in the proportion of IDU widows reporting pain raises the possibility that this is connected to improved mental health, although the same improvements were not apparent in relation to the other somatic domains.</p>", "<p>During the MSC process the participant and NGO panels gave differing emphasis to the reasons for selecting particular stories as representing the 'most significant change'. The former tended to value changes in the intra-personal, group and familial spheres, while the latter gave more focus to changes in the broader community sphere, perhaps reflecting their focus on sustainability of programs. This has implications for the design and evaluation of interventions. These will ideally meet the needs of both program participants and implementers if the contribution to HIV prevention is to be valued at all levels. The funding organisation is another stakeholder that could be involved in the MSC process, and they may well have a different perspective again. The MSC approach facilitates organisational learning for all connected to programs that aim to create change for participants.</p>", "<p>The intervention generally, and the MSC process specifically, enhanced awareness among the NGO leadership of the struggles faced by IDU widows, especially those infected with HIV. Many of the NGOs providing HIV-related services in north-east India are male-dominated because the epidemic initially affected male IDUs predominantly. Prior to the intervention, IDU widows were isolated and powerless, and struggled to gain access to the few NGO programs relevant to them. Providing an opportunity for the widows to come together around the theme of mental health promotion has helped them to mobilise. Additionally, awareness of mental health as an important health issue for all people, including those with HIV infection, has increased.</p>", "<p>While the number of women participating in the intervention was relatively small and the intervention relatively brief, these findings contribute to knowledge and understanding about the use of participatory interventions to improve the mental health of vulnerable women, and demonstrate the potential of this approach to contribute to HIV prevention.</p>", "<p>The study has several limitations. As no control group met without receiving the structured program designed to promote mental health, we cannot know to what extent the observed improvements were related to the program content. The act of coming together may in itself have induced the changes. The attrition in some groups was disappointing but not unexpected given the stressful and at times unpredictable nature of the widows' lives, and an overall retention rate of 80% is high in this context. The linguistic and ethnic diversity and different literacy levels of the participants presented challenges during the intervention and data collection (see below). While most of these were overcome successfully with extra time, effort and research, we recommend that the participants in future groups be relatively homogenous. In relation to the intervention, we would also recommend that some HIV prevention education be included as even though these women's husbands had mostly died from AIDS, their level of HIV knowledge was poorer than expected. Extending the intervention to include training in advocacy and community mobilisation is also worth considering.</p>", "<p>A relatively generous allowance was paid to the women to cover travel and childcare costs. This was important in motivating and recruiting women to participate in the intervention at the outset. While we do not know the extent to which the allowance influenced participation, six months after the last PAG meeting most of the groups are still meeting and some have expanded their activities. No allowance is now paid to the women, but support for the meetings is being given by NGOs.</p>", "<p>Capturing sensitive information such as engagement in HIV risk behaviours proved to be challenging. North-east India is characterised by deeply felt conservative values, which are likely to have made it difficult for the widows to respond to questions about paid or unsafe sex, even though we had processes in place to assure confidentiality. The impact of improved mental health on engagement in HIV risk behaviours is therefore difficult to assess. Further work is required to develop sensitive and valid measures of sexual behaviour among groups such as these: including exploration of the women's perspectives on non-threatening methods for data collection about sexual behaviours, and trying other approaches such as participatory methods or individual narrative interviews with a trusted peer interviewer. Developing more effective methods for gathering information about sexual risk behaviours is essential for future research into the relationship between mental health and engagement in HIV risk behaviours.</p>", "<p>Finally, the data collection tools had to be translated into the local language and back-translated. The meaning of individual questions in questionnaires such as the WHOQOL-BREF are nuanced and therefore not easy to exactly translate into languages that do not have the spectrum of words to communicate subtle differences in meaning. Imphal and Churachandpur in Manipur are only two hours apart by vehicle, but they do not share a common language, and two different translations were required for that state. Some people are able to read Manipuri when it is written in Manipuri script, while others can only read it in Bengali or transliterated Roman script. Nagamese is a spoken language with no official written form. Accurate translation of the data collection tools was an arduous task and further refinement and validation is warranted. This is typical of the many challenges encountered when trying to undertake quality research in remote settings.</p>" ]

[ "<title>Conclusion</title>", "<p>This pilot intervention study used a range of innovative approaches to program design, implementation and evaluation in order to reach a vulnerable group of women in a complex development setting with high HIV prevalence. The findings demonstrate that a participatory approach to mental health promotion can have a positive impact on the health and quality of life of vulnerable women. The intervention would benefit from further trialling and refinement and could be made available to other groups such as women living with HIV, the wives of IDUs, and specific sub-groups of widows such as sex workers and IDUs. Further research to evaluate the impact of the intervention on the lives of vulnerable women and to investigate the role of mental health promotion as a strategy for HIV prevention is warranted.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>HIV prevalence in north-east India is high and injecting drug use (IDU) is common. Due to HIV-related deaths there are increasing numbers of IDU widows, many of whom are HIV infected, and experiencing poor health, social isolation, discrimination and poverty, all factors likely to be compromising their mental health. There is increasing recognition of the links between HIV and mental health.</p>", "<title>Methods</title>", "<p>The aim of this study was to pilot a peer-facilitated, participatory action group (PAG) process and assess the impact of the intervention on the mental health of participants. The intervention consisted of 10 PAG meetings involving 74 IDU widows. Changes in quality of life (WHOQOL-BREF), mental health (GHQ12) and somatic symptoms were assessed. The value of the intervention from the perspective of the participants was captured using a qualitative evaluation method (Most Significant Change).</p>", "<title>Results</title>", "<p>Participants' quality of life, mental health and experience of somatic symptoms improved significantly over the course of the intervention, and the women told stories reflecting a range of 'significant changes'.</p>", "<title>Conclusion</title>", "<p>This pilot intervention study demonstrated that a participatory approach to mental health promotion can have a positive impact on the lives of vulnerable women, and the potential to contribute to HIV prevention. Further investigation is warranted.</p>" ]

[ "<title>Abbreviations</title>", "<p>AIDS: Acquired Immune Deficiency Syndrome; DFID: Department of International Development; GHQ12: General health Questionnaire 12; HIV: Human Immunodeficiency Virus; IDUs: Injecting Drug Use/Users; MSC: Most Significant Change; NGO: Non Governmental Organization; PAG: Participatory Action Group; PAR: Participatory Action Research; WHO: World health Organization; WHOQOL BREF: WHO Quality of Life Abbreviated tool.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>HH, AD and MK were involved in conception of the study. HH, AD, MK and PC were involved in design, implementation and data analysis of the study. BD and TG were involved in the implementation and data analysis of the study. MK drafted the paper with contributions from AD, HH and PC. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2458/8/294/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>The study was funded by the UK Department for International Development (DFID) through the Research and Learning Fund. The authors would like to acknowledge the contribution of the following north-east Indian organisations: Project ORCHID; Development Association of Nagaland (DAN), Guardian Angel (GA), Lamka Rehabilitation &amp; Research Centre Evangelical Baptist Convention (LRRC), Manipur Network of Positive People (MNP+), Naga Mothers Association HIV/AIDS Care and Support Hospice (NMA), Sneha Bhavan (SB) and the Emmanuel Hospital Association. We are also very grateful to the following people: Dr B Langkham, Dr Rebecca Sinate (Emmanuel Hospital Association); Mr Aviw-u Nienu (NMA); Mr Raphuba Paul Chawang (DAN); Ms Leishipem Shingnai (GA); Ms Kim and Ms Bina (MNP+); Ms Hatnu Simte (SB); Ms Biak Lun (LRRC); Ms Angela Rintoul (AIHI); and Ms N.Roopa (India). We thank Dr Jessica Dart (Clear Horizon) for her assistance with the Most Significant Change component of the study and Ms Suzanne O'Neill (Burnet Institute) for her contribution to the project design. Finally, we want to warmly acknowledge the contribution made by the peer facilitators and all the women who participated in the study. Without their courage and determination it would not have been possible.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Process of Most Significant Change story selection.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Two stories selected as representing the 'most significant change'.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Categories of valued changes identified using the Most Significant Change method.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Outline of the ten PAG meetings for widows of IDUs</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Session</bold></td><td align=\"left\"><bold>Outline</bold></td><td align=\"left\"><bold>Data Collection</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>1</bold></td><td align=\"left\"> • Introduction to the PAG process</td><td align=\"left\"> • Baseline questionnaires:</td></tr><tr><td/><td align=\"left\"> • Identifying members' expectations</td><td align=\"left\">  - WHO QOL-BREF</td></tr><tr><td/><td align=\"left\"> • Highlighting individual strengths and skills</td><td align=\"left\">  - GHQ12</td></tr><tr><td/><td/><td align=\"left\">  - Health Risk Questionnaire</td></tr><tr><td/><td/><td align=\"left\"> • Meeting summary report</td></tr><tr><td align=\"left\"><bold>2</bold></td><td align=\"left\"> • Concepts and determinants of mental health for widows of IDUs</td><td align=\"left\"> • Focus Group Discussion</td></tr><tr><td/><td/><td align=\"left\"> • Meeting summary report</td></tr><tr><td align=\"left\"><bold>3</bold></td><td align=\"left\"> • Mental health and mental health promotion</td><td align=\"left\"> • Meeting summary report</td></tr><tr><td align=\"left\"><bold>4</bold></td><td align=\"left\"> • Envisioning a positive future</td><td align=\"left\"> • Meeting summary report</td></tr><tr><td/><td align=\"left\"> • Promoting social inclusion</td><td/></tr><tr><td align=\"left\"><bold>5</bold></td><td align=\"left\"> • Addressing stigma and discrimination</td><td align=\"left\"> • Meeting summary report</td></tr><tr><td/><td align=\"left\"> • Relaxation techniques</td><td/></tr><tr><td align=\"left\"><bold>6</bold></td><td align=\"left\"> • Improving access to work and resources</td><td align=\"left\"> • Meeting summary report</td></tr><tr><td/><td align=\"left\"> • Prioritising ideas for action plan development</td><td/></tr><tr><td align=\"left\"><bold>7</bold></td><td align=\"left\"> • Developing action plans</td><td align=\"left\"> • Meeting summary report</td></tr><tr><td align=\"left\"><bold>8</bold></td><td align=\"left\"> • Developing action plans</td><td align=\"left\"> • Collection of MSC stories</td></tr><tr><td/><td align=\"left\"> • MSC approach</td><td align=\"left\"> • Meeting summary report</td></tr><tr><td align=\"left\"><bold>9</bold></td><td align=\"left\"> • Mental health and HIV</td><td align=\"left\"> • Focus Group Discussion</td></tr><tr><td/><td align=\"left\"> • Feedback of MSC stories</td><td align=\"left\"> • Meeting summary report</td></tr><tr><td align=\"left\"><bold>10</bold></td><td align=\"left\"> • Finalising action plans</td><td align=\"left\"> • Post-intervention questionnaires:</td></tr><tr><td/><td align=\"left\"> • Celebration</td><td align=\"left\">  - WHO QOL-BREF</td></tr><tr><td/><td/><td align=\"left\">  - GHQ12</td></tr><tr><td/><td/><td align=\"left\">  - Health Risk Questionnaire</td></tr><tr><td/><td/><td align=\"left\"> • Meeting summary report</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Retention in the intervention and reasons for dropping out</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Group</bold></td><td align=\"left\"><bold>No. started</bold></td><td align=\"left\"><bold>No. finished</bold></td><td align=\"left\"><bold>Retention (%)</bold></td><td align=\"left\"><bold>Reasons for attrition</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Manipur 1</bold></td><td align=\"left\">16</td><td align=\"left\">14</td><td align=\"left\">87.5</td><td align=\"left\">• Attending detoxification program</td></tr><tr><td/><td/><td/><td/><td align=\"left\">• Language barriers</td></tr><tr><td align=\"left\"><bold>Manipur 2</bold></td><td align=\"left\">14</td><td align=\"left\">13</td><td align=\"left\">92.8</td><td align=\"left\">• Illness</td></tr><tr><td align=\"left\"><bold>Manipur 3</bold></td><td align=\"left\">11</td><td align=\"left\">11</td><td align=\"left\">100</td><td align=\"left\">N/A</td></tr><tr><td align=\"left\"><bold>Nagaland 1</bold></td><td align=\"left\">12</td><td align=\"left\">7</td><td align=\"left\">58.3</td><td align=\"left\">• Returned to home village</td></tr><tr><td/><td/><td/><td/><td align=\"left\">• Death of participant</td></tr><tr><td/><td/><td/><td/><td align=\"left\">• Death of children</td></tr><tr><td/><td/><td/><td/><td align=\"left\">• Re-marriage</td></tr><tr><td/><td/><td/><td/><td align=\"left\">• Drug use</td></tr><tr><td align=\"left\"><bold>Nagaland 2</bold></td><td align=\"left\">9</td><td align=\"left\">5</td><td align=\"left\">55.5</td><td align=\"left\">• Ineligibility (not a widow)</td></tr><tr><td/><td/><td/><td/><td align=\"left\">• Decided not to participate</td></tr><tr><td/><td/><td/><td/><td align=\"left\">• Illness</td></tr><tr><td align=\"left\"><bold>Nagaland 3</bold></td><td align=\"left\">12</td><td align=\"left\">9</td><td align=\"left\">75.0</td><td align=\"left\">• Childbirth</td></tr><tr><td/><td/><td/><td/><td align=\"left\">• Illness</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\"><bold>TOTAL</bold></td><td align=\"left\"><bold>74</bold></td><td align=\"left\"><bold>59</bold></td><td align=\"left\"><bold>79.7</bold></td><td/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Demographics of the participants attending first PAG meeting</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Variable</bold></td><td align=\"center\"><bold>Manipur n = 41</bold></td><td align=\"center\"><bold>Nagaland n = 33</bold></td><td align=\"center\"><bold>Total n = 74</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Average age (range)</bold></td><td align=\"center\">33 yrs (23–46)</td><td align=\"center\">32 yrs (20–52)</td><td align=\"center\">32.5 yrs (20–52)</td></tr><tr><td align=\"left\"><bold>Average length of widowhood (range)</bold></td><td align=\"center\">4.5 yrs (3 mth-12.8 yrs)</td><td align=\"center\">4.2 yrs (1 mth-11 yrs)</td><td align=\"center\">4.3 yrs (1 mth-12.8 yrs)</td></tr><tr><td align=\"left\"><bold>Average no. children (range)</bold></td><td align=\"center\">2.2 (1–4)</td><td align=\"center\">2.7 (0–7)</td><td align=\"center\">2.4 (0–7)</td></tr><tr><td align=\"left\"><bold>Ethnicity – n (%)</bold></td><td/><td/><td/></tr><tr><td align=\"left\"> Naga</td><td align=\"center\">2 (5)</td><td align=\"center\">26 (79)</td><td align=\"center\">28 (38)</td></tr><tr><td align=\"left\"> Meitei</td><td align=\"center\">26 (63)</td><td align=\"center\">0</td><td align=\"center\">26 (35)</td></tr><tr><td align=\"left\"> Other</td><td align=\"center\">11 (27)</td><td align=\"center\">7 (21)</td><td align=\"center\">18(24)</td></tr><tr><td align=\"left\"> Missing</td><td align=\"center\">2 (5)</td><td align=\"center\">0</td><td align=\"center\">2 (3)</td></tr><tr><td align=\"left\"><bold>Religion – n (%)</bold></td><td/><td/><td/></tr><tr><td align=\"left\"> Christian</td><td align=\"center\">14 (34)</td><td align=\"center\">30 (91)</td><td align=\"center\">44 (59)</td></tr><tr><td align=\"left\"> Hindu</td><td align=\"center\">25 (61)</td><td align=\"center\">2 (6)</td><td align=\"center\">27 (37)</td></tr><tr><td align=\"left\"> Muslim</td><td align=\"center\">0</td><td align=\"center\">1 (3)</td><td align=\"center\">1 (1)</td></tr><tr><td align=\"left\"> Missing</td><td align=\"center\">2 (5)</td><td align=\"center\">0</td><td align=\"center\">2 (3)</td></tr><tr><td align=\"left\"><bold>Employment – n (%)</bold></td><td/><td/><td/></tr><tr><td align=\"left\"> Employed</td><td align=\"center\">2 (5)</td><td align=\"center\">10 (30)</td><td align=\"center\">12 (16)</td></tr><tr><td align=\"left\"> Unemployed</td><td align=\"center\">37 (90)</td><td align=\"center\">23 (70)</td><td align=\"center\">60 (81)</td></tr><tr><td align=\"left\"> Missing</td><td align=\"center\">2 (5)</td><td align=\"center\">0</td><td align=\"center\">2 (3)</td></tr><tr><td align=\"left\"><bold>Education – n (%)</bold></td><td/><td/><td/></tr><tr><td align=\"left\"> None</td><td align=\"center\">-</td><td align=\"center\">7 (21)</td><td align=\"center\">7 (9)</td></tr><tr><td align=\"left\"> Primary</td><td align=\"center\">5 (12)</td><td align=\"center\">1 (3)</td><td align=\"center\">6 (8)</td></tr><tr><td align=\"left\"> Secondary</td><td align=\"center\">15 (37)</td><td align=\"center\">24 (73)</td><td align=\"center\">39 (53)</td></tr><tr><td align=\"left\"> Tertiary</td><td align=\"center\">21 (51)</td><td align=\"center\">1 (3)</td><td align=\"center\">22 (30)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Mean scores (SD) for Quality of Life domains at the beginning and end of the intervention by state</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Domain</bold></td><td align=\"center\" colspan=\"3\"><bold>Manipur (n = 38)</bold></td><td align=\"center\" colspan=\"3\"><bold>Nagaland (n = 21)</bold></td><td align=\"center\" colspan=\"3\"><bold>Total (n = 59)</bold></td></tr></thead><tbody><tr><td/><td align=\"center\"><bold>Pre</bold></td><td align=\"center\"><bold>Post</bold></td><td align=\"center\"><bold>p value</bold></td><td align=\"center\"><bold>Pre</bold></td><td align=\"center\"><bold>Post</bold></td><td align=\"center\"><bold>p value</bold></td><td align=\"center\"><bold>Pre</bold></td><td align=\"center\"><bold>Post</bold></td><td align=\"center\"><bold>p value</bold></td></tr><tr><td colspan=\"10\"><hr/></td></tr><tr><td align=\"left\">Physical</td><td align=\"center\">42 (15.1)</td><td align=\"center\">58 (13.7)</td><td align=\"center\">&lt;0.01</td><td align=\"center\">55 (16.1)</td><td align=\"center\">57 (11.4)</td><td align=\"center\">NS</td><td align=\"center\">47 (16.7)</td><td align=\"center\">57 (12.8)</td><td align=\"center\">&lt;0.01</td></tr><tr><td align=\"left\">Psychological</td><td align=\"center\">51 (18.5)</td><td align=\"center\">59 (15.5)</td><td align=\"center\">&lt;0.01</td><td align=\"center\">53 (12.2)</td><td align=\"center\">57 (9.9)</td><td align=\"center\">NS</td><td align=\"center\">52 (16.5)</td><td align=\"center\">59 (13.7)</td><td align=\"center\">&lt;0.01</td></tr><tr><td align=\"left\">Social</td><td align=\"center\">43 (20.3)</td><td align=\"center\">49 (23.2)</td><td align=\"center\">NS</td><td align=\"center\">52 (12.8)</td><td align=\"center\">53 (17.2)</td><td align=\"center\">NS</td><td align=\"center\">46 (18.3)</td><td align=\"center\">51 (21.2)</td><td align=\"center\">NS</td></tr><tr><td align=\"left\">Environmental</td><td align=\"center\">34 (18.1)</td><td align=\"center\">45 (12.1)</td><td align=\"center\">&lt;0.01</td><td align=\"center\">45 (14.0)</td><td align=\"center\">47 (13.5)</td><td align=\"center\">NS</td><td align=\"center\">38 (17.5)</td><td align=\"center\">46 (12.5)</td><td align=\"center\">&lt;0.01</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Proportion of participants with a possible common mental disorder as assessed by GHQ12 (using 3/4 cut-off) at the beginning and end of the intervention by state</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\" colspan=\"3\"><bold>Manipur</bold></td><td align=\"center\" colspan=\"3\"><bold>Nagaland</bold></td><td align=\"center\" colspan=\"3\"><bold>Total</bold></td></tr></thead><tbody><tr><td align=\"center\"><bold>Pre</bold></td><td align=\"center\"><bold>Post</bold></td><td align=\"center\"><bold>p value</bold></td><td align=\"center\"><bold>Pre</bold></td><td align=\"center\"><bold>Post</bold></td><td align=\"center\"><bold>p value</bold></td><td align=\"center\"><bold>Pre</bold></td><td align=\"center\"><bold>Post</bold></td><td align=\"center\"><bold>p value</bold></td></tr><tr><td colspan=\"9\"><hr/></td></tr><tr><td align=\"center\">70%</td><td align=\"center\">49%</td><td align=\"center\">NS</td><td align=\"center\">70%</td><td align=\"center\">30%</td><td align=\"center\">0.04</td><td align=\"center\">70%</td><td align=\"center\">42%</td><td align=\"center\">&lt;0.01</td></tr><tr><td align=\"center\">26/37</td><td align=\"center\">18/37</td><td/><td align=\"center\">23/33</td><td align=\"center\">6/20</td><td/><td align=\"center\">49/70</td><td align=\"center\">24/57</td><td/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T6\"><label>Table 6</label><caption><p>Somatic complaints at the beginning and end of the intervention</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><italic><bold>Somatic symptom group</bold></italic></td><td align=\"center\" colspan=\"3\"><italic>Manipur</italic></td><td align=\"center\" colspan=\"3\"><italic>Nagaland</italic></td><td align=\"center\" colspan=\"3\"><italic>Total</italic></td></tr></thead><tbody><tr><td/><td align=\"left\"><bold>Pre </bold><break/><bold>(SD)</bold></td><td align=\"left\"><bold>Post </bold><break/><bold>(SD)</bold></td><td align=\"left\"><bold>p value</bold></td><td align=\"left\"><bold>Pre </bold><break/><bold>(SD)</bold></td><td align=\"left\"><bold>Post </bold><break/><bold>(SD)</bold></td><td align=\"left\"><bold>p value</bold></td><td align=\"left\"><bold>Pre </bold><break/><bold>(SD)</bold></td><td align=\"left\"><bold>Post </bold><break/><bold>(SD)</bold></td><td align=\"center\"><bold>p value</bold></td></tr><tr><td colspan=\"10\"><hr/></td></tr><tr><td align=\"left\">Pain</td><td align=\"left\">8.6 <break/>(2.9)</td><td align=\"left\">7.4 <break/>(3.2)</td><td align=\"left\">NS <break/>n = 26</td><td align=\"left\">10.0 <break/>(1.7)</td><td align=\"left\">8.7 <break/>(2.4)</td><td align=\"left\">= 0.04 <break/>n = 16</td><td align=\"left\">9.3 <break/>(2.5)</td><td align=\"left\">7.9 <break/>(3.0)</td><td align=\"center\">&lt;0.01 <break/>n = 42</td></tr><tr><td align=\"left\">Sensory sensations</td><td align=\"left\">16.9 <break/>(6.1)</td><td align=\"left\">15.6 <break/>(5.4)</td><td align=\"left\">NS <break/>n = 16</td><td align=\"left\">16.0 <break/>(4.7)</td><td align=\"left\">12.0 <break/>(3.5)</td><td align=\"left\">= 0.01 <break/>n = 8</td><td align=\"left\">16.7 <break/>(5.6)</td><td align=\"left\">14.5 <break/>(5.1)</td><td align=\"center\">NS <break/>n = 24</td></tr><tr><td align=\"left\">Bodily functions</td><td align=\"left\">7.0 <break/>(2.9)</td><td align=\"left\">6.2 <break/>(2.5)</td><td align=\"left\">NS <break/>n = 23</td><td align=\"left\">7.5 <break/>(3.6)</td><td align=\"left\">7.6 <break/>(1.9)</td><td align=\"left\">NS <break/>n = 8</td><td align=\"left\">7.1 <break/>(3.0)</td><td align=\"left\">6.6 <break/>(2.4)</td><td align=\"center\">NS <break/>n = 31</td></tr><tr><td align=\"left\">Reproductive health</td><td align=\"left\">3.3 <break/>(2.4)</td><td align=\"left\">5.4 <break/>(4.1)</td><td align=\"left\">= 0.02 <break/>n = 27</td><td align=\"left\">4.4 <break/>(2.7)</td><td align=\"left\">3.0 <break/>(2.5)</td><td align=\"left\">NS <break/>n = 11</td><td align=\"left\">3.7 <break/>(2.5)</td><td align=\"left\">4.7 <break/>(3.8)</td><td align=\"center\">NS <break/>n = 35</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>NS = not significant at p &lt; 0.05.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2458-8-294-1\"/>", "<graphic xlink:href=\"1471-2458-8-294-2\"/>", "<graphic xlink:href=\"1471-2458-8-294-3\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The Spanish National Health Service (NHS) provides universal cover and is financed, essentially, by general taxes. The system is divided into primary and secondary care. Primary care (PC) is organized as a network of primary health care teams (PHCT) that behave as geographical and administrative units where PC services are planned, managed and provided for a population ranging from 5,000 to 25,000 citizens. The PHCT staff includes: general practitioners (GPs), paediatricians, nurses, social workers, dentists and ancillary staff. GPs care for individuals older than 14 years (except in rural areas where they look after all members of the local population) and paediatricians look after those between 0 and 14 or 16 years, depending on the region. These clinicians act as gate-keepers for the rest of the public health care system. The PHCT works usually in a single health care centre but in rural areas there may have other small offices.</p>", "<p>Secondary care includes out-patient and in-patient hospital care and out-patient care in multi-disciplinary clinics.</p>", "<p>The Spanish NHS has been strongly decentralized into 17 Autonomous Communities or regions that configure the Spanish State. Each of these regions has its own governmental structure which, among other responsibilities, provides health-care services to the population. There are minor differences between them with respect to structure and administration. For example, in Aragon and Navarra all services are provided by the State-funded regional service while in Catalonia, the provision of primary care services is offered by different providers (state and non-state funded) among which the Catalan Health Institute (ICS) manages almost 80% of all PHCT. According to latest census of the year 2002, Aragon had a population of 1,209,888 inhabitants while Navarra had 560,235 inhabitants and Catalonia 6,418,387.</p>", "<title>Preventive services in Primary Care</title>", "<p>PC is the most accessible health-care level to the general population. In Spain, more than 95% of the population visit their GP at least once in every 5 years. No other health-care level is in a better position to evaluate the global health status of the individual and to decide when to act and what are the ideal measures to take in each specific situation [##UREF##0##1##].</p>", "<p>There is an increasing evidence of the health benefits achieved from the implementation of preventive measures in normal clinical practice. In the last quarter of the century several groups of experts such as the Canadian Task Force on Periodic Health Examination in 1979 [##REF##115569##2##] and the US Task Force in 1980 [##REF##6827744##3##,##UREF##1##4##] have published evidence-based recommendations regarding the relevance and the outcomes of the implementation of preventive interventions. Prestigious institutions such as WHO and the Royal College of General Practitioners [##UREF##2##5##] have highlighted the valuable role of health-care professionals within PCHTs in developing these services.</p>", "<p>In Spain, the Spanish Society of Family and Community Medicine <italic>[Sociedad Española de Medicina de Familia y Comunitaria; semFYC] </italic>launched in 1988 the Preventive Activities and Health Promotion Program <italic>[Programa de Actividades Preventivas y de Promoción de la Salud; PAPPS] </italic>that have as main objective promoting the implementation of preventive and health promotion services in PC [##UREF##3##6##]. The EUROPREV (European Network for Prevention and Health Promotion in Family Medicine and General Practice) was created to extend and coordinate the experiences from this program and to promote preventive services at the European level.</p>", "<p>In the period between 1989 and 2003 PAPPS program carried out biennial evaluations of a representative sample of patients attending the PHCTs participating in the program (671 PHCT since the beginning of the program). These evaluations have been the only evaluations of preventive services carried out nation-wide and provide information on the progress and effectiveness of the program over its 15 years of existence [##REF##8732325##7##, ####UREF##4##8##, ##UREF##5##9####5##9##]. From the studies we know that the best-implemented group of preventive services is the minimum package of the adult sub-program which includes screening for hypertension, smoking status and alcohol consumption (done in 92.8% of patients), followed by a second group that includes anti-influenza vaccination, screening for hypercholesterolemia and obesity and anti-tetanus vaccination (done in more than 70% of patients) [##UREF##4##8##].</p>", "<p>However, the progressive implementation of systematic evaluations of primary care performance promoted by the Autonomous health administrations and the progressive computerization of clinical records has brought-about a re-thinking of this evaluation model. The computerization of clinical records provides rich information based on individual data (not centre aggregated data) on a wider and more representative sample that avoids manual and voluntary recording of data. This allows a better study of the clinical practice variability (CPV) and its determining factors.</p>", "<title>Clinical practice variability in preventive services in Primary Care</title>", "<p>In Spain, there have been few nation-wide population studies evaluating CPV in PC. This has been due, in part, to the lack of reliable information systems. These studies have often been performed at a hospital level because of the existence of minimum basic data sets (MBDS) provided on discharge from hospital. Nevertheless, from the existing evaluations, PAPPs studies [##UREF##4##8##,##UREF##5##9##] and other national [##REF##11162322##10##, ####REF##15766491##11##, ##REF##11333638##12##, ##REF##12622982##13####12622982##13##]and international [##REF##7769470##14##, ####REF##11434075##15##, ##REF##12399352##16####12399352##16##] studies it is clear that there are important difficulties with respect to the correct implementation of preventive services and health promotion interventions and there is a high variability in their implementation. There is extensive literature on factors related to prevention of specific diseases but less related to implementation of a combination of preventive services as routine clinical practice in PC that may have a profound effect on different diseases. These factors vary considerably, depending on the type of study, the country and type of preventable disease. There are different factors associated with the supply and demand of services that result in elevated variability between and within centres: patient-related factors (socio-demographic and clinical factors), health professional associated factors (specialty, training and professional competence, number of years working with the PHCT), team related factors (existence of specific registries, workload, frequency of doctors encounters, teaching centre) and other health-care provision system characteristics [##REF##11333638##12##].</p>", "<p>High CPV translates into problems of clinical effectiveness and social efficiency in health-care provision and, as such, is of concern to health-care providers and to Society in general [##REF##9567282##17##,##REF##12585825##18##]. An analysis of CPV is therefore essential in decision making in the health-care provision politics and prioritization in clinical management (e.g. to generate guidelines for implementation in those areas of high clinical variability).</p>", "<p>Hence, it is essential to have studies that include a territory-wide sampling in order to achieve greater representativeness and which take into account methods of design and statistical analysis (e.g. cluster-based analyses) to obtain valid estimates and to explore the influence of patient, health-care professional and PHCT characteristics on the CPV of preventive services in Spain.</p>", "<title>Sources of information in Primary Care: the computerization of clinical records</title>", "<p>Primary Care uses electronic clinical records (ECRs) to monitor health problems and to register preventive care services. Although information collection intends to be comprehensive, the ECR systems are different for each region.</p>", "<p>Routinely-collected data have undisputed advantages in the study of CPV; they are available almost instantaneously and provide information on a large number of patients. The creation of registries or databases of health-care interest in PC was given a considerable boost with the computerization of clinical records. Spain, contrary to other countries [##REF##15171985##19##, ####REF##15119513##20##, ##REF##11052168##21##, ##REF##10213569##22##, ##UREF##6##23####6##23##], has little experience in the use of nation-wide databases containing electronic clinical records in PC. Recently, projects such as the BIFAP project in pharmacoepidemiological research [##REF##12525343##24##]<ext-link ext-link-type=\"uri\" xlink:href=\"http://www.bifap.org\"/> have been initiated in Spain.</p>", "<p>However, there are certain difficulties in their use: 1) The data was originally collected for a function different from current research requirements; 2) ECR systems are different in each region, with different standards and computer programs; 3) The heterogeneous degree of implementation in PHCTs; and 4) The degree of exhaustivity of data recorded (for example, some activities are carried out but are not registered). Nevertheless, the potential of this type of registry has been confirmed by different European experiences mentioned earlier [##REF##12829558##25##, ####REF##11238139##26##, ##REF##15073071##27####15073071##27##]. As such, the advantages associated to computerization of clinical records are beginning to be appreciated by the different regions and the richness of information provides enormous potential for the study of CPV in primary care.</p>", "<title>Intraclass correlation coefficient</title>", "<p>Reports of cluster randomised studies [##REF##15837446##28##] should include sample size calculations and statistical analyses that take clustering into account. The intraclass correlation coefficient (ICC) is a measure of the relatedness of clustered data [##REF##15209195##29##]. Compared with individual randomisation, cluster randomisation may substantially increase the sample size required to maintain adequate statistical power [##REF##12509376##30##].</p>", "<p>One of the problems that needs to be faced in designing cluster-based studies (area or organization-based studies) is that estimates of ICC and components of variance for the outcomes of interest need to be obtained from previous studies in order to determine the required sample size for the cluster design [##REF##12509376##30##, ####REF##7315838##31##, ##REF##7573621##32####7573621##32##]. There have been repeated calls for publication of ICCs for these types of studies to help others who are planning cluster-based studies [##REF##15209195##29##,##REF##9794847##33##, ####REF##15115554##34##, ##REF##9572764##35####9572764##35##].</p>", "<p>The differences [##REF##12509376##30##] in ICCs among potential outcome variables, the paucity of appropriate information [##REF##15485730##36##] concerning components of variance or ICC and the lack of data obtained from PC settings reinforce the need for valid estimates that would ensure proper study design.</p>", "<title>Study objectives</title>", "<p>The planned objectives in the present study are:</p>", "<p>- To describe the variability in the performance of 7 preventive services (detection of smoking status and excessive alcohol consumption; screening for hypertension, obesity and hypercholesterolemia; anti-influenza and anti-tetanus immunization) in a population older than 16 years of age receiving attention in years 2006 or 2007 and to analyse factors related to their implementation.</p>", "<p>- To analyse the evolution of 5 identified health problems (decrease in consumption or abstention from tobacco and alcohol and degree of control of hypertension, hypercholesterolemia and obesity) up to December 2008 in the population older than16 years of age attended by the reference PHCT at least once between 2006 and 2007 and followed up during 2008.</p>", "<p>- Estimate the ICC, together with the within-cluster and between-cluster components of variance for each of the outcome measures of interest at the PC level.</p>" ]

[ "<title>Methods/Design</title>", "<title>Design</title>", "<p>This is a multi-centered cross-sectional study of a randomised sample of PHCT stratified by region to describe the variability in the implementation of 7 preventive services and their related factors in the period of 2006 or 2007. In 2008, using another cross-sectional study, the evolution or degree of control of 5 of the identified health problems will be assessed, together with the determinants of variability. We use a longer period of study because we need more time to evaluate the impact on the degree of control of these health problems due to changes on life styles.</p>", "<title>Setting</title>", "<p>The study is set in PHCTs computerized prior to 1st of January 2005 in the regions of Aragón, Navarra, and Catalonia. So far computerization had occurred in 19% of 121 PHCTs in Aragón, 76% of 54 PHCTs in Navarra, and 97.5% of 278 PHCTs in Catalonia (at the Catalan Health Institute).</p>", "<title>Study population: PHCT and patients</title>", "<p>We performed random cluster sampling stratified by region with computerized PHCT used as unit of randomisation.</p>", "<p>The inclusion criteria of the PHCT are: 1) computerization of the clinical records before the 1st of January 2005 to ensure familiarity with the system's use; 2) wide use of ECR, and 3) agreement to participate in the study by most of the health-care professionals working in each PHCT (&gt; 80%).</p>", "<p>For each computerized PHCT we select the ECRs of all subjects older than16 years as the unit of data collection. For the assessment of variability, the ECR of subjects receiving attention at least once in each of the study years will be extracted from the database. It is estimated that the average number of subjects receiving attention per year is around 70%. Thus in a PHCT in Aragon this figure would be around 10,000 patients, in Navarra around 7,500 and in Catalonia around 12,000. For the analysis of the evolution of health problems and related factors, subjects visited at least on one occasion during the 2006–2007 period and on a further occasion in 2008 will be included.</p>", "<title>Sample size and randomisation</title>", "<p>To achieve the equivalent power of a patient-randomised design, in a cluster randomised design the effective sample size will need to be the standard sample size of a single random sampling multiplied by the design effect [##REF##7315838##31##,##REF##10982317##37##,##UREF##7##38##]. The design effect is a function of the average cluster (computerized PHCT) size and the (intraclass correlation coefficient (ICC) of the outcome:</p>", "<p></p>", "<p>where <italic>n </italic>is the average number of individuals sampled per cluster [##UREF##7##38##].</p>", "<p>sWith the cluster design we set an ICC of 0.05 on the assumption that the ICCs estimated for outcome variables from PC are less than 0.05 [##REF##15837446##28##,##REF##15485730##36##,##REF##11180309##39##] and that the average number of individuals sampled per centre is at least of 2,500. Therefore 20 computerized PHCTs are necessary from each region to guarantee a statistical power of 80%, assuming a 50% event rate with an alpha two-tailed significance level of 0.05. Finally, to allow any of the randomised computerized PHCTs declining our offer to participate, we increased the number of computerized PHCTs selected from each region to 22. As such, the total number of subjects included will be more than 300,000 in Aragón, 200,000 in Navarra and 350,000 in Catalunya i.e. more than 850,000 subjects in total. This volume of subjects ensures a good estimate of the different outcome measures in the sub-groups of each cluster.</p>", "<p>The selection of computerized PHCT to participate in the study is by a simple random sample within each region. Sample size calculation and computerized PHCT selection is performed with Epidat 3.1.</p>", "<title>Outcomes and others measures</title>", "<title>Primary outcome measures</title>", "<p>The <bold>primary outcome measures relating to objective 1 </bold>include the description of the implementation of preventive services according to the PAPPS criteria [##UREF##3##6##].</p>", "<p>Screening for smoking status: if there is information on the consumption of tobacco over the last two years in members of the population who previously were non-smokers.</p>", "<p>Screening for excessive alcohol consumption: if there is information on the consumption of alcohol over the last two years in members of the population who previously were not high consumers of alcohol.</p>", "<p>Screening for obesity: if there have been measurements of height and weight or body mass index (BMI) over the last 4 years in members of the population who previously were non-obese.</p>", "<p>Screening for hypertension: if there is information on systolic blood pressure (BPS) and diastolic blood pressure (BPD) during the last 4 years in members of the population aged between 16 and 40 years who previously did not have the diagnosis of hypertension; or over the past 2 years in members of the population older than 40 years who previously did not have the diagnosis of hypertension.</p>", "<p>Screening for hypercholesterolemia: if there is information on total blood cholesterol corresponding to the following criteria of age and gender: 1 measurement in males between 16–35 years old; 1 measurement in females between 16–45 years old; 1 measurement every 5 years in males between 35–75 years old; 1 measurement every 5 years in females between 45–75 years old; 1 measurement if there has not been a previous measurement in those older than 75; all of these performed in population that previously did not have the diagnosis of hypercholesterolemia.</p>", "<p>Anti-tetanus vaccination (ATV): if there is a registry of anti-tetanus vaccination over the previous 10 years.</p>", "<p>Anti-influenza vaccination (AIV): if there is a registry of annual anti-influenza vaccination in those older than 59 years of age.</p>", "<p>The <bold>primary outcome measures relating to objective 2 </bold>describes the clinical follow-up of the five health problems identified up to December 31^st2007 and evaluated up to December 31^st2008:</p>", "<p>Evolution of tobacco consumption: quit smoking or remaining an ex-smoker.</p>", "<p>Evolution of excessive alcohol consumption: cessation of consumption or decrease below the definition of excessive consumption (28 alcohol units/week in males and 17 alcohol units/week in females, where 1 alcohol unit = 10 grams of alcohol).</p>", "<p>Evolution of obesity: decrease in BMI to &lt;30 kg/m²or change in the diagnosis to non-obese.</p>", "<p>Evolution of hypertension: mean of the last 2 determinations of systolic BP/diastolic BP registered as being ≤ 130/80 mmHg in patients of high risk (diagnosed with diabetes, renal or cardiac insufficiency) and ≤ 140/90 mmHg in the rest of the population.</p>", "<p>Evolution of hypercholesterolemia: last measurement of LDL cholesterol registered as being ≤ 100 mg/dL in high-risk patients (diagnosed with diabetes, ischaemic heart disease or cerebrovascular accident) and ≤ 130 mg/dL in the rest of the population.</p>", "<title>Secondary outcome measures</title>", "<p>Number of different preventive services (among the 7 of the study) registered for each individual during the 2006 and 2007 year periods.</p>", "<title>Other measures</title>", "<p>We assess the following factors related to the outcome measures according to the levels of analysis:</p>", "<p>Computerized PHCT: variables of structure and organization of the PHCT (number of GPs and nurses, presence of specialists), adherence to PAPPS, training centre for family doctors, ECR system and starting year, assigned population.</p>", "<p>Health-care professionals (GPs, nurses): gender, age, specialty, years of employment in PC and in the PHCT, workload (mean number of visits/day), years of experience with ECR.</p>", "<p>Subjects: gender, age, comorbidity according to the Adjusted Clinical Groups grouper [##REF##1902278##40##,##UREF##8##41##], frequency of consultations (visits/year).</p>", "<title>Information source</title>", "<p>The individual data will be extracted directly from the central computer servers of the ECRs of the subjects included in the study. In Navarra and Aragón the system used is OMI-AP and in Catalonia the eCAP system. The data related to health-care professionals and to PHCT will be obtained from an on-line database for health-care professionals and from a person responsible in each centre.</p>", "<title>Data extraction and processing</title>", "<p>Using the specific interface for each specific computer system (OMI-AP and eCAP), each region will extract the study data and transfer them to a central database where they will be homogenized and processed into a specially-constructed final database.</p>", "<title>Confidentiality of data</title>", "<p>Data confidentiality will be assured during and subsequent to the extraction using encryption and dissociation of the individual's identification, health-care professionals, and of PHCTs.</p>", "<title>Ethical aspects</title>", "<p>The study has been favourably evaluated by the Research Committee of the IDIAP Jordi Gol.</p>", "<title>Quality control</title>", "<p>Quality controls will be carried out in the 2 extractions of data comparing agreement between the central database and the original data of some centres as for: number of patient, visits and diagnoses as well as data of prevalence of certain diseases. It will also be checked if the population data is appropriate by comparing it with data from the census and other official sources. In the case of items related to clinical activities (for example blood pressure measurement), any register with missing values will be considered as not performed.</p>", "<title>Statistical analyses</title>", "<p>For the first two objectives all the analyses will be performed on a global level and each region separately. Cluster level analysis (computerized PHCT and health-care professional) will be performed using standard parametric or non-parametric analytical methods, with cluster means or proportions as observations [##REF##10982317##37##,##UREF##9##42##]. For the adjusted analysis we will perform regression analysis of individual level data using methods for clustered data, multilevel, hierarchical or random effects models [##REF##10982317##37##,##UREF##10##43##]. This methodology facilitates simultaneous analyses of the effects of individual and cluster variables in order to evaluate whether the relationship between variables at the individual level varies in accordance with the characteristics of the cluster and whether the individual and cluster variability can be attributed to factors of the cluster and/or to the factors related to the individual. Analysis will be performed at 3 levels: PHCT, health-care professional and individual subjects.</p>", "<p>Multilevel linear (for continuous outcome variables) or logistic (for dichotomous outcome variables) regression analyses will be performed for the first two objectives. The analyses will be adjusted for explanatory factors at individual, health-care professional and PHCT level and for other variables that are clinically relevant. We will also adjust for possible confounders checking whether they will affect the outcomes. A nested or hierarchical structure will be considered: the attended individuals are nested within the health-care professional and the health-care professional within PHCT. The PHCT and the health-care professional will be considered random effects.</p>", "<p>Multilevel Poisson regression models will be used to evaluate the number of preventive activities recorded for each individual.</p>", "<p>For the third objective, the ICCs, their standard errors and the components of variance [##REF##15485730##36##,##REF##10982317##37##,##UREF##9##42##,##UREF##10##43##] for the different outcome measures will be calculated using methods that enable adjustment for covariables, together with random effect methods (health-care professionals and PHCT). Initially, we will estimate the coefficients of variance and ICCs using the outcome as the dependent variable and the health-care professional/PHCT as random effects. Subsequently, we will perform an adjusted analysis for the characteristics of the individual, the health-care professional and the PHCT.</p>", "<p>We will study interactions and collinearity. The collinearity of the maximal models will be evaluated using the criteria proposed by Belsley [##UREF##11##44##]. We will set the significance level at 1% (two-tailed).</p>", "<p>Analysis will be carried out using the SPSS statistical package for Windows, version 15 (SPSS Inc., Chicago, IL) and the SAS package 9.1.3 for Windows (SAS Institute Inc., Cary, NC, USA).</p>" ]

[]

[ "<title>Discussion</title>", "<p>The project results will provide very useful information on the degree of registration and performance of the 7 preventive services selected. These 7 activities have been widely implemented and have a high impact on the health of our population [##REF##115569##2##, ####REF##6827744##3##, ##UREF##1##4##, ##UREF##2##5##, ##UREF##3##6####3##6##]. The large population database of the study, with more than 850,000 people recorded, will increase considerably the external validity of the results and will reflect usual clinical practice at the Primary Care settings in the Spanish NHS.</p>", "<p>This fact may help to overcome the current limitations and to investigate CPV and the factors that have an impact on the implementation of preventive services.</p>", "<p>Certainly the source of information used and the ECRs may have limitations. The heterogeneity of current computer systems may complicate the extraction of outcomes and of variables. However, the design of the data extraction interfaces and data transfer will minimize the problems. The lack of exhaustivity in data recording may introduce some distortions in the information. Nevertheless, stability of computer systems at the centres will help to minimize this problem and the quality control measures will help to identify the magnitude of the problem.</p>", "<p>Studies based on cluster sampling allow important reduction in costs and greater easiness in administration but require knowledge of the ICCs. For this study ICCs has been estimated from studies conducted in other countries. In our study we will provide information concerning the magnitude of the ICC, evaluate factors that influence their magnitude, estimate within-cluster and between-cluster components of variance for the outcome measures under consideration and determine the sample size; all of these factors will aid the design and statistical analysis of future cluster-based studies set in PC.</p>", "<p>In conclusion, the present study will provide new knowledge in:</p>", "<p>- The identification of factors related to the variability in the implementation of preventive services, which will facilitate the design of strategies for improvement in the planning and administration of services in PC.</p>", "<p>- The assessment of the evolution of the detected health problems will improve the degree of existent evidence taking into account the results of screening and the control of those problems. This will be very useful for the evaluation of the PAPPS and other studies based on this source of information.</p>", "<p>- The estimation of the ICC will be of considerable help in the design, calculation of sample size and analysis of future studies based on the randomisation of clusters in our environment.</p>", "<p>- Finally, although not being an aim of the project, information will be obtained on the characteristics of the different computer systems and this information will generate recommendations for the standardization of systems for future applications.</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Preventive activities carried out in primary care have important variability that makes necessary to know which factors have an impact in order to establish future strategies for improvement. The present study has three objectives: 1) To describe the variability in the implementation of 7 preventive services (screening for smoking status, alcohol abuse, hypertension, hypercholesterolemia, obesity, influenza and tetanus immunization) and to determine their related factors; 2) To describe the degree of control of 5 identified health problems (smoking, alcohol abuse, hypertension, hypercholesterolemia and obesity); 3) To calculate intraclass correlation coefficients.</p>", "<title>Design</title>", "<p>Multi-centered cross-sectional study of a randomised sample of primary health care teams from 3 regions of Spain designed to analyse variability and related factors of 7 selected preventive services in years 2006 and 2007. At the end of 2008, we will perform a cross-sectional study of a cohort of patients attended in 2006 or 2007 to asses the degree of control of 5 identified health problems. All subjects older than16 years assigned to a randomised sample of 22 computerized primary health care teams and attended during the study period are included in each region providing a sample with more than 850.000 subjects. The main outcome measures will be implementation of 7 preventive services and control of 5 identified health problems. Furthermore, there will be 3 levels of data collection: 1) Patient level (age, gender, morbidity, preventive services, attendance); 2) Health-care professional level (professional characteristics, years working at the team, workload); 3) Team level (characteristics, electronic clinical record system). Data will be transferred from electronic clinical records to a central database with prior encryption and dissociation of subject, professional and team identity. Global and regional analysis will be performed including standard analysis for primary health care teams and health-care professional level. Linear and logistic regression multilevel analysis adjusted for individual and cluster variables will also be performed. Variability in the number of preventive services implemented will be calculated with Poisson multilevel models. Team and health-care professional will be considered random effects. Intraclass correlation coefficients, standard error and variance components for the different outcome measures will be calculated.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>BB and CP are the principal investigators responsible for the conception of the project and drafting the manuscript. TR was in charge of the design of sample calculation and statistical analyses. MPA, JM, MR, CV, MI, RM, SJ, JMP, and JA contributed to the description of the background, general design and definition of the different study variables and their adaptation to the different computerized clinical records systems.</p>", "<p>All authors have read and approved the final manuscript</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2458/8/281/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>This study was carried out with financial help from the Network of Preventive Activities and Health Promotion in Primary Care <italic>[Red de Actividades Preventivas y Promoción de la Salud en Atención Primaria; redIAPP] </italic>granted by the Carlos III Health Institute <italic>[Instituto de Salud Carlos III] </italic>(RD06/0018) and from another project grant (PI061737) in 2006 also from the Carlos III Health Institute. Financial help for translation was provided by the IDIAP Jordi Gol and editorial assistance by Josep Vidal Alaball.</p>" ]

[]

[]

[ "<disp-formula>design effect = 1 + (n-1)*ICC</disp-formula>" ]

[]

[]

[]

[]

[]

[]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Using the Internet to seek health information is becoming more common in Europe [##UREF##0##1##] as well as in the USA [##UREF##1##2##]. In France little data is available on the proportion of people who have ever used the Internet to seek health information [##REF##16902386##3##] but this proportion seems to be increasing rapidly (from 15% in 2002 [##UREF##0##1##] to 37% in 2005 [##UREF##2##4##]).</p>", "<p>People use the Internet to seek health information because of its advantages. The Internet is widely available (home, work, libraries), convenient (24 h a day) and anonymous. A recent review highlighted the main reasons of using the Internet to seek health information: to gather additional information after a consultation, to access more complex information about a symptom, a disease or a treatment, to look for information about healthy lifestyles or healthcare services, to participate in an online support group and to be aware of other treatment alternatives [##REF##16406474##5##]. Specific surveys have been carried out in samples of parents looking for health information. Many studies asked parents attending a paediatric hospital whether and why they used the Internet to seek heath information. A high proportion of parents attending outpatient departments seek health information online (from 53% to 64% in the most recent articles) [##REF##17270540##6##, ####REF##16925539##7##, ##REF##12677566##8####12677566##8##]. This proportion is even higher for parents of children with a chronic disease or condition (from 58% to 89%) [##REF##17065946##9##, ####REF##16417861##10##, ##REF##11875135##11####11875135##11##]. The authors highlighted that health professionals should advise a few selected websites to parents. Population-based surveys pointed out that mothers are high information seekers [##UREF##3##12##], especially during pregnancy and during the first few years following delivery look to find parenting advice and online clinical health information [##UREF##3##12##, ####REF##15658965##13##, ##REF##15111273##14####15111273##14##].</p>", "<p>The widespread utilisation of the Internet raises some questions about its impact on health behaviour, health services utilisation and finally on health outcomes. Although some characteristics of Internet users who seek health information have been well identified [##REF##16902386##3##,##REF##15312915##15##, ####REF##12746364##16##, ##REF##16129659##17####16129659##17##], no sufficient data is available to answer the above questions, in particular on the relation between seeking health information on the Internet and health care utilisation.</p>", "<p>Some observers have suggested that use of the Internet might actually decrease the cost of primary care services in systems with universal health care [##REF##16876378##18##]. In that case, one might expect a negative relationship between use of the Internet and primary care utilisation. Others might speculate that Internet usage represents just another channel for activated, information-seeking behaviour, in which case the prediction might be for a positive relationship with primary care utilisation. Results from the studies about the impact of the Internet on health care utilisation are heterogeneous, showing a positive relationship between Internet use and service utilisation [##REF##16325137##19##], a negative relationship [##REF##16901253##20##,##REF##11468503##21##] or no relationship at all [##REF##12509660##22##,##REF##11468502##23##]. All these studies but one [##REF##11468503##21##] were carried out within an adult population and none have been conducted elsewhere than in the USA.</p>", "<p>Since the health of young children is of particular concern to parents, we assume that parental use of the Internet to seek health information would be related to primary care utilisation for their children. As we previously stated, results from studies available in the literature were heterogeneous [##REF##16325137##19##, ####REF##16901253##20##, ##REF##11468503##21##, ##REF##12509660##22##, ##REF##11468502##23####11468502##23##]. Thus, we did not initially presume a positive or negative association.</p>", "<p>To test this association, we designed a cross-sectional study in a population of parents of pre-school children. In this paper, we examined the relationship between parental use of the Internet to seek health information and self-reported primary care consultation frequency for their children.</p>" ]

[ "<title>Methods</title>", "<title>Study design, study population and sample size</title>", "<p>We designed a cross-sectional survey carried out within a population of parents of pre-school children in the department of Vienne, France. In France, even if pre-school attendance is not compulsory, almost 100% of children aged 3 to 6 years attend pre-school (<italic>\"écoles maternelles\"</italic>). We defined 7 pre-school strata according to their private or public status and to their rural, semi-urban, urban or ZEP location [<italic>a ZEP school (Zone d'Education Prioritaire) is a school located in an underprivileged area. It benefits from additional resources to cope with academic and social problems. No ZEP private school exists in the Vienne area</italic>]. We selected certain schools in each of the 7 strata to ensure the representativeness of our sample according to the characteristics of the schools.</p>", "<p>To show a difference of at least one consultation in the last 12 months between those who use the Internet to seek health information (seekers) and those who do not (non-seekers), we needed 750 questionnaires (250 of seekers and 500 of non-seekers), expecting a proportion of one seeker for two non-seekers [##UREF##2##4##], to meet our objective (two-tailed hypothesis, 80% power, 5% alpha). Expecting a response rate of 35%, we sent out more than 2100 questionnaires. To meet the required sample size we selected 35 schools which represented a population of 2 197 children.</p>", "<title>Data collection</title>", "<p>In June 2007, with the help of pre-school principals, the parents of pre-school children were given a letter including an anonymous questionnaire, an explanatory note, and an envelope to return the questionnaire. Parents were asked to answer for their youngest schoolgoing child.</p>", "<p>The data collected was: parental characteristics, characteristics of their child, primary care consultation frequency for their child, and health information sources and methods used to seek health information on the Internet. The parental characteristics noted included age, household occupation according to the most advantaged occupation of either of the parents (disadvantaged (workers and unemployed), moderately advantaged (self-employed and employees) or advantaged (managers and executives)), parental education level according to the highest education level accomplished by either of the parents (elementary-secondary, high school diploma, lower tertiary or higher tertiary), annual family income (&lt;14 000 €, from 14 000 to 19 999 €, from 20 000 to 29 999 €, from 30 000 to 39 999 € and ≥ 40 000 €), single parent family, place of residence (urban, rural), stress level assessed with a 10-unit visual analogue rating scale (10 indicative of higher stress) and Internet access (home, work only, none). The characteristics of the child collected were school attended, age, birth order, gender, medical conditions if any (preterm infant, hospitalisation after birth, asthma, wearing of glasses, auditory disorders, allergy, behavioural disorder, surgical intervention and others), long-term use of medication if any, whether under homeopathic treatment, frequency of administration of over the counter drugs to their child without medical advice and whether the parents had taken advice from a pharmacist for their child within the last 12 months. We asked parents to self-report the number of primary care consultations for their child within the last 12 months (general practitioner, paediatrician or accidents and emergency department). For consultations and child's medical conditions, we asked parents to refer to their child's health record booklet if necessary. Parents were also asked about other health information sources that they had already used amongst the Internet, medical books, medical dictionaries, television, press, relatives working in the medical sector, and other relatives.</p>", "<p>The study as a whole had been previously approved by the consultative committee on the processing of information in medical research of CNIL, the French national commission on individual privacy (approval AR071193).</p>", "<title>Statistical analysis</title>", "<p>The dependant variable was the number of primary care consultations. The explanatory variables were child's age, child's birth order, child's medical condition and treatment, parental age, socio-economic position of the family, single parent family, parental stress level and health information sources. The main variable of interest was parental use of the Internet to seek health information.</p>", "<p>We assumed that the number of primary care consultations followed a negative binomial distribution – an extension of the Poisson distribution in case of over-dispersion [##REF##7501743##24##,##REF##12873651##25##]. We therefore used a negative binomial regression model to explain the variability of the number of consultations. We took the design effect (cluster effect) into account to avoid errors in the estimation of the parameters of the model [##REF##15209195##26##].</p>", "<p>To perform our analysis, we used the <italic>negbin </italic>function of STATA [##UREF##4##27##] along with the <italic>svy </italic>option to take the design effect into account. The variable \"parental use of the Internet to seek health information\" was forced into the model. Other variables were included in the initial regression model if they were associated with the number of consultations with a p-value &lt; 0.20 (using bivariate negative binomial regression). We then performed a multivariate analysis. From the initial regression model, variables were selected using a stepwise descending process. We tested the first-order interactions in the final model. Association between the number of consultations and variables of interest are rate ratios.</p>" ]

[ "<title>Results</title>", "<title>Response rate and characteristics of the study population</title>", "<p>Of the 2 197 questionnaires distributed, 1068 questionnaires were returned (49%). Characteristics of the population are presented in Table ##TAB##0##1##. The mean number of primary care consultations for a pre-school child within the last 12 months was 5.9 ± 4.6. Distribution of the self-reported number of primary care consultations is shown in Figure ##FIG##0##1##, which confirms the assumption of a negative binomial distribution. Data on the number of consultations were missing for 39 questionnaires (4%).</p>", "<p>The Internet was the most used health information source with 556 families (52%) who at least once had used it to seek health information. Relatives working in the medical sector and television were the second and third most common health information sources, with 518 (49%) and 397 (37%) families respectively having at least once used these sources (Figure ##FIG##1##2##).</p>", "<title>Use of the Internet to seek health information and number of primary care consultations for their child</title>", "<p>Mean numbers of consultations according to population characteristics together with results of the bivariate analysis are presented in Table ##TAB##1##2##. The multivariate analysis has been carried out on the 886 questionnaires (83%) for which no data was missing. Four variables were dropped due to the stepwise regression analysis (parental age, single parent family, allergy and television as health information source). None of the first-order interactions between explanatory variables of the final model were significant. Results of the multivariate analysis are presented in Table ##TAB##2##3##. No association was found between use of the Internet to seek health information and the number of consultations within the last 12 months (adjusted rate ratio 0.97; 95% CI 0.86 to 1.09). Seeking health information from relatives (whether they were from the medical sector or not) was associated with a slight increase in the number of consultations. The main variables related to the number of primary care consultations were child's age and medical condition. Number of consultations within the last 12 months decreased with child's age (with a decrease of 16%, 23% and 33% for children aged 4, 5 and 6 years respectively compared to children aged 2 or 3 years). Most of the children's medical conditions were positively related to the number of consultations. Some parental characteristics were related to a lower number of consultations: moderately advantaged occupation and annual income from 20 000€ and over. Parental stress was related to a higher number of consultations with a 3% increase in the number of consultations for every one-unit increase in the visual analogue rating scale.</p>", "<p>We also fitted a model without income on the 974 families (91%) for whom no data other than income was missing (data not shown). Results were very similar though associations of auditory disorders and homeopathic treatment with the number of consultations were not significant anymore.</p>" ]

[ "<title>Discussion</title>", "<p>We did not find any relationship between parental use of the Internet to seek health information and the number of self-reported consultations for their child. This finding runs counter to our initial assumption that parental use of the Internet to seek health information would be related to primary care utilisation, either in a positive or in a negative way.</p>", "<p>To our knowledge, this is the first study carried out elsewhere than in the USA to assess this relationship. Our results are consistent with an interventional study on mental health services utilisation, which did not find any significant difference in the number of mental health visits between a group that had Web site access and the control group [##REF##12509660##22##]. A quasi-experimental study carried out by Wagner et al. found a null association for parents [##REF##11468502##23##] and a negative association for children [##REF##11468503##21##]. Data from this study are not fully comparable with our study because the intervention was complex and not entirely Internet-based (self-care books, telephone advice nurses and computers). Within a population of Internet users, Eastin and Guinsler found an interaction between anxiety and Internet use to seek health information [##REF##16901253##20##]. Anxious individuals who used the Internet to seek health information had fewer consultations than anxious non-users, whereas such a difference was not found for less anxious individuals. Our findings do not corroborate this interaction, since we did not find any interaction between stress levels and parental use of the Internet to seek health information. Another American study conducted in 1999–2000 found a positive association with an increase of 1.6 consultations for women using computer-based resources [##REF##16325137##19##]. Differences in time period (1999–2000), geographical area (Baltimore metropolitan area, USA) and potential selection bias in both studies are likely to explain the differences between these results and our findings. Finally, findings from a cross-sectional study in seven European countries that investigated patterns of health-related Internet use and its consequences support our results. Only 6% of the sample claimed that they had made, cancelled or changed a doctor's appointment based on health related Internet activity [##REF##17425798##28##]. Even if the number of consultations was not collected in this study, only 6% of the sample claimed that they have made, cancelled or changed a doctor's appointment based on health related Internet activity.</p>", "<p>The association that we found between primary care utilisation and the child's age [##REF##1729585##29##, ####REF##3931044##30##, ##REF##11076248##31####11076248##31##], child's medical condition [##REF##1729585##29##], familial socio-economic position [##REF##10066207##32##] and psychological factors [##REF##8366679##33##] are consistent with previous findings. However, in most studies performed in the USA [##REF##1729585##29##,##REF##3931044##30##,##REF##8545220##34##], lower socio-economic position was associated with less frequent primary care utilisation, which is contrary to our results. Explanations are likely to come from the differences between the French and the American health insurance coverage for children. In France, health insurance coverage is nearly universal [##REF##12511380##35##]. In the USA, most of the studies have been carried out in the 1990's or before, when the problem of uninsured children was raised [##REF##1729585##29##,##REF##3931044##30##,##REF##8545220##34##]. At that time, children from the poorest families were more likely to be uninsured resulting in a lower number of primary care consultations. A few new factors associated with the number of primary care consultations have been identified in our study. Taking advice from a pharmacist, using relatives or friends as a health information source or using homeopathy for one's child could be explained by increased parental consciousness of health issues. These findings might reflect the \"familial context\" mentioned by Cardol [##REF##15772114##36##] which would explain about 20% of the variability of the number of consultations in primary care.</p>", "<p>The first limitation of our study was the overall response rate of 49%. However, details of response rates of each school gave us information to identify the bias due to non-responses. We found that response rates were lower in schools with a higher proportion of families of low socio-economic position and with a higher proportion of non-French speaking families. We therefore probably over-estimated the proportion of families who used the Internet to seek health information, and possibly under-estimated the mean number of consultations. The second limitation was that data on race/ethnicity was not asked in the questionnaire. According to the French population statistics, less than five percent of the inhabitants of the department of Vienne are non native French. In this context, race/ethnicity is not so important even if it is well established in the American context that disparities in Internet use for health information exist according to race/ethnicity [##REF##12885689##37##,##REF##14581904##38##].</p>", "<p>Another limitation was that the number of consultations was self-reported by the parents. Many studies have shown a tendency for underestimation when people were asked to report the frequency of their health care utilisation. Since we found a mean of 5.9 consultations per child within the last 12 months, which is consistent with the data of health care utilisation from the provider [##UREF##5##39##], the bias may be small. Missing data for parental income was another limitation, with 119 (11%) parents who did not report their annual family income. This omission in reporting annual family incomes is information probably not missing at random because it is more likely to occur when the income level is relatively high [##REF##16980149##40##].</p>" ]

[ "<title>Conclusion</title>", "<p>Even if our study had some limitations, we demonstrated that there was no relationship between parental use of the Internet to seek health information and primary care consultation of their children. The Internet seems to be used as a supplement to health services for some rather than as a replacement. Individuals are increasingly involved in the management of their own health and using the Internet to seek health information is one way to be involved actively. Some authors suggested that this would lead to saving in health costs [##REF##16876378##18##]. According to our findings, this may not be true from a short-term perspective. In our opinion, what is more likely to occur is an improvement in the health of those who use the Internet to seek health information, which, from a long-term perspective, would eventually lead to saving in health costs.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Using the Internet to seek health information is becoming more common. Its consequences on health care utilisation are hardly known in the general population, in particular among children whose parents seek health information on the Internet. Our objective was to investigate the relationship between parental use of the Internet to seek health information and primary care utilisation for their child.</p>", "<title>Methods</title>", "<p>This cross-sectional survey has been carried out in a population of parents of pre-school children in France. The main outcome measure was the self-reported number of primary care consultations for the child, according to parental use of the Internet to seek health information, adjusted for the characteristics of the parents and their child respectively, and parental use of other health information sources.</p>", "<title>Results</title>", "<p>A total of 1 068 out of 2 197 questionnaires were returned (response rate of 49%). No association was found between parental use of the Internet to seek health information and the number of consultations within the last 12 months for their child. Variables related to the number of primary care consultations were characteristics of the child (age, medical conditions, homeopathic treatment), parental characteristics (occupation, income, stress level) and consultation of other health information sources (advice from pharmacist, relatives).</p>", "<title>Conclusion</title>", "<p>We did not find any relationship between parental use of the Internet to seek health information and primary care utilisation for children. The Internet seems to be used as a supplement to health services rather than as a replacement.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>GB designed the study, collected the data and had primary responsibility for data analysis and manuscript preparation. VM helped with the implementation of the study, validated the methodology and contributed to data analysis and manuscript preparation. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2458/8/300/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>The authors would like to thank the school principals and parents who participated in the study, Dr Chantal Simmat who helped us for the study implementation in schools and Pr Pierre Ingrand for his support. We also wish to thank Dr Shivani Aguilera for the English revision.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Distribution of the self-reported number of primary care consultations for the child within the last 12 months.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Frequency (and 95% confidence intervals) of the different health information sources used by the 1 068 parents.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Characteristics of the 1068 pre-school children and their parents</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\" colspan=\"4\">Characteristics</td></tr></thead><tbody><tr><td align=\"left\">Characteristics of the children</td><td/><td align=\"right\">N</td><td align=\"right\">%</td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\">Age (n = 1067)</td><td align=\"left\">- 2 or 3 years old</td><td align=\"right\">227</td><td align=\"right\">21%</td></tr><tr><td/><td align=\"left\">- 4 years old</td><td align=\"right\">335</td><td align=\"right\">31%</td></tr><tr><td/><td align=\"left\">- 5 years old</td><td align=\"right\">328</td><td align=\"right\">31%</td></tr><tr><td/><td align=\"left\">- 6 years old</td><td align=\"right\">177</td><td align=\"right\">17%</td></tr><tr><td align=\"left\">Girls (n = 1068)</td><td/><td align=\"right\">537</td><td align=\"right\">50%</td></tr><tr><td align=\"left\">First in birth order (n = 1039)</td><td/><td align=\"right\">498</td><td align=\"right\">48%</td></tr><tr><td align=\"left\">Medical conditions (n = 1068)</td><td align=\"left\">- Preterm infant</td><td align=\"right\">96</td><td align=\"right\">9%</td></tr><tr><td/><td align=\"left\">- Hospitalisation after birth</td><td align=\"right\">81</td><td align=\"right\">8%</td></tr><tr><td/><td align=\"left\">- Asthma</td><td align=\"right\">156</td><td align=\"right\">15%</td></tr><tr><td/><td align=\"left\">- Wearing of glasses</td><td align=\"right\">146</td><td align=\"right\">14%</td></tr><tr><td/><td align=\"left\">- Auditory disorders</td><td align=\"right\">63</td><td align=\"right\">6%</td></tr><tr><td/><td align=\"left\">- Allergy</td><td align=\"right\">214</td><td align=\"right\">20%</td></tr><tr><td/><td align=\"left\">- Behavioural disorder</td><td align=\"right\">44</td><td align=\"right\">4%</td></tr><tr><td/><td align=\"left\">- At least one surgical intervention</td><td align=\"right\">225</td><td align=\"right\">21%</td></tr><tr><td/><td align=\"left\">- Other medical conditions</td><td align=\"right\">70</td><td align=\"right\">7%</td></tr><tr><td align=\"left\">Long term medication use (n = 1068)</td><td/><td align=\"right\">73</td><td align=\"right\">7%</td></tr><tr><td align=\"left\">Frequency of administrating over the counter drugs without medical advice (n = 1062)</td><td align=\"left\">- Never</td><td align=\"right\">168</td><td align=\"right\">16%</td></tr><tr><td/><td align=\"left\">- Rarely</td><td align=\"right\">311</td><td align=\"right\">29%</td></tr><tr><td/><td align=\"left\">- Often</td><td align=\"right\">509</td><td align=\"right\">48%</td></tr><tr><td/><td align=\"left\">- Almost every time my child is ill</td><td align=\"right\">74</td><td align=\"right\">7%</td></tr><tr><td align=\"left\">Under homeopathic treatment (n = 1043)</td><td/><td align=\"right\">549</td><td align=\"right\">53%</td></tr><tr><td align=\"left\">Advice from a pharmacist within the last 12 months (n = 1051)</td><td/><td align=\"right\">445</td><td align=\"right\">42%</td></tr><tr><td/><td/><td align=\"right\">Mean</td><td align=\"right\">SD</td></tr><tr><td align=\"left\">N. of primary care consultations within the last 12 months (n = 1029)</td><td/><td align=\"right\">5.9</td><td align=\"right\">4.6</td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\">Parental characteristics</td><td/><td align=\"right\">N</td><td align=\"right\">%</td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\">Single parent family (n = 1066)</td><td/><td align=\"right\">136</td><td align=\"right\">13%</td></tr><tr><td align=\"left\">Rural place of residence (n = 1064)</td><td/><td align=\"right\">217</td><td align=\"right\">20%</td></tr><tr><td align=\"left\">Occupation (n = 1056)</td><td align=\"left\">- Disadvantaged (workers, unemployed)</td><td align=\"right\">151</td><td align=\"right\">14%</td></tr><tr><td/><td align=\"left\">- Moderately advantaged (self-employed, employees)</td><td align=\"right\">563</td><td align=\"right\">53%</td></tr><tr><td/><td align=\"left\">- Advantaged (managers and executives)</td><td align=\"right\">342</td><td align=\"right\">33%</td></tr><tr><td align=\"left\">Education level (n = 1037)</td><td align=\"left\">- Elementary and secondary</td><td align=\"right\">285</td><td align=\"right\">27%</td></tr><tr><td/><td align=\"left\">- High school diploma (\"Baccalauréat\")</td><td align=\"right\">238</td><td align=\"right\">23%</td></tr><tr><td/><td align=\"left\">- Lower tertiary</td><td align=\"right\">235</td><td align=\"right\">23%</td></tr><tr><td/><td align=\"left\">- Higher tertiary</td><td align=\"right\">279</td><td align=\"right\">27%</td></tr><tr><td align=\"left\">Annual family income (n = 949)</td><td align=\"left\">- &lt; 14 000 €</td><td align=\"right\">168</td><td align=\"right\">18%</td></tr><tr><td/><td align=\"left\">- 14 000 to 19 999 €</td><td align=\"right\">203</td><td align=\"right\">21%</td></tr><tr><td/><td align=\"left\">- 20 000 to 29 999 €</td><td align=\"right\">286</td><td align=\"right\">30%</td></tr><tr><td/><td align=\"left\">- 30 000 to 39 999 €</td><td align=\"right\">170</td><td align=\"right\">18%</td></tr><tr><td/><td align=\"left\">- ≥ 40 000 €</td><td align=\"right\">122</td><td align=\"right\">13%</td></tr><tr><td/><td/><td align=\"center\">Mean</td><td align=\"center\">SD</td></tr><tr><td align=\"left\">Age of the respondent in year (n = 1058)</td><td/><td align=\"right\">34.0</td><td align=\"right\">5.0</td></tr><tr><td align=\"left\">Parental stress level (n = 1040)</td><td/><td align=\"right\">4.9</td><td align=\"right\">2.3</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Relation between mean number of primary care consultations for the child and population characteristics – bivariate negative regression analysis.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Explanatory variables (n = 1029 unless otherwise indicated)</td><td/><td/><td align=\"center\">Mean</td><td align=\"center\">SD</td><td align=\"center\">P</td></tr></thead><tbody><tr><td align=\"left\">Health information sources</td><td/><td/><td/><td/><td/></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\">Medical dictionary</td><td align=\"left\">No</td><td align=\"center\">n = 826</td><td align=\"center\">5.8</td><td align=\"center\">4.2</td><td align=\"center\">0.38</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 203</td><td align=\"center\">6.3</td><td align=\"center\">5.8</td><td/></tr><tr><td align=\"left\">Internet</td><td align=\"left\">No</td><td align=\"center\">n = 490</td><td align=\"center\">5.9</td><td align=\"center\">4.7</td><td align=\"center\">0.92</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 539</td><td align=\"center\">5.9</td><td align=\"center\">4.5</td><td/></tr><tr><td align=\"left\">Press</td><td align=\"left\">No</td><td align=\"center\">n = 753</td><td align=\"center\">5.8</td><td align=\"center\">4.3</td><td align=\"center\">0.20</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 276</td><td align=\"center\">6.3</td><td align=\"center\">5.2</td><td/></tr><tr><td align=\"left\">Medical books</td><td align=\"left\">No</td><td align=\"center\">n = 809</td><td align=\"center\">5.8</td><td align=\"center\">4.5</td><td align=\"center\">0.21</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 220</td><td align=\"center\">6.3</td><td align=\"center\">4.9</td><td/></tr><tr><td align=\"left\">Television</td><td align=\"left\">No</td><td align=\"center\">n = 646</td><td align=\"center\">5.7</td><td align=\"center\">4.2</td><td align=\"center\">0.05</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 383</td><td align=\"center\">6.2</td><td align=\"center\">5.1</td><td/></tr><tr><td align=\"left\">Relatives working in the medical sector</td><td align=\"left\">No</td><td align=\"center\">n = 528</td><td align=\"center\">5.6</td><td align=\"center\">4.2</td><td align=\"center\">0.004</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 501</td><td align=\"center\">6.3</td><td align=\"center\">5.0</td><td/></tr><tr><td align=\"left\">Other relatives</td><td align=\"left\">No</td><td align=\"center\">n = 779</td><td align=\"center\">5.7</td><td align=\"center\">4.4</td><td align=\"center\">0.006</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 250</td><td align=\"center\">6.6</td><td align=\"center\">5.0</td><td/></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\">Characteristics of the child</td><td/><td/><td/><td/><td/></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\">Age (n = 1028)</td><td align=\"left\">2–3 years old</td><td align=\"center\">n = 221</td><td align=\"center\">7.3</td><td align=\"center\">5.6</td><td align=\"center\">0.0001</td></tr><tr><td/><td align=\"left\">4 years</td><td align=\"center\">n = 319</td><td align=\"center\">6.0</td><td align=\"center\">3.8</td><td/></tr><tr><td/><td align=\"left\">5 years</td><td align=\"center\">n = 317</td><td align=\"center\">5.6</td><td align=\"center\">5.0</td><td/></tr><tr><td/><td align=\"left\">6 years</td><td align=\"center\">n = 171</td><td align=\"center\">4.7</td><td align=\"center\">3.1</td><td/></tr><tr><td align=\"left\">First in birth order (n = 1002)</td><td align=\"left\">No</td><td align=\"center\">n = 523</td><td align=\"center\">5.8</td><td align=\"center\">4.8</td><td align=\"center\">0.41</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 479</td><td align=\"center\">6.1</td><td align=\"center\">4.4</td><td/></tr><tr><td align=\"left\">Child gender</td><td align=\"left\">Boy</td><td align=\"center\">n = 514</td><td align=\"center\">6.0</td><td align=\"center\">4.9</td><td align=\"center\">0.71</td></tr><tr><td/><td align=\"left\">Girl</td><td align=\"center\">n = 515</td><td align=\"center\">5.9</td><td align=\"center\">4.2</td><td/></tr><tr><td align=\"left\">Medical conditions</td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">Preterm infant (n = 1025)</td><td align=\"left\">No</td><td align=\"center\">n = 933</td><td align=\"center\">5.9</td><td align=\"center\">4.5</td><td align=\"center\">0.21</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 92</td><td align=\"center\">6.6</td><td align=\"center\">5.4</td><td/></tr><tr><td align=\"left\">Hospitalisation after birth (n = 1023)</td><td align=\"left\">No</td><td align=\"center\">n = 948</td><td align=\"center\">5.9</td><td align=\"center\">4.5</td><td align=\"center\">0.34</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 75</td><td align=\"center\">6.6</td><td align=\"center\">5.7</td><td/></tr><tr><td align=\"left\">Asthma</td><td align=\"left\">No</td><td align=\"center\">n = 879</td><td align=\"center\">5.6</td><td align=\"center\">4.4</td><td align=\"center\">&lt;0.0001</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 150</td><td align=\"center\">7.9</td><td align=\"center\">5.1</td><td/></tr><tr><td align=\"left\">Wearing of glasses</td><td align=\"left\">No</td><td align=\"center\">n = 888</td><td align=\"center\">5.8</td><td align=\"center\">4.3</td><td align=\"center\">0.05</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 141</td><td align=\"center\">6.7</td><td align=\"center\">6.2</td><td/></tr><tr><td align=\"left\">Auditory disorders</td><td align=\"left\">No</td><td align=\"center\">n = 967</td><td align=\"center\">5.8</td><td align=\"center\">4.6</td><td align=\"center\">0.006</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 62</td><td align=\"center\">7.4</td><td align=\"center\">4.6</td><td/></tr><tr><td align=\"left\">Allergy</td><td align=\"left\">No</td><td align=\"center\">n = 825</td><td align=\"center\">5.6</td><td align=\"center\">4.6</td><td align=\"center\">&lt;0.0001</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 404</td><td align=\"center\">7.3</td><td align=\"center\">4.4</td><td/></tr><tr><td align=\"left\">Behavioural disorder</td><td align=\"left\">No</td><td align=\"center\">n = 987</td><td align=\"center\">5.9</td><td align=\"center\">4.6</td><td align=\"center\">0.88</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 42</td><td align=\"center\">5.8</td><td align=\"center\">4.2</td><td/></tr><tr><td align=\"left\">At least one surgical intervention</td><td align=\"left\">No</td><td align=\"center\">n = 817</td><td align=\"center\">5.7</td><td align=\"center\">4.6</td><td align=\"center\">0.0004</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 212</td><td align=\"center\">6.8</td><td align=\"center\">4.4</td><td/></tr><tr><td align=\"left\">Other medical conditions</td><td align=\"left\">No</td><td align=\"center\">n = 960</td><td align=\"center\">5.9</td><td align=\"center\">4.5</td><td align=\"center\">0.10</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 69</td><td align=\"center\">6.7</td><td align=\"center\">5.0</td><td/></tr><tr><td align=\"left\">Long term medication use</td><td align=\"left\">No</td><td align=\"center\">n = 961</td><td align=\"center\">5.7</td><td align=\"center\">4.5</td><td align=\"center\">&lt;0.0001</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 68</td><td align=\"center\">8.9</td><td align=\"center\">5.2</td><td/></tr><tr><td align=\"left\">Frequency of administrating over the counter drugs without medical advice (n = 1025)</td><td align=\"left\">Never</td><td align=\"center\">n = 158</td><td align=\"center\">5.8</td><td align=\"center\">5.0</td><td align=\"center\">0.63</td></tr><tr><td/><td align=\"left\">Rarely</td><td align=\"center\">n = 300</td><td align=\"center\">5.8</td><td align=\"center\">4.3</td><td/></tr><tr><td/><td align=\"left\">Often</td><td align=\"center\">n = 494</td><td align=\"center\">6.1</td><td align=\"center\">4.7</td><td/></tr><tr><td/><td align=\"left\">Almost every time my child is ill</td><td align=\"center\">n = 73</td><td align=\"center\">5.5</td><td align=\"center\">4.4</td><td/></tr><tr><td align=\"left\">Under homeopathic treatment (n = 1008)</td><td align=\"left\">No</td><td align=\"center\">n = 474</td><td align=\"center\">5.6</td><td align=\"center\">4.3</td><td align=\"center\">0.08</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 534</td><td align=\"center\">6.2</td><td align=\"center\">4.8</td><td/></tr><tr><td align=\"left\">Advice from a pharmacist within the last 12 months (n = 1027)</td><td align=\"left\">No</td><td align=\"center\">n = 592</td><td align=\"center\">5.4</td><td align=\"center\">4.2</td><td align=\"center\">0.0003</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 435</td><td align=\"center\">6.6</td><td align=\"center\">5.0</td><td/></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\">Parental characteristics</td><td/><td/><td/><td/><td/></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\">Single parent family (n = 1028)</td><td align=\"left\">No</td><td align=\"center\">n = 899</td><td align=\"center\">5.8</td><td align=\"center\">4.3</td><td align=\"center\">0.02</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 129</td><td align=\"center\">6.7</td><td align=\"center\">6.0</td><td/></tr><tr><td align=\"left\">Occupation (n = 1019)</td><td align=\"left\">Disadvantaged</td><td align=\"center\">n = 142</td><td align=\"center\">6.8</td><td align=\"center\">6.3</td><td align=\"center\">0.04</td></tr><tr><td/><td align=\"left\">Moderately advantaged</td><td align=\"center\">n = 542</td><td align=\"center\">5.9</td><td align=\"center\">4.6</td><td/></tr><tr><td/><td align=\"left\">Advantaged</td><td align=\"center\">n = 335</td><td align=\"center\">5.6</td><td align=\"center\">3.6</td><td/></tr><tr><td align=\"left\">Education level (n = 1002)</td><td align=\"left\">Elementary and secondary</td><td align=\"center\">n = 268</td><td align=\"center\">6.6</td><td align=\"center\">6.2</td><td align=\"center\">0.20</td></tr><tr><td/><td align=\"left\">High school diploma</td><td align=\"center\">n = 233</td><td align=\"center\">5.9</td><td align=\"center\">4.1</td><td/></tr><tr><td/><td align=\"left\">Lower tertiary</td><td align=\"center\">n = 232</td><td align=\"center\">5.6</td><td align=\"center\">3.6</td><td/></tr><tr><td/><td align=\"left\">Higher tertiary</td><td align=\"center\">n = 269</td><td align=\"center\">5.7</td><td align=\"center\">3.9</td><td/></tr><tr><td align=\"left\">Annual family income (n = 921)</td><td align=\"left\">&lt; 14 000 €</td><td align=\"center\">n = 159</td><td align=\"center\">7.2</td><td align=\"center\">7.4</td><td align=\"center\">0.03</td></tr><tr><td/><td align=\"left\">14 000 to 19 999 €</td><td align=\"center\">n = 195</td><td align=\"center\">5.8</td><td align=\"center\">3.9</td><td/></tr><tr><td/><td align=\"left\">20 000 to 29 999 €</td><td align=\"center\">n = 279</td><td align=\"center\">5.5</td><td align=\"center\">3.5</td><td/></tr><tr><td/><td align=\"left\">30 000 to 39 999 €</td><td align=\"center\">n = 167</td><td align=\"center\">5.9</td><td align=\"center\">3.9</td><td/></tr><tr><td/><td align=\"left\">≥ 40 000 €</td><td align=\"center\">n = 121</td><td align=\"center\">5.4</td><td align=\"center\">3.9</td><td/></tr><tr><td align=\"left\">Rural place of residence (n = 1025)</td><td align=\"left\">No</td><td align=\"center\">n = 815</td><td align=\"center\">5.9</td><td align=\"center\">4.7</td><td align=\"center\">0.88</td></tr><tr><td/><td align=\"left\">Yes</td><td align=\"center\">n = 210</td><td align=\"center\">5.9</td><td align=\"center\">4.3</td><td/></tr><tr><td/><td/><td/><td align=\"center\">Rate Ratio</td><td/><td align=\"center\">95% CI</td></tr><tr><td align=\"left\">Age of the respondent (n = 1020)</td><td/><td/><td align=\"center\">0.98</td><td/><td align=\"center\">[0.97 – 0.99]</td></tr><tr><td align=\"left\">Parental stress level (n = 1005)</td><td/><td/><td align=\"center\">1.03</td><td/><td align=\"center\">[1.01 – 1.06]</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Relation between number of primary care consultations for the child and population characteristics – multivariate analysis.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Explanatory variables</td><td/><td align=\"left\">Adjusted Rate Ratio*</td><td align=\"center\">95% CI</td></tr></thead><tbody><tr><td align=\"left\" colspan=\"4\">Health information sources</td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\">Internet</td><td align=\"left\">No</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Yes</td><td align=\"left\">0.97</td><td align=\"center\">[0.86–1.09]</td></tr><tr><td align=\"left\">Relatives working in the medical sector</td><td align=\"left\">No</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Yes</td><td align=\"left\">1.08</td><td align=\"center\">[1.01–1.16]</td></tr><tr><td align=\"left\">Other relatives</td><td align=\"left\">No</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Yes</td><td align=\"left\">1.12</td><td align=\"center\">[1.01–1.25]</td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\">Characteristics of the child</td><td/><td/><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\">Age</td><td align=\"left\">2–3 years old</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">4 years</td><td align=\"left\">0.84</td><td align=\"center\">[0.73–0.96]</td></tr><tr><td/><td align=\"left\">5 years</td><td align=\"left\">0.77</td><td align=\"center\">[0.65–0.90]</td></tr><tr><td/><td align=\"left\">6 years</td><td align=\"left\">0.67</td><td align=\"center\">[0.56–0.80]</td></tr><tr><td align=\"left\">Medical conditions</td><td/><td/><td/></tr><tr><td align=\"left\">Asthma</td><td align=\"left\">No</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Yes</td><td align=\"left\">1.28</td><td align=\"center\">[1.13–1.46]</td></tr><tr><td align=\"left\">Wearing of glasses</td><td align=\"left\">No</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Yes</td><td align=\"left\">1.19</td><td align=\"center\">[1.03–1.45]</td></tr><tr><td align=\"left\">Auditory disorders</td><td align=\"left\">No</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Yes</td><td align=\"left\">1.23</td><td align=\"center\">[1.03–1.46]</td></tr><tr><td align=\"left\">At least one surgical intervention</td><td align=\"left\">No</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Yes</td><td align=\"left\">1.12</td><td align=\"center\">[1.02–1.22]</td></tr><tr><td align=\"left\">Long term medication use</td><td align=\"left\">No</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Yes</td><td align=\"left\">1.26</td><td align=\"center\">[1.06–1.49]</td></tr><tr><td align=\"left\">Under homeopathic treatment</td><td align=\"left\">No</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Yes</td><td align=\"left\">1.13</td><td align=\"center\">[1.03–1.23]</td></tr><tr><td align=\"left\">Advice from a pharmacist within the last 12 months</td><td align=\"left\">No</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Yes</td><td align=\"left\">1.14</td><td align=\"center\">[1.05–1.24]</td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\">Parental characteristics</td><td/><td/><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\">Occupation</td><td align=\"left\">Disadvantaged</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Moderately advantaged</td><td align=\"left\">0.85</td><td align=\"center\">[0.75–0.96]</td></tr><tr><td/><td align=\"left\">Advantaged</td><td align=\"left\">0.87</td><td align=\"center\">[0.75–1.02]</td></tr><tr><td align=\"left\">Annual family income</td><td align=\"left\">&lt; 14 000 €</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">14 000 to 19 999 €</td><td align=\"left\">0.86</td><td align=\"center\">[0.73–1.01]</td></tr><tr><td/><td align=\"left\">20 000 to 29 999 €</td><td align=\"left\">0.80</td><td align=\"center\">[0.68–0.94]</td></tr><tr><td/><td align=\"left\">30 000 to 39 999 €</td><td align=\"left\">0.87</td><td align=\"center\">[0.76–0.99]</td></tr><tr><td/><td align=\"left\">≥ 40 000 €</td><td align=\"left\">0.80</td><td align=\"center\">[0.64–0.99]</td></tr><tr><td align=\"left\">Parental stress level</td><td/><td align=\"left\">1.03</td><td align=\"center\">[1.01–1.05]</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>* Adjusted on other health information sources (dictionary, press, books and television), characteristics of the child (birth order, gender, medical conditions – preterm infant, hospitalisation after birth, allergy, behavioral disorder – frequency of administrating over the counter drugs without medical advice) and parental characteristics (age, single parent, education level and place of residence).</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2458-8-300-1\"/>", "<graphic xlink:href=\"1471-2458-8-300-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The quality of medical care at the end of life has become of major importance in contemporary developed societies [##REF##10972435##1##, ####REF##11950744##2##, ##REF##12139768##3##, ##REF##17482035##4##, ##REF##12231124##5####12231124##5##]. In the past century there has been a significant shift in cause of death, away from acute deaths due to infectious disease towards deaths caused by chronic and degenerative illness such as cancer and cardiovascular disease [##REF##10972435##1##, ####REF##11950744##2##, ##REF##12139768##3####12139768##3##,##UREF##0##6##]. Combined with rising life expectancy and an ageing population, this epidemiological transition has resulted in an increased number of people experiencing a terminal illness phase at the end of life [##REF##10972435##1##,##UREF##0##6##].</p>", "<p>Parallel to these changes in the patterns of dying, advances in medical knowledge and technology have contributed considerably to the increase of treatment possibilities at the end of life. Physicians are now increasingly able to ensure effective treatment of pain and symptoms at the end of life, and to postpone a patient's death [##REF##10972435##1##,##REF##11950744##2##,##REF##17482035##4##,##UREF##0##6##]. However, in many cases a point is reached where those involved feel that prolonging life is no longer desirable as a certain minimal quality of life cannot always be maintained [##REF##10972435##1##,##REF##12231124##5##, ####UREF##0##6##, ##REF##10733450##7##, ##REF##14672110##8####14672110##8##]. This gives rise to decisions that possibly or certainly hasten the patient's death, i.e. end-of-life decisions (ELDs). These decisions include withholding or withdrawing potentially life-sustaining treatment, intensifying pain and/or symptom management with a possible life-shortening effect and administering drugs with the explicit intention of hastening death (i.e. physician-assisted suicide, life-ending without the patient's explicit request, and euthanasia).</p>", "<p>Past studies in different countries have revealed that these various end-of-life decisions are made in a significant proportion of deaths [##REF##14672110##8##, ####REF##1715962##9##, ##REF##8929370##10##, ##REF##9066548##11##, ##REF##9554861##12##, ##REF##11117913##13##, ##REF##12907005##14##, ##REF##12907015##15##, ##REF##12915432##16##, ##REF##15128690##17##, ##REF##16482752##18##, ##REF##17490798##19##, ##REF##17494928##20####17494928##20##], although incidence estimates vary somewhat across countries. According to the 2001 EURELD study in six European countries (Belgium, The Netherlands, Denmark, Italy, Sweden and Switzerland) the incidence of deaths preceded by an ELD ranges from 23% to 51% [##REF##12907005##14##]. In Belgium the incidence rate dropped slightly, although not statistically significantly, from 39,3% to 38,4% between 1998 and 2001 [##REF##17490798##19##]. These studies contributed to an ongoing ethical and legal debate concerning end-of-life decisions, culminating in Belgium in 2002 with the passing of the laws on palliative care, patients' rights and euthanasia (which permits euthanasia under strict conditions of prudent practice) [##UREF##1##21##, ####UREF##2##22##, ##UREF##3##23##, ##REF##14568754##24####14568754##24##].</p>", "<p>It is in this new legal context that a third ELD study in Belgium was undertaken. This study is part of the larger <bold>M</bold>onitoring quality of <bold>E</bold>nd-of-<bold>L</bold>ife <bold>C</bold>are (MELC) study in Flanders [##UREF##4##25##], and aims to obtain reliable incidence estimates of ELDs and their characteristics in Flanders for 2007, as well as to take a closer look at the decision-making process preceding ELDs and the treatment and care provided at the end of life. As a third measurement point for Flanders, one of the research aims is to permit a trend analysis of end-of-life decision making. Furthermore, the legalisation of euthanasia since the last ELD study in Flanders creates the opportunity of estimating the possible effects of the euthanasia law on the practice of euthanasia and other end-of-life practices [##REF##14568754##24##], and will shed light on the argument that legalising euthanasia will possibly lead to a slippery slope, e.g. a rise in life-ending acts without the patient's explicit request [##REF##17341228##26##,##REF##12882255##27##]. Comparison of the results of the Flemish study to the Dutch data from 2005 will put the findings in an international perspective [##REF##14568754##24##].</p>", "<p>To design an adequate methodology for a nationwide study of ELDs is not straightforward because of the sensitive nature of the issue and the specific difficulties involved in the organisation of such a survey. In this article we present the protocol of the 2007 Flemish ELD study, which was guided by four methodological questions: (1) which study design is most appropriate for obtaining reliable incidence estimates and descriptions of ELDs, even of rare ELDs, that are representative for all deaths in Flanders in 2007?; (2) how can comparability with earlier ELD studies in Flanders and other countries be ensured?; (3) how can strict anonymity of physicians and patients for ethical and judicial reasons be guaranteed?; and (4) how can a sufficient response rate for a survey on this sensitive subject be achieved?</p>", "<p>The study design we present in this article is based on a method, first developed in the Netherlands in 1990 [##REF##1715962##9##], that has been successfully used in several European countries to study the nationwide incidence and characteristics of ELDs [##REF##14672110##8##, ####REF##1715962##9##, ##REF##8929370##10####8929370##10##,##REF##11117913##13##, ####REF##12907005##14##, ##REF##12907015##15####12907015##15##,##REF##17490798##19##,##REF##17494928##20##]. However, this is the first time that this study design has been described in detail. We believe that presenting it will be useful to researchers in other countries who intend to embark on similar research. The methodology outlined in this article will also serve as a reference for future publications using data from this study.</p>" ]

[ "<title>Method design</title>", "<title>A retrospective survey based on death certificates</title>", "<p>Obtaining data from a representative sample of dying patients in a prospective study design is an impossible task, as this would entail following an excessively large number of patients in numerous care settings. Moreover, defining who is dying is never clear-cut, and the problems of patient burden and attrition or non-response of the sickest patients [##REF##16499468##28##,##REF##16484691##29##] rules out the option of a prospective study design. There is also a danger that a prospective study will influence the behaviour of physicians and other caregivers. A retrospective (post-mortem) study design was therefore the more favourable option for this study.</p>", "<p>Because the study aims to obtain reliable estimates of ELDs for all deaths in Flanders, it was desirable to take the death case as the unit of measurement as this evidently provides a clear epidemiological denominator for the entire population of deaths, as well as for the subpopulations of deaths, e.g. cancer deaths. This provides more reliable incidence estimates than incidence studies where the physician is the unit of measurement and a representative sample of physicians are asked to report on the last death under their supervision in e.g. the last 12 months [##REF##9066548##11##,##REF##9554861##12##,##REF##16482752##18##]. In these studies, the number of deaths per participating physician is often not taken into account, and ELD incidence rates on population level are estimated on the basis of physician characteristics. Also, recall bias can be considerable if the physician's last death occurred a long time before the study.</p>", "<p>Every death in Belgium must be registered via a death certificate issued by the civil registrar of the municipality where the death took place. The physician completes the first part of the death certificate, indicating the sex of the deceased, some medical information (such as causes of death), time and place of death and signs the certificate with full name and medical registration number. The second part of the death certificate (containing socio-demographic information about the residence, age, education, occupation, nationality, civil status and living situation of the deceased) is completed by the civil registrar of the municipality in which the death took place. The death certificates are first processed by the provinces where the death occurred before they are sent to the central administration authorities. For Flemish death certificates this is the Flemish Agency for Care and Health (part of the Flemish Ministry for Welfare, Public Health and Family). Death certificates are thus particularly suitable for a nationwide study of ELDs; because every death is represented by a death certificate, it is easy to draw a representative sample of deaths. Also, the certifying physician's identification details listed on the death certificates allow the physician to function as the observational unit for the study. Furthermore, the socio-demographic and morbidity data of the deceased are readily available on the certificates and can be included in the survey. We obtained permission from the Flemish Agency for Care and Health to conduct a cross-sectional postal survey among the certifying physicians of a representative sample of death certificates.</p>", "<title>Selection of deaths and sampling</title>", "<p>The selection of deaths and sampling procedure needed to provide a representative sample of all deaths in Flanders in 2007 and had to include a sufficient amount of deaths to yield reliable information on the characteristics of all types of ELDs. Inclusion criteria for the study were:</p>", "<p>- the death taking place in Flanders,</p>", "<p>- the deceased is a resident of Belgium at the time of death,</p>", "<p>- the deceased is aged one year or more at the time of death.</p>", "<p>The death must have occurred in Flanders as the aim of the study is to describe end-of-life practices in the Flemish region; the limited number of Flemish residents who died outside Flanders are thus not included. The criterion of residence in Belgium is necessary to exclude all deaths in Flanders of persons, with or without the Belgian nationality, who live abroad as their socio-demographic characteristics and medical history would not be available. The number of these deaths is very small anyway and the majority of them are caused by traffic accidents, indicating a low likelihood of an ELD preceding death. Deaths of neonates (under one year of age) are excluded because end-of-life decision making is a very different issue in this age group, requiring an adjusted questionnaire. ELD studies in neonates have been done in the past in Belgium and the Netherlands [##REF##14568754##24##,##REF##15124828##30##,##REF##17537009##31##], but were not necessary for the present study.</p>", "<p>We sampled a fraction of almost 25% in a six month period from June 1^stuntil November 30^th2007. This amounted to 6928 death cases, approximately 12% of all deaths in 2007 (percentages based on the mortality rate of Flemish deaths for 2006, the most recent reference year for which mortality statistics were available). The sample size and proportion are significantly larger than in the previous Flemish ELD studies [##REF##11117913##13##,##REF##12907005##14##,##REF##17490798##19##], ensuring the greater overall statistical power of the results. The sample size necessary to estimate accurately the incidence rates with a confidence level of 95% was calculated based on the response level of the previous Flemish ELD studies in 1998 (49%) in 2001 (59%) [##UREF##5##32##].</p>", "<p>The sample is proportioned for month of death and province of death (Flanders consists of five provinces). From the previous ELD studies we know that ELDs occur more frequently among patients with a certain cause of death [##REF##11117913##13##,##REF##12907005##14##]. We therefore adopted disproportionate sampling of deaths to include more patients with a cause of death known to have a higher likelihood of one or more ELDs. This should result in more cases in which an ELD preceded death, and should thus further increase the statistical power and reliability of the incidence estimates and descriptions, even for the less-prevalent ELDs. According to the underlying cause of death on the death certificates and the corresponding probability of an end-of-life decision being made (derived from the data of the Flemish 2001 ELD study) deaths are grouped into one of four strata and sampled disproportionately (see Table ##TAB##0##1##).</p>", "<p>Because end-of-life decision making in minors (1–17 years of age at death) may differ from that in adults, we also integrated a fifth stratum. As there are relatively few deaths of minors annually, all deaths of minors in the period June-November 2007 are sampled to guarantee reliable incidence estimates for deaths in this age category.</p>", "<title>Questionnaire (see additional file ##SUPPL##0##1##)</title>", "<p>In developing the questionnaire, attention was paid to issues of length, difficulty, clarity, term ambiguity and similarity of content to questionnaires in previous ELD studies. We developed a questionnaire which drew on those of the previous studies in Belgium, the Netherlands and other European countries, the first of which had been developed for the 1990 Dutch survey on ELDs [##REF##1715962##9##]. We used the same set of key questions to ask about the medical decisions that were made at the end of life, thereby making possible incidence estimates comparable to those in earlier studies. Secondary questions regarding the decision-making process preceding an ELD, treatments and care provided, pain and other symptoms present in the last 24 hours before death and the perceived quality of dying were altered or added. The questionnaire was thoroughly analysed and tested by several physicians to correct for any imperfections or ambiguities. Its length was limited to five pages and the difficulty of the questions was kept as low as possible, bearing in mind the complexity of the research subject. The original Flemish version of the questionnaire is provided as additional file ##SUPPL##0##1## to this manuscript.</p>", "<p>There are four sections to the questionnaire. In the first, general section the physicians fill in their occupation (general practitioner or specialist), whether they had contact with the patient before his or her death and whether or not the death was sudden and unexpected. The other sections are to be completed only if the treating physician had contact with the patient prior to death and death was not sudden and completely unexpected. The second section asks key questions concerning the medical decisions that were made at the end of the patient's life. Terms such as 'euthanasia' or 'physician-assisted suicide' are not used, as they are emotionally charged and subject to ambiguous and multidimensional definition. Instead, the types of ELDs are more validly determined by establishing (1) what act the physician initiated, (2) to which extent the physician intended life-shortening when initiating the act, and (3) if there had been an explicit request from the patient to initiate the act. Figure ##FIG##0##1## shows how a classification of ELDs is derived from the answers to the key questions. If more than one ELD was made, the decision with the most explicit intention of hastening death is given priority in the classification. And if there was more than one act with a similar intention to hasten death, the administering of drugs is chosen over the withholding or withdrawal of treatment. In the third section physicians can note the likely degree to which life was actually shortened and some characteristics of the decision-making process. We included additional questions also posed in the 2005 Dutch ELD study in this section: one about whether or not euthanasia cases were reported, as is required by the Euthanasia Law, and if not why they were not reported and another concerning the term physicians would use to describe their act. The fourth section comprises questions about the characteristics of care and treatment provided at the end of the patient's life, the symptoms observed in the last 24 hours as well as the perceived quality of the patient's death. Finally we integrated a set of questions in this section about palliative or terminal sedation (defined as continuous deep sedation until death). In addition to the types of drugs used for the sedation, the length of the sedation, and the withdrawal of food and fluids, the questionnaire asks about the presence of an explicit request by the patient or the family, possible alternatives to sedation and whether a life-shortening intention was present. Thus, the questionnaire can pursue the paramount question of whether sedation is performed as a treatment decision with no life-shortening intention whatsoever or as an ELD or even as an alternative to euthanasia, as suggested in the literature [##REF##10733450##7##,##REF##16606811##33##, ####REF##16636220##34##, ##REF##9019033##35####9019033##35##] and in recent research [##REF##17494928##20##].</p>", "<title>Mailing procedure and anonymity</title>", "<p>Safeguarding the anonymity of physicians and patients is not only necessary for obvious ethical reasons but also for judicial reasons. Some life-ending acts can be deemed unacceptable by the Belgian criminal law and if anonymity were not guaranteed physicians could risk criminal prosecution for end-of-life decisions reported in this study. Moreover, the response rate to the questionnaire as well as the reliability of the answers will only improve if physicians feel safe enough to answer. Therefore a rigorous procedure was implemented to guarantee that no completed questionnaire could be linked to a particular patient or physician and that both patients and physicians remained anonymous. This procedure has been used in past studies on ELDs, and has proved effective [##REF##1715962##9##,##REF##8929370##10##,##REF##11117913##13##,##REF##12907005##14##,##REF##17494928##20##]. To meet the requirement of anonymity, the different stages of the survey i.e. the sampling and mailing, receiving and processing of the questionnaires are spatially separated. Each stage is performed by different persons. Four parties are involved in the survey, each with specific functions. For a schematic overview of the procedure, see Figure ##FIG##1##2##.</p>", "<title>1. Flemish Agency for Care and Health (of the Flemish Ministry for Welfare, Public Health and Family)</title>", "<p>The Flemish Agency for Care and Health, the official department for processing death certificates, is responsible for the sampling of the death certificates, management of the sample database, and the mailing of the questionnaires. Each case is ascribed a unique sample number which is derived from the death certificate number using a fixed algorithm. These sample numbers are used at the end of the study to link the questionnaires to the patients' socio-demographic and morbidity data, derived from the death certificates, in a database provided by the Flemish Agency for Care and Health (cfr. infra). An accompanying letter is included with the questionnaire providing the physician with enough patient characteristics derived from the death certificates to identify the patient (i.e. sex, date of birth, date of death, and municipality of death). The researchers do not have access to the sample database as it contains identifying information of the patients and physicians.</p>", "<title>2. Physicians</title>", "<p>After identifying the patient by the patient characteristics in the accompanying letter, the physicians can fill out the questionnaires. They are advised to destroy the accompanying letter afterwards. No combination of answers given in the questionnaires can lead to identification of the patient or of the physician.</p>", "<p>In some cases, the certifying physician was not the treating physician for the patient in question. In such cases the physician is given a directive to transmit the questionnaire to the treating physician. If the identity of the treating physician is not known the case in question is discarded as being impossible to study. Also, some physicians no longer work at the hospital or practice where the patient died and can therefore not identify the patient or do not have access to the patient file. These cases are also removed from the sample.</p>", "<title>3. Lawyer</title>", "<p>The completed questionnaires are not returned to the Flemish Agency for Care and Health or to the researchers but instead to a sworn lawyer who is bound to professional confidentiality. The lawyer safeguards the anonymity of the questionnaires received by removing any possible identifying information from them such as notes, stamps and signatures. He also removes the sample numbers and reports them to the Flemish Agency for Care and Health. These cases are subsequently deleted from the sample database so that the certifying physician does not receive further reminders regarding this particular death.</p>", "<p>As removing the sample numbers from the questionnaires would make it impossible to link them to the corresponding patient's socio-demographic and morbidity data at the end of the study, the lawyer ascribes a new number to every questionnaire and keeps a database in which the original sample numbers and the corresponding numbers are linked to one another.</p>", "<p>The lawyer keeps the received questionnaires until the end of the survey. Afterwards, the Flemish Agency for Care and Health transmits the database of the patients' socio-demographic and morbidity characteristics to the lawyer. The lawyer links the cases in this database to their corresponding new numbers via the original sample numbers and then deletes the original sample numbers. When these sample numbers (derived from the death certificates) are deleted, the information in the database of patient characteristics and the information in the questionnaires can no longer be traced back to the corresponding death certificates.</p>", "<title>4. Research group</title>", "<p>After this complex procedure, the information in the database and questionnaires is strictly anonymous: the replacement of sample numbers by new numbers cuts the link between questionnaires and death certificates and neither the combination of patient characteristics nor the information provided in the questionnaires can lead to the identification of patients or physicians. The lawyer can thus transmit the questionnaires and the database with patient characteristics to the researchers, who combine the data from both (using the new numbers) into one database for analysis.</p>", "<title>Total Design Method (TDM)</title>", "<p>All efforts to attain a representative sample of deaths are ineffective if the response rate does not reach a minimal level, therefore some measures are taken to achieve this. We followed some prescriptions from Don A. Dillman's Total Design Method (TDM) for mail surveys [##UREF##6##36##]. Firstly, we established an intensive follow-up mailing. After the questionnaire is sent out, the physician receives a maximum of three reminders at an interval of 14 days until the questionnaire is returned. In the second reminder a new copy of the questionnaire is included, thus anticipating the possibility of the physician having lost the original.</p>", "<p>The TDM is based on the costs-benefits analysis of social action; in a social context a person only acts if there are advantages in that action. Given this principle, the probability of participating in a study will be greater if the study succeeds in keeping the costs (i.e. disadvantages, efforts, time) of participation as low as possible while at the same time maximising the gains for the respondent [##UREF##6##36##]. To minimise the costs of participation, the questionnaire was kept as short as possible, and the difficulty-level of the questions and answer options as low as possible considering the complexity of the study subject. A maximum of five death cases per physician was decided on to limit responder fatigue. A stamped return envelope was included with every questionnaire sent out. We feel that guaranteeing the anonymity of physicians and patients is also an important measure in minimising the potential costs of participation. To maximise the gains for the physicians, we stress the importance of the study for the medical field, as the results can contribute to better and more effective policies on end-of-life decisions in Belgium. Participation can thus ultimately result in better conditions for physicians to work in, as well as for patients at the end of their lives. We also deem it important to communicate the results of the study to the participants. All participating physicians are assured of an invitation to a seminar on the study after the data collection. As an extra incentive, a valuable artwork will be awarded to a randomly chosen participating physician. Due to the large number of participating physicians, the funds are not sufficient to reward every individual physician financially.</p>", "<p>Additionally to the involvement of the Vrije Universiteit Brussel and Ghent University in conducting the study, representatives of two other Flemish universities, the Katholieke Universiteit Leuven and Universiteit Antwerpen, and the Scientific Institute of Public Health support the study to increase its visibility. Also, the positive recommendation of the Belgian National Disciplinary Board of Physicians (cfr. infra) is mentioned in the accompanying letter.</p>", "<title>Non-response survey</title>", "<p>After the data collection a one-page questionnaire is sent to all non-responding physicians, asking about their reasons for non-response. Besides providing interesting information on non-response in general, these reasons can warrant the removal of some cases from the sample because of the physician's inability to fill out the questionnaire (e.g. the patient cannot be identified with the information provided, the physician no longer has access to the medical file, the certifying physician is not the attending physician and can not identify him or her, or the physician never received the questionnaire).</p>", "<title>Data analysis</title>", "<p>The researchers will prepare an SPSS 16.0 (SPSS Inc.) database with coding scheme for a certified data management company which will enter the data. Range and skip checks will prevent key-punching errors, and the data quality will further be improved partly via double data-entry and partly through extensive random sample checks. The researchers will perform data cleaning via SPSS syntax operations.</p>", "<p>The data will be weighted to correct for the disproportionate stratification of underlying causes of death and the deaths of minors. The influence of non-response on the representativity of the data will subsequently be checked and weighted through a comparison of proportionality of underlying causes of death and other patient characteristics (i.e. sex, age, educational level, marital status, living situation, province of residence, month of death and place of death) between deaths where responses have been received and deaths within the general population in 2007.</p>", "<p>Data will be analysed with descriptive statistics (valid percentages and 95% confidence intervals), as well as bi- and multivariate association statistics using SPSS version 16.0.</p>", "<title>Recommendations</title>", "<p>Positive recommendations for the anonymity procedure and study protocol were obtained from the Ethical Review Board of the University Hospital of the Vrije Universiteit Brussel, the Ethics Committee of the University Hospital of Ghent University, the Belgian National Disciplinary Board of Physicians and the Belgian Federal Privacy Commission.</p>" ]

[]

[ "<title>Discussion</title>", "<p>The 2007 Flemish ELD study aims to produce representative incidence estimates of end-of-life decisions in Flanders and to describe their characteristics as well as the circumstances under which they occur. A summary of the key characteristics of the study design is shown in Table ##TAB##1##2##.</p>", "<p>The four methodological challenges formulated at the outset of this article were addressed: (1) we opted for a retrospective study design based on death certificates, as this design provides the best chances of obtaining reliable incidence estimates of ELDs and their characteristics from a large and representative sample of deaths. Moreover, the disproportionate stratification based on the likelihood of an ELD preceding death further increases the statistical power of the results (2) comparability of the data to earlier studies is ensured by using the same set of key questions in the questionnaire, and by keeping the main characteristics of the study design constant (3) a rigorous procedure involving a lawyer as intermediary between physicians and researchers is employed to guarantee the anonymity of physicians and patients and (4) we use several measures from the Total Design Method to obtain a satisfactory response rate.</p>", "<p>The study has some strengths as well as weaknesses related to the use of death certificates and the study design in general.</p>", "<title>Strengths</title>", "<p>Most studies in end-of-life care research are limited with regard to sample size, care settings or illness types. This impedes the chances of obtaining representative population data in end-of-life care research. For example, one study examined end-of-life practices in a sample of dying patients but was set only in intensive care units [##REF##12915432##16##]. Using death certificates on the other hand facilitates the obtaining of robust data for the entire population, as a large sample of deaths can be drawn across care settings including all causes of death [##REF##16499468##28##,##REF##17922894##37##]. And, because of its nationwide scope, this study design is most suitable for international comparative research, as the EURELD six nations study has shown [##REF##12907005##14##]. Also, the retrospective nature of the study design does not encounter the problems of patient burden, attrition or non-response of the sickest patients found in prospective study designs [##REF##16499468##28##,##REF##16484691##29##], and it does not run the risk of influencing end-of-life practices, which is a realistic possibility in prospective studies.</p>", "<p>Because all deaths must be reported to the proper government authorities, death certificates also allow the use of the death as the unit of measurement, providing a clear denominator for reliable estimation of the incidence of ELDs [##REF##16499468##28##]. The reliability of these estimates is not guaranteed in studies based on the last deceased patient treated by a representative sample of physicians [##REF##9066548##11##,##REF##9554861##12##,##REF##16482752##18##], as the unit of measurement in these studies is the physician and the number of deaths preceded by an ELD is estimated on the basis of physician characteristics. Moreover, in contrast to the death certificate design, physicians in these studies are not guaranteed to have attended a death.</p>", "<p>Using death certificates also facilitates the anonymous linking of patient characteristics to the information provided in the questionnaires, allowing the study of associations between socio-demographic and morbidity characteristics of the patient on the one hand, and end-of-life decision making and provided care at the end of life on the other hand [##REF##18000128##38##].</p>", "<p>Another strength of the present study is that, whereas in other end-of-life research physicians can be inadvertently selected on the basis of their interest in or attitudes towards end-of-life practices, the use of death certificates excludes the possibility of a biased selection of physicians.</p>", "<title>Weaknesses</title>", "<p>The physician signing the death certificate is occasionally not the patient's treating physician, and therefore is not in a position to fill out the questionnaire. Despite the directive to transmit the questionnaire to the treating physician, some cases are impossible to study as the treating physician cannot be identified. In some instances not even the identity of the patient can be retrieved because the treating physician no longer has access to the patient file.</p>", "<p>Because death certificates have to be processed by the proper authorities before they can be made available for research, there can be a considerable delay between the patient's death and the study of that death [##REF##17922894##37##]. The delay in our study has reached as much as four months (there is variation across countries, ranging from two to six months). We can therefore not exclude some influence of recall bias. To address this issue, we encourage physicians to fill in their questionnaires using the patient files, which are mostly readily at their disposal.</p>", "<p>Given the death as the unit of measurement and the large number of deaths studied, one physician can receive several questionnaires. Despite a maximum of five cases for each physician, responder fatigue and diminishing response rates can result.</p>", "<p>A structured and semi-closed questionnaire can often overlook the intricacies of certain end-of-life decisions. Moreover, the questionnaire used in this study has, for reasons of response, been limited in length and time-consuming questions have been left out. There is a risk that in some cases vital information can be missed. To counter this problem, a section is provided at the end of the questionnaire in which the physician can comment or elaborate on the answers given.</p>", "<title>Opportunities for future research</title>", "<p>The present study is the third in a series of death certificate studies in Belgium. Keeping the study design and the questionnaire constant creates the opportunity for future studies to build on the comparable data obtained in the past and to identify accurately developments in the field of ELDs. The study design can be applied to research in other countries, so that data can be produced for international comparative research. Comparable data are already available in Belgium, the Netherlands, Italy, Switzerland, Denmark and Sweden [##REF##14672110##8##, ####REF##1715962##9##, ##REF##8929370##10####8929370##10##,##REF##11117913##13##, ####REF##12907005##14##, ##REF##12907015##15####12907015##15##,##REF##17490798##19##,##REF##17494928##20##]. Furthermore, the use of death certificates in end-of-life care research need not be limited to ELDs; they can also be applied in retrospective research on other issues in this field [##REF##17922894##37##].</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Reliable studies of the incidence and characteristics of medical end-of-life decisions with a certain or possible life shortening effect (ELDs) are indispensable for an evidence-based medical and societal debate on this issue. This article presents the protocol drafted for the 2007 ELD Study in Flanders, Belgium, and outlines how the main aims and challenges of the study (i.e. making reliable incidence estimates of end-of-life decisions, even rare ones, and describing their characteristics; allowing comparability with past ELD studies; guaranteeing strict anonymity given the sensitive nature of the research topic; and attaining a sufficient response rate) are addressed in a post-mortem survey using a representative sample of death certificates.</p>", "<title>Study design</title>", "<p>Reliable incidence estimates are achievable by using large at random samples of death certificates of deceased persons in Flanders (aged one year or older). This entails the cooperation of the appropriate administrative authorities. To further ensure the reliability of the estimates and descriptions, especially of less prevalent end-of-life decisions (e.g. euthanasia), a stratified sample is drawn. A questionnaire is sent out to the certifying physician of each death sampled. The questionnaire, tested thoroughly and avoiding emotionally charged terms is based largely on questions that have been validated in previous national and European ELD studies. Anonymity of both patient and physician is guaranteed through a rigorous procedure, involving a lawyer as intermediary between responding physicians and researchers. To increase response we follow the Total Design Method (TDM) with a maximum of three follow-up mailings. Also, a non-response survey is conducted to gain insight into the reasons for lack of response.</p>", "<title>Discussion</title>", "<p>The protocol of the 2007 ELD Study in Flanders, Belgium, is appropriate for achieving the objectives of the study; as past studies in Belgium, the Netherlands, and other European countries have shown, strictly anonymous and thorough surveys among physicians using a large, stratified, and representative death certificate sample are most suitable in nationwide studies of incidence and characteristics of end-of-life decisions. There are however also some limitations to the study design.</p>" ]

[ "<title>Abbreviations</title>", "<p>(Medical) end-of-life decisions: ELDs; Total Design Method: TDM; Monitoring quality of End-of-Life Care: MELC study.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>All authors contributed to the design and conceptual framework of the protocol. The manuscript was drafted by KC, with further input from all other authors. LD and FM are the project supervisors. All authors read, revised and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2458/8/299/prepub\"/></p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>The study is funded by the Institute for the Promotion of Innovation by Science and Technology Flanders. We are deeply indebted to the Flemish Agency for Care and Health. We would also like to thank all physicians who participated in testing and validating the questionnaire and the lawyer and his personnel who were prepared to act as an intermediary between respondents and researchers.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Questions to determine end-of-life decisions.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Schematic overview of the mailing and anonymity procedure.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Four strata for disproportionate stratification based on cause of death*</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\">Stratum 0</td></tr><tr><td align=\"left\">Cause of death implies that an ELD is certain</td></tr><tr><td align=\"left\">Included causes of death: euthanasia**.</td></tr><tr><td align=\"left\">Every death in this stratum is selected for the survey.</td></tr><tr><td/></tr><tr><td align=\"left\">Stratum 1</td></tr><tr><td align=\"left\">Cause of death implies that an ELD is probable</td></tr><tr><td align=\"left\">Included causes of death: neoplasms (ICD-10 codes: C, D00–D48).</td></tr><tr><td align=\"left\">One out of every two deaths in this stratum is selected for the survey.</td></tr><tr><td/></tr><tr><td align=\"left\">Stratum 2</td></tr><tr><td align=\"left\">Cause of death implies that an ELD is possible</td></tr><tr><td align=\"left\">Included causes of death: endocrine, nutritional and metabolic diseases; mental and behavioural disorders; diseases of the nervous system; diseases of the respiratory system; diseases of the digestive system; diseases of the genitourinary system (ICD-10 codes: E, F, G, J, K, N).</td></tr><tr><td align=\"left\">One out of every four deaths in this stratum is selected for the survey.</td></tr><tr><td/></tr><tr><td align=\"left\">Stratum 3</td></tr><tr><td align=\"left\">Cause of death implies that an ELD is improbable</td></tr><tr><td align=\"left\">All remaining causes of death are included in this stratum (ICD-10 codes: A, D50–D99, H, I, L, M, Q, R, S, T, U, V, Y).</td></tr><tr><td align=\"left\">One out of every eight deaths in this stratum is selected for the survey.</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Summary of the study design</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">DEATH CERTIFICATE SURVEY</td></tr><tr><td align=\"left\"> ✓ large sample of deaths</td></tr><tr><td align=\"left\"> ✓ nationwide (over care settings and causes of death)</td></tr><tr><td align=\"left\"> ✓ stratified disproportionately based on cause of death</td></tr></thead><tbody><tr><td align=\"left\">QUESTIONNAIRE</td></tr><tr><td align=\"left\"> ✓ short and validated questionnaire</td></tr><tr><td align=\"left\"> ✓ key questions of ELDs identical to those in earlier studies</td></tr><tr><td align=\"left\"> ✓ emotionally charged terms absent in key questions</td></tr><tr><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\">MAILING PROCEDURE</td></tr><tr><td align=\"left\"> ✓ guarantee of anonymity for physicians and patients</td></tr><tr><td align=\"left\"> ✓ response-increasing measures</td></tr><tr><td align=\"left\"> ✓ intensive follow-up mailing</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>Questionnaire of the 2007 Flemish ELD study.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>* Causes of death were grouped into strata based on the probability of an ELD as observed in the Flemish part of the EURELD six nations study (2001) [##REF##12907005##14##].</p><p>** Although there is no box to specify euthanasia in the death certificate, it is occasionally written down by the certifying physician in the section 'immediate cause of death'.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2458-8-299-1\"/>", "<graphic xlink:href=\"1471-2458-8-299-2\"/>" ]

[ "<media xlink:href=\"1471-2458-8-299-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Injuries constitute an important health problem worldwide and they are one of the major causes of death among people under 45 years old [##UREF##0##1##,##UREF##1##2##]. The majority of all injury-related deaths occur in low and middle-income countries [##UREF##0##1##, ####UREF##1##2##, ##REF##15623852##3##, ##UREF##2##4####2##4##] where knowledge is scarce regarding injury distribution, pattern and prevention [##UREF##3##5##]. Epidemiological studies have been conducted in some low and middle-income countries but, most often, traffic-related injuries and those occurring in urban settings have been in focus [##REF##9839745##6##, ####UREF##4##7##, ##REF##16147489##8####16147489##8##]. Yet, studies in rural areas have been conducted in countries in various continents, including Asia (Pakistan [##REF##17623066##9##], Bangladesh [##REF##10800153##10##,##REF##9490886##11##], India [##REF##17002729##12##], and Vietnam [##UREF##3##5##,##UREF##5##13##]), Africa (Kenya [##REF##17970004##14##], Ghana [##REF##10680242##15##], Uganda [##REF##11289535##16##] and Tanzania [##REF##15679887##17##]) and South America (Nicaragua [##REF##16260010##18##]). Those studies reveal that injuries constitute an important health problem in the rural areas.</p>", "<p>In Iran, where this research has been conducted, injury-related years of life lost are higher than for the worldwide average [##REF##17037707##19##,##UREF##6##20##]. Studies on injury epidemiology and prevention are limited and are mainly urban [##REF##11587698##21##, ####REF##18215528##22##, ##REF##9688197##23####9688197##23##]. Also people's experiences and opinions about injury prevention and control have rarely been addressed. Currently, approximately 33% of the total population live in rural areas [##UREF##7##24##], and people benefit from a well-established health network, consisting of village-based local \"health houses\", from which health workers (known as Behvarzes) work. The main function of the Behvarz is to offer primary health care services to the local population and to gather health information. Usually the Behvarzes are selected from their local community and can therefore establish a very close relationship with community members. This, in turn, can help to gather accurate data. Health house workers also contribute to the simple but well-integrated health information system [##REF##11794090##25##].</p>", "<p>In 2004, the Ministry of Health and Medical Education launched a program ultimately aiming at the reduction of injuries in rural areas. As a first step, an injury registration and surveillance system has been developed that forms part of the health information system and implies that the Behvarzes are responsible for the registration of all injuries leading to hospitalization (at least 6 hours as a standard criterion) or death. This data should provide information on the frequency of occurrence and characteristics of severe injuries and allow for the follow-up of future national and local interventions. Twiserkan district, where this study was conducted, is one of the districts selected for the pilot phase of the implementation of the surveillance system.</p>", "<p>In the current study, we take advantage of the reports gathered by the Behvarzes over a one year period to assess the incidence of rural injuries and, through interviews with injured people or their relatives, characterize those injuries' epidemiology and document the suggestions of people from affected families concerning injury prevention and control.</p>" ]

[ "<title>Methods</title>", "<p>The Twiserkan district is located in the Hamadan province (over 19 000 square kilometers), in western Iran. In 2002, Hamadan had over 1.7 million inhabitants, of whom 44% lived in rural areas [##UREF##8##26##,##UREF##9##27##]. In 2006, when this study was conducted, Twiserkan district had a population of about 110 000 inhabitants, of which 58% was rural. The number of rural households was 14,789 and the number of inhabitants amounted to 62,857.</p>", "<p>All unintentional injuries leading to hospitalization (more than 6 hours) or death, occurring over a one-year period (June 1, 2005 – May 31, 2006), were considered. These were first identified in the files compiled at the local health houses (n = 62) in the Twiserkan district (see below). Thereafter the household of each injured person was visited by a trained and experienced interviewer. Before any interview, the aim of the study and main content of the interview were explained to the interviewee who was also guaranteed confidentiality. Once verbal consent was given, a short face-to-face interview was conducted (in June 2006; 134 injuries in 117 households) with the family member identified as responsible of caring for the household (response rate 100%). To ensure as complete answers as possible, the injured family member took part in the interview any time she or he was present at the time of visit. This was the most common situation, except of course for fatal injuries. It can be underlined that the recall period ranged from a few days to one year post injury. About two-thirds of the injuries had occurred between 6 to 12 months before the interview and the remaining occurred either 3 to 6 (13.4%) or less than 3 months (18.7%).</p>", "<p>Prior to data collection, a structured questionnaire including both closed and open questions was developed and pre-tested by the research team. It included information about both the injured person (and his/her household) and the injury (type, nature and circumstances of occurrence). Open-ended questions were included so as to find out how people regarded the role of the community, the Behvarzes, and the authorities (health and others) with regard to injury prevention. For each actor, people were asked to give their opinion as to what more could be done to help reduce injury for the village residents. In households with more than one injury during the reference period, all injuries sustained were considered at once when addressing the questions about household opinions concerning the roles of different actors in injury prevention and control.</p>", "<p>The study was approved by the Iranian National Ethics Committee in Medical Research, Ministry of Health and Medical Education of Iran.</p>", "<p>Data were entered, processed and analyzed in Excel (version 2003). Injury incidence rates were estimated globally and for fatal and non-fatal injuries respectively. Injury characteristics were coded and categorized according to the WHO guidelines for injury survey and surveillance [##UREF##10##28##,##UREF##11##29##]. Thereafter, using descriptive statistics, the characteristics of the injured people (sex, age group, education and occupation), of the injury events (place of injury and injury mechanism) and of their consequences (nature, body region and injury severity/recovery) were highlighted. When a person sustained several injuries during the same injurious event, the most severe one was considered. This was made possible as space was provided in the questionnaire to identify the most severe injury, following the WHO guidelines and after discussion and consensus between two members of the research team.</p>", "<p>People's opinions about injury prevention were first entered as free text. Thereafter, answers were read by two members of the research team and key ideas/phrases were identified and discussed at different sessions. Meaningful categories were identified that represented specific and homogeneous domains of potential intervention or action. Since some people had several suggestions and as they were not asked to prioritize or rank them, all opinions expressed by each respondent were taken into account.</p>" ]

[ "<title>Results</title>", "<title>Injury incidence of non-fatal and fatal injuries</title>", "<p>A total of 134 injuries were reported by the Behvarzes during the study period. These were identified among 117 households (of 14,789 in total) and consisted of 26 fatal and 108 non-fatal injuries. The corresponding incidence rates of injuries per 10 000 person-years are therefore 21.4 injuries in total (95% CI 17.7–24.9), 4.1 fatal injuries per 10 000 person-years (95% CI 2.5–5.7) and 17.2 non-fatal injuries (95% CI 13.9–20.4).</p>", "<p>Injury deaths occurred most often in traffic crashes (n = 10), followed by burns (n = 6), poisoning (n = 5), falls (n = 4), and electrocution (n = 1). Of the 134 injury cases identified, 22 were attributable to seven injurious events: three traffic-related and four in the home (see below). In the remainder of the text, all injured people, even those injured in the same event, are considered as individual cases.</p>", "<title>Injured people's characteristics</title>", "<p>Table ##TAB##0##1## presents the characteristics of the injured people. In total, about three-quarters were males; 21.6% were aged 15 or less and an additional 22.4%, 56 and over. The majority had not completed high school (85.8%) and 26.1% were farmers.</p>", "<title>Fatal and non-fatal unintentional injury characteristics</title>", "<p>Figures ##FIG##0##1## and Figure ##FIG##1##2## show the characteristics of the fatal and non-fatal unintentional injury circumstances (place of injury and injury mechanism). The injuries occurred in similar proportions in the home or on the road, outside the village. Traffic injury was by far the most common injury mechanism (44.8%), followed by falls (26.1%) and thereafter burns (fire/flame/heat; 11.2%). More details of these three injury mechanisms are given in the text below.</p>", "<title>Road Traffic Injuries (RTIs; 60 injuries from 56 households)</title>", "<p>Of 60 RTIs, 48 were sustained by males and the bulk of them were among people of working age, most often 16–35 years (28 cases). Only 10 cases were reported among children 15 years and less. The most common mode of transportation among the injured people was by motorcycle (43.3%), followed by car (26.7%) and on foot (20%); most injured people were drivers (60%). The number of people injuries as motor vehicle passenger was almost equal between males and females (5 and 4 respectively). Injuries occurred above all on roads outside the village (68.3%).</p>", "<p>As mentioned above, traffic crashes were the primary cause of fatal injuries (n = 10). Two pedestrians were killed and eight motor-vehicle riders: five drivers (1 car, 2 motorcycle and 2 tractor drivers) and three car passengers. Three crashes involved more than one injured person with the following consequences: (1) the first had four injured people, three recovered and one died; (2) the second had two injured people, one recovered and one was disabled; (3) the last one had two injured people and both recovered completely.</p>", "<title>Falls (35 injuries from 35 households)</title>", "<p>Of 35 fall injuries, 28 were among males and 29 among adults – one in three (12 cases) were among people aged 65 years and above. Falls on the same level and from a roof were the most common kinds of falls reported (8 cases each), followed by fall from tree (n = 7) and from stairs (n = 5). Among older people, fall on the same level and from stairs were most common (4 cases each). Falls from trees occurred among adults and during work activities. This is common in rural Twiserkan, especially during the walnut harvest. Four deaths were fall-related: three occurred close to the injurious event (a man fell from a walnut tree, an older one fell from a roof, a woman aged 78 years on the stairs), and one some days later (a man aged 80 years following a pelvis fracture sustained after a fall when walking).</p>", "<title>Burns (15 injuries from 8 households)</title>", "<p>As many as 11 out of 15 burns were among males and 6 cases were paediatric burns (0–15 years). All burns occurred at home. Contact with flame was the most common cause (n = 12), several injurious events occurred when manipulating gas equipment used for cooking or heating. Other burns resulted from contact with hot liquids, steam or other gas (n = 3). Three events had more than one injured person: the first had three (two recovered and one died), the second had two (one recovered and the other one disabled) and the third one led to as many as five casualties.</p>", "<p>Burns were the second cause of fatal injuries: six casualties resulting from two injurious events. The first occurred in the evening while the family members (5 persons) were sitting in a room and one of them was trying to unscrew a small gas capsule (picnic gas stove), to prevent a gas leak. According to the interviewee, the capsule came off suddenly, caught fire and all five persons received serious burn injuries. They were taken to the hospital but all five died after a few days. The second event occurred in the morning and involved three people. In circumstances similar to the preceding case – a gas leak from a gas capsule – the house caught fire suddenly. The three family members were also taken to the hospital but the family's one-year old child died after a few days.</p>", "<title>Fatal and non-fatal unintentional injury consequences</title>", "<p>Table ##TAB##1##2## shows the characteristics of consequences of the fatal and non-fatal unintentional injuries, all injuries aggregated and for three injury mechanisms: traffic, fall, and burn. Fracture was the most frequent nature of injury, encompassing about half of the all cases and 86% of fall-related injuries. The category \"organ system injury/internal injury\" refers to damage of some vital internal systems such as respiratory system, blood circulation system. In cases where more than one nature of injury was reported, focus was placed on the most serious one, according to WHO guidelines [##UREF##11##29##].</p>", "<p>For all injuries aggregated, the first most common injured single body region was the lower limb (20.1%), followed by the head/face (14.2%) and the upper limb (13.4% respectively). The most commonly injured single body region for traffic injuries was head/face (28.3%) followed by lower limb (25%) and for fall-related injuries was upper limb (22.9%) followed by lower limb and pelvis/hip equally (each 20%).</p>", "<p>Whenever an injured person had more than one injured body region, it was considered and coded as \"multiple regions\"; which occurred in nearly one in three injuries (31.3%). As for the injury severity, as mentioned above, almost one in five injuries was fatal. Twice as many injured people recovered completely and 23.1% recovered partially.</p>", "<title>Interviewees' suggestions for prevention</title>", "<p>Interviewees came up with quite a lot of suggestions concerning what more could be done to contribute to injury prevention and these suggestions varied considerably in kind depending on what actor they were asked to reflect upon: the authorities, the Behvarzes or the people themselves. The suggestions proposed were organized in a number of different categories and are presented in Table ##TAB##2##3##, shown in percentages both by number of suggestions and by number of households and also considering all injuries aggregated and traffic, fall and burn injuries separately.</p>", "<title>Authorities</title>", "<p>Almost one-third (31.1%) of the suggestions concerned changes in the living and commuting environment e.g., engineering and/or building (including infrastructure such as road construction and asphalting; signalization and product design). Proposals regarding the infrastructure and modernization of the traffic were most common, followed by suggestions concerning the provision of daily services (19.7 %), e.g., piped gas and even fire station for burns prevention, outdoor lighting for fall prevention and telephone for rapid contact with the health services at the time of injury event. People also mentioned the need for financial support for safety improvement in the village and even for better housing (12%). Law enforcement was also raised, in particular for traffic injury prevention (10.4%).</p>", "<title>Behvarzes</title>", "<p>For the Behvarzes, focus was placed largely on the provision of education to the people (about 60–69%), including various forms of continuous education in the population and informing upwards in the health system. They pointed to safety education in general and also to different aspects of safety educations such as road traffic safety education for the young (especially in relation to motorcycling) and home safety (targeting childcare and supervision). The latter was also considered as a way to promote the development of safe behaviour in children. The second important category of suggestions concerning the Behvarzes was about post trauma care, including both health house and home visits (16.5 %). Greater availability and accessibility to the Behvarzes and also to drugs and equipment (10.1%) came in third place.</p>", "<title>People themselves</title>", "<p>About one in three respondents considered that people themselves could behave in a safer manner to help control and prevent injuries. Thereafter, respondents raised the issue of cooperation between people (18.6%), including assisting each other in the event of injury, e.g., by transferring injured people to the health facilities. The third category of suggestions was about people's cooperation with the authorities to improve safety in the village-environment (17.8%). People also raised issues related to compliance, in-house engineering, and being good role-models for children, including teaching them safe behaviours and practices.</p>" ]

[ "<title>Discussion</title>", "<p>The study reveals that injuries among people in rural areas affect mainly males and also people of working age, which is in line with the injury distribution in other settings [##UREF##1##2##,##UREF##3##5##,##REF##17623066##9##,##REF##9490886##11##,##REF##17970004##14##,##REF##15679887##17##,##REF##11035469##30##,##REF##17875213##31##]. One finding is the relative importance of injuries on the road, not only because of their severity (10 out 26 deaths), but also because of their frequency. They are indeed as numerous as injuries in the home. Similar results were obtained in earlier studies from Vietnam [##UREF##3##5##] and Nicaragua [##REF##16260010##18##].</p>", "<p>Thus, traffic injuries are not only a concern among urban people in Iran but also among rural dwellers [##REF##10793781##32##]. As motorized commuting is on the increase even in rural Iran, it is important to pay attention to road safety [##UREF##2##4##,##REF##10793781##32##]. In particular, the safety of motorcyclists must be carefully considered as motorcycles are a popular means of transport [##REF##18215528##22##] – as is the case in many Asian countries [##UREF##4##7##,##REF##17071249##34##].</p>", "<p>According to the opinions of the people interviewed – coming from households that have been affected by a severe RTI during the past year -, various things can be done: roads can be better designed and maintained, individual protection legislation (e.g. compulsory safety belt and helmet wearing) could be enforced, people could comply with and adopt safer behaviours, Behvahzes could educate the population and convey information of relevance for injury control and prevention upwards in the health system. Interestingly, many of those suggestions find an echo in the recommendations found in the WHO report on road traffic injuries, in particular concerning making road safety a political priority, enacting and enforcing legislation, managing infrastructure to promote safety for all, and campaigning for greater attention to road safety [##UREF##2##4##].</p>", "<p>Injuries at home are also a concern in rural areas just as they are for instance in Pakistan, a neighbouring country, where they form the majority of injuries [##REF##17623066##9##]. In the district studied herein, burns may require special attention, if not because of their frequency, because of their severity (see also earlier Iranian studies [##REF##15939446##35##, ####REF##17333829##36##, ##REF##16529866##37##, ##REF##17544221##38####17544221##38##]). Burns also affect children to a greater extent, which is consistent with an earlier study from Bangladesh that showed that the incidence of burns among rural children was more than four times higher than among urban children [##REF##18242869##39##].</p>", "<p>In the Twiserkan district – and perhaps even in Iran as a whole – gas equipment used for cooking or heating may warrant special attention. Some villages still do not have gas mains and people use gas capsules and/or other heating equipment that is poorly adapted to in-house use. At present, fire stations are far from most villages and people did mention this as a matter of concern.</p>", "<p>For its part, the prevention of falls is undermined by the vast number of different situations leading up to them, which is a challenge for community-based education programs. Potentially more severe falls from a height, e.g. from a roof and falls from a tree may constitute important targets. The latter occur during work activities, mainly during the walnut harvest. This work is done in a traditional, non-technical manner and every year some people fall from large trees and are injured or killed. The results of one study on safety assessment of agricultural machinery in Iran showed that in 60% of cases agricultural injuries were severe [##REF##17892070##40##]. It ought to be emphasized that an important number of falls affect older people, which has been also observed in an earlier study showing that falls from standing height, falls during walking and falls on stairs were important risk factors for hip fracture for older patients [##REF##16510325##41##]. Fall-related injury prevention may require not only environmental improvements in and around the house but also, in the long run, changes in health behaviours (e.g. eating, smoking, and exercising) so as to reduce individual susceptibility to fall and also recovery after fall.</p>", "<p>Generally, people frequently mentioned that Behvarzes could play an important role in safety education matters on the local level. Behvarzes already have face-to-face meetings with community members as part of their traditional duties. Also, in recent years, the Ministry of Health and Medical Education has introduced a number of home safety program [##UREF##6##20##,##REF##15939446##35##,##REF##16260011##42##], and provided Behvarzes with educational packages, which is consistent with international literature.</p>", "<p>To our knowledge, there is a dearth of studies conducted thus far in Iran or in other rural settings that have collected people's opinions and suggestions about injury control and prevention. This study shows that rural people have a lot of ideas which can be considered for the conception and implementation of context-relevant measures for injury prevention in their community. In particular, people from households where injuries have occurred during the past year consider that not only a change in their own behaviour but also environmental changes and the provision of information and education are needed. We hope that the suggestions highlighted, though not fully representative of the whole rural population, will be taken into account in future developments of safety measures and programs, in both the Twiserkan district and other districts.</p>", "<p>Because of the routines in place [##REF##11794090##25##] and the relatively small size of the catchment areas, we have good reasons to believe that the study offers an accurate coverage of the severe injuries incurred in the population under study during the study period. It is indeed very likely that health houses do have a complete coverage of injuries leading to hospitalization and death in their community [##REF##11794090##25##]. In spite of the fact that collecting injury data was a relatively new procedure when the cases were identified, we regard the likelihood of missing cases as very unlikely given that the injuries covered are relatively severe, that the Behvarzes are well anchored in their community, and that those communities are relatively small.</p>", "<p>Before concluding, it ought to be underlined that the study covers one district only and is limited to one year of observation. Because of this, it is not possible to extrapolate our results to any other time period or district. Yet, some results can be regarded as a matter for investigation in other districts as well (e.g., traffic related injuries or burns).</p>" ]

[ "<title>Conclusion</title>", "<p>Traffic injury is an important cause of severe and fatal injury among people from rural areas. Its prevention requires a variety of measures under the responsibility of different actors. Behvarzes may play an important role in both injury surveillance and in identifying context-relevant means of prevention.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Knowledge is sparse concerning injuries affecting rural populations in low and middle-income countries in general and in Iran in particular. This study documents the incidence and characteristics of severe injuries affecting rural people in the Iranian district of Twiserkan and it investigates these people's suggestions for injury prevention and control.</p>", "<title>Methods</title>", "<p>An interview-based investigation was undertaken that comprised all unintentional injuries leading to hospitalization (more than 6 hours) or death that had occurred within a twelve month period and that were identified in the files of the 62 \"health houses\" of the Twiserkan district. For each case, semi-structured interviews were conducted at the households of the injured people (134 injuries affecting 117 households were identified).</p>", "<title>Results</title>", "<p>The incidence rates of fatal and non-fatal injuries were respectively 4.1 and 17.2 per 10 000 person-years and, as expected, men were more affected than women (77.6% of all injury cases). Traffic injuries (in particular among motorcyclists) were as common as home-related injuries but they were far more fatal. Among common suggestions for prevention, people mentioned that the authorities could work on the design and engineering of the infrastructure in and around the village, that the rural health workers could contribute more with local information and education and that the people themselves could consider behaving in a safer manner.</p>", "<title>Conclusion</title>", "<p>Not only domestic injuries but also those in traffic are an important cause of severe and fatal injury among rural people. Health workers may play an important role in injury surveillance and in identifying context-relevant means of prevention that they or other actors may then implement.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>FR-S has made substantial contributions to the conception and design of the study, took responsibility for and coordinated the acquisition of data, which she analyzed. She took part actively in the analysis of the data and in the writing up of the manuscript. LL, MN and MS contributed to the conception and design of the study. LL and MN were closely involved in the data collection process and took active part in the data analysis, result interpretation and manuscript writing. MS contributed to the study design, data acquisition and results interpretation. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2458/8/269/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>This study was sponsored by the Ministry of Health and Medical Education of Islamic Republic of Iran.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Places of occurrence of fatal and non-fatal unintentional injury events (June 1, 2005 – May 31, 2006).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Mechanisms of fatal and non-fatal unintentional injuries (June 1, 2005 – May 31, 2006).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Characteristics of the fatal and non-fatal unintentional injured people (June 1, 2005-May 31, 2006)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Characteristics</bold></td><td align=\"right\"><bold>Number</bold></td><td align=\"right\"><bold>%</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Sex</bold></td><td/><td/></tr><tr><td align=\"left\">Male</td><td align=\"right\">104</td><td align=\"right\">77.6</td></tr><tr><td align=\"left\">Female</td><td align=\"right\">30</td><td align=\"right\">22.4</td></tr><tr><td/><td/><td/></tr><tr><td align=\"left\"><bold>Age group (in years)</bold></td><td/><td/></tr><tr><td align=\"left\">&lt; 1</td><td align=\"right\">1</td><td align=\"right\">0.7</td></tr><tr><td align=\"left\">1–5</td><td align=\"right\">8</td><td align=\"right\">6.0</td></tr><tr><td align=\"left\">6–15</td><td align=\"right\">20</td><td align=\"right\">14.9</td></tr><tr><td align=\"left\">16–25</td><td align=\"right\">25</td><td align=\"right\">18.7</td></tr><tr><td align=\"left\">26–35</td><td align=\"right\">17</td><td align=\"right\">12.7</td></tr><tr><td align=\"left\">36–45</td><td align=\"right\">17</td><td align=\"right\">12.7</td></tr><tr><td align=\"left\">46–55</td><td align=\"right\">16</td><td align=\"right\">11.9</td></tr><tr><td align=\"left\">56–65</td><td align=\"right\">13</td><td align=\"right\">9.7</td></tr><tr><td align=\"left\">66+</td><td align=\"right\">17</td><td align=\"right\">12.7</td></tr><tr><td/><td/><td/></tr><tr><td align=\"left\"><bold>Education*</bold></td><td/><td/></tr><tr><td align=\"left\">High school graduate (high school is grades 9 to 12) or above</td><td align=\"right\">9</td><td align=\"right\">6.7</td></tr><tr><td align=\"left\">Secondary school graduate (secondary school is grades 6 to 8) and/or some high school</td><td align=\"right\">24</td><td align=\"right\">17.9</td></tr><tr><td align=\"left\">Completed primary school (primary school is up to grade 5) and/or some secondary school</td><td align=\"right\">34</td><td align=\"right\">25.4</td></tr><tr><td align=\"left\">Primary school not completed</td><td align=\"right\">20</td><td align=\"right\">14.9</td></tr><tr><td align=\"left\">No schooling</td><td align=\"right\">37</td><td align=\"right\">27.6</td></tr><tr><td align=\"left\">N/A (children under 6 years old)</td><td align=\"right\">10</td><td align=\"right\">7.5</td></tr><tr><td/><td/><td/></tr><tr><td align=\"left\"><bold>Occupation</bold></td><td/><td/></tr><tr><td align=\"left\">Farmer</td><td align=\"right\">35</td><td align=\"right\">26.1</td></tr><tr><td align=\"left\">Other self employed</td><td align=\"right\">15</td><td align=\"right\">11.2</td></tr><tr><td align=\"left\">Student</td><td align=\"right\">24</td><td align=\"right\">17.9</td></tr><tr><td align=\"left\">Housewife</td><td align=\"right\">20</td><td align=\"right\">14.9</td></tr><tr><td align=\"left\">Unemployed</td><td align=\"right\">14</td><td align=\"right\">10.4</td></tr><tr><td align=\"left\">Labourer</td><td align=\"right\">13</td><td align=\"right\">9.7</td></tr><tr><td align=\"left\">Retired</td><td align=\"right\">10</td><td align=\"right\">7.5</td></tr><tr><td align=\"left\">Other (governmental employee and conscript)</td><td align=\"right\">3</td><td align=\"right\">2.2</td></tr><tr><td align=\"left\">N/A (children under 10 years old)</td><td align=\"right\">10</td><td align=\"right\">7.5</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Consequences of the fatal and non-fatal unintentional injuries by mechanism (June 1, 2005-May 31, 2006)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Consequences</bold><break/><break/></td><td align=\"center\"><bold>All </bold><break/><bold>(n = 134)</bold><break/><bold>Number (%)</bold></td><td align=\"center\"><bold>Traffic </bold><break/><bold>(n = 60)</bold><break/><bold>Number (%)</bold></td><td align=\"center\"><bold>Fall </bold><break/><bold>(n = 35)</bold><break/><bold>Number (%)</bold></td><td align=\"center\"><bold>Burn* </bold><break/><bold>(n = 15)</bold><break/><bold>Number</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Physical nature</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Fracture</td><td align=\"center\">66 (49.3)</td><td align=\"center\">32 (53.3)</td><td align=\"center\">30 (85.7)</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Concussion</td><td align=\"center\">17 (12.7)</td><td align=\"center\">15 (25.0)</td><td align=\"center\">2 (5.7)</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Cut, bite or other open wound</td><td align=\"center\">15 (11.2)</td><td align=\"center\">4 (6.7)</td><td align=\"center\">1 (2.9)</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Burn</td><td align=\"center\">15 (11.2)</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">15</td></tr><tr><td align=\"left\">Poisoning</td><td align=\"center\">7 (5.2)</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Organ system injury/internal injury</td><td align=\"center\">6 (4.5)</td><td align=\"center\">3 (5.0)</td><td align=\"center\">-</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Bruise or superficial injury</td><td align=\"center\">4 (3.0)</td><td align=\"center\">4 (6.7)</td><td align=\"center\">-</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Sprain/strain</td><td align=\"center\">3 (2.2)</td><td align=\"center\">1 (1.7)</td><td align=\"center\">2 (5.7)</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Unspecified</td><td align=\"center\">1 (0.7)</td><td align=\"center\">1 (1.7)</td><td align=\"center\">-</td><td align=\"center\">-</td></tr><tr><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>Body region</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Head/face</td><td align=\"center\">19 (14.2)</td><td align=\"center\">17 (28.3)</td><td align=\"center\">1 (2.9)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Upper limb</td><td align=\"center\">18 (13.4)</td><td align=\"center\">5 (8.3)</td><td align=\"center\">8 (22.9)</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Lower limb</td><td align=\"center\">27 (20.1)</td><td align=\"center\">15 (25.0)</td><td align=\"center\">7 (20.0)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Neck/shoulder/lower back/rib</td><td align=\"center\">10 (7.4)</td><td align=\"center\">4 (6.7)</td><td align=\"center\">3 (8.6)</td><td align=\"center\">2</td></tr><tr><td align=\"left\">Pelvis/hip</td><td align=\"center\">8 (6.0)</td><td align=\"center\">1 (1.7)</td><td align=\"center\">7 (20.0)</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Internal system</td><td align=\"center\">7 (5.2)</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Multiple regions</td><td align=\"center\">42 (31.3)</td><td align=\"center\">16 (26.7)</td><td align=\"center\">9 (25.7)</td><td align=\"center\">10</td></tr><tr><td align=\"left\">Unspecified</td><td align=\"center\">3 (2.2)</td><td align=\"center\">2 (3.3)</td><td align=\"center\">-</td><td align=\"center\">1</td></tr><tr><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>Severity/recovery</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Complete recovery</td><td align=\"center\">54 (40.3)</td><td align=\"center\">25 (41.7)</td><td align=\"center\">16 (45.7)</td><td align=\"center\">3</td></tr><tr><td align=\"left\">Partial recovery</td><td align=\"center\">31 (23.1)</td><td align=\"center\">14 (23.3)</td><td align=\"center\">10 (28.6)</td><td align=\"center\">2</td></tr><tr><td align=\"left\">Disability</td><td align=\"center\">22 (16.4)</td><td align=\"center\">11 (18.3)</td><td align=\"center\">5 (14.3)</td><td align=\"center\">4</td></tr><tr><td align=\"left\">Death</td><td align=\"center\">26 (19.4)</td><td align=\"center\">10 (16.7)</td><td align=\"center\">4 (11.4)</td><td align=\"center\">6</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Household-based suggestions about activities that can be undertaken by various actors in order to control and prevent injuries</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>All</bold></td><td align=\"center\"><bold>Traffic</bold></td><td align=\"center\"><bold>Fall</bold></td><td align=\"center\"><bold>Burn</bold></td></tr><tr><td/><td align=\"center\">no. household=117</td><td align=\"center\">no. household=56</td><td align=\"center\">no. household=35</td><td align=\"center\">no. household=8</td></tr><tr><td align=\"left\"><bold>People's opinions/suggestions</bold></td><td align=\"center\">% suggestions <break/><italic>(% households)</italic></td><td align=\"center\">% suggestions <break/><italic>(% households)</italic></td><td align=\"center\">% suggestions <break/><italic>(% households)</italic></td><td align=\"center\">Number<break/></td></tr></thead><tbody><tr><td align=\"left\"><bold>Authorities</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Engineering/building (infrastructure/signalization/product)</td><td align=\"center\">31.1 (48.7)</td><td align=\"center\">37.4 (60.7)</td><td align=\"center\">34.8 (45.7)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Provision of daily services (accessibility/availability)</td><td align=\"center\">19.7 (30.8)</td><td align=\"center\">19.8 (32.1)</td><td align=\"center\">19.6 (25.7)</td><td align=\"center\">3</td></tr><tr><td align=\"left\">Financial support</td><td align=\"center\">12.0 (18.8)</td><td align=\"center\">7.7 (12.5)</td><td align=\"center\">23.9 (31.4)</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Enforcement</td><td align=\"center\">10.4 (16.2)</td><td align=\"center\">12.1 (19.6)</td><td align=\"center\">6.5 (8.6)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Provision of emergency services (accessibility/availability)</td><td align=\"center\">8.7 (13.7)</td><td align=\"center\">5.5 (8.9)</td><td align=\"center\">6.5 (8.6)</td><td align=\"center\">5</td></tr><tr><td align=\"left\">Education</td><td align=\"center\">8.2 (12.8)</td><td align=\"center\">8.8 (14.3)</td><td align=\"center\">4.3 (5.7)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Maintenance/repair</td><td align=\"center\">7.7 (12.0)</td><td align=\"center\">8.8 (14.3)</td><td align=\"center\">2.2 (2.9)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Other</td><td align=\"center\">2.2 (3.4)</td><td align=\"center\">-</td><td align=\"center\">2.2 (2.9)</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Total number of suggestions</td><td align=\"center\">183</td><td align=\"center\">91</td><td align=\"center\">46</td><td align=\"center\">12</td></tr><tr><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>Behvarzes</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Instruction/education/information</td><td align=\"center\">62.4 (58.1)</td><td align=\"center\">69.4 (60.7)</td><td align=\"center\">63.3 (54.3)</td><td align=\"center\">3</td></tr><tr><td align=\"left\">Post trauma care (health house/home visit)</td><td align=\"center\">16.5 (15.4)</td><td align=\"center\">14.3 (12.5)</td><td align=\"center\">13.3 (11.4)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Availability/accessibility of Behvarz all day round/drug and equipment of health house</td><td align=\"center\">10.1 (9.4)</td><td align=\"center\">10.2 (8.9)</td><td align=\"center\">10.0 (8.6)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">As good as it can be</td><td align=\"center\">9.2 (8.5)</td><td align=\"center\">4.1 (3.6)</td><td align=\"center\">13.3 (11.4)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Other (Insist on of financial support by authorities and pass people's problem to them)</td><td align=\"center\">1.8 (1.7)</td><td align=\"center\">2.0 (1.8)</td><td align=\"center\">-</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Total number of suggestions</td><td align=\"center\">109</td><td align=\"center\">49</td><td align=\"center\">30</td><td align=\"center\">7</td></tr><tr><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>People</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Behave in a safe manner</td><td align=\"center\">34.1 (37.6)</td><td align=\"center\">35 (37.5)</td><td align=\"center\">30.0 (34.3)</td><td align=\"center\">3</td></tr><tr><td align=\"left\">Cooperation together</td><td align=\"center\">18.6 (20.5)</td><td align=\"center\">13.3 (14.3)</td><td align=\"center\">20.0 (22.9)</td><td align=\"center\">3</td></tr><tr><td align=\"left\">Cooperation (with authorities)</td><td align=\"center\">17.8 (19.7)</td><td align=\"center\">20.0 (21.4)</td><td align=\"center\">20.0 (22.9)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Compliance</td><td align=\"center\">12.4 (13.7)</td><td align=\"center\">18.3 (19.6)</td><td align=\"center\">7.5 (8.6)</td><td align=\"center\">-</td></tr><tr><td align=\"left\">Engineering/building</td><td align=\"center\">8.5 (9.4)</td><td align=\"center\">3.3 (3.6)</td><td align=\"center\">17.5 (20.0)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Education (pay attention to planned education for them and/or pass to their children)</td><td align=\"center\">8.5 (9.4)</td><td align=\"center\">10.0 (10.7)</td><td align=\"center\">5 (5.7)</td><td align=\"center\">1</td></tr><tr><td align=\"left\">Total number of suggestions</td><td align=\"center\">129</td><td align=\"center\">60</td><td align=\"center\">40</td><td align=\"center\">9</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>*In decreasing order according to the Iranian system.</p></table-wrap-foot>", "<table-wrap-foot><p>*As the number of burns amounted to 15, no percentages are presented.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2458-8-269-1\"/>", "<graphic xlink:href=\"1471-2458-8-269-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>It has been evident for decades that there is a strong association between alcohol use and risk of tuberculosis (TB). Prevalence of alcohol use disorders among TB patients have ranged from 10% to 50% in studies carried out in Australia, Canada, Russia, Switzerland, and the USA [##REF##14217248##1##, ####REF##5493339##2##, ##REF##982112##3##, ##REF##1928942##4##, ##REF##2786330##5##, ##REF##17398230##6##, ##REF##16704041##7####16704041##7##]. Similar evidence of a strong link emerges from studies in which population groups with high prevalence of alcohol use disorders have been screened for TB. Jones et al[##REF##13174381##8##] found that the prevalence of active pulmonary TB among social service clients (among whom alcohol use disorders was the main problem) in the USA in the 1950s was 55 times the prevalence of the general population (2,220/100,000 vs. 40/100,000). Friedman et al[##REF##3674581##9##] reported a 46 times higher prevalence among people with alcohol use disorders (who did not abuse other drugs) in New York in the early 1980s (1,500/100,000 vs. 32/100,000). In a cohort of persons with alcohol use disorders who were followed prospectively for 8 years, the TB incidence was 464/100,000 person-years, which was 9 times the age-matched incidence among the general population in New York[##REF##8596549##10##]. However, these studies did not control for potential confounders.</p>", "<p>Possible causal pathways include specific social mixing patterns among people with alcohol use disorders, leading to higher risk of infection [##REF##10325918##11##, ####REF##14723363##12##, ##REF##11825968##13####11825968##13##], or weakened immune system leading to higher risk of break down from infection to TB disease. The latter may be through direct toxic effects of alcohol on the immune system [##UREF##0##14##, ####REF##15706761##15##, ##UREF##1##16##, ##REF##15102763##17##, ##UREF##2##18####2##18##], or indirectly through micro- and macronutrient deficiency[##REF##15139466##19##], or other alcohol-related medical conditions such as malignancies[##UREF##3##20##] and depression[##UREF##4##21##,##REF##17804063##22##].</p>", "<p>This paper reviews analytical epidemiological studies with individual-level data on alcohol exposure and TB disease status, with the aim to determine if there is a likely causal association between alcohol use and risk of TB disease. The paper also attempts to estimate the strength of such an association.</p>" ]

[ "<title>Methods</title>", "<title>Inclusion criteria</title>", "<p>The review included case-control and cohort studies that reported individual level data on alcohol exposure (amount of alcohol intake or a clinical diagnosis of an alcohol use disorder) and active TB disease, and which reported either crude or adjusted odds ratio, or crude data from which odds ratios could be calculated.</p>", "<title>Search strategy</title>", "<p>Initially, all 16,527 articles in a comprehensive private collection of scientific tuberculosis publications (compiled by Dr Hans Rieder) of which a copy is kept at the Stop TB Department at the World Health Organization, were screened using Reference Manager™, with the keywords \"alcohol\" or \"alcoholism\". Next, PubMed was searched using the keywords \"alcohol OR alcoholism AND tuberculosis\", which revealed a total of 2,007 abstracts. Titles were initially screened, followed by screening of abstracts. In addition, we screened a report of a systematic review of the association between smoking and tuberculosis[##UREF##5##23##], which included detailed information about all covariates that were analysed in 50 reviewed studies. All studies in which alcohol was a listed covariate were reviewed in detail. Finally, the reference list of all reviewed articles were screened.</p>", "<title>Study assessments and analysis</title>", "<p>A total of 21 studies [##REF##10490003##24##, ####UREF##6##25##, ##REF##15875929##26##, ##UREF##7##27##, ##REF##14044850##28##, ##REF##1546917##29##, ##REF##7977912##30##, ##REF##8844044##31##, ##REF##10206497##32##, ##UREF##8##33##, ##REF##12733370##34##, ##REF##11829102##35##, ##REF##12365575##36##, ##REF##15139448##37##, ##REF##15860066##38##, ##REF##15914505##39##, ##REF##15921928##40##, ##REF##15801638##41##, ##UREF##9##42##, ##REF##16466042##43##, ##REF##16339219##44##, ##REF##17705978##45####17705978##45##] fulfilled the inclusion criteria and were further assessed with regards to setting, inclusion criteria of study subjects, definition of exposure and outcome, mechanisms for ascertainment of exposure and outcome, and confounders controlled for (table ##TAB##0##1##).</p>", "<p>The studies were initially grouped in three categories with regards to exposure level. The low-exposure category (4 studies) included those studies that defined exposure as alcohol use above a cut-off point that was set at a level below 40 g (or 50 ml) alcohol per day. This is the upper cut-off point for low-risk (for chronic harm) alcohol consumption for men [##UREF##10##46##]. The high-exposure category (5 studies) included studies that defined exposure as alcohol consumption above a cut-off set at a level above 40 g per day. The third category included 6 studies that had ascertained a diagnosis of alcohol use disorder from medical records. None of the studies included details about any ICD classification used for the diagnosis. Therefore, this categorization is imprecise and does not allow for further subgroup analysis with regard to alcohol use disorders. One study[##UREF##7##27##] included data which allowed calculation of crude odds ratio for both alcohol consumption above 40 g/day, and for consumption between 10 and 40 g/day. The data were included in the low- and high-exposure categories respectively (labelled Brown I and Brown II respectively). Seven studies that did not report how exposure had been defined were excluded from this categorization.</p>", "<p>Adjusted odds ratios for the odds of active TB disease among people with a particular level/type of alcohol exposure vs. no such exposure were extracted from the original papers. If no adjusted odds ratio (results reported from multivariate or stratified analyses) were reported, crude odds ratios were either extracted, or calculated from absolute numbers reported in the paper. Heterogeneity was assessed using Cochrane's Q statistic and the I² statistic which estimates the percentage of the total variation across studies that is due to heterogeneity rather than chance [##REF##12958120##47##]. Pooled effect sizes were calculated using both fixed and random effect models for each sub-category of studies.</p>", "<p>The pooled effect size across the five high-exposure category studies did not differ significantly from the pooled effect size across the six studies that had ascertained a diagnosis of an alcohol use disorder. Therefore, these two categories were combined into one high-exposure/alcohol use disorder category (herewith termed \"high-exposure category\") for the further analysis. Those 11 studies were further grouped with regards to which constellation of confounders had been controlled for (age, sex, HIV, smoking, socioeconomic status (SES), and infection status), and according to types of TB studied (table ##TAB##1##2##). None of the studies reported disaggregated analysis by type of TB. Therefore, further subgroup analyses with regards to type of TB was not possible. In Table ##TAB##1##2## we have reported the results of models with fixed and random effects but we will refer to the results of the random effects models in the following discussion. Although the random effects model gives slightly higher estimates of the effect sizes than does the fixed effects model it also gives wider confidence limits and the confidence limits for the latter are always contained within the confidence limits for the former.</p>", "<p>Funnel plots (log odds ratio plotted against the standard error of the log odds ratio for each study) were constructed to examine potential publication bias. Publication bias was suspected if relatively few studies with high standard errors and odds ratios close to one were identified[##UREF##11##48##]. Since publication bias was suggested by the funnel plots, we excluded the three studies with the highest standard errors for the analysis of pooled effect across studies in the high exposure category.</p>" ]

[ "<title>Results</title>", "<p>Among the 21 reviewed studies, three studies were cohort studies and 18 were case control studies. Table ##TAB##0##1## summarizes the characteristics of these studies. Figure ##FIG##0##1## displays effect sizes in the 21 studies.</p>", "<p>The pooled odds ratio across the 11 studies in the high-exposure category was 3.50 (95% CI: 2.01–5.93). The pooled odds ration across the four studies in the low-exposure category was (1.08, 95% CI: 0.82–1.40) (Table ##TAB##1##2##).</p>", "<p>Funnel plots indicated that there was under-representation of small studies with weak or absent association, both for all studies combined, as well as for the studies in the high-exposure category (figure ##FIG##1##2##).</p>", "<p>After exclusion of the three studies that had the highest standard error, because of suspected publication bias, the pooled effect sizes for studies in the high-exposure category was 2.94 (95% CI 1.89–4.59). There was significant heterogeneity across these studies. When further excluding the two studies with the highest and lowest effect sizes respectively (Brown I and Kim), there was no heterogeneity and the pooled effect size was 2.76 (95% CI 2.09–3.64).</p>", "<p>Studies that included only pulmonary TB cases had higher pooled odds ratio than studies that included all types of TB. The difference was of borderline statistical significance (OR 4.16, 95% CI: 2.99–5.80 vs. 2.55, 95% CI: 2.02–3.23). After excluding the three smallest studies, the difference between the pooled odds ratio for these two categories decreased and was not statistically significant (3.67, 95% CI: 2.58–5.22, vs. 2.52, 95% CI: 1.98–3.19, table ##TAB##1##2##). Studies in the high-exposure category that had controlled for different sets of important confounders had similar or higher, but not significantly different, pooled effect size compared to all studies in this category combined (table ##TAB##1##2##).</p>" ]

[ "<title>Discussion</title>", "<p>This review suggest that low to moderate alcohol intake is not associated with increased risk of TB disease. However, there seem to be a substantial risk increase among people who drink more than 40 g alcohol per day, and/or have an alcohol use disorder. The pooled effect size across studies in the high-exposure category was 2.94 (95% CI 1.89–4.59) after excluding the three studies with largest standard errors in order to make a crude adjustment for the suspected publication bias. There was a tendency that studies that included only pulmonary TB cases reported higher odds ratios than studies that included all types of TB, but the difference was not statistically significant when small studies had been excluded to adjust for possible publication bias.</p>", "<p>The original heterogeneity across all 21 studies decreased after subdividing studies into low and high-exposure level studies. However, there was remaining significant heterogeneity in the high-exposure category, which we could not explain through further subgroup analysis. Varying degree of misclassification of exposure across the studies may explain some of the heterogeneity. However, there was insufficient information in the reviewed studies to explore this further. Underestimation of level of alcohol intake may have biased the results in several studies. It is reasonable to assume that underestimation of alcohol intake by study subjects was either non-differential, or more pronounced among the cases in the case-control studies. In either case this would have led to an underestimation of the risk increase.</p>", "<p>Bias caused by different approaches for the selection of controls in the case control studies may also have contributed to the heterogeneity. Several of the case control studies used hospital controls or controls recruited among other groups, such as prisoners and social service clients, that are likely to have higher alcohol intake levels than the general population. This may have biased, to various degrees across the studies, the odds ratios towards one.</p>", "<p>Studies that had controlled for potential confounding effects (either by design or in the analysis) of important factors such as age, sex, HIV, some measure of socioeconomic status, and smoking, had similar or somewhat higher, but not significantly different pooled effect sizes. The degree to which important confounders were controlled for varied considerably across studies. There might have been residual confounding that could have biased the pooled estimate across the studies. Socioeconomic status is difficult to measure and fully control for. Furthermore, there are some risk factors for TB disease that have not been assessed in most of the reviewed studies. For example, malnutrition[##REF##15139466##19##], diabetes[##REF##17822539##49##] and indoor air pollution[##REF##17227135##50##] respectively are associated with higher risk of TB disease. Mental health disorders may also be associated with higher risk of TB through impact on the immune system[##UREF##4##21##,##REF##17804063##22##]. However, confounding effect of these factors would have to be of considerable magnitude to offset the relatively strong association found in this review. Moreover, it may not be correct to control for some factors, since they may be on the causal pathway. For example, alcohol use disorders can lead to social downward drift and it can cause or contribute to malnutrition. Other known risk factors for TB such as silicosis, malignancies and immunosuppressant treatment are probably too rare to have influenced the results significantly.</p>", "<p>The pooled effect size across high-exposure studies that had controlled for infection status (OR 4.21, 95% CI: 2.73–6.48), suggest that one possible causal pathway through which alcohol operates as a risk factor for TB, is through increased risk of progression from infection to disease. It is somewhat surprising that this pooled effect size is larger than for the pooled effect size across all studies. However, the confidence interval is wide, and overlaps that of the pooled effect of the other studies. Furthermore, it is possible that part of the risk increase is due to increased risk of re-infection, since none of the studies was designed to distinguish re-infection from re-activation. Four studies in the high-exposure category were designed as cohort studies, or nested case control studies, in a way that allowed controlling for infection status. All four studies defined exposure as an alcohol use disorder noted in medical records in a way that allowed ascertaining the temporal sequence between exposure and outcome. All four studies reported adjusted risk ratios controlled for age, sex, some indicator of socioeconomic status. Three of these studies also controlled for smoking and three controlled for HIV. Two of the studies used recurrent TB as study outcome, and both controlled for type of treatment, treatment duration, and adherence.</p>", "<p>Alcohol may assert a direct toxic effects on the immune system rendering the host more susceptible to TB disease. Animal studies suggest that cell mediated immunity and macrophage functions (which are essential for the host response to <italic>M. tuberculosis </italic>infection) are directly impaired by chronic and acute alcohol consumption[##UREF##0##14##,##REF##15706761##15##]. One mechanisms may be through inhibited tumour necrosis factor (TNF) response[##UREF##2##18##]. Alcohol may also reduce the NO system response to mycobacterial infection, which may prevent the destruction of mycobacteria. Furthermore, at least in mice, alcohol can inhibit granuloma formation, IL-2 production, IFN-gamma production, and CD4⁺proliferation[##REF##15102763##17##]. Alcohol use disorders may also cause impaired immunity indirectly through micro- and macronutrient deficiency, or through other alcohol-related disorders such as malignancies[##UREF##3##20##].</p>", "<p>The association between alcohol use and TB could also be explained by specific social mixing patterns, which may increase the risk of exposure to people with infectious TB disease in settings such as bars, shelters for homeless, prisons, and social institutions. This is supported by a few molecular-epidemiological studies. A study in a high incidence areas of Western Cape Province, South Africa, 1993–1996, suggested that most of the TB transmission took place outside the households, and found that 58% the identified contacts outside the household took place while drinking in social groups[##REF##10325918##11##]. Zolnir-Dovc[##REF##14723363##12##] demonstrated an increased risk of belonging to a TB cluster among people with alcohol use disorders, indicating increased risk of recent transmission. Diel et al[##REF##11825968##13##] investigated a TB outbreak in Hamburg, Germany during 1997–2002. They demonstrated that transmission between people who were socialising in a specific bar was an important factor behind the dissemination and perpetuation of the outbreak. The increased risk of TB transmission in prisons has been well established[##REF##15906874##51##]. The prevalence of TB among people in social service institutions have been found to be very high [##REF##13174381##8##, ####REF##3674581##9##, ##REF##8596549##10####8596549##10##].</p>", "<p>The strength of the association is likely to vary between settings, due to varied social context of alcohol use and different mix of other risk factors that could modify the effect of alcohol use. The random effect model for pooling effect sizes may be appropriate in this context, since it assumes an underlying variation of the true effect size across different settings. However, this variation may not be random, and we cannot draw conclusions about which factors might modify the effects in a systematic way. Therefore, generalization to a specific setting, even based on the confidence limits of the random effect model, should be done with caution.</p>", "<p>Nevertheless, the pooled effect size can be used to obtain an indicative estimate of population level importance of alcohol use as risk factors for TB disease. There is a huge variation in prevalence of drinking more than 40 gram alcohol per day (for men, and &gt; 20 g per day for women) across the world, ranging from 0.1% in parts of the Eastern Mediterranean Region to 18.6% in parts of Eastern Europe [##UREF##10##46##]. The population attributable fraction can be calculated from these prevalence estimates and from the odds ratio obtained in the present study (2.9, 95% CI: 1.9–4.6): It ranges from close to zero in parts of the Eastern Mediterranean Region to more than 30% in parts of Europe.</p>", "<p>Future research on the association between alcohol use and risk of TB should carefully assess both potential confounding effects and interaction between alcohol use and other TB risk factors. Possible difference in the risk of pulmonary vs. non-pulmonary TB should also be investigated. Furthermore, there is a need to better understand the possible causal pathways with regards to risk of infection and risk of break down from infection to disease.</p>" ]

[ "<title>Conclusion</title>", "<p>There is a three-fold risk increase of active TB associated with consumption of more than 40 g alcohol per day, and/or having an alcohol use disorder. This could be due to both increased risk of infection related to specific social mixing patterns associated with alcohol use, as well as influence on the immune system of alcohol itself and of alcohol related conditions. These findings have implications for TB control strategies globally, particularly in countries where a high proportion of TB can be attributed to alcohol use.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>It has long been evident that there is an association between alcohol use and risk of tuberculosis. It has not been established to what extent this association is confounded by social and other factors related to alcohol use. Nor has the strength of the association been established. The objective of this study was to systematically review the available evidence on the association between alcohol use and the risk of tuberculosis.</p>", "<title>Methods</title>", "<p>Based on a systematic literature review, we identified 3 cohort and 18 case control studies. These were further categorized according to definition of exposure, type of tuberculosis used as study outcome, and confounders controlled for. Pooled effect sizes were obtained for each sub-category of studies.</p>", "<title>Results</title>", "<p>The pooled relative risk across all studies that used an exposure cut-off level set at 40 g alcohol per day or above, or defined exposure as a clinical diagnosis of an alcohol use disorder, was 3.50 (95% CI: 2.01–5.93). After exclusion of small studies, because of suspected publication bias, the pooled relative risk was 2.94 (95% CI: 1.89–4.59). Subgroup analyses of studies that had controlled for various sets of confounders did not give significantly different results and did not explain the significant heterogeneity that was found across the studies.</p>", "<title>Conclusion</title>", "<p>The risk of active tuberculosis is substantially elevated in people who drink more than 40 g alcohol per day, and/or have an alcohol use disorder. This may be due to both increased risk of infection related to specific social mixing patterns associated with alcohol use, as well as influence on the immune system of alcohol itself and of alcohol related conditions.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests. Four of the authors (KL, BW, EJ and CD) are staff members of the World Health Organization. The authors alone are responsible for the views expressed in this publication and they do not necessarily represent the decisions or policies of the World Health Organization. No external funding was provided for this research.</p>", "<title>Authors' contributions</title>", "<p>All authors contributed to the conceptualisation of the paper. KL and SS did the initial review, the selection of abstracts, and the identification of papers to be included in the final review. All authors contributed to the assessment of papers. BGW and KL did the statistical analysis. All authors reviewed the results of the analysis. KL drafted the manuscript, and all authors contributed to its completion. KL is the guarantor.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2458/8/289/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We are grateful to Vladimir Poznyak and Dag Rekve for their comments on early drafts of the manuscript.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Forest plot of all 21 studies.</bold> Bars indicate 95% confidence interval. Filled squares represent point estimate for studies in the high exposure/alcoholism category, white squares represent studies in the low exposure category, and grey circles studies that did not report a well-defined exposure level.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Funnel plot of the odds-ratio against the precision of the estimates.</bold> Points to the right of the dashed line are significant at the 5% level. The apex of the funnel gives the point estimate. Points outside the funnel differ from the point estimate at the 5% level and suggest heterogeneity in the estimates. If there is no bias in the selection of studies for publication, the points should be evenly scattered to the left and right. Squares represent the three studies with largest standard error that were excluded in the category \"Excluding three smallest studies\" in table 2 (Mori et al 1992, Spletter 2000, and Riekstina et al 2005). The two filled circles that are outside the funnel represent the two additional studies that were excluded in the category \"Excluding three smallest and Brown I and Kim\" in table 2</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Summary of study characteristics</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\" colspan=\"14\"><bold><italic>Cohort studies</italic></bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>First Author, Year, setting</bold></td><td/><td align=\"left\"><bold>Cohort</bold></td><td/><td align=\"left\"><bold>Outcome measure</bold></td><td/><td align=\"left\"><bold>Exposure Measure</bold></td><td/><td align=\"left\"><bold>Confounders Controlled for</bold></td><td/><td align=\"left\"><bold>Effect size (95% confidence interval)</bold></td><td/><td align=\"left\"><bold>Comments</bold></td></tr><tr><td colspan=\"14\"><hr/></td></tr><tr><td align=\"left\">Hemilä et al, 1999, Finland, 198–1993</td><td/><td align=\"left\">26,975 male smokers participating in RCT on the effect of nutritional support with a-tocopherol, P-carotene, or a-tocopherol + P-carotene for cancer prevention</td><td/><td align=\"left\">Clinical diagnoses of TB ascertained from the discharge register of hospitals. 167 incident cases of TB registered from 1985 to 1993.</td><td/><td align=\"left\">Self reported at baseline. Alcohol use categorized as 30 gram alcohol per day or more.</td><td/><td align=\"left\">Age, BMI, martial status, education, residential neighbourhood, smoking, nutritional intervention</td><td/><td align=\"left\">Adjusted relative risk: 1.03 (95% CI: 0.70–1.53)</td><td/><td align=\"left\">Eight years follow up and change in drinking pattern not ascertained.<break/>Prevalence of exposure among controls: 20%</td></tr><tr><td align=\"left\">Moran-Mendoza, 2004, British Columbia, Canada, 1990–2000</td><td/><td align=\"left\">33,146 contacts of active TB cases recorded in division of disease control 1990–2000, who had a TST performed, excluding those with TB history and those with HIV, followed until 2001</td><td/><td align=\"left\">Any type of TB, registered in the division of TB control database. 228 active cases identified.</td><td/><td align=\"left\">Alcoholism as noted in medical record</td><td/><td align=\"left\">Age, sex, Canadian-born, aboriginal, DM, malnutrition, malignancy, immunosuppressant treatment, BCG, no of contacts, type of contact, TST size, SES (geographical location), latent TB treatment, intravenous drug use, recent arrival from high TB incidence country</td><td/><td align=\"left\">Adjusted relative risk: 2.9 (1.3–6.5)</td><td/><td align=\"left\">Entire study population are TB infected. RR reflect risk of progress to active disease.<break/>Prevalence of alcoholism among whole cohort: 0.8%</td></tr><tr><td align=\"left\">Thomas et al 2005, Tiruvallur district, Tamil Nadu, India, 2000–2001</td><td/><td align=\"left\">503 cured new smear positive pulmonary patients as per TB district register, followed prospectively</td><td/><td align=\"left\">TB recurrence within 18 months (62 recurrencess recorded)</td><td/><td align=\"left\">Self reported during initial treatment. Exposure was \"Habitual drinking\", which was not defined in terms of amounts or frequency</td><td/><td align=\"left\">Adjusted OR from multivariate analysis not reported.<break/>Factors accounted for were sex, age, occupation, education, smoking, adherence, drug sensitivity, smear conversion, initial weight</td><td/><td align=\"left\">Crude relative risk: 2.3 (1.3–4.1)</td><td/><td align=\"left\">Level of exposure not provided, but since the prevalence of exposure of \"habitual drinking\" in the cohort was 33% in this rural Indian district, it not likely to correspond to high level consumption.</td></tr><tr><td colspan=\"14\"><hr/></td></tr><tr><td align=\"left\" colspan=\"14\"><bold><italic>Case control studies</italic></bold></td></tr><tr><td colspan=\"14\"><hr/></td></tr><tr><td align=\"left\"><bold>Author (Year), Setting</bold></td><td/><td align=\"left\"><bold>Cases and controls</bold></td><td/><td align=\"left\"><bold>Exposure Measure</bold></td><td/><td align=\"left\"><bold>Confounders Controlled for</bold></td><td/><td align=\"left\"><bold>Effect size (95% confidence interval)</bold></td><td/><td align=\"left\"><bold>Comments</bold></td><td/></tr><tr><td colspan=\"14\"><hr/></td></tr><tr><td align=\"left\">Brown and Campbell. 1961, Hospital for ex-servicemen, Victoria, Australia, 1950s</td><td/><td align=\"left\"><underline>Cases</underline> (100): All consecutive new admissions<break/><underline>Controls</underline> (100): Randomly selected from surgical ward in same hospital (excluding orthopedic cases)</td><td/><td align=\"left\">Self reported daily consumption<break/>Moderate to heavy drinking defined as 26 ml alcohol per day or more. Crude numbers for different level of exposure were reported, allowing calculation of association also at the &gt; 50 ml (40 g) and other cut-off points.</td><td/><td align=\"left\">Stratified by smoking status. All subjects were men. All ex-army staff in the age bracket 20–70. Age distribution very similar between cases and control. Pre- HIV era</td><td/><td align=\"left\">Crude OR of moderate to heavy alcohol vs. none/low: 4.88 (95% CI: 2.59–9.24)<break/>For &gt; 50 ml vs =&lt; 50 ml: OR 8.18 (4.05–16.53)<break/>For 1–50 ml vs. none: 1.98 (0.89–4.43)<break/>Significant (p &lt; 0.0001) dose response relationship:<break/><underline>OR</underline><break/>0 (reference) 1.00<break/>10–25 ml/day: 1.66<break/>26–50 ml/day: 2.38<break/>51–75 ml/day: 9.27<break/>76–100 ml/day: 8.50<break/>101–125 ml/day: 27.82<break/>126- ml/day: 43.27</td><td/><td align=\"left\">OR not analysed in original study. The ORs reported here are calculated based on crude data reported in the paper<break/>Smoking possibly effect modifier. Stratified for none smokers and smokers respectively (any alcohol vs. no alcohol):<break/>Non smokers: 2.25 (0.54–9.86)<break/>Smokers: 5.22 (1.83–15.61)<break/>Prevalence of \"moderate to heavy alcohol intake\" in controls: 39%</td><td/></tr><tr><td align=\"left\">Lewis and Chamberlain, 1963, Hospital, London, 1962</td><td/><td align=\"left\"><underline>Cases</underline> (100): Male, active cases of pulmonary TB<break/><underline>Controls</underline> (200): Matched for age and social class: A (100): From medical and surgical wards at the same hospital. B (100): From emergency department at another, general, hospital</td><td/><td align=\"left\">Self-reported average daily consumption 6 months before symptoms started<break/>\"Regular drinking\" defined as the equivalent of 2 or more pints per day.</td><td/><td align=\"left\">Only men, stratified by age, social class, marital status and smoking. Pre-HIV era.</td><td/><td align=\"left\">Crude OR for regular drinkers vs. not regular drinker 2.64 (95% CI: 1.50-4-66)<break/>Did not change when stratified for smoking status: OR 2.68 and 2.61 in respective stratum</td><td/><td align=\"left\">OR not analysed in original study.<break/>Social class effect modifier? Stratified for SES:<break/>Class I-II: OR = 1.16 (0.42–3.22)<break/>Class III-V: OR = 4.07 (1.98–8.41)<break/>Prevalence of \"regular drinkers\" among controls: 19.5%<break/>UK pint = 568 ml. 2 pints of 5% beer contains about 45 g alcohol</td><td/></tr><tr><td align=\"left\">Mori et al, 1992, Indian Health Service hospital, Pine Ridge Reservation, South Dakota, USA</td><td/><td align=\"left\"><underline>Cases</underline> (46): All new, active, adult (18 years and above), cases registered between 1983–1989.<break/><underline>Controls</underline> (46). Randomly selected, matched for age and residence, from health care register in Reservation, where all residents are included</td><td/><td align=\"left\">Chart review: Alcohol abuse/alcoholism listed in medical record, or alcohol related admission within 10 years or outpatient visit within 5 years</td><td/><td align=\"left\">Matched by age and residence.<break/>OR adjusted for sex, isoniazid profylaxis, and diabetes<break/>All study subjects from same Indian community.</td><td/><td align=\"left\">Adjusted OR (AOR) for alcohol abuse vs. no alcohol abuse: 3.8 (1.15–12.3)</td><td/><td align=\"left\">Prevalence in control group: 32%</td><td/></tr><tr><td align=\"left\">Buskin, et al, 1994, Seattle, King County Tuberculosis Clinic, Washington State, 1988–1990</td><td/><td align=\"left\"><underline>Cases</underline> (151): Active TB cases, aged &gt; 17 registered at TB clinic 1988–1990<break/><underline>Controls</underline> (545): Individuals seeking care at the clinic, but no TB diagnosed</td><td/><td align=\"left\">Self reported frequency of drinking and amount consumed.<break/>Heavy drinkers defined as 3 or more drinks/day or more than 5 drinks on average on each drinking occasion.</td><td/><td align=\"left\">OR adjusted for age and smoking.<break/>Sex, SES, BMI, and race were analysed, but did not influence result</td><td/><td align=\"left\">Adjusted OR heavy drinking vs. non-drinkers 2.0 (95% CI: 1.1; 3.7)</td><td/><td align=\"left\">1 US standard drink is 14 gram, thus 3 standard drinks is 42 gram.<break/>Prevalence of heavy drinking in control group: 12.5%</td><td/></tr><tr><td align=\"left\">Rosenman et al,1996, New Jersey, USA, 1985–1987</td><td/><td align=\"left\"><underline>Cases</underline> (148): All active male, HIV-negative, cases over age of 35, born in USA, notified 1985–87<break/><underline>Controls</underline> (290): From Medicade finance administration files, matched for age and race</td><td/><td align=\"left\">Self reported. \"Heavy drinking\" defined as &gt; 22 alcohol equivalents/week</td><td/><td align=\"left\">Only HIV- men in study, controls matched for age and race. Alcohol association not controlled for other variables in study, since alcohol was treated purely as confounder</td><td/><td align=\"left\">Crude OR: 3.33 (1.99.5.59)</td><td/><td align=\"left\">Prevalence \"heavy drinkers\" among controls: 14% 1 US standard drink is 14 gram, thus &gt; 22 drinks per week = &gt; 44 grams per day</td><td/></tr><tr><td align=\"left\">Schluger et al, 1999, Social services agencies and chest clinic, NY, USA. 1994–1997</td><td/><td align=\"left\"><underline>Cases</underline> (20): Persons screened positive for active TB among 3,828 individuals seeking social services<break/><underline>Controls</underline> (3,245): Those not screened positive for active TB</td><td/><td align=\"left\">Self reported \"moderate to heavy alcohol use\". This was not defined further</td><td/><td align=\"left\">None, but all subjects are social service clients</td><td/><td align=\"left\">Crude OR 2.38 (0.88–6.58)</td><td/><td align=\"left\">The authors did analyse, the study as a case control study.<break/>Considering that the subjects were all social service clients and alcohol problem was common in this group, it can be assumed that \"moderate to heavy\" correspond to at least 40 g per day and/or alcohol abuse<break/>Prevalence among controls: 43%</td><td/></tr><tr><td align=\"left\">Spletter, 2000, TB Control Clinic, Phoenix, Ariziona, USA, 1993–1999</td><td/><td align=\"left\"><underline>Cases</underline> (43): active pulmonary TB, 25–64 years old, excluding refugees, HIV positive, and comorbidity such as gastrectomy, jejunuilial bypass, DM, silicosis, renal failure, immunosuppressive treatment, malignancies.<break/><underline>Controls</underline> (258): Patients infected with <italic>M.tuberculosis</italic>, but active disease ruled out.</td><td/><td align=\"left\">Medical record review: Heavy drinking defined as those with chart entries indicating alcohol abuse or alcohol history recorded as \"heavy drinking\"</td><td/><td align=\"left\">See list of exclusion criteria. Controlled for age, sex, smoking, race, US born, high risk residence, illicit drug use.</td><td/><td align=\"left\">Adjusted OR for heavy alcohol use vs. no heavy alcohol use: 6.1 (1.4; 26.2):</td><td/><td align=\"left\">Entire study population are TB infected. OR reflect risk of progress to active disease.<break/>Prevalence of heavy alcohol consumption in controls: 2.3%</td><td/></tr><tr><td align=\"left\">Dong et al, 2001, 12 communes in Chengdu, China, 1996–97</td><td/><td align=\"left\"><underline>Cases</underline> (174): All active TB cases recorded between March 1996 and March 1997<break/><underline>Controls</underline> (174): Random sample from community (population registry), matched for age, sex, and place of residence</td><td/><td align=\"left\">Self reported use.<break/>Definition of alcohol use or amounts not reported.</td><td/><td align=\"left\">Matched for age and sex and district. Smoking, crowding, darkness in dwelling, air-pollution and BMI are reported variables, but not reported what was actually controlled for in the logistic regression</td><td/><td align=\"left\">Adjusted OR (alcohol vs no alcohol): 1.76 (0.90–3.42)</td><td/><td/></tr><tr><td align=\"left\">Tocque et al 2001, Liverpool, UK, 1989–1996</td><td/><td align=\"left\"><underline>Cases</underline> (112): All notified in the city<break/><underline>Controls</underline> (198): From Liverpool general practitioner database, matched for sex, age and residential area</td><td/><td align=\"left\">Self reported, high consumption defined as &gt; 30 units per week (&gt; 4.3/day), both at time of interview and 2 years prior to diagnosis</td><td/><td align=\"left\">Matched for age, sex, and residence area<break/>Alcohol not included in multivariate analysis</td><td/><td align=\"left\">Crude OR for drinkers vs. non-drinkers: 1.01 (0.67–1.70, at 2 years before diagnosis</td><td/><td align=\"left\">One UK alcohol unit is 8 gram, thus 4.3 units/day = 34 gram</td><td/></tr><tr><td align=\"left\">Tekkel et al, 2002, Hospital, Tallinn, Estonia, 1999–2000</td><td/><td align=\"left\"><underline>Cases</underline> (248): consecutive, incident pulmonary TB cases admitted to one hospital in Tallinn<break/><underline>Controls</underline> (248): From population registry, matched for age, sex, and country of residence.</td><td/><td align=\"left\">Self reported frequency of drinking during last year. Not defined in amounts of alcohol</td><td/><td align=\"left\">Age, sex, and country of residence matched for. OR adjusted for smoking, drug abuse, nutrition, weight loss, contact with TB, place of birth, marital status, and education</td><td/><td align=\"left\">Adjusted OR for people who consumed alcohol several times a week/day vs. rarely: 13.63 (4.63–40.10);</td><td/><td align=\"left\">Prevalence of alcohol consumptions several times per week: 7.3%</td><td/></tr><tr><td align=\"left\">Crampin et al, 2004, Karonga district, Malawi, 1996–2001</td><td/><td align=\"left\"><underline>Cases</underline> (598):All new TB cases, aged &gt; 15, residing in district<break/><underline>Controls</underline> (992): Random sample from community register, matched for age, sex and areas of residence.</td><td/><td align=\"left\">Self reported as current (1/week or &lt; 1/week), past, or never</td><td/><td align=\"left\">Matched for age, sex, area of residence. Adjusted for SES, HIV, TB contacts, BCG</td><td/><td align=\"left\">Adjusted OR for current 1/w vs. never: 0.9 (0.5–1.7)</td><td/><td align=\"left\">Prevalence of drinking 1/week among controls: 11%</td><td/></tr><tr><td align=\"left\">Kim and Crittenden, 2005, County Prison, USA, 1992–1998</td><td/><td align=\"left\"><underline>Cases</underline> (441): All inmates screened positive for active TB 1992–1998<break/><underline>Control</underline> (478): Sex matched, random sample from prison pop.</td><td/><td align=\"left\">Alcohol abuse as recorded in prison health record</td><td/><td align=\"left\">Sex, age, ethnicity, marital status, education, homelessness, IV drug use, HIV, length of stay in prison, type of crime.</td><td/><td align=\"left\">Adjusted OR for alcohol abuse vs. no alcohol abuse: 1.59 (p &lt; 0.01, no confidence interval reported)</td><td/><td align=\"left\">Prevalence of alcohol abuse among controls: 40.2%</td><td/></tr><tr><td align=\"left\">Lienhardt et al 2005, Multicenter, Guinée, Guniea Bissau, and The Gambia, 1999–2001</td><td/><td align=\"left\"><underline>Cases</underline> (687): Newly detected smear positive TB<break/><underline>Controls</underline>: For each case: A (687): Age-matched household control, and: B (687): Residence area matched community control</td><td/><td align=\"left\">Self reported as never/past/current</td><td/><td align=\"left\">A large set of host related and environmental factors</td><td/><td align=\"left\">Crude OR for current/past vs. never: 1.84 (1.28–2.66)</td><td/><td align=\"left\">When controlling for age, sex, family history of TB, HIV and smoking, this association was no longer significant. However, no adjusted OR is reported in paper.<break/>Prevalence of current/past use among controls: 19%</td><td/></tr><tr><td align=\"left\">Selassie et al: 2005, South Carolina, USA, 1970–2002</td><td/><td align=\"left\"><underline>Cases</underline> (437): All recurrent pulmonary TB cases, after at least 12 months from time of treatment completion between 1970 and 2001<break/><underline>Controls</underline> (442): Random sample of people who remained free of TB &gt; 12 months after completion, matched for year of initial diagnosis</td><td/><td align=\"left\">Medical records reviewed. \"Alcoholism\" as recorded in medical record</td><td/><td align=\"left\">Age, sex, race, treatment duration, adherence, regimen, HIV/AIDS, other chronic condition, country of residence, initial sputum, reported side effects.</td><td/><td align=\"left\">Adjusted OR for alcoholism vs. no alcoholism: 3.90 (2.49–6.12)</td><td/><td align=\"left\">Entire study population are TB infected and previously successfully treated. OR reflect risk of recurrent TB.<break/>Prevalence of recorded alcoholism among controls: 12.4%</td><td/></tr><tr><td align=\"left\">Riekstina, et al 2005, Latvia, 1996</td><td/><td align=\"left\"><underline>Cases</underline> (48): New pulmonary cases who had early (within 4 years) recurrence after successful treatment, adults only, excluding those with any resistance to first line drugs, and prisoners<break/><underline>Control</underline> (96):successful treatment, no recurrence, matched for sex and bacteriological status</td><td/><td align=\"left\">Alcohol problem according to medical records</td><td/><td align=\"left\">Sex and bacteriological status matched for. Age, sex, unemployment, treatment facility, treatment interruption</td><td/><td align=\"left\">Adjusted OR for alcohol problems vs. no alcohol problem: 16.63 (3.63–76.10)</td><td/><td align=\"left\">Entire study population are TB infected. OR reflect risk of progress to active disease.<break/>Prevalence among controls (all TB patients): 23%</td><td/></tr><tr><td align=\"left\">Shetty et al, 2006, Medical college hospital, Bangalore, India, 2001–2003</td><td/><td align=\"left\"><underline>Cases</underline> (189): all consecutive new active pulmonary TB<break/><underline>Controls</underline> (189): age and sex matched relatives of non-TB patients in same hospital</td><td/><td align=\"left\">Self reported as never, past (&gt; 6 months ago), or current use. Amounts not reported.</td><td/><td align=\"left\">Age and sex matched. Education, income, crowding, religion, marital status, BMI, cooking fuel, smoking, chronic illness.</td><td/><td align=\"left\">Adjusted OR for current vs.- never use 2.37 (0.95–5.93)</td><td/><td align=\"left\">Prevalence of current alcohol use in control group: 11.1%</td><td/></tr><tr><td align=\"left\">Coker et al, 2006, TB clinic, Samara town, Russia, 2003</td><td/><td align=\"left\"><underline>Cases</underline> (334): Culture confirmed pulmonary TB<break/><underline>Controls</underline> (334): Age and sex matched from population registry</td><td/><td align=\"left\">Self reported \"heavy drinking\" at least once per month during last year, but \"heavy drinking\" not further defined</td><td/><td align=\"left\">Age and sex matched. Adjusted for exposure (family contact and drinking unpasteurized milk)</td><td/><td align=\"left\">Adjusted OR for heavy drinking at least once a month vs. no drinking: 2.43 (1.22–4.85)</td><td/><td align=\"left\">Not clear if also smoking, illicit drug use, imprisonment, and household assets were controlled for. Alcohol not included in final multivariate analysis, reason not reported, alcohol listed as \"not appropriate\" in table.</td><td/></tr><tr><td align=\"left\">Kolappan et al, 2007, Prevalence survey 2001–2003, Rural district, Tamil Nadu, India</td><td align=\"center\"/><td align=\"left\"><underline>Cases</underline> (429): Bacteriologically positive cases, aged &gt; = 15, detected during prevalence survey<break/><underline>Controls</underline> (93,516): Those not diagnosed with TB in the prevalence survey, aged &gt; = 15</td><td align=\"center\"/><td align=\"left\">Self reported, alcohol intake in ml. Alcoholism not defined.</td><td align=\"center\"/><td align=\"left\">Age, sex, smoking</td><td align=\"center\"/><td align=\"left\">Adjusted OR for alcoholism vs. no alcoholism: 1.5 (1.2–2.0)</td><td align=\"center\"/><td align=\"left\">Prevalence among controls: 11%</td><td align=\"center\"/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Pooled effect sizes for different sub-categories of studies.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Study category</bold></td><td align=\"center\"><bold>No of studies</bold></td><td align=\"center\"><bold>Hetero-geneity test </bold><break/>Cochrane's Q p-value (I²)</td><td align=\"center\"><bold>Pooled, fixed effect assumption</bold><break/><bold>(95% confidence interval)</bold></td><td align=\"center\"><bold>Pooled, random effect assumption</bold><break/><bold>(95% confidence interval)</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Level of exposure</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">High exposure</td><td align=\"center\">11</td><td align=\"center\">&lt; 0.01 (0.82)</td><td align=\"center\">2.90 (2.39–3.51)</td><td align=\"center\">3.50 (2.01–5.93)</td></tr><tr><td align=\"left\">Low exposure</td><td align=\"center\">4</td><td align=\"center\">0.46 (0.00)</td><td align=\"center\">1.08 (0.82–1.40)</td><td align=\"center\">1.08 (0.82–1.40)</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\"><bold>High-exposure studies</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Controlled* for HIV status</td><td align=\"center\">7</td><td align=\"center\">0.03 (0.57)</td><td align=\"center\">2.93 (2.37–3.61)</td><td align=\"center\">3.26 (2.26–4.70)</td></tr><tr><td align=\"left\">Controlled* age, sex, SES, smoking</td><td align=\"center\">5</td><td align=\"center\">0.04 (0.61)</td><td align=\"center\">3.27 (2.38–4.50)</td><td align=\"center\">3.49 (2.06–5.90)</td></tr><tr><td align=\"left\">Controlled* HIV, age, sex, SES, smoking</td><td align=\"center\">4</td><td align=\"center\">0.07 (0.42)</td><td align=\"center\">3.92 (2.70–5.71)</td><td align=\"center\">4.08 (2.49–6.68)</td></tr><tr><td align=\"left\">Controlled* infection, age, sex, SES</td><td align=\"center\">4</td><td align=\"center\">0.23 (0.30)</td><td align=\"center\">4.11 (2.84–5.94)</td><td align=\"center\">4.21 (2.73–6.48)</td></tr><tr><td align=\"left\">Excluding three smallest studies</td><td align=\"center\">8</td><td align=\"center\">0.03 (0.59)</td><td align=\"center\">2.75 (2.19–3.46)</td><td align=\"center\">2.94 (1.89–4.59)</td></tr><tr><td align=\"left\">Excluding three smallest and Brown I and Kim</td><td align=\"center\">6</td><td align=\"center\">0.32 (0.15)</td><td align=\"center\">2.76 (2.34–3.81)</td><td align=\"center\">2.96 (2.28–3.85)</td></tr><tr><td align=\"left\">Pulmonary TB cases only**</td><td align=\"center\">2</td><td align=\"center\">0.49 (0.00)</td><td align=\"center\">3.67 (2.58–5.22)</td><td align=\"center\">3.67 (2.58–5.22)</td></tr><tr><td align=\"left\">All types of TB**</td><td align=\"center\">6</td><td align=\"center\">&lt; 0.01 (0.83)</td><td align=\"center\">2.52 (1.98–3.19)</td><td align=\"center\">2.87 (1.47–5.58)</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>OR = Odds Ratio, DM = Diabetes Mellitus, BMI = Body Mass Index, TST = Tuberculin Skin Test, SES = Socioeconomic Status</p></table-wrap-foot>", "<table-wrap-foot><p>*Controlled for respective covariates, either by design (e.g. through inclusion/exclusion criteria) or in the analysis (stratification or multivariate analysis)</p><p>**Excluding three smallest studies</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2458-8-289-1\"/>", "<graphic xlink:href=\"1471-2458-8-289-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Messenger RNA (mRNA) sequences encode the amino acid sequence of the protein but may also bear additional information. For example, certain synonymous codons may improve translation [##REF##3916708##1##, ####REF##11675589##2##, ##REF##9159473##3####9159473##3##] and a variety of motifs may regulate expression at the level of translation, cellular localization, decay or splicing [##UREF##0##4##, ####REF##15572659##5##, ##REF##15601259##6##, ##REF##16213112##7##, ##UREF##1##8##, ##REF##12446149##9####12446149##9##]. Many of these motifs are secondary structures, and eukaryotic mRNAs contain regulatory structures in their 5' and 3' UTRs [##REF##16373489##10##, ####REF##3322328##11##, ##UREF##2##12##, ##UREF##3##13##, ##UREF##4##14##, ##REF##9927459##15####9927459##15##], or introns [##REF##7568070##16##,##REF##12972637##17##]. However, it remains unclear whether secondary structures in the coding regions are of functional importance. Laboratory studies suggest that local secondary structures within coding regions can interfere with translation [##REF##16554824##18##,##REF##1507219##19##], and one may therefore expect selection against structures that are too stable. Surprisingly, however, several bioinformatic studies find that RNA structures within the protein coding regions are more stable than expected by chance [##REF##10075987##20##, ####REF##12952875##21##, ##REF##16168082##22##, ##REF##16682450##23####16682450##23##] (but see [##REF##17263882##24##] for opposing result). These studies used various algorithms to predict the secondary structures of mRNA sequences, and then compared the free energy values of these structures to the values for randomized sequences.</p>", "<p>Here, I test for selection on mRNA secondary structure using another approach. Instead of testing for unexpected structural stabilities, I directly compare the selective constraint of sites that differ in their importance for the predicted secondary structure. I.e. I predict the secondary structure of coding yeast sequences using different algorithms, and for each nucleotide, I identify whether it is paired with another nucleotide, or unpaired. I assume paired sites are more important for the predicted secondary structure than unpaired sites. If there is selection for secondary structures, one might expect higher structural constraint at paired than at unpaired sites. Such constraint would affect synonymous codon use and substitution rates. In <italic>S. cerevisiae </italic>a relationship between codon use, tRNA abundance and expression level indicates that codon use is affected by selection for translationally optimal codons [##REF##3916708##1##]. If there is selection for mRNA structure, structurally important sites may be under conflicting selection pressures: a codon might support the preferred mRNA structure that is translationally non-optimal. Under structural selection, one might expect lower numbers of optimal codons at paired than at unpaired sites. If mRNA structure is conserved across species, one might further expect lower numbers of synonymous substitutions at paired than at unpaired sites; possible compensatory substitutions however may make the latter test predictions less clear-cut. When a mutation occurs at a paired site and disrupts the pairing ability, a second compensatory mutation on the corresponding paired site may restore the pairing ability [##UREF##5##25##,##REF##10791023##26##]. Compensatory mutations may increase substitution numbers at paired sites. Innan and Stephan [##REF##11560913##27##] show however, that unless selection against deleterious intermediates is very small, substitutions should occur only very slowly in paired regions [##REF##11560913##27##].</p>", "<p>Accurate structure prediction is obviously crucial for these tests. In several studies [##REF##10075987##20##, ####REF##12952875##21##, ##REF##16168082##22####16168082##22##], mRNA structures are predicted by thermodynamic properties using the minimum free energy (MFE) algorithm [##REF##6163133##28##] only although taking the whole ensemble of possible structures and comparative information into account is known to increase predictive accuracy [##REF##15272118##29##, ####UREF##6##30##, ##REF##15458580##31##, ##UREF##7##32####7##32##]. I therefore predict the secondary structures by thermodynamic and comparative information (RNA- and ALIfold [##REF##12824340##33##]), using the minimum free energy (MFE) algorithm and McCaskill's partition function of the thermodynamic equilibrium [##REF##1695107##34##].</p>", "<p>Results of this study are consistent with selection upon mRNA structures: numbers of conserved optimal codons and synonymous substitutions are reduced at structurally important sites.</p>" ]

[ "<title>Methods</title>", "<title>Choice of study organism &amp; data</title>", "<p>I focus on <italic>Saccharomyces cerevisiae</italic>, as this model eukaryote is well studied, with genome sequences available for it and several related species. Importantly yeast allows using optimal codon numbers to investigate alternative selective constraints while controlling for effects of base composition. This is because (i) translational selection has been investigated extensively and supported in yeast [##REF##3916708##1##, ####REF##11675589##2##, ##REF##9159473##3####9159473##3##]: certain translationally \"optimal\" codons increase in frequency with expression level and correspond to the most abundant tRNAs in the cell or to the tRNA with which they form the strongest binding. (ii) Crucially, translationally optimal codons in yeast are not biased towards GC-ending codons, as in many other Eukaryotic organisms. In yeast 12 optimal codons end with G or C (-GC), 12 with A or T (-AT). To control for base composition is important as RNA secondary structure predictions are – at least partly- based on thermodynamic properties and will therefore be affected by GC content: GC nucleotides form the most stable binding with three hydrogen bonds and will consequently more likely be paired in the structure. From the yeast alignments provided by Kellis et al. [##REF##12748633##35##] comparing <italic>Saccharomyces cerevisiae </italic>with <italic>S. paradoxus</italic>, <italic>S. mikatae </italic>and <italic>S. bayanus</italic>, I use 492 genes that have start and stop codons but no premature stop codons or frame-shifting indels in all four species.</p>", "<title>Secondary structure</title>", "<p>I predict the secondary structure of the coding sequences using the below methods and identify for each nucleotide whether it is paired with another nucleotide, or unpaired. I assume paired sites are more important for the predicted <italic>secondary </italic>structure than unpaired sites. Note however, that unpaired sites may well be important for maintaining the mRNA's <italic>tertiary </italic>structure.</p>", "<title>Secondary structure prediction methods</title>", "<p>The thermodynamic stability of a secondary structure is measured as the amount of free energy released or used by forming base pairs. Positive free energy requires work to form a structure, negative free energy releases stored work. Free energy parameters are estimated from chemical melting experiments. The widely used Minimum Free Energy (MFE) algorithm [##REF##6163133##28##] computes the one single structure with the most negative energy value, that thermodynamically is hence the most likely to be formed. The MFE algorithm seems fairly accurate for short RNA sequences, for which ~73% of paired sites are accurately predicted. mRNAs however are likely to be present in a population of structures [##UREF##8##36##,##REF##7929367##37##]. Often 5–10% of structures share very similar free energy values [##UREF##9##38##], and the predicted MFE structure might just be one out of many thermodynamically similar structures. Taking all possible secondary structures of the thermodynamic equilibrium into account, McCaskill's algorithm [##REF##1695107##34##] computes the most probable structure and calculates the probability that each site is paired. When taking base pairings with high probabilities, the accuracy of the prediction increases [##REF##15272118##29##]. Another benefit of McCaskill's algorithm is that it is less affected by small but reasonable variations in the underlying energy parameters – while the MFE prediction is very sensitive [##UREF##10##39##,##REF##15673712##40##]. I used the RNAfold (<underline>Vienna RNA Secondary Structure</underline>[##REF##12824340##33##,##UREF##11##41##]) package to predict structures of the four yeasts separately using the MFE and McCaskill's algorithms. When using McCaskill's algorithm, I consider sites to be paired that pair with high probability (&gt;2/3) across the structure ensemble; all other sites are considered as unpaired. With increasing sequence lengths predictive accuracy decreases presumably because of the enormous increase in the number of potential base pairings that can be made as sequence length increases [##REF##10329189##42##]. I therefore look at both the complete set of genes, and at the subset of genes shorter than 800 bp.</p>", "<p>To predict the secondary structure, one can also assume structural conservation, and compute the one consensus structure that allows the largest amount of structural conservation across homologous sequences. Especially supportive of structural conservation are sites that vary at the sequence level but retain potential of Watson-Crick pairings in the structure (co-variations). Structures predicted with the aid of comparative data appear to be more accurate than those based on thermodynamic properties alone [##UREF##6##30##, ####REF##15458580##31##, ##UREF##7##32####7##32##]. I use the ALIfold package [##REF##12824340##33##,##REF##12079347##43##] that integrates comparative information in the prediction made with either MFE or McCaskill's algorithm and predict the consensus structures of the four yeasts together using the ALIfold default settings for co-variation weight (Φ₁= 1, and Φ₂= 1).</p>", "<title>Optimal codon use</title>", "<p>Codon identification is based on the <italic>S. cerevisiae </italic>sequence. Optimal codons are defined as in Kliman et al. (2003) [##REF##12962310##44##]. The relative frequency of optimal codons (F_op[##REF##6175758##45##]) is the ratio of optimal codons to synonymous codons. I compute the relative frequency of optimal codons for each amino acid and gene separately. For amino acids with both one AT- as well as one GC-ending optimal codon (thr, val, ile, ser), I compute the relative optimal codon frequencies of the two optimal codons per amino acid separately. Throughout the paper, the terms \"optimal\" and \"suboptimal\" will refer to translational selection.</p>", "<title>Tests</title>", "<p>If there is selection for secondary structures, one may expect higher constraint at structurally important (paired) than at structurally less important (unpaired) sites.</p>", "<p>(1) Under translational selection one may expect lower numbers of translationally optimal codons at paired compared to unpaired sites. Note that the analysis is restricted to those codons that are conserved across the four yeast species and are likely to experience stronger selection pressures. Restricting the analysis to conserved sites is crucial for the ALIfold measure, as it incorporates substitutions in its prediction: ALIfold may tend to pair conserved sites, and under translational selection conserved sites tend to have higher optimal codon use than non-optimal sites. This could generate an artificial positive correlation between optimal codon numbers and structure when considering all codons. As GC-ending optimal codons are more likely to be paired, I look at GC- and AT-ending optimal codons separately. I do this for the four yeast species separately (using RNAfold) as well as for their consensus structure (using ALIfold), using MFE as well as McCaskill's algorithm for both methods.</p>", "<p>(2) If mRNA structures are conserved across species one may further expect lower numbers of substitutions at paired compared to unpaired sites. As ALIfold incorporates comparative information, this test is only meaningful for structures predicted by RNAfold. Codons experiencing non-synonymous substitutions are excluded from this analysis as one may expect possible selection on mRNA structure will mainly affect synonymous substitutions, while non-synonymous substitutions will be more constrained for other reasons. To check the structural similarity and potential conservation of predicted structures across species, I first compute the relative number of base pairings per gene that are consistently, i.e. unambiguously, predicted to be paired or unpaired across species. To estimate structural constraint at synonymous sites, I count for each synonymous optimal and non-optimal codon how often the respective third codon position is paired and unpaired in the <italic>S. cerevisiae </italic>structure (RNAfold) and how often the codon is conserved or experiences a synonymous substitution compared to <italic>S. paravensis</italic>. Note that translational selection and structural selection may be counter-balancing with respect to synonymous substitution numbers. I.e. unpaired sites with high numbers of optimal codons may experience reduced synonymous substitution numbers due to translational selection while paired sites with high numbers of non-optimal codons may experience reduced synonymous substitution numbers due to structural selection. To disentangle structural selection from translational selection, I look at optimal and non-optimal codons separately as. I further look at GC- and AT-ending codons separately as mutational processes and gene conversion events may be compositionally biased [##REF##12082137##46##].</p>", "<title>Statistics</title>", "<p>Each of our analyses generates a set of 2 × 2 contingency tables per gene and per amino acid or codon. These are divided according to whether the site is paired or unpaired in the predicted secondary structure, and whether (1) the codon is optimal or non-optimal, and whether (2) the codon is conserved or synonymous polymorphic across the four species. To combine these independent 2 × 2 tables, I use the Mantel-Haenszel Z statistic according to Sokal and Rohlf [##UREF##12##47##]. I compute joint probabilities for all tables or certain subsets. To disentangle an effect of GC content on synonymous codon use at paired sites, I combine amino acids with AT-ending ending and amino acids with GC-ending optimal codons. I exclude contingency tables when expected values were zero, tested for homogeneity and computed the joint odds ratio (W_MH) and its significance, including the continuity correction. I orient the odds ratio such that selection in favour of mRNA secondary structure is indicated by W_MH&lt;1: i.e. lower numbers of optimal codons, and lower numbers of synonymous substitutions at paired sites.</p>" ]

[ "<title>Results</title>", "<p>1) Conserved optimal codon numbers are significantly lower at paired compared to unpaired sites, irrespective of the method (RNA- and ALIfold) and algorithm (MFE and McCaskill's) used to predict secondary structure. Crucially, the tendency remains whether I consider amino acids with AT- or GC-ending optimal codons (Tables ##TAB##0##1##, ##TAB##1##2##).</p>", "<p>The tendency is true for most amino acids separately; even GC-ending optimal codons that are more likely to be paired for thermodynamic reasons tend to be less frequent at paired sites (Table ##TAB##1##2##). Notable exceptions however are leu, lys, ile, and for RNAfold additionally phe (Table ##TAB##1##2##). One explanation for these exceptions may be that selection strength for translationally optimal codons is stronger in these amino acids, for example translational errors may be more likely or more costly. Considering prediction accuracy may decrease with gene length, I first restricted the data to genes shorter than 800 bp; including all genes however does not change the result.</p>", "<p>(2) I first check the similarity and potential conservation of structures predicted by RNAfold. The major parts of mRNAs do not seem conserved in structure across species or prediction accuracy is low: 75% of sites are ambiguous, i.e. predicted to be paired in one or more species, but predicted to be unpaired in the remaining species (Table ##TAB##2##3##). When looking pairwise on average 41% of sites are ambiguous; number of ambiguous sites is only slightly lower for short genes. The ambiguity of predicted structural status will introduce considerable noise and may cause non-significant results. Despite high ambiguity in structure prediction the numbers of synonymous substitutions are consistently lower (W_MH&lt;1) at paired sites (Table ##TAB##3##4##). The tendencies remain when restricting the data to genes shorter than 800 bp. Other species comparisons lead to similar results (data not presented). Results become significant for GC-ending (optimal and non-optimal) codons (when structure is predicted using McCaskill's algorithm). G and C nucleotides do not only form stronger bonds and are more likely to be paired and structurally important, they are also more likely to be unambiguously predicted paired than A and T nucleotides (means GC: 0.263, AT: 0.153, t = 29.8409, df = 866.05, ***). This could reduce the level of noise and cause the significance of results for GC-ending codons.</p>" ]

[ "<title>Discussion</title>", "<p>I tested for evidence of selective constraint acting on mRNA secondary structures in protein coding yeast genes. Predicted secondary structures differ greatly according to the prediction method used and between species. Nevertheless, there are significantly fewer conserved optimal codons and consistently fewer synonymous substitutions at paired sites for all predicted secondary structures. The results of this study are consistent with purifying selection on mRNA secondary structures.</p>", "<p>Similar tendencies of codon use have been reported for Drosophila and humans: mRNA stability seems high when optimal codon use is low in Drosophila [##REF##11606539##48##] and paired sites contain an excess of rare codons in humans [##REF##16275783##49##]. Note that in this study, the comparison of optimal codon use is restricted to conserved sites. Besides the methodological need for ALIfold (see Material and Methods), the restriction to conserved sites restricts the analysis to sites potentially under considerable strong selection. For RNAfold structures, results become non-significant when not restricting the data to these conserved sites (data not presented). Strong conflicting selection pressures seem to act on certain sites while the remaining sites seem less constrained for structure. Selection on <italic>local </italic>and not <italic>global </italic>structures may explain these results and contribute to the low structural similarity across species. Selection on local mRNA structures in coding regions of eukaryotic genes has been suggested before [##REF##16275783##49##]. Beside the low structural similarity also compensatory substitutions may contribute to the non-significant results when comparing substitution numbers at paired and unpaired sites.</p>", "<p>Previous bioinformatic studies that focussed on whether or not the thermodynamic <italic>stability </italic>of mRNA structures of various organisms is selected for or against [##REF##10075987##20##, ####REF##12952875##21##, ##REF##16168082##22##, ##REF##16682450##23####16682450##23##,##REF##16275783##49##,##REF##10572183##55##] lead to partly inconsistent results and controversies about the accurate randomization procedure. In these studies, the observed MFE is compared to the expected MFE, which is estimated by taking the mean MFE of randomized versions of the same sequence, and a significant deviation is taken as evidence for selection for or against thermodynamic stability of the structure. The randomization of sequences can be performed in a number of different ways holding various properties of the sequence constant, while randomizing others. The properties are of biological importance; variables that are affected by forces other than selection for mRNA structure – for example the amino acid sequence – should be fixed. Which variables should remain free to vary however may not always be obvious, while the results are very sensitive to them. Di-nucleotide content for example might be selected for its effect on stability and should be allowed to vary for randomized sequences argue Chamary and Hurst [##REF##16168082##22##]. However, di-nucleotides might well be affected by mutation bias, or selected for some other reason [##REF##12952875##21##], in which case, di-nucleotide content should be kept fixed. The control of di-nucleotides in fact renders significant results non-significant [##REF##10075987##20##, ####REF##12952875##21##, ##REF##16168082##22##, ##REF##16682450##23####16682450##23##,##REF##10572183##55##].</p>", "<p>Note that in contrast to comparing observed and expected MFE values, the comparison of constraint at paired and unpaired sites does not indicate that selection acts for or against the thermodynamic <italic>stability </italic>of the structure, but that the very predicted <italic>structure </italic>is under selection. With respect to selection for or against <italic>stability </italic>of structures, ALIfold results indicate that the thermodynamically most stable global structure is not conserved across the four yeasts: ALIfold consensus energy value is much higher i.e. less stable compared to the average energy value of the single sequences [see also Washietl et al. [##REF##15665081##50##] for approach]. This is conform with results of Babak et al. [##REF##17263882##24##] which support selection against stability of structures in coding regions. It is reasonable to expect selection on mRNA structures may act against too stable structures because too stable and un-flexible mRNA structures may interfere for instance with translation [##REF##16554824##18##] and some mRNAs flexibility may allow their specific and dynamic complexes with other factors. mRNAs lead a complex life [##UREF##13##51##] and besides thermodynamic stability, selection on mRNA structure may also exist to maintain specific local or global mRNA structures that allow binding and interaction with other factors and thus effect biological functioning. Not only structural targets may be of effect, also accessibility of sequence targets may depend on global or local mRNA structures.</p>", "<p>While results of this study are consistent with selection upon mRNA structures in coding regions and support laboratory studies that report synonymous substitutions are functionally important with respect to mRNA structure and translation in humans [##REF##10393914##52##, ####UREF##14##53##, ##REF##17185601##54####17185601##54##], two considerations should be made. First, we do not know whether thermodynamic mRNA structure predictions predict the mRNA structures that are formed in the cell. mRNAs are generally associated with other factors [##UREF##13##51##], and effects of mRNA-associated microRNAs and proteins on the structure are hard to predict. Also, kinetics of mRNA folding and pseudo-knots are not considered here. Even with the comparative method, mRNA structures may remain at best approximations of the real mRNA structures in the cell. Secondly, the predicted and also the real structure will be affected by certain DNA patterns – however whether or not the respective DNA patterns are selected for mRNA structure or another reason may be hard to judge. There are several DNA patterns one may consider. (i) Di-nucleotide content of naturally occurring sequences leads to higher than expected thermodynamic stability [e.g. [##REF##12952875##21##,##REF##16682450##23##,##REF##10572183##55##]]. Di-nucleotide content may be selected for its effect on mRNA structure but it may also be affected by mutation bias, or selected for some other reason, for example for nucleosome positioning [##REF##16862119##56##, ####REF##17006463##57##, ##REF##8831784##58####8831784##58##] or transcription pause sites [##REF##16777599##59##]. (ii) Frequency of polypurine tracts is increased in exons and may affect thermodynamic structure. Again, polypurine tracts may be selected with respect to mRNA structure but also for other reasons such as enhancing splicing [##REF##17211550##60##]. (iii) Translational protein folding into alpha-helix and beta sheets may affect synonymous codon use [##REF##17185560##61##] and periodic DNA patterns may affect mRNA structure. If thermodynamic predictions correspond to any other force such as selection on nucleosome positioning and transcription or co-translational pause sites, the observed patterns may be a consequence of that and inference of selection acting directly upon on mRNA structure may be incorrect.</p>", "<p>Alternative selection upon mRNA structures (or any other selective target) may counterbalance translational selection and explain why the bias towards translationally optimal codon is never complete and even in highly expressed genes non-optimal codons are used. Alternative selection may also contribute to the discrepancy between expected and observed codon bias [##REF##8831784##58##], and may lead to systematic underestimates of selection strength for optimal codons. As selection for mRNA structures may be acting stronger on GC-ending codons, in organisms in which potential translationally optimal codons are biased towards GC, such as Drosophila, mammals, C. elegans, estimates of selective strength for optimal codons may also be overestimated. It will be worth considering effects of alternative selection and disentangling the different targets of selection.</p>" ]

[ "<title>Conclusion</title>", "<p>I tested for evidence of selective constraint acting on mRNA secondary structures in yeast. Predicted structures differ greatly according to the prediction method used and between species. Nevertheless, there are significantly fewer conserved optimal codons and consistently fewer synonymous substitutions at paired sites for all predicted secondary structures. These results are consistent with purifying selection on mRNA secondary structures in protein coding yeast sequences and suggest their biological importance. One should consider however that accuracy of structure prediction is unknown for mRNAs and interrelated selective forces may contribute. Selective pressures alternative to translational selection seem affect synonymous and optimal codon use in yeast. Depending on strength and direction of translational selection in an organism, such alternative selective forces may lead to under- or over-estimates of selective strength on optimal codon use.</p>" ]

[ "<title>Background</title>", "<p>Eukaryotic mRNAs often contain secondary structures in their untranslated regions that are involved in expression regulation. Whether secondary structures in the protein coding regions are of functional importance remains unclear: laboratory studies suggest stable secondary structures within the protein coding sequence interfere with translation, while several bioinformatic studies indicate stable mRNA structures are more frequent than expected.</p>", "<title>Results</title>", "<p>In contrast to several studies testing for unexpected structural stabilities, I directly compare the selective constraint of sites that differ in their structural importance. I.e. for each nucleotide, I identify whether it is paired with another nucleotide, or unpaired, in the predicted secondary structure. I assume paired sites are more important for the predicted secondary structure than unpaired sites. I look at protein coding yeast sequences and use optimal codons and synonymous substitutions to test for structural constraints. As expected under selection for secondary structures, paired sites experience higher constraint than unpaired sites, i.e. significantly lower numbers of conserved optimal codons and consistently lower numbers of synonymous substitutions. This is true for structures predicted by different algorithms.</p>", "<title>Conclusion</title>", "<p>The results of this study are consistent with purifying selection on mRNA secondary structures in yeast protein coding sequences and suggest their biological importance. One should be aware, however, that accuracy of structure prediction is unknown for mRNAs and interrelated selective forces may contribute as well. Note that if selection pressures alternative to translational selection affect synonymous (and optimal) codon use, this may lead to under- or over-estimates of selective strength on optimal codon use depending on strength and direction of translational selection.</p>" ]

[]

[ "<title>Acknowledgements</title>", "<p>I first would like to thank Adam Eyre-Walker very much; he has provided plenty valuable advice, many helpful comments and suggestions at various stages of the manuscript; I am very grateful for his continuous presence for discussions and encouragement. I thank John Welch very much for an early helpful discussion and encouragement, as for many helpful comments later on the manuscript. I thank Wolfgang Stephan for friendly welcome me in his RNA discussion group, including Andreas Brunnert, Mihaela Martis and Sebastian Strempel who are all acknowledged for helpful discussions on RNA prediction methods and Perl codes. I am very grateful for support from Joachim Hermisson and his respect for my independent PhD projects. Finally, I thank both anonymous reviewers for heir helpful comments. NS was partly supported by an Emmy-Noether grant to J. Hermisson.</p>" ]

[]

[ "<table-wrap id=\"T1\" position=\"float\"><label>Table 1</label><caption><p>Comparison of conserved optimal codon numbers at paired and unpaired sites.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th align=\"left\">Method</th><th align=\"left\">Algorithm</th><th align=\"left\">ALL</th><th align=\"left\">GC-ending</th><th align=\"left\">AT-ending</th><th align=\"left\">GC leu &amp; lys</th></tr></thead><tbody><tr><td align=\"left\" colspan=\"6\"><bold>Genes shorter than 800 bp</bold></td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\">RNAfold <italic>S. cerevisiae</italic></td><td align=\"left\">MFE</td><td align=\"left\">0.646 ***</td><td align=\"left\">0.624 **</td><td align=\"left\">0.503 ***</td><td align=\"left\">1.353 ***</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">0.542 ***</td><td align=\"left\">0.567 ***</td><td align=\"left\">0.422 ***</td><td align=\"left\">0.910 ***</td></tr><tr><td align=\"left\">RNAfold <italic>S. paravensis</italic></td><td align=\"left\">MFE</td><td align=\"left\">0.667 ***</td><td align=\"left\">0.608 ***</td><td align=\"left\">0.542 NS</td><td align=\"left\">1.137 ***</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">0.560 ***</td><td align=\"left\">0.571 ***</td><td align=\"left\">0.407 **</td><td align=\"left\">1.119 ***</td></tr><tr><td align=\"left\">RNAfold <italic>S. mikitae</italic></td><td align=\"left\">MFE</td><td align=\"left\">0.653 ***</td><td align=\"left\">0.590 ***</td><td align=\"left\">0.544 *</td><td align=\"left\">1.172 ***</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">0.572 ***</td><td align=\"left\">0.623 ***</td><td align=\"left\">0.411 ***</td><td align=\"left\">1.131 ***</td></tr><tr><td align=\"left\">RNAfold <italic>S. bayanus</italic></td><td align=\"left\">MFE</td><td align=\"left\">0.640 ***</td><td align=\"left\">0.597 **</td><td align=\"left\">0.502 ***</td><td align=\"left\">1.181 ***</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">0.537 **</td><td align=\"left\">0.557 ***</td><td align=\"left\">0.401 ***</td><td align=\"left\">1.003 ***</td></tr><tr><td align=\"left\">ALIfold</td><td align=\"left\">MFE</td><td align=\"left\">0.638 ***</td><td align=\"left\">0.577 ***</td><td align=\"left\">0.465 NS</td><td align=\"left\">1.499 ***</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">0.468 **</td><td align=\"left\">0.444 ***</td><td align=\"left\">0.326 ***</td><td align=\"left\">0.997 ***</td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\" colspan=\"6\"><bold>All genes</bold></td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\">RNAfold <italic>S. cerevisiae</italic></td><td align=\"left\">MFE</td><td align=\"left\">0.920 ***</td><td align=\"left\">0.863 ***</td><td align=\"left\">0.751 ***</td><td align=\"left\">1.584 ***</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">0.841 ***</td><td align=\"left\">0.866 ***</td><td align=\"left\">0.676 ***</td><td align=\"left\">1.436 ***</td></tr><tr><td align=\"left\">RNAfold <italic>S. paravensis</italic></td><td align=\"left\">MFE</td><td align=\"left\">0.878 ***</td><td align=\"left\">0.826 ***</td><td align=\"left\">0.733 ***</td><td align=\"left\">1.497 ***</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">0.819 ***</td><td align=\"left\">0.822 ***</td><td align=\"left\">0.659 ***</td><td align=\"left\">1.460 ***</td></tr><tr><td align=\"left\">RNAfold <italic>S. mikitae</italic></td><td align=\"left\">MFE</td><td align=\"left\">0.912 ***</td><td align=\"left\">0.814 ***</td><td align=\"left\">0.790 NS</td><td align=\"left\">1.160 ***</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">0.839 ***</td><td align=\"left\">0.869 ***</td><td align=\"left\">0.740 NS</td><td align=\"left\">1.404 ***</td></tr><tr><td align=\"left\">RNAfold <italic>S. bayanus</italic></td><td align=\"left\">MFE</td><td align=\"left\">0.904 ***</td><td align=\"left\">0.855 ***</td><td align=\"left\">0.742 ***</td><td align=\"left\">1.590 ***</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">0.833 ***</td><td align=\"left\">0.840 ***</td><td align=\"left\">0.675 ***</td><td align=\"left\">1.455 ***</td></tr><tr><td align=\"left\">ALIfold</td><td align=\"left\">MFE</td><td align=\"left\">0.899 ***</td><td align=\"left\">0.759 **</td><td align=\"left\">0.739 NS</td><td align=\"left\">1.937 ***</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">0.770 ***</td><td align=\"left\">0.724 ***</td><td align=\"left\">0.645 ***</td><td align=\"left\">1.529 ***</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"T2\" position=\"float\"><label>Table 2</label><caption><p>Comparison of conserved optimal codon numbers at paired and unpaired sites.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th/><th align=\"center\" colspan=\"2\">RNAfold (<italic>S. cerevisiae</italic>)</th><th align=\"center\" colspan=\"2\">ALIfold</th></tr><tr><th/><th colspan=\"2\"><hr/></th><th colspan=\"2\"><hr/></th></tr><tr><th/><th align=\"left\">MFE</th><th align=\"left\">Mc</th><th align=\"left\">MFE</th><th align=\"left\">Mc</th></tr></thead><tbody><tr><td align=\"left\" colspan=\"5\"><bold>Amino acids with GC-ending optimal codons</bold></td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\"><bold>Leu</bold>_{<bold>TTG</bold>}</td><td align=\"left\">1.380 ***</td><td align=\"left\">1.273 ***</td><td align=\"left\">1.811 ***</td><td align=\"left\">1.456 ***</td></tr><tr><td align=\"left\"><bold>Lys</bold>_{<bold>AAG</bold>}</td><td align=\"left\">1.772 ***</td><td align=\"left\">1.654 ***</td><td align=\"left\">2.019 ***</td><td align=\"left\">1.654 ***</td></tr><tr><td align=\"left\"><bold>Phe</bold>_{<bold>TTC</bold>}</td><td align=\"left\">1.075 ***</td><td align=\"left\">1.120 ***</td><td align=\"left\">0.786 NS</td><td align=\"left\">0.864 **</td></tr><tr><td align=\"left\"><bold>Tyr</bold>_{<bold>TAC</bold>}</td><td align=\"left\">0.809 NS</td><td align=\"left\">0.720 NS</td><td align=\"left\">0.741 NS</td><td align=\"left\">0.631 NS</td></tr><tr><td align=\"left\"><bold>His</bold>_{<bold>CAC</bold>}</td><td align=\"left\">0.707 NS</td><td align=\"left\">0.600 NS</td><td align=\"left\">0.568 NS</td><td align=\"left\">0.657 *</td></tr><tr><td align=\"left\"><bold>Asp</bold>_{<bold>GAC</bold>}</td><td align=\"left\">0.813 NS</td><td align=\"left\">0.838 NS</td><td align=\"left\">0.656 ***</td><td align=\"left\">0.746 **</td></tr><tr><td align=\"left\"><bold>Asn</bold>_{<bold>AAC</bold>}</td><td align=\"left\">0.891 *</td><td align=\"left\">0.831 NS</td><td align=\"left\">0.725 **</td><td align=\"left\">0.637 NS</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\" colspan=\"5\"><bold>Amino acids with one GC- and one AT-ending optimal codon</bold></td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\"><bold>Ile</bold>_{<bold>ATC</bold>}</td><td align=\"left\">1.318 ***</td><td align=\"left\">1.254 ***</td><td align=\"left\">1.555 ***</td><td align=\"left\">1.049 ***</td></tr><tr><td align=\"left\"><bold>Ile</bold>_{<bold>ATT</bold>}</td><td align=\"left\">1.272 ***</td><td align=\"left\">0.978 ***</td><td align=\"left\">1.561 ***</td><td align=\"left\">1.009 ***</td></tr><tr><td align=\"left\"><bold>Val</bold>_{<bold>GTC</bold>}</td><td align=\"left\">0.585 NS</td><td align=\"left\">0.852 ***</td><td align=\"left\">0.734 ***</td><td align=\"left\">0.560 ***</td></tr><tr><td align=\"left\"><bold>Val</bold>_{<bold>GTT</bold>}</td><td align=\"left\">0.705 *</td><td align=\"left\">0.794 NS</td><td align=\"left\">0.806 *</td><td align=\"left\">0.686 NS</td></tr><tr><td align=\"left\"><bold>Thr</bold>_{<bold>ACC</bold>}</td><td align=\"left\">0.838 **</td><td align=\"left\">0.921 ***</td><td align=\"left\">0.811 ***</td><td align=\"left\">0.750 ***</td></tr><tr><td align=\"left\"><bold>Thr</bold>_{<bold>ACT</bold>}</td><td align=\"left\">0.940 ***</td><td align=\"left\">0.876 ***</td><td align=\"left\">0.932 ***</td><td align=\"left\">0.655 ***</td></tr><tr><td align=\"left\"><bold>Ser</bold>_{<bold>TCC</bold>}</td><td align=\"left\">0.614 NS</td><td align=\"left\">0.668 NS</td><td align=\"left\">0.583 NS</td><td align=\"left\">0.591 NS</td></tr><tr><td align=\"left\"><bold>Ser</bold>_{<bold>TCT</bold>}</td><td align=\"left\">0.794 NS</td><td align=\"left\">0.816 NS</td><td align=\"left\">0.816 ***</td><td align=\"left\">0.985 ***</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\" colspan=\"5\"><bold>Amino acids with AT-ending optimal codons</bold></td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\"><bold>Ala</bold>_{<bold>GCT</bold>}</td><td align=\"left\">0.968 **</td><td align=\"left\">0.904 ***</td><td align=\"left\">1.073 ***</td><td align=\"left\">0.883 ***</td></tr><tr><td align=\"left\"><bold>Arg</bold>_{<bold>AGA,CGT</bold>}</td><td align=\"left\">0.588 ***</td><td align=\"left\">0.545 ***</td><td align=\"left\">0.518 ***</td><td align=\"left\">0.460 ***</td></tr><tr><td align=\"left\"><bold>Gly</bold>_{<bold>GGT</bold>}</td><td align=\"left\">0.894 NS</td><td align=\"left\">0.779 NS</td><td align=\"left\">0.901 ***</td><td align=\"left\">0.752 NS</td></tr><tr><td align=\"left\"><bold>Gln</bold>_{<bold>CAA</bold>}</td><td align=\"left\">0.442 ***</td><td align=\"left\">0.353 ***</td><td align=\"left\">0.293 ***</td><td align=\"left\">0.278 ***</td></tr><tr><td align=\"left\"><bold>Glu</bold>_{<bold>GAA</bold>}</td><td align=\"left\">0.946 ***</td><td align=\"left\">0.350 ***</td><td align=\"left\">0.386 ***</td><td align=\"left\">0.315 ***</td></tr><tr><td align=\"left\"><bold>Pro</bold>_{<bold>CCA</bold>}</td><td align=\"left\">0.851 NS</td><td align=\"left\">0.734 NS</td><td align=\"left\">0.729 **</td><td align=\"left\">0.689 *</td></tr><tr><td align=\"left\"><bold>Cys</bold>_{<bold>TGT</bold>}</td><td align=\"left\">0.500 NS</td><td align=\"left\">0.382 NS</td><td align=\"left\">0.755 ***</td><td align=\"left\">0.403 NS</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"T3\" position=\"float\"><label>Table 3</label><caption><p>Similarity of predicted structures for species pairs.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th align=\"left\">Species comparison</th><th align=\"left\">Prediction Method</th><th align=\"left\">(P+U)/all</th><th align=\"left\">P/all</th><th align=\"left\">U/all</th></tr></thead><tbody><tr><td align=\"left\" colspan=\"5\"><bold>Genes shorter than 800 bp</bold></td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\"><bold>Across all 4 yeasts</bold></td><td align=\"left\">MFE</td><td align=\"left\">27% ± 0.6</td><td align=\"left\">17% ± 0.3</td><td align=\"left\">10% ± 0.5</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">27% ± 0.4</td><td align=\"left\">9% ± 0.2</td><td align=\"left\">18% ± 0.3</td></tr><tr><td align=\"left\"><bold><italic>S. cerevisiae – S. paravensis</italic></bold></td><td align=\"left\">MFE</td><td align=\"left\">63% ± 0.6</td><td align=\"left\">36% ± 0.3</td><td align=\"left\">27% ± 0.3</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">64% ± 0.5</td><td align=\"left\">22% ± 0.4</td><td align=\"left\">42% ± 0.5</td></tr><tr><td align=\"left\"><bold><italic>S. cerevisiae – S. mikitae</italic></bold></td><td align=\"left\">MFE</td><td align=\"left\">59% ± 0.3</td><td align=\"left\">34% ± 0.2</td><td align=\"left\">25% ± 0.2</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">61% ± 0.4</td><td align=\"left\">20% ± 0.4</td><td align=\"left\">41% ± 0.6</td></tr><tr><td align=\"left\"><bold><italic>S. cerevisiae – S. bayanus</italic></bold></td><td align=\"left\">MFE</td><td align=\"left\">42% ± 0.4</td><td align=\"left\">23% ± 0.5</td><td align=\"left\">19% ± 0.1</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">59% ± 0.3</td><td align=\"left\">20% ± 0.0</td><td align=\"left\">40% ± 0.5</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\" colspan=\"5\"><bold>All genes</bold></td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\"><bold>Across all 4 yeasts</bold></td><td align=\"left\">MFE</td><td align=\"left\">25% ± 0.3</td><td align=\"left\">8% ± 0.1</td><td align=\"left\">16% ± 0.2</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">25% ± 0.4</td><td align=\"left\">16% ± 0.2</td><td align=\"left\">9% ± 0.1</td></tr><tr><td align=\"left\"><bold><italic>S. cerevisiae – S. paravensis</italic></bold></td><td align=\"left\">MFE</td><td align=\"left\">61% ± 0.4</td><td align=\"left\">36% ± 0.2</td><td align=\"left\">25% ± 0.2</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">63% ± 0.3</td><td align=\"left\">22% ± 0.3</td><td align=\"left\">41% ± 0.4</td></tr><tr><td align=\"left\"><bold><italic>S. cerevisiae – S. mikitae</italic></bold></td><td align=\"left\">MFE</td><td align=\"left\">58% ± 0.2</td><td align=\"left\">35% ± 0.2</td><td align=\"left\">23% ± 0.1</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">60% ± 0.3</td><td align=\"left\">20% ± 0.3</td><td align=\"left\">40% ± 0.4</td></tr><tr><td align=\"left\"><bold><italic>S. cerevisiae – S. bayanus</italic></bold></td><td align=\"left\">MFE</td><td align=\"left\">57% ± 0.2</td><td align=\"left\">34% ± 0.1</td><td align=\"left\">23% ± 0.1</td></tr><tr><td/><td align=\"left\">Mc</td><td align=\"left\">58% ± 0.2</td><td align=\"left\">20% ± 0.2</td><td align=\"left\">39% ± 0.4</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"T4\" position=\"float\"><label>Table 4</label><caption><p>Comparison of synonymous substitution numbers at paired and unpaired sites.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th/><th/><th align=\"left\"><bold>AT</bold>_{<bold>Opt</bold>}</th><th align=\"left\"><bold>AT</bold>_{<bold>Nopt</bold>}</th><th align=\"left\">AT</th><th align=\"left\"><bold>GC</bold>_{<bold>Opt</bold>}</th><th align=\"left\"><bold>GC</bold>_{<bold>Nopt</bold>}</th><th align=\"left\">GC</th><th align=\"left\">All</th></tr></thead><tbody><tr><td align=\"left\"><bold>(1)</bold></td><td align=\"left\"><bold>MFE</bold></td><td align=\"left\">0.511<break/>NS</td><td align=\"left\">0.542<break/>NS</td><td align=\"left\">0.600<break/>NS</td><td align=\"left\">0.488<break/>NS</td><td align=\"left\">0.296<break/>NS</td><td align=\"left\">0.414<break/>NS</td><td align=\"left\">0.481<break/>NS</td></tr><tr><td/><td align=\"left\"><bold>Mc</bold></td><td align=\"left\">0.567<break/>NS</td><td align=\"left\">0.528<break/>NS</td><td align=\"left\">0.546<break/>NS</td><td align=\"left\">0.544**</td><td align=\"left\">0.327*</td><td align=\"left\">0.458**</td><td align=\"left\">0.505<break/>NS</td></tr><tr><td align=\"left\"><bold>(2)</bold></td><td align=\"left\"><bold>MFE</bold></td><td align=\"left\">0.255<break/>NS</td><td align=\"left\">0.201<break/>NS</td><td align=\"left\">0.232<break/>NS</td><td align=\"left\">0.140<break/>NS</td><td align=\"left\">0.257<break/>NS</td><td align=\"left\">0.213<break/>NS</td><td align=\"left\">0.225<break/>NS</td></tr><tr><td/><td align=\"left\"><bold>Mc</bold></td><td align=\"left\">0.255<break/>NS</td><td align=\"left\">0.230<break/>NS</td><td align=\"left\">0.238<break/>NS</td><td align=\"left\">0.264<break/>NS</td><td align=\"left\">0.167<break/>NS</td><td align=\"left\">0.222<break/>NS</td><td align=\"left\">0.232<break/>NS</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>I combine contingency tables for all amino acids and genes (ALL) and subsets of amino acids with GC- and AT- ending optimal codons (leu and lys are treated separately, as these two GC-ending amino acids behave very opposing, see below Table 2). Mantel Haenzsel estimators and significances are presented, W_MH&lt;1 = lower optimal codon use at paired than at unpaired sites.</p><p>* &lt; 0.05, ** &lt; 0.01, *** &lt; 0.005, NS = not significant. Structure prediction is based on ALIfold and RNAfold using MFE and McCaskill's (Mc) algorithm.</p></table-wrap-foot>", "<table-wrap-foot><p>Separately for each amino acid, I combine contingency tables of the different genes. Mantel Haenzsel estimators and significances are presented, with W_MH&lt;1 = lower optimal codon use at paired than at unpaired sites. Structure prediction is based on ALIfold and RNAfold using MFE and McCaskill's (Mc) algorithm.</p><p>* &lt; 0.05, ** &lt; 0.01, *** &lt; 0.005, NS = not significant</p></table-wrap-foot>", "<table-wrap-foot><p>The average percentages of sites (± variances) unambiguously predicted to be paired (P/all) and/or unpaired ((P+U)/all, U/all) for the respective species comparison using RNAfold MFE and McCaskill's (Mc) algorithm are presented.</p></table-wrap-foot>", "<table-wrap-foot><p>Looking at <italic>S. cerevisiae </italic>and <italic>S. paravensis </italic>, I compare numbers of each codon in <italic>S. cerevisiae </italic>being either synonymous non-conserved or conserved at paired or unpaired sites. Structure prediction is based on RNAfold upon the <italic>S. cerevisiae </italic>sequence using MFE and McCaskill's (Mc) algorithm. Mantel Haenzsel estimators and significances are presented. W_MH&lt;1 = lower numbers of synonymous substitutions at paired sites. (1) All genes, (2) Genes that are shorter than 800 bp.</p><p>* &lt; 0.05, ** &lt; 0.01, *** &lt; 0.005, NS = not significant</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The TRIM gene family encodes proteins involved in a broad range of biological processes and characterized by the presence of the <underline>tri</underline>partite <underline>m</underline>otif (hence the name TRIM), which consists of a <underline>R</underline>ING domain, one or two <underline>B</underline>-box motifs and a <underline>C</underline>oiled-<underline>c</underline>oil region (RBCC) [##REF##11331580##1##,##REF##11428128##2##]. The tripartite motif is always present at the N-terminus of the TRIM proteins. The order of the domains that compose the motif is also conserved: a RING finger domain precedes the B-box motif(s), and a Coiled-coil (CC) region invariably follows. Even if one of the domains is missing, the order of the remaining ones is maintained. Different C-terminal domains are associated with the tripartite motif in the TRIM family [##REF##11331580##1##, ####REF##11428128##2##, ##REF##16175175##3##, ##REF##16434393##4####16434393##4##].</p>", "<p>Both RING and B-boxes are cysteine-rich zinc-binding domains. The RING finger domain is present, in combination with other domains, in hundreds of proteins and is defined by a linear series of conserved cysteine and histidine residues that represent zinc coordination sites [##REF##10662664##5##]. The B-boxes are the critical determinants of the TRIM family and can be present as B-box1 and B-box2, which share a similar but distinct pattern of cysteine and histidine residues [##REF##11331580##1##]. When both B-box domains are present, type 1 always precedes type 2; when only one B-box domain is present, it is always type 2 [##REF##11331580##1##]. While the tripartite motif is restricted to this protein family, the C-terminal domains are also found in unrelated proteins. A limited choice of C-terminal domains is found in association with the tripartite motif and determined the recent classification of the TRIM proteins in subfamilies [##REF##16434393##4##]. This conserved multi-domain structure appears to behave as an integrated module, rather than a collection of separate motifs, suggesting a possible common function [##REF##11331580##1##,##REF##16237670##6##].</p>", "<p>We previously classified 35 human RBCC-containing proteins as a gene-protein family and named it TRIM. We observed that these proteins have strong self-association ability, mainly mediated by their CC region, which results in the formation of large protein complexes. In most cases the TRIM proteins identify different discrete nuclear and/or cytoplasmic sub-cellular structures [##REF##11331580##1##].</p>", "<p>The presence of the RING domain and recent experimental evidence indicate that these proteins can act as E3 ubiquitin ligases, the proteins responsible for mediating the transfer of the ubiquitin moiety to the specific targets [##REF##11007473##7##, ####REF##12075357##8##, ##REF##11685209##9##, ##REF##12878161##10####12878161##10##]. Alteration of their activity within ubiquitylation processes might be responsible for the clinical manifestation observed in human diseases caused by mutations in TRIM genes [##REF##16175175##3##,##REF##16237670##6##]. <italic>PML</italic>, <italic>RFP</italic>, <italic>TIF</italic>, and <italic>EFP </italic>are implicated in tumor insurgence and progression [##REF##1652369##11##, ####REF##8753810##12##, ##REF##7744009##13##, ##REF##10781795##14####10781795##14##]. Other TRIM genes are involved in Mendelian inherited disorders: <italic>MID1 </italic>and <italic>MUL </italic>are altered in two developmental genetic diseases, Opitz Syndrome and Mulibrey nanism, respectively [##REF##9354791##15##,##REF##10888877##16##]. <italic>TRIM32 </italic>is involved in both a form of muscular dystrophy and a form of Bardet-Biedl Syndrome (BBS11), and <italic>MURF-1 </italic>is implicated in muscular atrophy [##REF##11822024##17##, ####REF##16606853##18##, ##REF##11679633##19####11679633##19##]. Ro52 is the target antigen of auto-antibodies in both Sjogren syndrome and Systemic Lupus Erythematosus [##REF##3485431##20##]. Finally, TRIM5α has been identified as the major factor restricting HIV-1 during the early phase of infection in Old World monkey cells [##REF##16175175##3##,##REF##14985764##21##].</p>", "<p>The TRIM family represents one of the largest classes of putative single protein RING-finger E3 ubiquitin ligases, strongly suggesting that the tripartite motif was selectively maintained to carry out a specialized basic common function within the ubiquitylation process. We used a genomic approach to complete the identification of all human TRIM genes and to study their evolutionary relationships in vertebrate and invertebrate organisms. We observed a general paradigm for the evolution of this family and propose a possible relationship between the evolution of TRIM genes and that of their function.</p>" ]

[ "<title>Methods</title>", "<title>Gene searching</title>", "<p>Known mammalian TRIM and TRIM-like gene/protein sequences were retrieved from the National Center for Biotechnology Information (NCBI) and re-defined by searching against their respective genome assemblies using BLAT at the UCSC genome browser (<ext-link ext-link-type=\"uri\" xlink:href=\"http://genome.ucsc.edu\">http://genome.ucsc.edu</ext-link>). All the 'corrected' sequences were then used as queries to search potential novel TRIM genes within the human, mouse, rat, cow, and dog genomes, using BLAT at UCSC and TBLASTN at NCBI [##UREF##1##67##] genome browsers, respectively. All searches have been performed in several iterations using default parameters. Human nr and EST databases were screened using the PHI-BLAST in several iterations using the patterns previously defined for B-box1 and B-box2 [##REF##11331580##1##]. We also searched the human, mouse, rat, cow, and dog proteomes using the B-box2 as a bait in five iterations of the PHI-BLAST program [##UREF##1##67##], which provides a highly sensitive analysis, taking advantage of the fact that the B-box2 is a peculiar constituent of TRIM proteins and does not produce a large amount of background noise in this analysis. The retrieved amino acid sequences were subsequently used to search the respective genomes for identifying their encoding loci. Representative B-box2 sequences were also used as queries for TBLASTN search of the five mammalian genomes to identify all the potential loci encoding TRIM and TRIM-like proteins. All the retrieved genomic sequences were aligned to the available cDNA/EST sequences to infer the gene architectures. For genes that lacked a transcript counterpart in public databases, we performed a careful manual examination of the genomic sequences BLAST-comparing them to the putative more closely related ortholog or paralog, looking for splicing donor and acceptor signals to define the exon-intron boundaries. Constructed open reading frames (ORFs) were conceptually translated into amino acid sequences and checked against their closest homologs. The original genome sequencing traces (Traces-WGS), which are available at the NCBI web site, were checked when the constructed coding sequences presented either stop codons or ORF frame-shifts. When a difference between WGS traces and the genomic assembly was evident, the constructed sequence was properly corrected. Comparison of human, mouse, rat, cow, and dog orthologous TRIM and TRIM-like genes showed that &gt;99% of splicing acceptor and donor sites were conserved at the same relative position in the coding sequence in all species, i.e. the gene structure of TRIM genes is identical among different mammals. To retrieve TRIM genes form ciona (<italic>Ciona intestinalis</italic>), chick (<italic>Gallus gallus</italic>), pufferfish (<italic>Tetraodon nigroviridis</italic>) and zebrafish (<italic>Danio rerio</italic>) we used a combination of PHI-BLAST and TBLASTN against nr protein and nucleotide databases at NCBI, respectively. The following genome releases have been used for this work: <italic>Homo sapiens</italic>, May 2004 assembly (NCBI Build 35); <italic>Mus musculus</italic>, February 2006 assembly (NCBI Build 36); <italic>Rattus norvegicus</italic>, June 2003 assembly (Baylor College of Medicine HGSC v.3.1); Bos taurus, March 2005 assembly (Baylor College of Medicine Btau_2.0); <italic>Canis familiaris</italic>, May 2005 assembly (Broad Intitute CanFam2.0). <italic>Drosophila melanogaster </italic>TRIM genes: <italic>CG1624</italic>, <italic>CG5206</italic>, <italic>CG12218</italic>, <italic>CG8419</italic>, <italic>CG5071</italic>, <italic>CG10719</italic>, <italic>CG31721</italic>. <italic>Caenorabditis elegans </italic>TRIM genes: arc1, B0281, ZK1240.1, F43C11.8, ZK1240.2, F43C11.7, ZK1240.9, ZK1240.3, ZK1240.8, ZK1240.6, C28G1.6, K09F6.7, lin41, C39F7, nhl-2, nhl-3, ncl-1, F47G9.</p>", "<title>Protein domain analyses</title>", "<p>To identify and analyze the domain composition of the TRIM and TRIM-like protein products we used the major alternative splicing isoforms, if more than one was available, and utilized different domain prediction programs. First, we submitted the TRIM amino acid sequences to the SMART tool [##UREF##2##68##] where we analyzed the sequence against contemporary the SMART and Pfam [##UREF##3##69##] domains databases. The denotation of the C-terminal domains found within the TRIM sequences are the following: MATH, SM00061; PHD, SM00249; BROMO, SM00297; IGFLMN, SM00557; EXOIII, SM00479; FN3, SM00060; PRY, SM00589; SPRY, SM00449; ARF, SM00177; NHL, PF01436 (Pfam). The tripartite motif domains were additionally analyzed as described below. Besides the SMART results, the RING and B-boxes domains were also defined in each TRIM and TRIM-like protein by hand using the previously published patterns [##REF##11331580##1##]. In order to obtain a new profiling, the sequences corresponding to each domain were then aligned using the PRATT 2.1 program [##UREF##4##70##] and the best scoring consensi were selected and integrated by hand. The order of the sequences in the alignment shown reflects their degree of sequence conservation. The region of each TRIM and TRIM-like protein immediately after the last Cys or His of the B-box2 domain was analyzed for Coiled-coil prediction with the Coil 2.2 program [##UREF##5##71##]. Analysis was performed with MTIDK and MTK matrices and both the weighted option, which takes into account the polarity of the residue within the predicted Coiled-coil heptad repeat, as well as the unweighted option. When differences of around 20–30% in Coiled-coil prediction were observed between the different methods utilized, the prediction was considered bad and not indicated in the list of Additional file ##SUPPL##1##2##. Moreover, only percentages of prediction higher than 50% were considered using two residue windows, 21 and 28 amino acids.</p>", "<p>Plant B-box containing proteins (from <italic>A. thaliana</italic>, <italic>O. sativa</italic>, <italic>P. sativum</italic>, <italic>B. nigra</italic>) were retrieved from the SMART B-box database [##UREF##2##68##].</p>", "<title>Evolutionary analyses</title>", "<p>To perform phylogenetic analysis, TRIM and TRIM-like protein sequences were aligned using MultAlin [##REF##2849754##72##] in a multi-step process. Only proteins containing the complete module R-B1-B2-CC were aligned in a first step, eliminating from each sequence the portion downstream of the coiled-coil domain and all segments that caused a gap to interrupt the alignment. A first phylogenetic tree was produced starting from this multi-alignment. In a successive step, each of the remaining protein sequences was singularly added to the multi-alignment, edited for exceeding amino acids, and assigned to a TRIM subgroup after inspection of the topology of the resulting phylogenetic tree. Once all TRIM and TRIM-like protein sequences were assigned to a subgroup, phylogenetic analyses were performed independently for each subgroup. TRIM37 was not included in any subgroup and therefore was used as an outgroup in all final analyses. Nucleotide sequences of TRIM5/6/22/34 and related non-human sequences were also aligned using MultAlin [##REF##2849754##72##], but in this case a gap-removal step was not necessary due to the high similarity among all considered sequences. Neighbor-Joining and bootstrap analyses were performed with Phylo_win [##REF##9021275##73##], computing the distances among sequences with all the methods available in the package (protein analysis: observed divergence with and without Poisson correction; PAM distance. DNA analysis: observed divergence; Jukes and Cantor distance; Kimura distance; Tajima and Nei distance; HKY distance; Galtier and Gouy distance; and LogDet distance) [##REF##9021275##73##] (and references therein). Gap-removal was set as pairwise rather than global to minimize information loss. Bootstrap values were computed over 1000 repetitions. All tree topologies resulted to coincide in the different methods for branches with a bootstrap value &gt;50. Evaluation of Ka/Ks values for pairs of human-mouse TRIM- and TRIM-like-coding sequences was performed at the Norwegian bioinformatics platform [##UREF##6##74##].</p>", "<p>Comparison of the two groups quantitative parameters (gene lengths, exon number, amino acid identity and Ka/Ks ratios) were analyzed using the two samples t-test (two-tail test assuming unequal variances) by the Microsoft Excel statistical package. The comparison of the two groups Ka/Ks distribution has been analyzed using a two-sample Kolmogorov-Smirnov test.</p>" ]

[ "<title>Results</title>", "<title>Defining the complete set of TRIM genes in humans and other mammals</title>", "<p>To search for all TRIM genes in humans, mouse, rat, cow, and dog, we screened their genomic sequences, using all known mammalian TRIM sequences as queries, with the BLAST and BLAT algorithms at the NCBI and UCSC genome browsers. We also performed a Pattern-Hit Initiated-Blast (PHI-Blast) search against both redundant and non-redundant databases using the sequence patterns that we previously defined for the two B-box domains as query [##REF##11331580##1##]. In addition, we used representative B-box1 and B-box2 sequences to perform TBLASTN genome screening aimed at identifying all the potential loci encoding for B-box-containing proteins. Each retrieved genomic sequence was compared to available EST/cDNA sequences to infer gene architecture. For those genes that lacked a transcript counterpart, we performed a careful manual examination of the genomic sequences by aligning them to the most closely related TRIM of the same or other species to define exon boundaries.</p>", "<p>By combining these methods in several iterations, we retrieved the entire set of human TRIM genes. While most of these have been recently reported in the context of other studies, we also report some novel TRIM genes [##REF##16175175##3##,##REF##16434393##4##,##REF##18213395##22##] (<ext-link ext-link-type=\"uri\" xlink:href=\"http://TRIMbase.tigem.it\">http://TRIMbase.tigem.it</ext-link>). Some of the genes we found are present as perfect or almost perfect multiple duplications in the pericentromeric region of chromosome 11 and it is difficult in these cases to establish whether they represent expressed genes (see also below). We also annotated the TRIM complement in mouse, rat, cow, and dog. The inventory of these sets and their comparisons are available at (<ext-link ext-link-type=\"uri\" xlink:href=\"http://TRIMbase.tigem.it\">http://TRIMbase.tigem.it</ext-link>).</p>", "<title>Domain composition of human TRIM proteins</title>", "<p>The majority of the human proteins reported in <ext-link ext-link-type=\"uri\" xlink:href=\"http://TRIMbase.tigem.it\">http://TRIMbase.tigem.it</ext-link> fulfill the TRIM rule of domain order and composition (RING, B-box(es), CC, C-terminal domain(s)). During our searches, we also found genes encoding 'incomplete' TRIM proteins, i.e. lacking one of the domains present within the tripartite motif (RING, B-box, or CC). Differently from the RING and CC domains, our analysis clearly indicated that B-box domains are virtually always present within the tripartite motif in metazoans. However, there are a few exceptions in which the B-box(es) domain is associated with only one of the domains belonging to the tripartite motif: in humans 6 proteins that possess B-box(es) lack the RING domain (B-box and CC) and 2 have a very short sequence after the B-box and almost entirely lack the CC region (RING and B-box).</p>", "<p>In the evolutionary analyses reported in this study we included the 68 genes listed in Table ##TAB##0##1##: the 'orthodox' TRIM genes and the 8 'incomplete' TRIM genes or TRIM-like genes (possessing the B-box domain associated with either the CC or the RING domains). The 'incomplete' TRIM-like genes are included in Table ##TAB##0##1## and mentioned within the text with their non-TRIM names to remark their non full adherence to the strict definition of TRIM member; within the databases they are also annotated with a TRIM name, which is reported (Table ##TAB##0##1##). Moreover, of the chromosome 11 pericentromeric clusters we included in the analyses only representative members.</p>", "<p>With the identification of the entire complement of the human TRIM and TRIM-like family, we confirmed and extended the domain composition features of these proteins. Within the TRIM modular structure we found that the spacing between adjacent domains is conserved. In fact, the distance between the RING domain and the first B-box, either type 1 or type 2, ranges from 35 to 55 residues; the distance between the two B-box domains ranges from 13 to 20 amino acids; and the spacing, partly occupied by the CC region, between the B-box2 and the C-terminal domain is usually 170–220 residue-long. The maintenance of the domain scaffold, order and spacing clearly indicates that the TRIM structure is a functional module.</p>", "<title>The Tripartite Motif</title>", "<p>We aligned the sequences of the RING finger domain of all human TRIM proteins to define a general TRIM-specific RING pattern (Additional file ##SUPPL##0##1##). Besides the cysteine and histidine residues, which coordinate the two zinc atoms, we found clear preferences for specific residues in positions that are probably required to maintain the cross-brace structure of the RING domain [##REF##7729428##23##]. The loop delimited by the second and third Cys residues has a tighter length range within the TRIM family (on average 11 residues) than within other RING-containing proteins [##REF##10662664##5##]. The second loop, bounded by Cys6 and Cys7, is frequently longer than the 48 residues of the general RING consensus [##REF##10662664##5##]. The RING domain has been associated with the ubiquitylation process and is mainly responsible for the interaction with the ubiquitin conjugating enzymes (E2) in the ubiquitylation cascade process [##REF##11007473##7##]. The different length and composition of the intervening sequences of the RING loops may underlie the binding specificity towards the different E2 enzymes.</p>", "<p>The comparison of the B-box domains from all TRIM and TRIM-like sequences confirmed that the pattern of Cys and His is similar, although clearly distinct, in the two types of B-boxes (Additional file ##SUPPL##0##1##) [##REF##11331580##1##,##REF##11428128##2##]. The B-box1 has a short and tight consensus in which, besides the Cys and His that coordinate two atoms of zinc [##UREF##0##24##], only two positions show a clear preference for a limited choice of residues. B-box2 sequences are longer than type 1 and their consensus is looser. The cysteine and histidine residues at all 8 possible coordination positions are highly conserved consistent with the recently reported B-box2 structure definition that revealed the coordination of 2 zinc atoms [##REF##17428496##25##] in contrast with previous data [##REF##8846787##26##]. Moreover, additional non-polar or hydrophobic residues are also maintained in defined positions. Twenty-two out of 60 TRIM and 8 TRIM-like proteins possess both B-boxes, with B-box1 always preceding B-box2, whilst the remaining proteins have a single type 2 B-box domain (Table ##TAB##0##1##).</p>", "<p>We observed that the third component of the tripartite motif, the Coiled-coil (CC) region, follows the B-box2 in all bona fide human TRIM proteins as well as in six of the eight TRIM-like proteins. Only RNF101/TRIM48* and RNF102/TRIM52* do not possess this Coiled-coil region as they are truncated immediately after the B-box2. In all other cases, the CC region is always confined within 120 amino acids from the end of the B-box2 domain and in approximately 50% of the human TRIM proteins is bipartite (Additional file ##SUPPL##1##2##).</p>", "<title>TRIM C-terminal domains</title>", "<p>The C-terminal domains found in the TRIM and TRIM-like family members are not an exclusive property of this family but are also present in otherwise unrelated proteins [##REF##11331580##1##,##REF##16434393##4##]. The definition of the full complement of human TRIM and TRIM-like proteins allowed us to update the occurrence of C-terminal domains displayed by these proteins (Table ##TAB##0##1##).</p>", "<p>The majority of the human TRIM and TRIM-like proteins (40 members) possess either the SPRY domain or the association of PRY and SPRY domain, also known as B30.2 or RFP-like domain. Table ##TAB##0##1## reports the presence of the PRY and SPRY domains that we found using the domain detection tools described in Methods and detailed in the legend but, due to the complicated and still debated relationship between PRY-SPRY and B30.2 domains, we will herein simply refer to them as SPRY [##REF##9204703##27##, ####REF##16313355##28##, ##REF##16498413##29####16498413##29##]. The SPRY domain in turn can be associated with Fibronectin type III repeat (FN3) [##REF##8548820##30##] and COS microtubule binding domain in different combinations [##REF##16434393##4##]. Five TRIM proteins display NHL and IGFLMN domains, either in association or alone [##REF##9868369##31##,##REF##10518939##32##]. TRIM56 C-terminal region shares sequence similarity with this domain although no clear NHL repeats are detected. Three TRIM and one TRIM-like proteins contain a PHD associated with a BROMO domain, a combination that was demonstrated to cooperate in nucleosome binding [##REF##15033350##33##]. Other domains are present in only one member of the TRIM family: the MATH domain in TRIM37; the ARF domain in TRIM23; and the EXOIII domain in TRIM19/PML [##REF##10888877##16##,##REF##8700863##34##,##REF##16460575##35##]. Fifteen TRIM and TRIM-like proteins do not possess a defined C-terminal domain. In these cases, either their coding region is limited to the tripartite motif or the C-terminal portion is not similar to any other known domains (Table ##TAB##0##1##).</p>", "<p>This comprehensive review of TRIM and TRIM-like associated C-terminal domains confirmed that a discrete number of motifs have been selected downstream of the tripartite motif in humans.</p>", "<title>The TRIM modular structure is metazoan-specific</title>", "<p>To trace back in evolution the origin of the TRIM family, we used human B-box1 and B-box2 sequences as queries to investigate the occurrence of these domains in the genomes of prokaryotic and eukaryotic representative species. We did not find any sequences similar to the B-box domains in prokaryotes. B-box sequences are present in plants with a consensus that is more similar to B-box1 than B-box2 (Fig. ##FIG##0##1A##). We examined 50 B-box containing proteins from 4 plant species (<italic>A. thaliana</italic>, <italic>O. sativa</italic>, <italic>P. sativum</italic>, <italic>B. nigra</italic>): the B-box is found alone or associated with a second B-box, with the CCT (<underline>C</underline>ONSTANS, <underline>C</underline>O-like, and <underline>T</underline>OC1) domain [##REF##10926537##36##], or with both. Differently from mammals, proximal and distal plant B-boxes (we analyzed a total of 60 B-box sequences) are very similar to each other and, consistently, do not separate in distinct branches in phylogenetic analysis (data not shown). No association with RING or Coiled-coil domains was detected in all the plant proteins analyzed.</p>", "<p>Besides plants and metazoans, we found B-box domains in some unicellular eukaryotes (unpublished observation). These protist species possess B-box domains that resemble either the plant or metazoan consensi, but the difficulty in attributing these lineages to specific clades compounds the tracing of the evolution of their B-box domain. In addition, since many of these protists are parasites of metazoans, we cannot rule out the possibility that horizontal gene transfer might have occurred [##REF##16344004##37##].</p>", "<p>Among the metazoans we also searched the genomes of two invertebrate species, <italic>Drosophila melanogaster </italic>and <italic>Caenorhabditis elegans</italic>, for the presence of B-box domains. Distinct proximal (B-box1) and distal (B-box2) domains are found in these species, sharing with mammals the same B-box1 pattern and a similar B-box2 consensus (Fig. ##FIG##0##1A##). The B-box domains in these species associate with a RING domain and a Coiled-coil region in a tripartite motif as in mammals. The tripartite motif is therefore exclusive to metazoans, despite the fact that its constitutive elements are not.</p>", "<p>However, these invertebrate organisms have TRIM complements that differ significantly from mammals: the fruitfly has 7 TRIM genes and the worm 18, 12 of which code only for a Tripartite motif (Fig. ##FIG##0##1B##). Fruitfly and worm TRIM proteins share many of the C-terminal domains found in humans, however, their proportion varies among these species, highlighting lineage-specific expansions, e.g. SPRY in humans (Fig. ##FIG##0##1B##).</p>", "<title>The human TRIM family can be subdivided in two distinct groups: group 1 and group 2</title>", "<p>Given the numerical and structural complexity of TRIM genes in humans, we sought to characterize the relatedness among members of the family. The presence of different combinations of domains characterized by spaced cysteine and histidine residues rendered a global and reliable alignment of all TRIM and TRIM-like proteins along their entire length difficult. We therefore performed an initial alignment using the B-box2 and Coiled-coil portion. The unrooted phylogenetic tree generated from this alignment supports a recent expansion of the genes that contain the SPRY domain and suggests a preliminary separation of the human TRIM proteins in two main groups based on domain composition and branch topology (Fig. ##FIG##1##2##). Group 1, composed of 34 proteins (29 TRIM and 5 TRIM-like proteins), includes a high proportion of members with a RING-B1-B2-CC module in combination with all the C-terminal domains found in TRIM proteins. Group 2 is composed of the remaining 34 proteins (31 TRIM and 3 TRIM-like proteins), which possess only the B-box2 domain and are mostly organized as RING-B2-CC-SPRY; the 5 proteins of this group that lack the SPRY domain consist of the tripartite motif alone (Fig. ##FIG##1##2##).</p>", "<p>The alignment of either single or combination of domains that compose the tripartite motif produces similar tree topologies (Additional file ##SUPPL##2##3##). This suggests that co-evolution of the domains present within the tripartite motif has occurred, i. e. this module mainly evolved as a single block, with no evidence of large rearrangements leading to domain acquisition or swapping among different TRIM family members; the only exceptions are the incomplete TRIM proteins that have lost one of the tripartite motif domain.</p>", "<p>We investigated whether the two groups show differences at the level of their genomic organization. Considering solely the coding region, group 2 genes span on average 10.3 kbp split in 5.7 coding exons compared to the 45.4 kbp and 8.3 exons of the genes that belong to group 1, differences that are statistically significant (<italic>P </italic>&lt; 0.01 for gene length and exon number). Besides the average values, the homogeneity of group 2 with respect to these two parameters is striking. In fact, group 2 gene lengths range from 1.4 to 27 kbp, with only 2 genes larger than 20 kbp, whereas group 1 genes are distributed within a larger range, 1.2 to 143 kbp. Homogeneity of group 2 is also observed for the distribution of the number of coding exons: about two thirds of group 2 TRIM genes are composed of 6 or 7 exons and only in one case they span over 10 exons; again the distribution for group 1 is broader, ranging from 1 to 20 exons.</p>", "<p>Taken together, group 2 genes are smaller and less complex than group 1. Interestingly, several group 2 genes are clustered in small chromosomal regions, especially within the Major Histocompatibilty Complex region in 6p21.33 (Table ##TAB##0##1## and Additional file ##SUPPL##3##4##) [##REF##9196055##38##,##REF##12622776##39##]. The high homogeneity of group 2 genes and their organization in clusters suggest a more recent origin of this group.</p>", "<title>Group 2 TRIM genes evolve more rapidly than group 1</title>", "<p>To investigate whether group 1 and 2 have different evolutionary dynamics, we compared the members of the human TRIM and TRIM-like set to their mouse counterparts (see <ext-link ext-link-type=\"uri\" xlink:href=\"http://TRIMbase.tigem.it\">http://TRIMbase.tigem.it</ext-link> and below for the definition of orthologous pairs). Ten out of 60 human TRIM and 8 TRIM-like genes (<italic>TRIM4</italic>, <italic>5</italic>, <italic>22</italic>, <italic>43</italic>, <italic>48*</italic>, <italic>49</italic>, <italic>52*</italic>, <italic>64</italic>, <italic>73</italic>, <italic>74</italic>) do not have a murine ortholog, whereas TRIM31, 15, 20, and 61 are divergent at the level of entire domains compared to the mouse. Interestingly, these 14 non-conserved or divergent genes fall within group 2 and represent an important proportion of this group (41%). Conversely, all human group 1 genes have a mouse ortholog.</p>", "<p>The degree of conservation of the remaining 54 human/mouse pairs is highly variable, ranging from 49% to more than 99% amino acid identity (peptide comparisons are available at <ext-link ext-link-type=\"uri\" xlink:href=\"http://TRIMbase.tigem.it\">http://TRIMbase.tigem.it</ext-link>). Group 2 pairs present on average 78% amino acid identity versus 89.2% of group 1 (<italic>P </italic>&lt; 0.01). Furthermore, the majority of group 1 proteins show 90–100% amino acid identity against 50–90% of most group 2 proteins (Fig. ##FIG##2##3A##).</p>", "<p>Fast-evolving genes tend to have a higher ratio of nonsynonymous substitutions per nonsynonymous site (Ka) to synonymous substitutions per synonymous site (Ks) than the slow-evolving ones. The average Ka/Ks ratio between human and rodent coding sequences is 0.18 [##REF##9689093##40##]. We analyzed the Ka/Ks ratio in human-mouse TRIM and TRIM-like orthologous pairs and found that group 1 genes present on average a ratio of 0.13 versus 0.29 of group 2 (<italic>P </italic>&lt; 0.01). The Ka/Ks distribution is also significantly different (<italic>P </italic>= 0.01091) between the two groups: the majority of group 1 genes display values below 0.1 while most of group 2 show values above 0.1 (Fig. ##FIG##2##3B##). Furthermore, only members of group 2 (TRIM38, 40, 60, 75) exceed a Ka/Ks ratio value of 0.5 (Fig. ##FIG##2##3B##).</p>", "<p>Taken together, our analyses indicate that group 2 genes evolve more rapidly compared to group 1. This suggests that the two groups may be subject to different evolutionary constraints, likely underlying species-specific adaptations.</p>", "<title>Group 2 is absent in invertebrates and includes species-specific genes in mammals</title>", "<p>Preliminary rounds of sequence alignment and evolutionary analyses allowed us to divide the TRIM and TRIM-like family members into major classes and subsequently generate their phylogenetic trees separately using the full-length protein sequences from man, mouse, fruitfly and worm (see Methods for details). The results show that these proteins are evolutionarily organized in two groups that coincide with group 1 and group 2 (Fig. ##FIG##3##4## and ##FIG##4##5##). The only discrepancy with the previous subdivision is TRIM62, which segregates with group 2 proteins in the evolutionary analysis. Within group 1, TRIM37 did not segregate with any subgroups in preliminary studies and therefore it was used as an outgroup in all phylogenetic analyses.</p>", "<p>Group 1 is further divided into subgroups that grossly match with the domains downstream of the tripartite motif (Fig. ##FIG##3##4##). This analysis confirms that members of group 1 are present in both human and mouse with a strict 1:1 orthologous correspondence and are represented in invertebrates (Fig. ##FIG##3##4##). By means of this phylogenetic analysis we could appreciate better the mammals-invertebrates TRIM relationship and, since the worm and the fruitfly possess many of the C-terminal domains present in mammals, we could follow the late evolution of the TRIM modular structure as discussed below. Interestingly, no invertebrate sequences are found to be homologous to members of group 1 subgroup E, which includes TRIM and TRIM-like proteins with B1, B2, and SPRY in various combinations (Fig. ##FIG##3##4E## and Table ##TAB##0##1##). Group 2 proteins are not represented in invertebrate species as well, and have either a complete or a truncated RING-B2-CC-SPRY domain composition (Fig. ##FIG##4##5## and Table ##TAB##0##1##). The major structural difference between subgroup E and group 2 is the presence of the B-box1, which indicates that group 2 proteins could have derived from a subgroup E member upon loss of B-box1.</p>", "<p>Interestingly, the evolutionary analysis of group 2 proteins showed, in addition to 24 pairs of orthologs, the presence of species-specific TRIM and TRIM-like proteins, not only in humans (see above) but also in mouse (Fig. ##FIG##4##5##), indicating high dynamicity in the evolution of this group compared to group 1.</p>", "<p>The analysis of the TRIM and TRIM-like complements of rat, cow, and dog showed that these species have the same set of group 1 genes as humans and mouse. Conversely, the sets of group 2 genes are different and specific to each species, including closely related ones such as mouse and rat, with only 17 genes (50–70% of group 2 genes) shared among all species (<ext-link ext-link-type=\"uri\" xlink:href=\"http://TRIMbase.tigem.it\">http://TRIMbase.tigem.it</ext-link>). These differences are due to events of gene duplication, deletion, or degeneration that likely occurred during the evolution of each single lineage. Remarkably, in three cases the same gene has undergone degeneration or deletion independently in different lineages: human <italic>TRIM38 </italic>orthologs, found in mouse and cow, are pseudogenes in rat and dog; human <italic>TRIM60 </italic>has orthologous loci in all the examined mammals but has become pseudogenes in rat, cow, and dog; human <italic>TRIM61 </italic>orthologs are found to be pseudogenes or absent in all organisms but mouse. Furthermore, two TRIM genes are present in the five species in three different states: <italic>TRIM15 </italic>is intact in human and cow, presents premature stop codons in mouse and rat, and is a pseudogene in dog; similarly, <italic>TRIM31 </italic>is intact in mouse and rat, has distal frameshifts in humans and cow, and is absent in dog. Both premature stop codons and frameshifts result in truncated proteins that have lost the SPRY domain. These losses may represent a degenerative step towards the complete inactivation of these genes.</p>", "<p>Of note, massive gene duplications and rearrangements had occurred at the level of several genes of group 2 that cluster in the same chromosomal location (Table ##TAB##0##1##). A thorough characterization of the TRIM-rich 6p21.33 locus is reported in human and chicken [##REF##12622776##39##,##REF##15744538##41##,##REF##9866204##42##]. Indeed, in addition to the human genes here presented, extra-copies of group 2 genes <italic>TRIM43</italic>, <italic>48*</italic>, <italic>49</italic>, and <italic>64 </italic>are clustered at 2q11.1, 11q11.1, and 11q14.3 for a total of 11 predicted genes and 14 pseudogenes. These clustered loci have paralogs, but not orthologs, in some of the other examined mammalian species (<ext-link ext-link-type=\"uri\" xlink:href=\"http://TRIMbase.tigem.it\">http://TRIMbase.tigem.it</ext-link>).</p>", "<p>A further example of recent gene evolution is the genomic cluster at 11p15.4 containing <italic>TRIM5</italic>, a gene involved in HIV-1 viral restriction in some primates [##REF##14985764##21##]. This cluster includes <italic>TRIM5 </italic>and <italic>22</italic>, which are currently considered primate-specific, in addition to <italic>TRIM6 </italic>and <italic>34</italic>, which are not [##REF##16648259##43##,##REF##15857996##44##]. By means of comparative and evolutionary analyses (Additional file ##SUPPL##4##5##) we found that the entire cluster, including <italic>TRIM5 </italic>and <italic>22</italic>, was present in the last common ancestor of humans, cow, and dog, supporting a common origin for <italic>TRIM5 </italic>and its cow functional ortholog <italic>LOC516599 </italic>rather than evolutionary convergence as previously proposed [##REF##16648259##43##].</p>", "<title>TRIM genes in other vertebrate species</title>", "<p>Given the situation in mammals, we asked whether the TRIM complements evolved likewise in other vertebrate species. We searched the databases for TRIM and TRIM-like sequences in representative aves and fish species, chick (<italic>Gallus gallus</italic>) and a pufferfish (<italic>Tetraodon nigroviridis</italic>), respectively. In addition, we included in our analysis the urochordata <italic>Ciona intestinalis</italic>, a representative of the early chordate lineage from which the vertebrates originated. The searches were performed combining different iterations of PHI-BLAST (using the B-box2 pattern) and TBLASTN and BLASTP against nr protein and nucleotide databases at NCBI, starting from both human TRIM proteins and TRIM sequences found in the above species. We found 10 TRIM sequences in ciona, 37 in chick and 58 in pufferfish; all the sequences we retrieved were present in the databases as assessed or predicted genes and only some of them (especially in chicken) were already annotated as TRIM genes (the sequences retrieved for these species are reported at <ext-link ext-link-type=\"uri\" xlink:href=\"http://TRIMbase.tigem.it\">http://TRIMbase.tigem.it</ext-link>). These novel TRIM complements should not necessarily be regarded as complete since in these species there are still regions not yet sequenced or unequivocally assembled. We compared these additional vertebrate sequences to the human sequences using multiple protein alignments and phylogenetic tree constructions using as a paradigm the subdivision into groups and subgroups of Figure ##FIG##3##4## and ##FIG##4##5##. Detailed analyses of these novel genes, at the level of transcript, protein and genomic locus, are required and will be addressed elsewhere, therefore the phylogenetic analyses we present show relationships and are not intended to represent precise evolutionary distances.</p>", "<p>In all species analyzed, we found clear orthologs for TRIM23 and TRIM37 (Fig. ##FIG##5##6A##). With respect to subgroup C, there is one representative of this subgroup in ciona, closely related to the fruitfly member. All the known mammalian members of this subgroup are found in chick with recognizable orthologous relationships (Fig. ##FIG##5##6B##). This is not completely true for tetraodon in which 3 members belonging to this group have been found: one orthologous to TRIM33 and the others representing fish-specific duplications related to the TRIM33-24 clade. Genes strictly related to KIAA0298/TRIM66* and TRIM19 have not been found in tetraodon (Fig. ##FIG##5##6B##). In ciona there are 5 representative members of subgroup A, 3 representing the ancestors of different subclades (TRIM9-67; TRIM1-18; TRIM54-55-63) and 2 apparently more ancestral genes. DIBP/TRIM44* appears to be uniquely represented in mammals. All the other members are present in chick and fish although orthology is clear only for TRIM9, 67, 1, 18, and 36. TRIM46, if truly absent, may have been lost in chick while of the MURF group (TRIM54, 55, 63) 2 orthologues are present in chick while 4 members with no direct orthologous relationship are present in fish (Fig. ##FIG##5##6C##). An analogous situation is observed for subgroup D. In fact, ciona has two representative members; orthologous relationship with human is observed for both chick and fish for four members (TRIM2, 3, 45, 71). TRIM32 might have been lost in chick and TRIM56 in both chick and fish (Fig. ##FIG##5##6D##). For the above group 1 subgroups, we therefore found that most of the mammalian components are present in chick often with clearly recognizable orthologous relationship. Although the number of members within these subgroups is similar also in tetraodon, the TRIM and TRIM-like genes in this species, consistent with a larger evolutionary distance, have duplicated and diverged more extensively, sometimes obscuring orthologies.</p>", "<p>A different situation is observed with the group 1 subgroup E and, as expected, with group 2. As for the other invertebrates, ciona genes are not represented in this subgroup and in group 2. Orthologous relationship among mammals, chick and fish is observed only for two genes of subgroup E (ATDC/TRIM29* and TRIM65). The other genes within the subgroup are more conserved in chicken while in tetraodon many independent duplication events occurred (Fig. ##FIG##5##6E##). Even more extensive duplication events and independent divergences are observed for group 2 genes. In this case, close homology among the three species is only recognizable for TRIM35 and 62. A couple of other mammalian TRIM genes are conserved in chick (TRIM7 and 41) but the remaining genes in the three species have been subjected to independent duplications and evolution (Fig. ##FIG##5##6F##).</p>", "<p>To confirm in fish the presence of so many TRIM sequences poorly related to the human TRIM genes, we search TRIM and TRIM-like genes also in zebrafish (<italic>Danio rerio</italic>) using the same criteria and methods described for the chick and tetraodon. Differently from tetraodon, zebrafish presents an elevated number of TRIM and TRIM-like genes; we found 240 entries corresponding to independent genes (the list of zebrafish genes is reported at <ext-link ext-link-type=\"uri\" xlink:href=\"http://TRIMbase.tigem.it\">http://TRIMbase.tigem.it</ext-link>). Also in this case, the number of genes encoding for group 1 TRIM (excluding subgroup E) is comparable to the number in mammals, chick and pufferfish (1 subgroup B; 5 subgroup C; 16 subgroup A; 12 subgroup D) although in many cases clear duplication events occurred. However, the great expansion of the TRIM genes in the zebrafish is associated with members belonging to the group 1 subgroup E and group 2 genes (data not shown).</p>", "<p>Analyses of TRIM complements in aves and fish corroborate the high conservation of group 1 genes during evolution and highlight the generation of unique sets of group 2 genes in each vertebrate species analyzed. Moreover, the data in non-mammalian vertebrates, especially in tetraodon, confirm that members of the group 1 subgroup E very likely gave rise to group 2 genes.</p>", "<p>In conclusion, our study indicates the presence of two distinct TRIM gene groups. Group 1 is evolutionary more ancient than group 2 and is likely to contain basic functions that are essential to both vertebrate and invertebrate species. On the other hand, group 2 is younger and more dynamic, possibly acting as a sort of TRIM genes \"reservoir\" to develop novel species-specific functions.</p>" ]

[ "<title>Discussion</title>", "<p>Here, we report the identification and genomic characterization of the full complement of the human TRIM family examined from an evolutionary perspective by comparison with several vertebrate and invertebrate species.</p>", "<p>We definitively assessed that the B-box domain is only present within the tripartite module in metazoans, with the few exceptions mentioned above and discussed below, and is therefore the defining domain of the TRIM family. We redefined the B-box1 and B-box2 consensi as well as the TRIM specific RING finger pattern using all human sequences. Within these domains we found conservation not only of the residues putatively involved in metal coordination, but also of other amino acids that compose the novel consensi. It will be interesting to model the RING, B-box1 and B-box2 sequences of the TRIM proteins on these structures to study the possible role of the conserved residues in relation to the ubiquitylation cascade [##REF##11007473##7##, ####REF##12075357##8##, ##REF##11685209##9####11685209##9##,##UREF##0##24##,##REF##17428496##25##].</p>", "<p>Based on our analyses, we propose a general model of TRIM structure evolution (Fig. ##FIG##6##7##). Our studies suggest that the origin of the B-box domain is quite ancient and probably dates back to a common ancestor of plants and metazoans. The plants maintained either one or two B-boxes without apparent sequence differentiation into proximal and distal. Conversely, metazoans differentiated a pair of B-boxes into a proximal and a distal type. In an early step of metazoan evolution, a RING domain and a Coiled-coil region associated with the B-box(es) to generate a solid tripartite module that has been maintained from invertebrates to mammals. From then onward, the tripartite motif has evolved as a unique block and it is frequently encoded by a single exon. Interestingly, we found different mammalian and invertebrate B-box2 patterns. This may underlie functional coupling with specific interactors in each lineage, which would have forced convergence at specific sites. A similar species-specific evolutionary convergence was recently described for sulfatase enzymes and their common post-translational modification factor [##REF##16174644##45##].</p>", "<p>Before invertebrate-vertebrate lineage split, the tripartite module has been associated with a discrete number of C-terminal domains. Addition and loss of C-terminal domains and structure remodeling have then occurred in the various evolutionary lineages (Fig. ##FIG##6##7##). In at least one case the same domain acquisition has occurred independently in primates and arthropods. In fact, in some species of New World monkeys (<italic>Aoutus </italic>genus), Cyclophilin A (CypA) is fused with the tripartite motif of TRIM5 [##REF##15243629##46##]. TRIM5 confers a potent block to HIV-1 infection in Old World primates, while Cyclophilin A (CypA) enhances infection by direct interaction with the HIV capsid [##REF##14985764##21##,##REF##12897779##47##]. HIV-1 blockage in <italic>Aoutus </italic>cells was explained by the exclusive presence of the <italic>TRIM5-CypA </italic>chimeric gene [##REF##15243629##46##,##REF##16203999##48##,##REF##16282502##49##]. Interestingly, we found that a tripartite motif is associated with a Cyclophilin domain also in fruitfly CG5071 indicating evolutionary convergence. Independent events of gene fusion are considered a hallmark of functional coupling that can be also present in those species in which a similar gene fusion is not observed [##REF##10573422##50##]. Similar processes of fusion between functionally associated domains may have been one of the mechanisms underlying the selection of C-terminal domains during TRIM evolution.</p>", "<p>The early association of the B-box modules with the RING finger, a domain linked to the ubiquitylation process, has eventually brought the proteins possessing a tripartite motif to exert a common basic biochemical function, i.e. ubiquitin ligase [##REF##16237670##6##]. The large number of proteins belonging to this family in mammals highlights the success of this module to undertake its task. Since the TRIM family represents one of the largest RING finger classes, it is tempting to speculate that, among the myriads of cellular E3 substrates, a large proportion is demanding the unique tripartite structure for reasons yet to be discovered. Whereas the tripartite motif may provide the catalytic E3 activity and the ability to form the scaffold of the TRIM-defined sub-cellular compartments [##REF##11331580##1##], the C-terminal region may contribute to select the specific substrate and/or direct the tagged substrate towards downstream pathways. The PHD-BROMO domain, for example, determines the association with chromatin, and the TRIM and TRIM-like proteins containing the PHD-BROMO domain are consistently involved in chromatin remodeling [##REF##12096914##51##,##REF##15820681##52##]. Along the same way, MID1/TRIM18 and the related TRIM proteins that possess a COS microtubule-binding domain exert their role on the cytoskeleton [##REF##16434393##4##].</p>", "<p>It should be reaffirmed that not all the proteins we included in our study retain an entire tripartite motif. As mentioned in the 'Results section', we decided to include all the genes encoding for proteins with B-box motifs and that in human correspond to the 'complete' TRIM proteins (with RING, B-box and CC) and few 'incomplete' TRIM proteins (or TRIM-like) presenting only two of the three tripartite motif composing domains (with either 'RING and B-box' or 'B-box and CC'). This might pose a formal problem on what should be classified as a TRIM protein. In the classic definition, a protein family is composed of proteins that have a common phylogenetic origin and share a degree of amino acid identity/similarity above an established threshold. In the case of the TRIM family, the initial definition of family was based on the observation that most members of this protein family share the tripartite arrangement at their N-terminal portion [##REF##11331580##1##]. What was not clear was whether the TRIM proteins had a common origin or rather they were the result of domain swapping from evolutionarily unrelated proteins. Our analyses demonstrate that all the proteins with a RING-B-box-CC module actually have a common evolutionary origin. Therefore, what was raised as an 'operative' definition is now demonstrated to be perfectly adherent to the classic definition of a gene/protein family. Our data allow the same conclusion to be drawn for the 'incomplete' TRIM proteins which possess the B-box motif and which we found evolutionarily belonging to the TRIM family. In support of that, some of the 'incomplete' TRIM proteins also present C-terminal domains characteristic of the 'complete' TRIM proteins. This parallel cannot be used for other domains present within the tripartite motif, e.g. the RING domain has been 'used' to build many different protein families in association with several TRIM unrelated domains [##REF##18213395##22##]. On the other hand, we think that a strict definition of TRIM family based only on function is not feasible at present. As discussed above, the presence of the RING domain suggests a role as E3 ubiquitin ligases for the TRIM proteins. Experimentally, this has been proven for some TRIM proteins and we cannot exclude that some of them, although containing the RING domain in the proper tripartite motif, might have a different biochemical role. What is the role of the 6 RING-less proteins we included in our study? They may be involved in ubiquitylation as well by, for example, acting as regulators of orthodox TRIM proteins through hetero-interaction. Given that the recent solution of the B-box1 and B-box2 domains revealed a strong structural similarity with the RING domain [##UREF##0##24##,##REF##17428496##25##], it is tempting to speculate that these domains may interact with components of the ubiquitylation machinery and attribute these RING-less proteins the role of E3 ubiquitin ligases. Coherently with these observations, we propose to include within the TRIM family all the proteins that are phylogenetically related to established TRIM members and that have a tripartite motif at their N-terminus, including the few examples in which part of this motif has been lost.</p>", "<p>The relatively small number of TRIM genes in lower eukaryotes compared to mammals suggests rapid and recent changes of the TRIM family. Our study revealed the presence of two main groups of mammalian TRIM genes that show distinct evolutionary features and that we named group 1 and group 2. Group 1, that is in turn subdivided in several subgroups, is composed of genes that are present in human, mouse, rat, dog, and cow with a one to one relationship. Although orthology with mammals is not always recognizable, this group of genes is highly conserved also in other vertebrates (chick and fish) in number and structure. Our data on the Ka/Ks ratio of human and mouse group 1 genes suggest that they are subject to purifying selection aimed at conserving their function. It is conceivable that group 1 consists of diversified and essential TRIM and TRIM-like functions for which little or no redundancy is present. Consistently, many group 1 genes are involved in basic cellular processes, such as cell cycle progression and transcriptional regulation, and result, when mutated, in developmental disorders, muscular phenotypes, cancer insurgence, etc. [##REF##11428128##2##,##REF##16237670##6##]. Some group 1 TRIM genes have also been found to be involved in viral response, namely <italic>TRIM1 </italic>[##REF##15249690##53##], <italic>TRIM19/PML </italic>[##REF##16175175##3##,##REF##11704856##54##,##REF##8277273##55##] and <italic>TRIM32 </italic>[##REF##7778269##56##]. TRIM19/PML, besides its involvement in acute promyelocytic leukemia, has been shown to interfere with the replicative cycle of many DNA and RNA viruses and evidence indicate that it may represent a broad-spectrum cellular defence factor [##REF##16175175##3##].</p>", "<p>The important increase of TRIM number in vertebrates is primarily due to the buildup of the genes that constitute group 2. Group 2 is in fact evolutionarily more recent than group 1, is not represented in invertebrates, and evolves at a faster rate. Interestingly, many TRIM proteins that belong to group 2 have been recently associated with cellular innate immunity towards viral infection. In addition to <italic>TRIM5α</italic>, other members are being investigated as potential retrovirus restriction factors. Among them, <italic>TRIM21</italic>, <italic>22 </italic>and <italic>34 </italic>are regulated by interferons, a family of secreted proteins that exert antiviral and immunomodulatory activities [##REF##16289103##57##]. Moreover, other group 2 genes, <italic>PYRIN</italic>/<italic>TRIM20* </italic>and <italic>TRIM21 </italic>are involved in immuno-related diseases [##REF##3485431##20##,##REF##9288758##58##,##REF##9288094##59##]. Interestingly, group 2 proteins share the same C-terminal motif, the SPRY domain. In the case of TRIM5α, this domain is responsible for the species-specific HIV-1 restriction and is subject to positive selection in primates, underlying its possible role in directing and specifying capsid recognition [##REF##15857996##44##,##REF##15689398##60##]. Of note, the SPRY domain is also present in SOCS proteins, involved in cytokine signaling and innate immunity, and in the BTN family of lymphoid expressed proteins, possibly involved in immune regulation [##REF##16313355##28##,##REF##9419338##61##,##REF##15259006##62##]. It has been proposed that the sharing of the SPRY domain between the TRIM and BTN family members located within the MHC locus is somewhat linked to their immunological function [##REF##15744538##41##]. The SPRY domain might therefore confer to group 2 proteins the ability to specifically recognize viral capsids and interfere with early steps of viral infection. Differently from anti-viral group 2 proteins, group 1 TRIM19/PML interferes with general mechanisms of viral replication common to various viruses and consistently is not subject to positive selection [##REF##16460575##35##].</p>", "<p>Our comparative analysis in five mammalian species shows that subsets of group 2 TRIM and TRIM-like genes are different and specific in each examined lineage. This is more evident in the three non-mammalian vertebrates analyzed, where large numbers of newly identified group 2 genes mainly lie on species-specific evolutionary clades. This observation might underlie dispensability/redundancy of some group 2 genes, which could have provided the basis for novel species-specific roles during evolution. The presence of clusters composed of massively duplicated group 2 genes, in mammals but also in chick, suggests that they may be hot-spots for TRIM gene production and remodeling. In the case of the teleost fish species some of the duplications may be the remnants of the whole genome duplication event early in the teleost lineage. Global duplication is not however enough to explain the large and independent expansion of subgroup E and group 2 genes in the teleosts and a different cause must underlie these expansions. It is interesting to note that similar to the Group 2 TRIM and TRIM-like genes, other families of genes involved in innate immune response and in particular the components that interact with pathogens have been subject to similar large lineage specific expansions in the teleost fish [##REF##18039395##63##]. Moreover, since we observed that group 2 genes tend to have a human/mouse Ka/Ks ratio higher than group 1 genes, it is tempting to speculate that some group 2 genes other than <italic>TRIM5α </italic>may be subject in some species to positive selection at specific sites to counteract species-specific battles against viral infections, as it has been shown for other family of genes involved in innate cellular immunity [##REF##2283951##64##, ####REF##15578175##65##, ##REF##15728394##66####15728394##66##].</p>" ]

[ "<title>Conclusion</title>", "<p>We found that the TRIM domain structure is an innovation of metazoans. The growing evidence for a common biochemical function of the TRIM proteins as ubiquitin ligases justifies the maintenance of their basic modular structure throughout evolution. Our studies indicate the presence of two distinct TRIM gene groups. Group 1 is evolutionary more ancient than group 2 and is likely to contain basic functions that are essential to both vertebrate and invertebrate species. On the other hand, group 2 is younger and more dynamic, possibly acting as a sort of TRIM genes \"reservoir\" to develop novel functions. Since some of the TRIM genes that belong to this group are implicated in innate immune response, we propose that the different selection we observed for this group of genes underlies pressure towards rapid changes necessary to counteract species-specific battles against viral infection.</p>" ]

[ "<title>Background</title>", "<p>The TRIM family is composed of multi-domain proteins that display the Tripartite Motif (RING, B-box and Coiled-coil) that can be associated with a C-terminal domain. TRIM genes are involved in ubiquitylation and are implicated in a variety of human pathologies, from Mendelian inherited disorders to cancer, and are also involved in cellular response to viral infection.</p>", "<title>Results</title>", "<p>Here we defined the entire human TRIM family and also identified the TRIM sets of other vertebrate (mouse, rat, dog, cow, chicken, tetraodon, and zebrafish) and invertebrate species (fruitfly, worm, and ciona). By means of comparative analyses we found that, after assembly of the tripartite motif in an early metazoan ancestor, few types of C-terminal domains have been associated with this module during evolution and that an important increase in TRIM number occurred in vertebrate species concomitantly with the addition of the SPRY domain. We showed that the human TRIM family is split into two groups that differ in domain structure, genomic organization and evolutionary properties. Group 1 members present a variety of C-terminal domains, are highly conserved among vertebrate species, and are represented in invertebrates. Conversely, group 2 is absent in invertebrates, is characterized by the presence of a C-terminal SPRY domain and presents unique sets of genes in each mammal examined. The generation of independent sets of group 2 genes is also evident in the other vertebrate species. Comparing the murine and human TRIM sets, we found that group 1 and 2 genes evolve at different speeds and are subject to different selective pressures.</p>", "<title>Conclusion</title>", "<p>We found that the TRIM family is composed of two groups of genes with distinct evolutionary properties. Group 2 is younger, highly dynamic, and might act as a <italic>reservoir </italic>to develop novel TRIM functions. Since some group 2 genes are implicated in innate immune response, their evolutionary features may account for species-specific battles against viral infection.</p>" ]

[ "<title>Authors' contributions</title>", "<p>MS carried out the evolutionary studies and contributed to the identification of TRIM genes in different species; he contributed to the design of the experiments and to the writing of the manuscript. SC identified the entire set of TRIM genes in human and performed the single domains alignments. BF collected all the data in the database and performed statistical analyses. AB contributed to the interpretation of the data and to the drafting the manuscript. GM conceived, designed and coordinated the study and wrote the manuscript.</p>", "<title>Additional files</title>", "<p>The complete sets of human, mouse, rat, dog and cow TRIM and TRIM-like genes and pseudogenes as well as their sequence comparisons are available at <ext-link ext-link-type=\"uri\" xlink:href=\"http://TRIMbase.tigem.it\">http://TRIMbase.tigem.it</ext-link>. At the same site are the TRIM related sequences from ciona, chick, tetraodon, and a list with the accession numbers of the zebrafish TRIM-like genes. See Additional files ##SUPPL##0##1## to ##SUPPL##4##5##.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Mario Traditi and Angelo Raggioli for informatic assistance and Luciana Esposito and Adriana Zagari for helpful discussion; we are grateful to Graciana Diez-Roux, Anna Savoia, Elena I. Rugarli, and Henrik Kaessmann for critical reading of the manuscript. This work was supported by the 'Italian Telethon Foundation' (TGMP4.2 to GM).</p>" ]

[ "<fig id=\"F1\" position=\"float\"><label>Figure 1</label><caption><p>TRIM domains in evolution. <bold>A</bold>) Logo representation of the sequences of Plant B-box (60 B-box sequences; representative species: <italic>A. thaliana</italic>, <italic>O. sativa</italic>, <italic>P. sativum</italic>, <italic>B. nigra</italic>); Metazoan B-box1 (all the B-box1 sequences in representative species: <italic>H. sapiens</italic>, <italic>D. melanogaster</italic>, and <italic>C. elegans</italic>); Mammalian B-box2 (all B-box2 sequences in representative species: <italic>H. sapiens</italic>) and Invertebrate B-box2 (all B-box2 sequences in representative species: <italic>D. melanogaster</italic>, and <italic>C. elegans</italic>). The overall height of each position is proportional to its information content and, within a given position, the conservation of each residue is represented as the relative height of amino acid symbols. Shaded columns indicate the residues involved in the coordination of zinc atoms. Blue bars represent amino acid segments of variable length; the mean value for each segment is reported. Red bars represent segments of fixed amino acid length that are present only in a proportion (indicated in red above the bar) of proteins. <bold>B) </bold>TRIM complements of humans, fruitfly (<italic>D. melanogaster</italic>) and worm (<italic>C. elegans</italic>). The total number of TRIM and TRIM-like genes in each species is indicated (top). The presence of the TRIM associated C-terminal domains is indicated with the same color code (bottom). The length of each bar in the bottom part is proportional to the number (also indicated) of the relative TRIM C-terminal domains found in each species.</p></caption></fig>", "<fig id=\"F2\" position=\"float\"><label>Figure 2</label><caption><p>Relatedness of the human TRIM and TRIM-like proteins. <bold>A</bold>) Unrooted phylogenetic tree generated upon alignment of the B-box2 and Coiled-coil region of human TRIM and TRIM-like proteins. The numbers indicate the TRIM family members; numbers with an asterisk indicate the 'incomplete' TRIM proteins named with their alternative TRIM name (see Table 1); light blue circles indicate the presence of the B-box1 domain; the colored open circles represent the different C-terminal domains as indicated in the figure. Partition in groups 1 and group 2 is indicated. <bold>B</bold>) Representative protein structures of the two groups obtained in A); dashed parentheses indicate that B-box 1 may not be present.</p></caption></fig>", "<fig id=\"F3\" position=\"float\"><label>Figure 3</label><caption><p>Group 1 and group 2 TRIM gene conservation in human and mouse. <bold>A</bold>) Distribution of the percentage of amino acid identity between human TRIM and TRIM-like proteins and their murine counterparts. Group 2 (yellow) and group 1 (grey). The bars represent the number of human TRIM genes (Y axis) for each percentage of identity interval (X axis); NO indicates absence of a murine counterpart. <bold>B</bold>) Distribution of the Ka/Ks ratios observed in human-mouse orthologous TRIM pairs considering the two groups separately. The bars represent the number of TRIM pairs (Y axis) for each Ka/Ks value interval (X axis).</p></caption></fig>", "<fig id=\"F4\" position=\"float\"><label>Figure 4</label><caption><p>Phylogenetic analysis of human (h, blue), mouse (m, blue), fruitfly (CGs, red), and worm (green) group 1 TRIM and TRIM-like proteins. Human and mouse TRIM proteins are indicated with their TRIM numbers; 'incomplete' TRIM proteins are indicated with their alternative TRIM number with an asterisk (see Table 1); fruitfly and worm sequences are indicated with GenBank accession numbers. Bootstrap support values above 50% based on 1000 replicates are shown. The main domain distal to the tripartite motif is indicated on the right; ND indicates no known domain detected. Panels A-E show the evolutionary relationships among the members of TRIM subgroups that belong to group 1; the worm TRIM genes composed of the R-B2-CC motif only (Figure 2B) represent a separate group related to group 1 and are not shown in the figure. TRIM37 (C-terminal domain: MATH) did not segregated within any subgroups in preliminary analyses and was therefore used as an outgroup in all phylogenetic analyses. The trees were drawn to the scale of amino acid sequence divergence indicated at the bottom right corner. A) Subgroup A includes FN3 and FN3-related TRIM sequences. Fruitfly CG31721 and its worm ortholog C39F7 are the only invertebrate proteins present in the FN3 subgroup and segregate with mammalian TRIM9 and 67. B) Subgroup B includes ARF-related TRIM sequences. Genes encoding a protein homologous to TRIM23 are found in worm and in the honeybee <italic>Apis mellifera </italic>(ENS10667 = ENSAPMT00000010667) but not in <italic>D. melanogaster</italic>, suggesting that the ARF domain has been acquired by a tripartite-gene precursor before vertebrate-invertebrate lineage separation, and has occasionally been lost in some species. C) Subgroup C includes PHD-BROMO and PHD-BROMO-related TRIM sequences. Fruitfly <italic>CG5206 </italic>behaves as an outgroup for all human and mouse PHD-BROMO proteins, suggesting that it may be regarded as an ortholog of their protein ancestor. D) Subgroup D includes IGFLMN-related TRIM sequences. This subgroup is the only example of TRIM expansion in invertebrates, because it includes worm and fly genes that do not have any direct correspondent in mammals. E) Subgroup E includes TRIM proteins with B1, B2, and SPRY in various combinations (see Table 1). No invertebrate sequences are found within this subgroup.</p></caption></fig>", "<fig id=\"F5\" position=\"float\"><label>Figure 5</label><caption><p>Phylogenetic analysis of human (h) and mouse (m) TRIM and TRIM-like proteins from Group 2. TRIM proteins are indicated with their TRIM number ('incomplete' TRIM proteins are indicated with their alternative TRIM number with an asterisk, see Table 1). No invertebrate TRIM proteins are represented in this group. Bootstrap support values above 50% based on 1000 replicates are shown. Group 1 TRIM37 sequences are used as outgroup. The scale of amino acid sequence divergence is indicated at the bottom right corner. Twenty-four pairs of orthologs can be identified (gray-shadowed branches). The remaining proteins can be subdivided in three different types, based on the phylogenetic relationship with their neighbors: (i) Proteins that are present only in one species and apparently started to diverge from their paralogs before human-mouse split (green clades); (ii) Clades of paralogous proteins that are present only in human and share a single homologous counterpart in mouse (red-shadowed branches); (iii) Clades of paralogous proteins that are present only in mouse and do not have any obvious homologous counterpart in humans (blue-shadowed branches).</p></caption></fig>", "<fig id=\"F6\" position=\"float\"><label>Figure 6</label><caption><p>Phylogenetic analysis of TRIM and TRIM-like proteins of representative species of mammals, aves, and fish. Human (Hs, dark blue), chicken (Gg, dark green), tetraodon (Tn, light blue); ciona (Ci, orange), fruitfly (red) and worm (light green) are included. Bootstrap support values based on 1000 replicates are shown. Group 1 TRIM37 sequences are used as outgroups. The scale of amino acid sequence divergence is indicated at the bottom right corner.</p></caption></fig>", "<fig id=\"F7\" position=\"float\"><label>Figure 7</label><caption><p>Proposed model for TRIM structure evolution (see text). The C-terminal domains probably derived from a single ancestor domain are indicated with the same color.</p></caption></fig>" ]

[ "<table-wrap id=\"T1\" position=\"float\"><label>Table 1</label><caption><p>Human TRIM genes included in our study^a</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th/><th align=\"left\">DOMAINS^b</th><th align=\"left\">ACC. No.</th><th align=\"left\">MAPPING</th><th align=\"left\">Alternative names</th></tr></thead><tbody><tr><td align=\"left\">TRIM1</td><td align=\"left\">R-B1-B2-CC-COS-FN3-SPRY</td><td align=\"left\">NM_052817</td><td align=\"left\">Xq22.3</td><td align=\"left\">MID2</td></tr><tr><td align=\"left\">TRIM2</td><td align=\"left\">R-B2-CC-IGFLMN-NHL(6)</td><td align=\"left\">NM_015271</td><td align=\"left\">4q31.3</td><td align=\"left\">RNF86, KIAA0517, Narf</td></tr><tr><td align=\"left\">TRIM3</td><td align=\"left\">R-B2-CC-IGFLMN-NHL(6)</td><td align=\"left\">NM_006458</td><td align=\"left\">11p15.4</td><td align=\"left\">BERP, HAC1, RNF22, RNF97</td></tr><tr><td align=\"left\">TRIM4</td><td align=\"left\">R-B2-CC-PRY-Spry</td><td align=\"left\">NM_033091</td><td align=\"left\">7q22.1</td><td align=\"left\">RNF87</td></tr><tr><td align=\"left\">TRIM5</td><td align=\"left\">R-B2-CC-PRY-Spry</td><td align=\"left\">NM_033034</td><td align=\"left\">11p15.4</td><td align=\"left\">RNF88</td></tr><tr><td align=\"left\">TRIM6</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_001003818</td><td align=\"left\">11p15.4</td><td align=\"left\">RNF89</td></tr><tr><td align=\"left\">TRIM7</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_203293</td><td align=\"left\">5q35.3</td><td align=\"left\">GNIP, RNF90</td></tr><tr><td align=\"left\">TRIM8</td><td align=\"left\">R-B1-B2-CC-nd</td><td align=\"left\">NM_030912</td><td align=\"left\">10q24.32</td><td align=\"left\">GERP, RNF27</td></tr><tr><td align=\"left\">TRIM9</td><td align=\"left\">R-B1-B2-CC-COS-FN3-SPRY</td><td align=\"left\">NM_015163</td><td align=\"left\">14q22.1</td><td align=\"left\">RNF91, SPRING, KIAA0282</td></tr><tr><td align=\"left\">TRIM10</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_006778</td><td align=\"left\">6p21.33</td><td align=\"left\">RNF9, HERF1, RFB30</td></tr><tr><td align=\"left\">TRIM11</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_145214</td><td align=\"left\">1q42.13</td><td align=\"left\">BIA1, RNF92</td></tr><tr><td align=\"left\">TRIM13</td><td align=\"left\">R-B2-CC-TM</td><td align=\"left\">NM_005798</td><td align=\"left\">13q14.2</td><td align=\"left\">CAR, LEU5, RNF77, RFP2</td></tr><tr><td align=\"left\">TRIM15</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_033229</td><td align=\"left\">6p21.33</td><td align=\"left\">RNF93, ZNFB7</td></tr><tr><td align=\"left\">TRIM17</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_016102</td><td align=\"left\">1q42.13</td><td align=\"left\">RBCC, terf, RNF16</td></tr><tr><td align=\"left\">TRIM18</td><td align=\"left\">R-B1-B2-CC-COS-FN3-PRY-SPRY</td><td align=\"left\">NM_000381</td><td align=\"left\">Xp22.22</td><td align=\"left\">MID1, FXY, OSX, XPRF, GBBB1, RNF59, ZNFXY,</td></tr><tr><td align=\"left\">TRIM19</td><td align=\"left\">R-B1-B2-CC-EXOIII</td><td align=\"left\">NM_033238</td><td align=\"left\">15q24.1</td><td align=\"left\">PML, MYL, RNF71</td></tr><tr><td align=\"left\">TRIM21</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_003141</td><td align=\"left\">11p15.4</td><td align=\"left\">RO/SSA, SSA, RO52, RNF81</td></tr><tr><td align=\"left\">TRIM22</td><td align=\"left\">R-B2-CC-Spry</td><td align=\"left\">NM_006074</td><td align=\"left\">11p15.4</td><td align=\"left\">STAF50, RNF94</td></tr><tr><td align=\"left\">TRIM23</td><td align=\"left\">R-B1-B2-CC-ARF</td><td align=\"left\">NM_001656</td><td align=\"left\">5q12.3</td><td align=\"left\">ARD1, RNF46</td></tr><tr><td align=\"left\">TRIM24</td><td align=\"left\">R-B1-B2-CC-PHD-BROMO</td><td align=\"left\">NM_015905</td><td align=\"left\">7q33-q34</td><td align=\"left\">PTC6, TF1A, RNF82, TIF1A, hTIF1, TIF1ALPHA</td></tr><tr><td align=\"left\">TRIM25</td><td align=\"left\">R-B1-B2-CC-PRY-SPRY</td><td align=\"left\">NM_005082</td><td align=\"left\">17q23.2</td><td align=\"left\">EFP, Z147, RNF147</td></tr><tr><td align=\"left\">TRIM26</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_003449</td><td align=\"left\">6p21.33</td><td align=\"left\">AFP, RNF95, ZNF173</td></tr><tr><td align=\"left\">TRIM27</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_006510</td><td align=\"left\">6p22.1</td><td align=\"left\">RFP, RNF76</td></tr><tr><td align=\"left\">TRIM28</td><td align=\"left\">R-B1-B2-CC-PHD-BROMO</td><td align=\"left\">NM_005762</td><td align=\"left\">19q13.43</td><td align=\"left\">KAP1, TF1B, RNF96, TIF1B</td></tr><tr><td align=\"left\">TRIM31</td><td align=\"left\">R-B2-CC-nd</td><td align=\"left\">NM_007028</td><td align=\"left\">6p22.1</td><td align=\"left\">RING, RNF, HCG1, HCGI, C6orf13</td></tr><tr><td align=\"left\">TRIM32</td><td align=\"left\">R-B2-CC-NHL(5)</td><td align=\"left\">NM_012210</td><td align=\"left\">9q33.1</td><td align=\"left\">HT2A, TATIP, LGMD2H</td></tr><tr><td align=\"left\">TRIM33</td><td align=\"left\">R-B1-B2-CC-PHD-BROMO</td><td align=\"left\">NM_015906</td><td align=\"left\">1p13.2</td><td align=\"left\">TIF1g, PTC7, RFG7, TIF1G, FLJ11429, KIAA1113</td></tr><tr><td align=\"left\">TRIM34</td><td align=\"left\">R-B2-CC-SPRY</td><td align=\"left\">NM_021616</td><td align=\"left\">11p15.4</td><td align=\"left\">IFP1, RNF21</td></tr><tr><td align=\"left\">TRIM35</td><td align=\"left\">R-B2-CC-PRY-Spry</td><td align=\"left\">NM_015066</td><td align=\"left\">8p21.2</td><td align=\"left\">HLS5, MAIR, KIAA1098, MGC17233</td></tr><tr><td align=\"left\">TRIM36</td><td align=\"left\">R-B1-B2-CC-COS-FN3-Spry</td><td align=\"left\">NM_018700</td><td align=\"left\">5q22.3</td><td align=\"left\">RNF98, RBCC728, haprin</td></tr><tr><td align=\"left\">TRIM37</td><td align=\"left\">R-B2-CC-MATH</td><td align=\"left\">NM_015294</td><td align=\"left\">17q23.2</td><td align=\"left\">MUL, TEF3, KIAA0898</td></tr><tr><td align=\"left\">TRIM38</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_006355</td><td align=\"left\">6p22.2</td><td align=\"left\">RNF15</td></tr><tr><td align=\"left\">TRIM39</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_021253</td><td align=\"left\">6p21.33</td><td align=\"left\">RNF23</td></tr><tr><td align=\"left\">TRIM40</td><td align=\"left\">R-B2-CC</td><td align=\"left\">NM_138700</td><td align=\"left\">6p21.33</td><td align=\"left\">RNF35</td></tr><tr><td align=\"left\">TRIM41</td><td align=\"left\">R-B2-CC-PRY-Spry</td><td align=\"left\">NM_033549</td><td align=\"left\">5q35.3</td><td align=\"left\">MGC1127, MGC31991</td></tr><tr><td align=\"left\">TRIM42</td><td align=\"left\">R-B1-B2-CC-COS-FN3</td><td align=\"left\">NM_152616</td><td align=\"left\">3q23</td><td align=\"left\">FLJ40097</td></tr><tr><td align=\"left\">TRIM43</td><td align=\"left\">R-B2-CC-Spry</td><td align=\"left\">NM_138800</td><td align=\"left\">2q11.1</td><td/></tr><tr><td align=\"left\">TRIM45</td><td align=\"left\">R-B1-B2-CC-IGFLMN</td><td align=\"left\">NM_025188</td><td align=\"left\">1p13.1</td><td align=\"left\">RNF99, FLJ13181</td></tr><tr><td align=\"left\">TRIM46</td><td align=\"left\">R-B1-B2-CC-COS-FN3-Spry</td><td align=\"left\">NM_025058</td><td align=\"left\">1q22</td><td align=\"left\">FLJ23229, TRIFIC</td></tr><tr><td align=\"left\">TRIM47</td><td align=\"left\">R-B1-B2-CC-PRY-Spry</td><td align=\"left\">NM_033452</td><td align=\"left\">17q25.1</td><td align=\"left\">GOA, RNF100</td></tr><tr><td align=\"left\">TRIM49</td><td align=\"left\">R-B2-CC-Spry</td><td align=\"left\">NM_020358</td><td align=\"left\">11q14.3</td><td align=\"left\">RNF18</td></tr><tr><td align=\"left\">TRIM50</td><td align=\"left\">R-B2-CC-PRY-Spry</td><td align=\"left\">NM_178125</td><td align=\"left\">7q11.23</td><td/></tr><tr><td align=\"left\">TRIM54</td><td align=\"left\">R-B2-CC-COS</td><td align=\"left\">NM_032546</td><td align=\"left\">2p23.3</td><td align=\"left\">RNF30, MURF1</td></tr><tr><td align=\"left\">TRIM55</td><td align=\"left\">R-B2-CC-COS</td><td align=\"left\">NM_184085</td><td align=\"left\">8q13.1</td><td align=\"left\">RNF29</td></tr><tr><td align=\"left\">TRIM56</td><td align=\"left\">R-B1-B2-CC-nd</td><td align=\"left\">NM_030961</td><td align=\"left\">7q22.1</td><td align=\"left\">RNF109, DKFZP667O116</td></tr><tr><td align=\"left\">TRIM58</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_015431</td><td align=\"left\">1q44</td><td align=\"left\">BIA2, FLJ38869, DKFZp434c091</td></tr><tr><td align=\"left\">TRIM59</td><td align=\"left\">R-B2-CC-TM</td><td align=\"left\">NM_173084</td><td align=\"left\">3q25.33</td><td align=\"left\">RNF104, TSBF1</td></tr><tr><td align=\"left\">TRIM60</td><td align=\"left\">R-B2-CC-PRY-Spry</td><td align=\"left\">NM_152620</td><td align=\"left\">4q32.3</td><td align=\"left\">RNF129, FLJ35882</td></tr><tr><td align=\"left\">TRIM61</td><td align=\"left\">R-B2-CC-PRY-Spry</td><td align=\"left\">NM_001012414</td><td align=\"left\">4q32.3</td><td/></tr><tr><td align=\"left\">TRIM62</td><td align=\"left\">R-B2-CC-PRY-Spry</td><td align=\"left\">NM_018207</td><td align=\"left\">1p35.1</td><td align=\"left\">FLJ16558</td></tr><tr><td align=\"left\">TRIM63</td><td align=\"left\">R-B2-CC-COS</td><td align=\"left\">NM_032588</td><td align=\"left\">1p36.11</td><td align=\"left\">RNF28, MURF2, IRIS</td></tr><tr><td align=\"left\">TRIM64</td><td align=\"left\">R-B2-CC-Spry</td><td align=\"left\">XM_061890</td><td align=\"left\">11q14.3</td><td/></tr><tr><td align=\"left\">TRIM65</td><td align=\"left\">R-B2-CC-Spry</td><td align=\"left\">NM_173547</td><td align=\"left\">17q25.1</td><td/></tr><tr><td align=\"left\">TRIM67</td><td align=\"left\">R-B1-B2-CC-COS-FN3-Spry</td><td align=\"left\">NM_001004342</td><td align=\"left\">1q42.2</td><td align=\"left\">TNL</td></tr><tr><td align=\"left\">TRIM68</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">NM_018073</td><td align=\"left\">11p15.4</td><td align=\"left\">SS-56, DKFZp686D0513, RNF137, FLJ10369</td></tr><tr><td align=\"left\">TRIM71</td><td align=\"left\">R-B1-B2-CC-IGFLMN-NHL(6)</td><td align=\"left\">DQ232881</td><td align=\"left\">3p23</td><td align=\"left\">hLIN41</td></tr><tr><td align=\"left\">TRIM72</td><td align=\"left\">R-B2-CC-PRY-Spry</td><td align=\"left\">NM_001008274</td><td align=\"left\">16p11.2</td><td/></tr><tr><td align=\"left\">TRIM73</td><td align=\"left\">R-B2-CC</td><td align=\"left\">XM_353628</td><td align=\"left\">7q11.23</td><td align=\"left\">TRIM50B</td></tr><tr><td align=\"left\">TRIM74</td><td align=\"left\">R-B2-CC</td><td align=\"left\">NM_198853</td><td align=\"left\">7q11.23</td><td align=\"left\">TRIM50C</td></tr><tr><td align=\"left\">TRIM75</td><td align=\"left\">R-B2-CC-PRY-SPRY</td><td align=\"left\">XM_939332</td><td align=\"left\">4q32.3</td><td/></tr><tr><td align=\"left\">KIAA0129</td><td align=\"left\">B2-CC-PRY-SPRY</td><td align=\"left\">NM_033220</td><td align=\"left\">9q22.33</td><td align=\"left\">TRIM14, Pub</td></tr><tr><td align=\"left\">EBBP</td><td align=\"left\">B1-B2-CC-PRY-SPRY</td><td align=\"left\">NM_006470</td><td align=\"left\">17p12</td><td align=\"left\">TRIM16</td></tr><tr><td align=\"left\">PYRIN</td><td align=\"left\">PAAD-B2-CC-PRY-SPRY</td><td align=\"left\">NM_000243</td><td align=\"left\">16p13.3</td><td align=\"left\">TRIM20, MARENOSTRIN, FMF, MEFV</td></tr><tr><td align=\"left\">ATDC</td><td align=\"left\">B1-B2-CC-nd</td><td align=\"left\">NM_012101</td><td align=\"left\">11q23.3</td><td align=\"left\">TRIM29</td></tr><tr><td align=\"left\">DIPB</td><td align=\"left\">B1-B2-CC</td><td align=\"left\">NM_017583</td><td align=\"left\">11p13</td><td align=\"left\">TRIM44</td></tr><tr><td align=\"left\">KIAA0298</td><td align=\"left\">B1-B2-CC-PHD-BROMO</td><td align=\"left\">XM_084529</td><td align=\"left\">11p15.4</td><td align=\"left\">TRIM66</td></tr><tr><td align=\"left\">RNF101</td><td align=\"left\">R-B2</td><td align=\"left\">NM_024114</td><td align=\"left\">11q11</td><td align=\"left\">TRIM48, MGC4827</td></tr><tr><td align=\"left\">RNF102</td><td align=\"left\">R-B2</td><td align=\"left\">NM_032765</td><td align=\"left\">5q35.3</td><td align=\"left\">TRIM52, MGC16175</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>Includes the alignments of the RING, B-box1 and B-box2 domains of all the human TRIM and TRIM-like proteins, alignments from which the consensi for these domains have been generated.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional file 2</title><p>Reports the values of Coiled-coil predictions for all the human TRIM and TRIM-like proteins.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S3\"><caption><title>Additional file 3</title><p>Shows the unrooted phylogenetic trees generated from the alignments of single domains of the tripartite motif.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S4\"><caption><title>Additional file 4</title><p>Shows a schematic representation of the human TRIM genomic clusters.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S5\"><caption><title>Additional file 5</title><p>Shows comparative and evolutionary analyses of the cluster of <italic>TRIM5</italic>, <italic>6</italic>, <italic>22</italic>, and <italic>34 </italic>in mammals.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>^aThe 68 genes listed include 60 TRIM genes and the 8 genes with an incomplete tripartite motif (see text).</p><p>^bR, RING finger; B1, B-box1 domain; B2, B-box2 domain; CC, Coiled-coil; FNIII, Fibronectin type III repeat; PRY, PRY domain; SPRY domain; Spry, SPRY domain detected only by PFAM; NHL, NHL repeats; IGFLMN, Filamin type immunoglobulin domain; MATH, MATH domain; PHD, Plant Homeodomain; BROMO, BROMO domain; ARF, ARF domain; PAAD, PAAD domain; COS, COS-box; ExoIII, Exonuclease III motif; nd, no known domain detected.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2148-8-225-1\"/>", "<graphic xlink:href=\"1471-2148-8-225-2\"/>", "<graphic xlink:href=\"1471-2148-8-225-3\"/>", "<graphic xlink:href=\"1471-2148-8-225-4\"/>", "<graphic xlink:href=\"1471-2148-8-225-5\"/>", "<graphic xlink:href=\"1471-2148-8-225-6\"/>", "<graphic xlink:href=\"1471-2148-8-225-7\"/>" ]

[ "<media xlink:href=\"1471-2148-8-225-S1.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2148-8-225-S2.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2148-8-225-S3.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2148-8-225-S4.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2148-8-225-S5.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The regulatory factor X (RFX) gene family transcription factors (TFs) were first detected in mammals as the regulatory factor that binds to a conserved <italic>cis</italic>-regulatory element called the X-box motif about 20 years ago [##REF##3133120##1##]. The X-box motifs, which are typically 14-mer DNA sequences, were initially identified as a result of alignment and inspection of the promoter regions of major histocompatibility complex (MHC) class II genes for conserved DNA elements [##REF##3114745##2##,##REF##3495805##3##]. Further investigations revealed that the X-box motif is highly conserved in the promoter regions of various MHC class II genes [##REF##2054109##4##]. The first RFX gene (RFX1) was later characterized as a candidate major histocompatibility complex (MHC) class II promoter binding protein [##REF##2498880##5##]. RFX1 was later found to function also as a transactivator of the hepatitis B virus enhancer [##REF##8413236##6##]. Subsequent studies revealed that RFX1 is not alone. Instead, it became the founding member of a novel family of homodimeric and heterodimeric DNA-binding proteins, which also includes RFX2 and RFX3 [##REF##8289803##7##]. More members of this gene family were subsequently identified. A fourth RFX gene (RFX4) was discovered in a human breast tumor tissue [##REF##1603086##8##] and the fifth, RFX5, was identified as a DNA-binding regulatory factor that is mutated in primary MHC class II deficiency (bare lymphocyte syndrome, BLS) [##REF##7744245##9##]. The identification of RFX1-5 and RFX genes in other genomes including the genomes of lower eukaryote species <italic>Saccharomyces cerevisiae </italic>[##REF##9741624##10##] and <italic>Schizosaccharomyces pombe </italic>[##REF##7862141##11##], and higher eukaryote species the nematode <italic>Caenorhabdits elegans </italic>[##REF##10882127##12##] helped understand both the evolution of the RFX gene family and the DNA binding domains [##REF##8600444##13##]. Notably, while previous studies reported five RFX genes (RFX1-5) in human, only one RFX gene has been identified in most invertebrate animals and yeast. In contrast, the fruit fly (<italic>Drosophila melanogaster</italic>) genome has been found to have two RFX genes, dRFX [##REF##12403718##14##] and dRFX2 [##REF##15494451##15##]. All of these RFX genes are transcription factors possessing a novel and highly conserved DNA binding domain (DBD) called RFX DNA binding domain [##REF##8600444##13##], the defining feature of all members belonging to the RFX gene family, suggesting that these RFX TFs all bind to the X-box motifs.</p>", "<p>In addition to the defining DBD domains in all of these RFX genes, most of these previously identified RFX genes also contain other conserved domains including B, C, and D domains [##REF##8600444##13##]. The D domain is also called the dimerization domain [##REF##8600444##13##]. The B and C domains also play a role in dimerization and are thus called the extended dimerization domains [##REF##9733744##16##]. Another important domain found in many members of the RFX family is the RFX activation domain (AD). For instance, RFX1 contains a well defined AD [##REF##9733744##16##]. However, AD is not found in many other members of the RFX family including the human RFX5 and <italic>C. elegans </italic>DAF-19 [##REF##8600444##13##]. Outside of these conserved domains, RFX genes from different species or even from same species show little similarity in other regions, which is quite consistent with their diverse functions and distinct expression profiles.</p>", "<p>In humans, RFX1 is primarily found in the brain with high expression in cerebral cortex and Purkinje cells [##REF##16723357##17##]. RFX2 [##REF##16676351##18##] and RFX4 [##REF##11682486##19##] are found to be heavily expressed in the testis. RFX4 is also expressed in the brain [##REF##12925582##20##]. RFX3 is expressed in ciliated cells and is required for growth and function of cilia including pancreatic endocrine cells [##REF##17229940##21##], ependymal cells [##REF##16930429##22##], and neuronal cells [##REF##15121860##23##]. RFX3-deficient mice show left-right (L-R) asymmetry defects [##REF##15121860##23##], developmental defect, diabetes [##REF##17229940##21##], and congenital hydrocephalus in mice [##REF##16930429##22##]. RFX5 is the most extensively studied RFX gene so far primarily since it serves as a transcription activator of the clinically important MHC II genes [##REF##10779326##24##] and mediates a enhanceosome formation, which results in a complex containing RFXANK (also known as RFX-B), RFXAP, CREB, and CIITA [##REF##11244040##25##]. Mutation in any one of these complex members leads to bare lymphocyte syndrome (BLS) [##REF##11244040##25##]. In <italic>C.elegans </italic>and <italic>S.cerevisae </italic>only one copy of the RFX gene exists. In <italic>C. elegans </italic>it is called DAF-19 and in <italic>S.cerevisae </italic>it is called Crt1. DAF-19 is involved in regulation of sensory neuron cilium whereas Crt-1 is involved in regulating DNA replication and damage checkpoint pathways [##REF##9741624##10##,##REF##10882127##12##]. In <italic>D.melanogaster</italic>, two of RFX genes have been identified, one is called dRFX and the other is called dRFX2. dRFX is expressed in the spermatid and brain and is necessary for ciliated sensory neuron differentiation [##REF##12403718##14##,##REF##11335126##26##]. dRFX2 has not been studied extensively and as such its function in Drosophila still remains unclear; however, there is evidence suggesting that dRFX2 plays a role in cell-cycle of the eye imaginal discs [##REF##15494451##15##].</p>", "<p>In this project, we have identified and characterized two novel RFX genes in genomes of human and many other mammals, which have now been sequenced, annotated, and analyzed.</p>" ]

[ "<title>Methods</title>", "<title>Data source and data mining</title>", "<p>Gene sets were obtained from the FTP site of the ENSEMBL database <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.ensembl.org/index.html\">http://www.ensembl.org/index.html</ext-link>[##UREF##1##42##]. In this project, the genomes of six mammals were analyzed. They are human (<italic>Homo sapiens</italic>, NCBI36.44), chimpanzee (<italic>Pan troglodytes</italic>, CHIMP2.1.44), dog (<italic>Canis familiaris</italic>, BROADD2.44), monkey (<italic>Macaca mulatta</italic>, MMUL_1.44), mouse (<italic>Mus musculus</italic>, NCBIM36.44), and rat (<italic>Rattus norvegicus</italic>, RGSC3.4.44). DBD sequences in human RFX1-5 were manually identified and extracted to a file. The sequences were aligned using ClustalW [##REF##12824352##43##]. The alignment was used as input to the profile building program hmmbuild, which is a program in the HMMER package <ext-link ext-link-type=\"uri\" xlink:href=\"http://hmmer.janelia.org\">http://hmmer.janelia.org</ext-link>[##UREF##0##29##]. The resulting profile was used for searching curated proteomes of the six mammals described above using hmmsearch, another program in the HMMER package.</p>", "<title>Gene model improvement</title>", "<p>All RFX genes except one–dog (Cfa) RFX7–show good alignment with their corresponding orthologs. Dog RFX7 gene is truncated at the N-terminus, missing 37 residues compared to other RFX7 genes. We attempted to use GeneWise <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.ebi.ac.uk/Wise2/\">http://www.ebi.ac.uk/Wise2/</ext-link>[##REF##15123596##44##,##REF##10779496##45##] to remodel this RFX gene. Using human (Hsa) RFX7 as the reference protein sequence and GeneWise, we recovered the missing residues. However, the first codon so identified was not the typical Met. Extending the coding sequence upstream did not help. This is likely due to a sequencing error.</p>", "<title>Protein domain analysis</title>", "<p>We retrieved DBDs and ADs from RFX genes using InterProScan (version 4.3.1) [##REF##18025686##32##]. To identify B, C, D domains, we used the HMMER program [##UREF##0##29##] as described above. Briefly, for HMMER searches, we used sequences of B, C, and D domains from known RFX genes (RFX1-3) to generate profiles for these domains respectively. We then searched for candidate B, C, and D domains in RFX6 and RFX7 using these profiles.</p>", "<title>RFX interactome network analysis</title>", "<p>Data were obtained at the HiMAP <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.himap.org/\">http://www.himap.org/</ext-link> database [##REF##16082366##46##] following online search instructions. All types of interactions were selected for searching. All seven interactions between RFX6 and other genes (DAT1, DTX2, FHL3, SS18L1, CCNK, RFX2, and RFX3) were previously reported by Rual <italic>et al</italic>[##REF##16189514##34##].</p>", "<title>Sequence alignment and phylogenetic analysis</title>", "<p>Multiple-sequence alignment was carried out using the program ClustalW (version 1.83) [##REF##12824352##43##]. Phylogenetic tree construction was performed using PHYLIP <ext-link ext-link-type=\"uri\" xlink:href=\"http://evolution.genetics.washington.edu/phylip.html\">http://evolution.genetics.washington.edu/phylip.html</ext-link> (Version 3.66). Briefly, sequence alignment in PHYLIP format was first created using ClustalW (Version1.83) [##REF##12824352##43##]. The alignment was used as input for PHYLIP. Programs utilized in the PHYLIP, in their respective order, were seqboot, protdist, neighbor, and consense. The phylogenetic tree file was visualized using Tree View <ext-link ext-link-type=\"uri\" xlink:href=\"http://taxonomy.zoology.gla.ac.uk/rod/treeview.html\">http://taxonomy.zoology.gla.ac.uk/rod/treeview.html</ext-link>.</p>", "<title>Expression profile of mammalian RFX genes using</title>", "<title>ESTs and SAGE libraries</title>", "<p>The EST database from NCBI was used to perform tblastn. The queries used for this tblastn were RFX1-7 of <italic>H. sapiens, M. musculus</italic>, and <italic>R. norvegicus</italic>. Hits with identity greater than or equal to 95% were selected.</p>" ]

[ "<title>Results and discussions</title>", "<p>With the current version of the human genome [##REF##11237011##27##,##REF##11181995##28##], we explored whether additional members of the RFX TF family could be identified and characterized in the human genome. We applied a Hidden Markov Model (HMM) based search method [##UREF##0##29##] and used DBD domain sequences of known human RFX TFs to search the entire human proteome. In addition to retrieving all known human RFX genes–RFX1-5, we identified two additional genes in the human genome that contain well conserved RFX DBDs. These two genes were previously assigned as RFXDC1 and RFXDC2 by the HUGO Gene Nomenclature Committee (HGNC, <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.genenames.org/\">http://www.genenames.org/</ext-link>); this nomenclature was based solely on an initial bioinformatic analyses. There are no previous publications describing these two genes. Here, we demonstrate that these two genes are also RFX gene family members closely related to RFX1-5, and our phylogenetic analysis suggests two separate recent gene duplications leading to the generation of these two genes. Thus, we proposed new gene nomenclature of RFX6 and RFX7 (Table ##TAB##0##1##), respectively. Our proposal has been accepted by the HGNC.</p>", "<p>Because all known human RFX genes–RFX1-5–are well conserved and have been identified in other mammalian genomes, we hypothesized that orthologs of RFX6 and RFX7 also exist in other mammalian genomes. As expected, we have retrieved all seven RFX genes in the genomes of five other mammalian species including chimpanzee (<italic>Pan troglodytes</italic>), monkey (<italic>Macaca mulatta</italic>), dog (<italic>Canis familiaris</italic>), mouse (<italic>Mus musculus</italic>), and rat (<italic>Rattus norvegicus</italic>) with only one exception. In the rat genome, all except RFX2 were found despite extensive searches (Additional file ##SUPPL##0##1##). Most identified RFX genes are expressed and their transcripts can be found in existing EST libraries. Interestingly, existing EST evidence suggests that RFX6 and RFX7 have no or very few alternative isoforms similar to RFX1. In contrast, RFX2-4 usually have more alternative isoforms (Additional file ##SUPPL##0##1##).</p>", "<p>To confirm that the two novel human RFX genes–RFX6 and RFX7 are indeed RFX TFs, we further examined their DBDs by aligning them with DBDs from RFX1-5 protein sequences. As expected, the DBDs of RFX6 and RFX7 align well with those of RFX1-5 (Figure ##FIG##0##1##). RFX TFs belong to the winged-helix family of DNA binding proteins because their DBDs are related in structure and function to the helix-turn-helix bacterial transcriptional regulatory proteins [##REF##10742162##30##]. DBDs from RFX6 and RFX7 each contain one wing (W1), which is the same as DBDs from RFX1-5. W1 interacts with the major groove and another conserved fold H3 (helix 3) interacts with the minor groove of DNA. In particular, the nine residues in DBDs (Figure ##FIG##0##1##, indicated with arrow heads) that make direct or water-mediated DNA contacts [##REF##10706293##31##] are almost entirely conserved in RFX6 and RFX7 (Figure ##FIG##0##1##) with a couple of minor exceptions. Of the nine residues, the human RFX7 DBD has two residues different from most of the other RFX DBDs. The first different residue is the first of the nine indicated residues. It is Lys in RFX7 DBD and RFX5 DBD, compared to Arg in DBDs of other RFX genes. Thus this difference is shared with the RFX5 DBD. The other different residue is the third of the nine residues. It is Lys in RFX7, compared to Arg at this site for DBDs of all other RFX genes. Because both Lys and Arg are basic amino acids, such substitutions are not expected to have dramatic impacts on the binding between the DBDs and their cognate binding sites. This high degree of conservation suggests that RFX6 and RFX7 may bind to similar if not identical <italic>cis</italic>-regulatory elements, i.e., the X-box motif [##REF##3133120##1##]. Hence RFX6 and RFX7 are new members of the human RFX gene family with conserved DBDs.</p>", "<p>In addition to the highly conserved DBDs, other domains including ADs, B, C, and D domains (also known as dimerization domain) [##REF##8600444##13##] have been described in human RFX1-3 (Figure ##FIG##1##2##). Among these functional domains, ADs have been identified in RFX1-3. However, ADs have not been identified RFX4-5. The B and C domains, which are usually called extended dimerization domains, play supporting roles in dimerization [##REF##9733744##16##]. B, C, and D domains have also been identified in RFX4 but are missing from RFX5. Using InterProScan [##REF##18025686##32##] and HMMER [##UREF##0##29##], we have found that RFX6 possesses B, C, and D domains, but not AD (Figure ##FIG##1##2##). The motif composition of RFX6 is similar to RFX4, which also has B, C, and D domains but lacks AD. In contrast, we failed to identify B, C, and D domains or AD in RFX7. None of these domains can be found in RFX5 as well. Because these C-terminal domains–B, C, and D domains–have been shown to mediate dimerization as well as transcriptional repression [##REF##10556033##33##], RFX6, which contains B, C, D domain, and RFX7, which does not possess B, C, or D domains, may therefore play different role in transcriptional regulation.</p>", "<p>Characterization of the functional domain composition of RFX genes will provide insights into how different RFX TFs function. In particular, how do RFX6 and RFX7, as well as RFX4 and RFX5, function in transcription considering that they do not have identified ADs? There are two possible mechanisms. First, because RFX TFs are known to form dimers and bind to same or similar binding sites (the X-box motifs) in DNA [##REF##10706293##31##], they may function together with RFX genes (RFX1-3) that do have ADs. Examination of a recently available proteome-scale map of the human protein-protein interaction network [##REF##16189514##34##], which was constructed using yeast-two-hybrid technique, has shown that RFX6 and RFX1-4 interact with each other and also interact with many other genes (Figure ##FIG##2##3##). RFX6 interacts directly with RFX2 and RFX3, the latter of which has been shown to be expressed and to function in the pancreas [##REF##17229940##21##], as well as many other tissues. The interaction between RFX6 and other RFX TFs provides further supporting evidence that RFX6 is indeed a member of the RFX gene family. Interactions between RFX7 and other genes were not observed, which is likely due to the incomplete coverage of the human protein-protein interactions analyzed in this study. Second, RFX TFs may function by interacting with many other non-RFX TFs. For example, it has been demonstrated that mammalian RFX 5 forms a complex (\"enhanceosome\") with RFXANK (also known as RFX-B), RFXAP, CREB, and CIITA to regulate expression of MHC class II genes [##REF##11244040##25##]. Notably, all of the five genes shown to interact with RFX6 (DTX1, DTX2, FHL3, CCNK, and SS18L1) (Figure ##FIG##2##3##) except only one–SS18L1–are also putative TFs.</p>", "<p>To explore the relationship between RFX6 and RFX7 and the known RFX family members RFX1-5, we have constructed a phylogenetic tree that contains all mammalian RFX genes described above (Additional file ##SUPPL##0##1##, Figure ##FIG##0##1##), as well as <italic>C. elegans </italic>RFX gene <italic>daf-19 </italic>product DAF-19 [##REF##10882127##12##], which has been extensively studied, for comparison. We used the DBD sequence of the yeast <italic>Saccharomyces cerevisiae </italic>RFX gene Crt-1[##REF##9741624##10##] as an out group in the phylogenetic tree construction. From the phylogenetic tree (Figure ##FIG##3##4##), all seven genes show perfect one-to-one orthologous relationships between different mammalian genomes. It is clear that the seven mammalian RFX genes fall into three subgroups (Figure ##FIG##3##4##). The first subgroup contains RFX1-3; the second RFX4 and RFX6; while the third RFX5 and RFX7. It is likely that RFX4 and RFX6 resulted from one gene duplication that predated the split of these mammalian species, while RFX5 and RFX7 resulted from another similar independent duplication. This hypothesis is generally consistent with the gene models of these RFX genes (Additional file ##SUPPL##1##2##). RFX6 has 19 exons, which is similar to the number of exons contained in RFX4 (18 exons); while RFX7 has 6 exons, which is similar to the number of exons contained in RFX5 (9 exons). The <italic>C. elegans </italic>RFX gene, DAF-19 clusters together with RFX1-3 genes, supporting a previously proposed hypothesis that the divergence of the subgroup RFX1-3 from other two subgroups likely predated the divergence between mammals and the nematodes [##REF##8600444##13##]. This hypothesis predicts that <italic>C. elegans </italic>should have orthologous RFX TFs to RFX4-7 [##REF##8754849##35##]. However, only one <italic>C. elegans </italic>RFX gene–<italic>daf-19</italic>–has been reported so far and our extensive search has concluded that <italic>daf-19 </italic>is the only RFX TF in <italic>C. elegans</italic>. One possible explanation is that additional RFX TFs were lost in evolution. Alternatively, RFX4-7 may have undergone positive selection in mammals to accommodate additional functional complexity in mammalian gene regulation, while RFX1-3 and <italic>daf-19 </italic>remained highly conserved due to purifying evolution. Interestingly, although the phylogenetic tree was constructed based only on DBDs, the grouping of these mammalian RFX genes is also consistent with the composition of other conserved domains. In particular, RFX1-3 all contain DBDs, ADs, Bs, Cs and Ds, while RFX4 and RFX6 have all of these domains except ADs, and RFX5 and RFX7 have only DBDs (Figures ##FIG##1##2## and ##FIG##3##4##).</p>", "<p>To gain insight into the function of these two newly identified RFX genes, we explored the expression profiles of RFX6 and RFX7 and compared them to those of RFX1-5. We analyzed two independent datasets. First, we searched the dbEST database in genBank <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.ncbi.nlm.nih.gov/dbEST/\">http://www.ncbi.nlm.nih.gov/dbEST/</ext-link>[##REF##9116859##36##] to examine which EST libraries express transcripts of these RFX genes. The results indicate that the expression profile of RFX1-5 matches well with previously published data (see INTRODUCTION): RFX1 is found in many different tissue types including white blood cells, heart, eye, testis, and cancerous cell; RFX2 appears to be expressed in testis and brain; RFX3 appears to be expressed in the placenta and brain (<italic>i.e</italic>., medulla); RFX4 is found in the brain, as well as in testis as RFX2; and RFX5 expression has been observed in various different tissues including thymus, T-cells, kidney, brain, and lymph. The consistency of expression for RFX1-5 obtained from the dbEST database with previous observations suggests that dbEST provides good estimations of RFX genes' expression profiles. Using the same method, we found that RFX6 is primarily expressed in pancreas, with minor expression in liver, while RFX7 is widely and heavily expressed in many different tissue types including kidney (tumor tissues), thymus, brain, and placenta.</p>", "<p>Second, to gain a quantitative understanding of the expression of RFX genes, we took advantage of the recent availability of serial analysis of gene expression (SAGE) libraries constructed by the Mouse Atlas of Gene Expression Project <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.mouseatlas.org/\">http://www.mouseatlas.org/</ext-link>[##REF##16352711##37##]. To start with, we tested the hypothesis that the expression of mouse RFX TFs approximates the expression of human RFX TFs. We analyzed 196 mouse SAGE libraries, each of which was produced by using a RNA library prepared from different tissue types (some of which are duplicates). Different SAGE libraries contain slightly different number of total SAGE tags. To ensure that SAGE tags and tag counts were comparable between different SAGE libraries all the libraries were normalized to 1,000,000 SAGE tags. Qualitatively, expression profiles of mouse RFX genes obtained from SAGE analysis are consistent with the expression profiles of human RFX genes obtained from the dbEST database analysis, as well as previous publications about human RFX gene expressions (Figure ##FIG##4##5##). In contrast to all other RFX genes–RFX1-5 and RFX7, which are heavily expressed in the brain, RFX6 is clearly absent from all types of brain tissues (Figure ##FIG##4##5##). RFX6 is primarily found in the pancreas (Figure ##FIG##4##5##) which is consistent with results obtained from analyzing dbEST. Low level expression of RFX6 is found in liver (also detected in dbEST) and heart. In addition to the high tissue-specificity, RFX6 has the lowest overall expression level among all seven RFX genes, suggesting that RFX6 may be under tighter regulatory control. In contrast, RFX7 has the highest relative expression level among all seven mouse RFX genes. Similar to RFX1 and RFX5, RFX7 is found in essentially all types of tissues that were examined (Figure ##FIG##4##5##).</p>", "<p>Examining additional gene expression databases, including publicly available Genomics Institute of the Novartis Research Foundation (GNF) Gene Expression Database <ext-link ext-link-type=\"uri\" xlink:href=\"http://symatlas.gnf.org/SymAtlas/\">http://symatlas.gnf.org/SymAtlas/</ext-link>, revealed very similar results.</p>" ]

[ "<title>Results and discussions</title>", "<p>With the current version of the human genome [##REF##11237011##27##,##REF##11181995##28##], we explored whether additional members of the RFX TF family could be identified and characterized in the human genome. We applied a Hidden Markov Model (HMM) based search method [##UREF##0##29##] and used DBD domain sequences of known human RFX TFs to search the entire human proteome. In addition to retrieving all known human RFX genes–RFX1-5, we identified two additional genes in the human genome that contain well conserved RFX DBDs. These two genes were previously assigned as RFXDC1 and RFXDC2 by the HUGO Gene Nomenclature Committee (HGNC, <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.genenames.org/\">http://www.genenames.org/</ext-link>); this nomenclature was based solely on an initial bioinformatic analyses. There are no previous publications describing these two genes. Here, we demonstrate that these two genes are also RFX gene family members closely related to RFX1-5, and our phylogenetic analysis suggests two separate recent gene duplications leading to the generation of these two genes. Thus, we proposed new gene nomenclature of RFX6 and RFX7 (Table ##TAB##0##1##), respectively. Our proposal has been accepted by the HGNC.</p>", "<p>Because all known human RFX genes–RFX1-5–are well conserved and have been identified in other mammalian genomes, we hypothesized that orthologs of RFX6 and RFX7 also exist in other mammalian genomes. As expected, we have retrieved all seven RFX genes in the genomes of five other mammalian species including chimpanzee (<italic>Pan troglodytes</italic>), monkey (<italic>Macaca mulatta</italic>), dog (<italic>Canis familiaris</italic>), mouse (<italic>Mus musculus</italic>), and rat (<italic>Rattus norvegicus</italic>) with only one exception. In the rat genome, all except RFX2 were found despite extensive searches (Additional file ##SUPPL##0##1##). Most identified RFX genes are expressed and their transcripts can be found in existing EST libraries. Interestingly, existing EST evidence suggests that RFX6 and RFX7 have no or very few alternative isoforms similar to RFX1. In contrast, RFX2-4 usually have more alternative isoforms (Additional file ##SUPPL##0##1##).</p>", "<p>To confirm that the two novel human RFX genes–RFX6 and RFX7 are indeed RFX TFs, we further examined their DBDs by aligning them with DBDs from RFX1-5 protein sequences. As expected, the DBDs of RFX6 and RFX7 align well with those of RFX1-5 (Figure ##FIG##0##1##). RFX TFs belong to the winged-helix family of DNA binding proteins because their DBDs are related in structure and function to the helix-turn-helix bacterial transcriptional regulatory proteins [##REF##10742162##30##]. DBDs from RFX6 and RFX7 each contain one wing (W1), which is the same as DBDs from RFX1-5. W1 interacts with the major groove and another conserved fold H3 (helix 3) interacts with the minor groove of DNA. In particular, the nine residues in DBDs (Figure ##FIG##0##1##, indicated with arrow heads) that make direct or water-mediated DNA contacts [##REF##10706293##31##] are almost entirely conserved in RFX6 and RFX7 (Figure ##FIG##0##1##) with a couple of minor exceptions. Of the nine residues, the human RFX7 DBD has two residues different from most of the other RFX DBDs. The first different residue is the first of the nine indicated residues. It is Lys in RFX7 DBD and RFX5 DBD, compared to Arg in DBDs of other RFX genes. Thus this difference is shared with the RFX5 DBD. The other different residue is the third of the nine residues. It is Lys in RFX7, compared to Arg at this site for DBDs of all other RFX genes. Because both Lys and Arg are basic amino acids, such substitutions are not expected to have dramatic impacts on the binding between the DBDs and their cognate binding sites. This high degree of conservation suggests that RFX6 and RFX7 may bind to similar if not identical <italic>cis</italic>-regulatory elements, i.e., the X-box motif [##REF##3133120##1##]. Hence RFX6 and RFX7 are new members of the human RFX gene family with conserved DBDs.</p>", "<p>In addition to the highly conserved DBDs, other domains including ADs, B, C, and D domains (also known as dimerization domain) [##REF##8600444##13##] have been described in human RFX1-3 (Figure ##FIG##1##2##). Among these functional domains, ADs have been identified in RFX1-3. However, ADs have not been identified RFX4-5. The B and C domains, which are usually called extended dimerization domains, play supporting roles in dimerization [##REF##9733744##16##]. B, C, and D domains have also been identified in RFX4 but are missing from RFX5. Using InterProScan [##REF##18025686##32##] and HMMER [##UREF##0##29##], we have found that RFX6 possesses B, C, and D domains, but not AD (Figure ##FIG##1##2##). The motif composition of RFX6 is similar to RFX4, which also has B, C, and D domains but lacks AD. In contrast, we failed to identify B, C, and D domains or AD in RFX7. None of these domains can be found in RFX5 as well. Because these C-terminal domains–B, C, and D domains–have been shown to mediate dimerization as well as transcriptional repression [##REF##10556033##33##], RFX6, which contains B, C, D domain, and RFX7, which does not possess B, C, or D domains, may therefore play different role in transcriptional regulation.</p>", "<p>Characterization of the functional domain composition of RFX genes will provide insights into how different RFX TFs function. In particular, how do RFX6 and RFX7, as well as RFX4 and RFX5, function in transcription considering that they do not have identified ADs? There are two possible mechanisms. First, because RFX TFs are known to form dimers and bind to same or similar binding sites (the X-box motifs) in DNA [##REF##10706293##31##], they may function together with RFX genes (RFX1-3) that do have ADs. Examination of a recently available proteome-scale map of the human protein-protein interaction network [##REF##16189514##34##], which was constructed using yeast-two-hybrid technique, has shown that RFX6 and RFX1-4 interact with each other and also interact with many other genes (Figure ##FIG##2##3##). RFX6 interacts directly with RFX2 and RFX3, the latter of which has been shown to be expressed and to function in the pancreas [##REF##17229940##21##], as well as many other tissues. The interaction between RFX6 and other RFX TFs provides further supporting evidence that RFX6 is indeed a member of the RFX gene family. Interactions between RFX7 and other genes were not observed, which is likely due to the incomplete coverage of the human protein-protein interactions analyzed in this study. Second, RFX TFs may function by interacting with many other non-RFX TFs. For example, it has been demonstrated that mammalian RFX 5 forms a complex (\"enhanceosome\") with RFXANK (also known as RFX-B), RFXAP, CREB, and CIITA to regulate expression of MHC class II genes [##REF##11244040##25##]. Notably, all of the five genes shown to interact with RFX6 (DTX1, DTX2, FHL3, CCNK, and SS18L1) (Figure ##FIG##2##3##) except only one–SS18L1–are also putative TFs.</p>", "<p>To explore the relationship between RFX6 and RFX7 and the known RFX family members RFX1-5, we have constructed a phylogenetic tree that contains all mammalian RFX genes described above (Additional file ##SUPPL##0##1##, Figure ##FIG##0##1##), as well as <italic>C. elegans </italic>RFX gene <italic>daf-19 </italic>product DAF-19 [##REF##10882127##12##], which has been extensively studied, for comparison. We used the DBD sequence of the yeast <italic>Saccharomyces cerevisiae </italic>RFX gene Crt-1[##REF##9741624##10##] as an out group in the phylogenetic tree construction. From the phylogenetic tree (Figure ##FIG##3##4##), all seven genes show perfect one-to-one orthologous relationships between different mammalian genomes. It is clear that the seven mammalian RFX genes fall into three subgroups (Figure ##FIG##3##4##). The first subgroup contains RFX1-3; the second RFX4 and RFX6; while the third RFX5 and RFX7. It is likely that RFX4 and RFX6 resulted from one gene duplication that predated the split of these mammalian species, while RFX5 and RFX7 resulted from another similar independent duplication. This hypothesis is generally consistent with the gene models of these RFX genes (Additional file ##SUPPL##1##2##). RFX6 has 19 exons, which is similar to the number of exons contained in RFX4 (18 exons); while RFX7 has 6 exons, which is similar to the number of exons contained in RFX5 (9 exons). The <italic>C. elegans </italic>RFX gene, DAF-19 clusters together with RFX1-3 genes, supporting a previously proposed hypothesis that the divergence of the subgroup RFX1-3 from other two subgroups likely predated the divergence between mammals and the nematodes [##REF##8600444##13##]. This hypothesis predicts that <italic>C. elegans </italic>should have orthologous RFX TFs to RFX4-7 [##REF##8754849##35##]. However, only one <italic>C. elegans </italic>RFX gene–<italic>daf-19</italic>–has been reported so far and our extensive search has concluded that <italic>daf-19 </italic>is the only RFX TF in <italic>C. elegans</italic>. One possible explanation is that additional RFX TFs were lost in evolution. Alternatively, RFX4-7 may have undergone positive selection in mammals to accommodate additional functional complexity in mammalian gene regulation, while RFX1-3 and <italic>daf-19 </italic>remained highly conserved due to purifying evolution. Interestingly, although the phylogenetic tree was constructed based only on DBDs, the grouping of these mammalian RFX genes is also consistent with the composition of other conserved domains. In particular, RFX1-3 all contain DBDs, ADs, Bs, Cs and Ds, while RFX4 and RFX6 have all of these domains except ADs, and RFX5 and RFX7 have only DBDs (Figures ##FIG##1##2## and ##FIG##3##4##).</p>", "<p>To gain insight into the function of these two newly identified RFX genes, we explored the expression profiles of RFX6 and RFX7 and compared them to those of RFX1-5. We analyzed two independent datasets. First, we searched the dbEST database in genBank <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.ncbi.nlm.nih.gov/dbEST/\">http://www.ncbi.nlm.nih.gov/dbEST/</ext-link>[##REF##9116859##36##] to examine which EST libraries express transcripts of these RFX genes. The results indicate that the expression profile of RFX1-5 matches well with previously published data (see INTRODUCTION): RFX1 is found in many different tissue types including white blood cells, heart, eye, testis, and cancerous cell; RFX2 appears to be expressed in testis and brain; RFX3 appears to be expressed in the placenta and brain (<italic>i.e</italic>., medulla); RFX4 is found in the brain, as well as in testis as RFX2; and RFX5 expression has been observed in various different tissues including thymus, T-cells, kidney, brain, and lymph. The consistency of expression for RFX1-5 obtained from the dbEST database with previous observations suggests that dbEST provides good estimations of RFX genes' expression profiles. Using the same method, we found that RFX6 is primarily expressed in pancreas, with minor expression in liver, while RFX7 is widely and heavily expressed in many different tissue types including kidney (tumor tissues), thymus, brain, and placenta.</p>", "<p>Second, to gain a quantitative understanding of the expression of RFX genes, we took advantage of the recent availability of serial analysis of gene expression (SAGE) libraries constructed by the Mouse Atlas of Gene Expression Project <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.mouseatlas.org/\">http://www.mouseatlas.org/</ext-link>[##REF##16352711##37##]. To start with, we tested the hypothesis that the expression of mouse RFX TFs approximates the expression of human RFX TFs. We analyzed 196 mouse SAGE libraries, each of which was produced by using a RNA library prepared from different tissue types (some of which are duplicates). Different SAGE libraries contain slightly different number of total SAGE tags. To ensure that SAGE tags and tag counts were comparable between different SAGE libraries all the libraries were normalized to 1,000,000 SAGE tags. Qualitatively, expression profiles of mouse RFX genes obtained from SAGE analysis are consistent with the expression profiles of human RFX genes obtained from the dbEST database analysis, as well as previous publications about human RFX gene expressions (Figure ##FIG##4##5##). In contrast to all other RFX genes–RFX1-5 and RFX7, which are heavily expressed in the brain, RFX6 is clearly absent from all types of brain tissues (Figure ##FIG##4##5##). RFX6 is primarily found in the pancreas (Figure ##FIG##4##5##) which is consistent with results obtained from analyzing dbEST. Low level expression of RFX6 is found in liver (also detected in dbEST) and heart. In addition to the high tissue-specificity, RFX6 has the lowest overall expression level among all seven RFX genes, suggesting that RFX6 may be under tighter regulatory control. In contrast, RFX7 has the highest relative expression level among all seven mouse RFX genes. Similar to RFX1 and RFX5, RFX7 is found in essentially all types of tissues that were examined (Figure ##FIG##4##5##).</p>", "<p>Examining additional gene expression databases, including publicly available Genomics Institute of the Novartis Research Foundation (GNF) Gene Expression Database <ext-link ext-link-type=\"uri\" xlink:href=\"http://symatlas.gnf.org/SymAtlas/\">http://symatlas.gnf.org/SymAtlas/</ext-link>, revealed very similar results.</p>" ]

[ "<title>Conclusion</title>", "<p>Our results show that we have identified two novel RFX genes in the human genome, RFX6 and RFX7, thus expanding the human RFX gene family from five members (RFX1-5) to seven members (RFX1-7). In addition to their possession of highly conserved DBDs, RFX6 and RFX7 show similarity to known human RFX TFs in their functional domains. In particular, RFX6 and RFX4 all have B, C, and D domains, while RFX7 and RFX5 only have DBDs. Studies carried out over the past 20 years have demonstrated that RFX1-5 are critical for development and many additional biological processes and play an important role in various devastating disease conditions. For example, RFX3-deficient mice show left-right (L-R) asymmetry defects [##REF##15121860##23##], developmental defects, diabetes [##REF##17229940##21##], and congenital hydrocephalus [##REF##16930429##22##]. RFX3 may regulate the transcription of many genes that, when mutated, cause cilia defects and many disease conditions collectively called ciliopathies [##REF##16722803##38##]. Many known ciliopathy genes, including Bardet-Biedle syndrome (BBS) genes, are well conserved and the transcription of their <italic>C. elegans </italic>orthologs are regulated by the only RFX gene in <italic>C. elegans</italic>–DAF-19 [##REF##10882127##12##,##REF##15916950##39##, ####REF##17187676##40##, ##REF##15790967##41####15790967##41##]. Mutation in any one of the RFX5 enhanceosome members–RFXANK, RFXAP, CREB, and CIITA–leads to bare lymphocyte syndrome (BLS) [##REF##11244040##25##]. We hypothesize that RFX6 and RFX7 are equally important as RFX1-5. The fact that RFX6 is primarily expressed in pancreatic tissues and is expressed at a low level compared to all other RFX genes (Figure ##FIG##4##5##) is particularly interesting. RFX6 may function as a key component of a transcriptional regulatory complex that regulates pancreas development and function.</p>" ]

[ "<title>Background</title>", "<p>Five regulatory factor X (RFX) transcription factors (TFs)–RFX1-5–have been previously characterized in the human genome, which have been demonstrated to be critical for development and are associated with an expanding list of serious human disease conditions including major histocompatibility (MHC) class II deficiency and ciliaophathies.</p>", "<title>Results</title>", "<p>In this study, we have identified two additional RFX genes–RFX6 and RFX7–in the current human genome sequences. Both RFX6 and RFX7 are demonstrated to be winged-helix TFs and have well conserved RFX DNA binding domains (DBDs), which are also found in winged-helix TFs RFX1-5. Phylogenetic analysis suggests that the RFX family in the human genome has undergone at least three gene duplications in evolution and the seven human RFX genes can be clearly categorized into three subgroups: (1) RFX1-3, (2) RFX4 and RFX6, and (3) RFX5 and RFX7. Our functional genomics analysis suggests that RFX6 and RFX7 have distinct expression profiles. RFX6 is expressed almost exclusively in the pancreatic islets, while RFX7 has high ubiquitous expression in nearly all tissues examined, particularly in various brain tissues.</p>", "<title>Conclusion</title>", "<p>The identification and further characterization of these two novel RFX genes hold promise for gaining critical insight into development and many disease conditions in mammals, potentially leading to identification of disease genes and biomarkers.</p>" ]

[ "<title>Authors' contributions</title>", "<p>NS conceived of the study, participated in experimental design. SA, LS and JSCC carried out the analysis. SA and NS wrote the manuscript. All authors read and approved the final manuscript.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>This project was supported by an NSERC Discovery Award to NC. SA is supported by a NSERC USRA. LS is a Pacific Century Graduate Scholar. JSCC is supported by a Hemingway Nelson Architects Graduate Scholarship and a Weyerhaeuser Molecular Biology Graduate Scholarship. NC is also a Michael Smith Foundation for Health Research Scholar. We thank Drs. Robert Johnsen and Maja Tarailo for critical reading of the manuscript and insightful suggestions.</p>" ]

[ "<fig id=\"F1\" position=\"float\"><label>Figure 1</label><caption><p><bold>Mammalian RFX DBDs are highly conserved</bold>. DBDs from six mammalian RFX genes were aligned using ClustalW. The conservation of amino acid is depicted by a color gradient from the color yellow, which indicates low conservation, to red, which indicates high conservation. Nine residues that make direct or water-mediated DNA contacts are indicated with arrow heads. The species names included in this figure are abbreviated. They are: Mus–mouse (<italic>Mus musculus</italic>); Rno–Rat (<italic>Rattus norvegicus</italic>); Cfa–dog (<italic>Canis familiaris</italic>); Ptr–chimpanzee (<italic>Pan troglodytes</italic>); Mmu–monkey (<italic>Macaca mulatta</italic>) and Hsa–human (<italic>Homo sapiens</italic>).</p></caption></fig>", "<fig id=\"F2\" position=\"float\"><label>Figure 2</label><caption><p><bold>Functional domains in the known and novel human RFX genes</bold>. The functional domains, AD, DBD, B, C, and D are indicated using color-coded boxes. Genes are represented using horizontal lines, which are proportional to the protein lengths. The domain lengths and positions are also proportional to their actual lengths. The graphs are aligned based on the position of the DBDs.</p></caption></fig>", "<fig id=\"F3\" position=\"float\"><label>Figure 3</label><caption><p><bold>RFX interactome</bold>. Circles depict gene products and lines depict protein-protein interactions. The interactions between RFX6 and its direct interactors were obtained using yeast-two-hybrid method in a large-scale human protein-protein interaction study [##REF##16189514##34##]. Additional interactions were constructed by Rhodes <italic>et al</italic>[##REF##16082366##46##]. The network was generated using program available at the HiMap website <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.himap.org/\">http://www.himap.org/</ext-link>[##REF##16082366##46##].</p></caption></fig>", "<fig id=\"F4\" position=\"float\"><label>Figure 4</label><caption><p><bold>Phylogenetic analysis of mammalian RFX genes</bold>. This phylogenetic tree was constructed based on DBDs of RFX genes for six mammalian species and <italic>C. elegans </italic>using yeast RFX gene product Crt1 as the out-group. The phylogenetic tree was bootstrapped for 100 times with the numbers at each internal node being the bootstrap values. Each ortholog group is colored differently. The species names included in this figure are abbreviated. They are: Mus–mouse (<italic>Mus musculus</italic>); Rno–Rat (<italic>Rattus norvegicus</italic>); Cfa–dog (<italic>Canis familiaris</italic>); Ptr–chimpanzee (<italic>Pan troglodytes</italic>); Mmu–monkey (<italic>Macaca mulatta</italic>) and Hsa–human (<italic>Homo sapiens</italic>).</p></caption></fig>", "<fig id=\"F5\" position=\"float\"><label>Figure 5</label><caption><p><bold>Relative expression of human RFX genes revealed by SAGE</bold>. Original SAGE libraries were generated by the Mouse Atlas Project [##REF##16352711##37##]. X-axis shows different tissue types, while Y-axis shows relative SAGE tag frequency.</p></caption></fig>" ]

[ "<table-wrap id=\"T1\" position=\"float\"><label>Table 1</label><caption><p>Names and Protein ID of Representative RFX genes.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th align=\"left\">Gene names</th><th align=\"left\">Accession Number (RefSeq)</th><th align=\"left\">ESEMBL protein ID</th><th align=\"center\" colspan=\"4\">Genomic coordinates</th><th align=\"right\">Protein lengths</th><th align=\"right\">Number of exons</th><th align=\"right\">Number of isoforms</th></tr><tr><th/><th/><th/><th colspan=\"4\"><hr/></th><th/><th/><th/></tr><tr><th/><th/><th/><th align=\"left\">chromosome</th><th align=\"right\">start</th><th align=\"right\">end</th><th align=\"right\">strand</th><th/><th/><th/></tr></thead><tbody><tr><td align=\"left\">RFX1</td><td align=\"left\">NM_002918</td><td align=\"left\">ENSP00000254325</td><td align=\"left\">19</td><td align=\"right\">13933353</td><td align=\"right\">13978097</td><td align=\"right\">-1</td><td align=\"right\">979</td><td align=\"right\">21</td><td align=\"right\">1</td></tr><tr><td align=\"left\">RFX2</td><td align=\"left\">NM_000635</td><td align=\"left\">ENSP00000306335</td><td align=\"left\">19</td><td align=\"right\">5944175</td><td align=\"right\">6061554</td><td align=\"right\">-1</td><td align=\"right\">723</td><td align=\"right\">18</td><td align=\"right\">2</td></tr><tr><td align=\"left\">RFX3</td><td align=\"left\">NM_134428</td><td align=\"left\">ENSP00000371434</td><td align=\"left\">9</td><td align=\"right\">3208297</td><td align=\"right\">3515983</td><td align=\"right\">-1</td><td align=\"right\">749</td><td align=\"right\">18</td><td align=\"right\">8</td></tr><tr><td align=\"left\">RFX4</td><td align=\"left\">NM_213594</td><td align=\"left\">ENSP00000350552</td><td align=\"left\">12</td><td align=\"right\">105501163</td><td align=\"right\">105680710</td><td align=\"right\">1</td><td align=\"right\">744</td><td align=\"right\">18</td><td align=\"right\">4</td></tr><tr><td align=\"left\">RFX5</td><td align=\"left\">NM_000449</td><td align=\"left\">ENSP00000357864</td><td align=\"left\">1</td><td align=\"right\">149581060</td><td align=\"right\">149586457</td><td align=\"right\">-1</td><td align=\"right\">616</td><td align=\"right\">11</td><td align=\"right\">3</td></tr><tr><td align=\"left\">RFX6</td><td align=\"left\">NM_173560</td><td align=\"left\">ENSP00000332208</td><td align=\"left\">6</td><td align=\"right\">117305068</td><td align=\"right\">117351384</td><td align=\"right\">1</td><td align=\"right\">928</td><td align=\"right\">19</td><td align=\"right\">2</td></tr><tr><td align=\"left\">RFX7</td><td align=\"left\">NM_022841</td><td align=\"left\">ENSP00000373793</td><td align=\"left\">15</td><td align=\"right\">54166958</td><td align=\"right\">54222377</td><td align=\"right\">-1</td><td align=\"right\">1281</td><td align=\"right\">7</td><td align=\"right\">1</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional File 1</title><p>Gene names and Protein ID of mammalian RFX genes.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional File 2</title><p>Gene models of human RFX genes, including RFX1-5 and newly identified RFX6-7. (a) Exon-intron structures of human RFX genes. Exons are represented using boxes, while introns are represented using lines. Both exons and introns shown in this panel are proportional to their real lengths. (b) Illustration of exon-intron structures of human RFX-genes. In this panel, while exons are proportional to their real lengths, for better visualization, introns are represented using lines of same lengths, regardless of their real lengths.</p></caption></supplementary-material>" ]

[]

[ "<graphic xlink:href=\"1471-2148-8-226-1\"/>", "<graphic xlink:href=\"1471-2148-8-226-2\"/>", "<graphic xlink:href=\"1471-2148-8-226-3\"/>", "<graphic xlink:href=\"1471-2148-8-226-4\"/>", "<graphic xlink:href=\"1471-2148-8-226-5\"/>" ]

[ "<media xlink:href=\"1471-2148-8-226-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2148-8-226-S2.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Cooperation is central to transitions in individuality [##UREF##0##1##, ####UREF##1##2##, ##UREF##2##3##, ##REF##12689724##4####12689724##4##]. Full individuality is achieved when components cooperate and relinquish their autonomy to the larger whole. Depending on the type of transition, this may necessitate the division of labor in growth, reproduction, development, feeding, movement, and protection against external aggression and internal conflict [##UREF##3##5##,##UREF##4##6##]. In the evolution of multicellularity, the chain of events from autonomous individuals at one level to the incorporation of these individuals into a more complex entity remains unclear [##UREF##3##5##]. However, some of the putative forces are likely to be general, since multicellularity has arisen many different times in evolutionary history [##UREF##5##7##,##UREF##6##8##]. Moreover, that many groupings do not show sophisticated specialization and are characterized by substantial levels of internal conflict [##REF##15714559##9##,##REF##11127911##10##], suggests that incomplete multicellularity may be a frequent outcome. What mechanisms are essential to generate individuality? We believe that a general theory needs to explain both full and incomplete transitions towards multicellular individuals.</p>", "<p>Previous work highlights group and kin selection [##UREF##3##5##,##REF##11127911##10##,##REF##1054490##11##], organism size [##UREF##7##12##,##REF##12547910##13##], and the reorganization of fitness and specialization tradeoffs [##REF##16751277##14##] as playing roles in the evolution of multicellularity. A feature common to these mechanisms is the establishment and maintenance of cooperative behaviors amongst subunits through, for example, conflict mediation (e.g. [##UREF##8##15##,##UREF##9##16##]). Based on a recent literature review, Grosberg and Strathmann [##UREF##6##8##] argued that for cooperation to emerge and favor the specialization of subunits, groups of cells need to reduce genetic conflicts arising in cell lineages [##REF##11127911##10##]. They conclude that several mechanisms can limit such conflicts, perhaps the most important being development from a single cell (e.g., [##UREF##3##5##,##UREF##9##16##]).</p>", "<p>A key type of subunit specialization in multicellular organisms is the separation of germ and soma [##UREF##0##1##,##UREF##3##5##,##REF##11127911##10##,##REF##2402151##17##,##UREF##10##18##]. Separating germ and somatic functions amongst individual cells or cell lineages requires that each sacrifice autonomy. Theory predicts that such specialization is promoted by non-mutually exclusive mechanisms such as cooperation and relatedness amongst cell lineages [##REF##11127911##10##], cheater control [##UREF##0##1##,##REF##9766963##19##,##UREF##11##20##] and adaptive responses to tradeoffs between survival and reproductive functions, i.e. a covariance effect augmenting the fitness of the group over the average fitness of its members [##REF##16751277##14##]. It is not known whether the alignment of fitness interests in emerging soma and germ lines tends to occur before, during or after other types of specialization characteristic of multicellular organisms [##UREF##7##12##].</p>", "<p>A pervasive feature in a diverse array of social systems is that individuals not contributing to the common good either act as dispersers, or are either rewarded for, or coerced into, cooperating. Examples range from bacteria (e.g. <italic>Pseudomonas fluorescens</italic>) through protozoa (e.g. <italic>Volvox carteri</italic>) to metazoans, like eusocial insects and mammals (see Additional file ##SUPPL##0##1##). For example, in naturally occurring <italic>Dictyostelium </italic>slime molds prespores secrete a chlorinated hexaphenone (DIF-1) inhibiting redifferentiation of prestalk cells into prespores, which would transpose them from \"cooperative\" stalk building to \"cheating\" spore production (i.e. a transition into the dispersing and perennial germ line; [##REF##7601348##21##,##REF##18707497##22##]). Cheating is further curtailed by pleiotropic effects of a gene required to permit receipt of this signal, which affects also the probability of spore formation [##REF##15470429##23##]. In tunicates such as <italic>Botryllus schlosseri</italic>, natural chimeras consisting of genetically nonhomogenous organisms often show reproducible germ cell parasitism that is sexually inherited, with \"parasitic forms\" being expressed only in the germ line, i.e. in the dispersing entities [##UREF##12##24##]. In the cooperatively breeding cichlid fish <italic>Neolamprologus pulcher</italic>, brood care helpers of both sexes are forced to pay rent for being tolerated in a safe territory [##UREF##13##25##,##UREF##14##26##]. To avoid being punished they preemptively appease dominants by cooperative and submissive behavior [##UREF##15##27##]. Typically, in these cichlids and in cooparatively breeding meerkats <italic>Suricata suricatta</italic>, subordinates preparing for dispersal reduce helping [##UREF##16##28##,##UREF##17##29##], which might be explained by reduced costs of potential punishment by eviction [##REF##15038984##30##,##REF##15729644##31##]. In eusocial mole rats (<italic>Heterocephalus glaber </italic>and <italic>Cryptomys damarensis</italic>) non-reproductive helpers and hardly helping dispersers coexist [##REF##8602260##32##, ####REF##11087866##33##, ##REF##16598257##34####16598257##34##]. Policing of subordinates by dominant breeders may simultaneously maintain social order and stimulate cooperative behaviors [##UREF##18##35##,##UREF##19##36##]. This distinction of roles between individuals is particularly obvious in the separation between soma and germ that has apparently evolved many times independently [##UREF##5##7##]. Nevertheless, there are examples where cooperative behaviors are associated with enhanced group dispersal (see Additional file ##SUPPL##0##1##). For example, in the soil-dwelling social bacterium <italic>Myxococcus xanthus</italic>, individualistic cell movement ('A-motility') promotes swarming on hard surfaces, whereas swarming on soft surfaces is a group function driven primarily by individually costly S-motility [##REF##12955143##37##].</p>", "<p>These empirical patterns merit explanation, and we take a first step by employing optimization techniques to evaluate the conditions leading to associations between dispersal and social strategy. Sociality in our models takes the form of cooperation in the production of a public good. Previous study of public goods has shown how cheating, if left unchecked, potentially leads to a \"tragedy of the commons\" [##REF##5699198##38##,##REF##17981363##39##], whereby individual selection tends to favor exploitation of the public good at some concurrent or future detriment of the group. Several non-mutually exclusive mechanisms may promote cooperation and group persistence, including kin selection (e.g., [##UREF##20##40##, ####UREF##21##41##, ##UREF##22##42####22##42##]), rewards and sanctions (e.g., [##UREF##23##43##,##REF##12778541##44##]), spatial and network structure (e.g., [##REF##9750181##45##, ####REF##11935015##46##, ##UREF##24##47####24##47##]), and signals involving kin or non-kin (e.g., [##UREF##25##48##, ####REF##12643576##49##, ##REF##15446434##50####15446434##50##]). Recent reviews and perspectives can be found in Crespi [##REF##11245940##51##], Sachs and colleagues [##REF##15232949##52##], Lehmann and Keller [##REF##16910958##53##], and West and coworkers [##REF##17714660##54##].</p>", "<p>We develop a model based on kin selection that incorporates dispersal specialization, as suggested by the case studies in Table S1 (see Additional file ##SUPPL##0##1##). We employ the terms \"soma\" and \"germ\" to represent the functions of within-group growth and dispersal leading to the founding of new groups, respectively. Our use of the terms \"cooperators\" and \"cheaters\" refers to social behaviors within the commons (e.g., soma), and this should be distinguished from the frequent usage of \"cheaters\" as cooperative somatic lineages trying to gain access to germ line (e.g., [##UREF##0##1##,##UREF##3##5##,##UREF##6##8##,##REF##11127911##10##,##REF##18707497##22##]). Specifically, cooperators contribute to the public good within a distinct group at an individual cost, and cheaters exploit the public good. Cooperators and/or cheaters may be selected to either remain in a group, or to disperse (potentially founding new groups). Our theory proposes a mechanism leading to high overall cooperation, based on dispersal specialization. In addition to increasing our understanding of cooperative and dispersal behaviors, it could apply to the evolution of multicellularity in a range of contexts, including physiologically integrated organisms [##REF##10322523##55##,##REF##21680404##56##], organisms with both solitary and integrated life-styles (e.g., [##UREF##26##57##]), and complex societies [##REF##2811397##58##].</p>" ]

[ "<title>Methods</title>", "<p>We formalize our verbal arguments given above by developing and analyzing a model of coevolution between exploitation of the commons and dispersal. From the outset, we stress that our model is a highly simplified representation of this process, and not aimed to make quantitative predictions for any given system. Rather, our goal is to identify the qualitatively important drivers in the coevolution of individual strategies and the evolution of multicellularity.</p>", "<p>In our model the focal units of selection are individuals themselves, rather than the higher-level unit. A transition to multicellularity is favored when the interests of the individual and the higher-level (the group) are aligned [##UREF##3##5##,##UREF##6##8##,##UREF##8##15##]. Previous models investigating the transition to multicellularity invoke a framework where the group is the focal unit of selection (see, for example [##UREF##8##15##]). However, focusing on the higher-level as the focal unit does not easily allow the investigation of optimization at the lower level [##REF##17211805##59##], and the individually-selected conditions leading to a major transition [##UREF##27##60##]. Grosberg and Strathmann [##UREF##6##8##] have argued that many of the requirements for transitions to multicellularity exist in unicellular organisms (for social groups, see [##UREF##28##61##]). Once a transition is in progress, and the \"group\" begins to behave as an individual entity, one can begin to treat this unit as an evolving individual in itself.</p>", "<p>We analyze an optimization model that takes into account the effect of both the phenotype of the focal individual and the average phenotype of the group in which it lives, on the fitness of the focal individual (see Table ##TAB##0##1## for descriptions of parameters and variables). The approach is based on the direct fitness method [##UREF##22##42##,##REF##8763356##62##] in that, by considering the effects of both individual and average group phenotypes on the fitness of a focal individual, we can apply the Price Equation to partition these effects as weighted by the relatedness of the focal individual to other members of the group [##UREF##22##42##]. We can then assess the relative impacts of (1) costs and benefits of individual behaviors and (2) kin structure, on associations between exploitative strategy within a group, and dispersal to found new groups. Nevertheless, our model oversimplifies the complexity of social behavior and dispersal decisions (for review, see [##UREF##29##63##]), and should thus be viewed as a preliminary attempt to identify patterns.</p>", "<p>Our model makes several assumptions. First, we do not explicitly consider dynamics, such as group founding, group numbers, individual emigration and immigration, and competition for limiting resources within or between groups. Rather, we assume negligible variation in inter-group competition. Second, our model does not explicitly incorporate genetic polymorphisms, meaning that the heritable traits are probabilities to adopt alternatives of each strategy (disperse or stay; cooperate or cheat) depending on environmental and/or social conditions [##UREF##0##1##,##REF##11127911##10##,##REF##8602260##32##,##UREF##30##64##, ####UREF##31##65##, ##UREF##32##66####32##66##]. Third, there is a simple direct tradeoff between an individual's viability (growth, survival and reproduction) within the group and its ability to disperse and found new groups. This is based on the well established life-history trade-off between reproduction and dispersal (see [##UREF##33##67##]), probably best studied in insects (on the physiological scale e.g. [##UREF##34##68##, ####UREF##35##69##, ##REF##16609924##70####16609924##70##]; on the ecological scale e.g. [##UREF##36##71##,##UREF##37##72##]). Whereas growth and reproduction within the group impacts the production and consumption of the public good, the tendency to disperse reduces these impacts because of the limited presence of dispersers in the source group.</p>", "<title>Life cycle and fitness equations</title>", "<p>We assume that a group's life-cycle has three sequential stages: colonization, growth, reproduction and survival of individuals within the group; exhaustion of resources; and the dispersal of survivors. Some of the survivors may stay at the same site of the source group, and others disperse as colonists to other sites.</p>", "<p>The model tracks the fitness contribution of a mutant individual <italic>i</italic>, within group <italic>j </italic>[##UREF##22##42##,##REF##8763356##62##]. Fitness effects are partitioned between cooperators and cheats–who have positive and negative impacts on the public good, respectively–and amongst dispersal strategies. Thus four strategies are possible: (1) cooperate and remain in group, (2) cooperate and disperse, (3) cheat and remain in group, and (4) cheat and disperse. Only the first and third strategies affect the public good.</p>", "<p>The proportion of cooperators in the group is <italic>n</italic>_{<italic>i </italic>}(for simplicity, hereafter we denote individual <italic>i </italic>within group <italic>j </italic>using the subscript <italic>i </italic>only), which can take continuous values between 0 and 1. Moreover, our model incorporates two dispersal strategies based on whether the dispersing individual is a cooperator or a cheater. We define <italic>y</italic>_{<italic>i </italic>}as the investment of a cooperator in dispersal and <italic>z</italic>_{<italic>i </italic>}as the investment of a given cheater in dispersal. Both of these quantities take on continuous values between zero and one. The mean proportions of dispersing cooperators and cheaters in group <italic>j</italic> are <italic>y</italic>_{<italic>j</italic>}<italic>n</italic>_{<italic>j </italic>}and <italic>z</italic>_{<italic>j</italic>}(1-<italic>n</italic>_{<italic>j</italic>}), respectively and overall investment in dispersal is <italic>d</italic>_{<italic>j </italic>}= <italic>y</italic>_{<italic>j</italic>}<italic>n</italic>_{<italic>j </italic>}+ <italic>z</italic>_{<italic>j</italic>}(1-<italic>n</italic>_{<italic>j</italic>}).</p>", "<p>The fitness equation takes the form</p>", "<p></p>", "<p>where the functions <italic>D </italic>and <italic>E</italic>, respectively, represent the contribution of selection on dispersal and the exploitation of the public good of individual <italic>i </italic>in group <italic>j </italic>to its own fitness. Function <italic>G </italic>is the overall investment in the public good in group <italic>j</italic>.</p>", "<p>Dispersal is modeled by considering the fitness contributions of both individuals that stay at the site previously occupied by the group and others that disperse [##UREF##38##73##]. We assume that the costs of dispersal may differ between cooperators (<italic>c</italic>) and cheaters (<italic>e</italic>). Small costs would indicate abundant new sites for group establishment and high disperser survival. Although we consider different cases in the analysis, our general expectation is that the costs of cooperation will extend to dispersal, such that <italic>c </italic>&gt; <italic>e</italic>.</p>", "<p>The function, <italic>D</italic>, takes the form</p>", "<p></p>", "<p>The first term in square brackets describes the fitness of a non-disperser (1 - <italic>z</italic>_{<italic>i </italic>}(1-<italic>n</italic>_{<italic>i</italic>}) - <italic>y</italic>_{<italic>i </italic>}<italic>n</italic>_{<italic>i</italic>}) relative to the average non-disperser (1 - <italic>z</italic>_{<italic>j </italic>}(1-<italic>n</italic>_{<italic>j</italic>}) - <italic>y</italic>_{<italic>j </italic>}<italic>n</italic>_{<italic>j</italic>}) and immigrants ((1-<italic>e</italic>) <italic>z </italic>(1-<italic>n</italic>) + (1-<italic>c</italic>)<italic>y n</italic>). The second term describes the fitness of a disperser ((1-<italic>e</italic>) <italic>z</italic>_{<italic>i </italic>}(1-<italic>n</italic>_{<italic>i</italic>}) + (1-<italic>c</italic>)<italic>y</italic>_{<italic>i </italic>}<italic>n</italic>_{<italic>i</italic>}) given the competition it faces with residents (1 - <italic>z </italic>(1-<italic>n</italic>) - <italic>y n</italic>) and migrants ((1-<italic>e</italic>) <italic>z </italic>(1-<italic>n</italic>) + (1-<italic>c</italic>)<italic>y n</italic>) in another group. The terms <italic>n</italic>, <italic>z </italic>and <italic>y </italic>(i.e., without subscripts) are population-wide means. The denominator in both terms represents the amount of competition faced either in the original group, in the case of a non-disperser, or in a new group, in the case of the disperser. Note that in the limit of no dispersal, individual fitness can still be positive under the assumption that groups survive indefinitely.</p>", "<p>All non-dispersing individuals are selected to exploit, but given our assumption that there is a cost of cooperation (<italic>s</italic>), this will weight selection to favoring cheaters, all else being equal. The function, <italic>E</italic>, describes the contribution of individual <italic>i </italic>to its own fitness through exploitation of the public good and is given by</p>", "<p></p>", "<p>where the subscript <italic>j </italic>indicates mean group levels, and the constant <italic>s </italic>measures the cost to individual cooperators in producing the public good.</p>", "<p>The overall effect of group investment in the public good on individual fitness is described by</p>", "<p></p>", "<p>where it is assumed that non-dispersing cooperators have a positive effect on the public good (scaled by P) as their frequency, <italic>n</italic>_{<italic>j</italic>}, increases [##REF##15278839##74##,##REF##11935014##75##], whereas cheaters have a net negative effect on the public good (scaled by Q) as their frequency, 1-<italic>n</italic>_{<italic>j</italic>}, increases. Note that in the absence of cooperators, cheats can persist as long as their impact on the commons is sufficiently low (<italic>z </italic>Q&lt; 1). Alternatively, when group effects are nil (i.e. P = Q = 0), the notion of a group is a collection of autonomous individuals.</p>", "<title>Relatedness and numerical simulation methods</title>", "<p>We analyze the model by employing the Price Equation, which enables us to express possible fitness maxima as a function of constant parameters and variables, and the relatedness, <italic>r</italic>, between individuals. Taylor and Frank [##REF##8763356##62##] give methods for finding the equilibrium, such that for any trait <italic>v </italic>we have</p>", "<p></p>", "<p>from which we can find a steady state(s) when d<italic>w</italic>_{<italic>i</italic>}/d<italic>v</italic>_{<italic>i </italic>}= 0 to find any or all <italic>v* </italic>= <italic>y</italic>*, <italic>z</italic>*, <italic>n*</italic>.</p>", "<p>In our model, <italic>r </italic>can either be a parameter (referred to as an \"open model\" by Gardner and West [##REF##16911000##76##]) or can emerge from the underlying structure of the population (referred to as a \"closed\" model in [##REF##16911000##76##]). In the latter case, we may derive <italic>r </italic>from the dispersal of individuals in the population with the recursion relation (e.g., [##REF##17981363##39##,##REF##16911000##76##])</p>", "<p></p>", "<p>This recursion tracks the probability that a given focal individual is identical by descent to another randomly picked individual at time <italic>t</italic>. The parameter <italic>k </italic>is the effective number of individuals in the group, and can be viewed as a measure of genetic diversity due to individual aggregation in group founding and habitat structure. [Note however that our model does not explicitly track the actual number of individuals in the group]. Low <italic>k </italic>is indicative of group founding by single individuals, group resistance to immigration, and abundant open sites for group founding [##REF##11127911##10##,##UREF##39##77##].</p>", "<p>In the recursion above, the term 1/<italic>k </italic>represents the probability that the randomly picked individual is the focal individual itself. The second term represents the probability that the randomly picked individual is different to the focal individual, and that neither have dispersed (represented by (1-<italic>d</italic>)²). This is multiplied by the relatedness from the previous round. Solving this recursion relation yields the equilibrium relatedness, which is</p>", "<p></p>", "<p>As we assume weak selection, the probability that a given individual disperses depends on the probability that it is a cooperator and disperses, plus the probability that it is a cheater and disperses, so <italic>d </italic>= <italic>yn</italic>+<italic>z</italic>(1-<italic>n</italic>) in this case. Under the assumptions of weak selection, we evaluate this recursion for the case when <italic>v</italic>_{<italic>i </italic>}= <italic>v</italic>_{<italic>j </italic>}= <italic>v</italic>, where <italic>v </italic>is the trait in question.</p>", "<p>Optimal strategies were solved numerically. This consisted of iterating equation (5) with steps of 0.05 or smaller for a total of 100,000 steps, which was sufficient to identify the steady state in all cases. We found that whereas initial levels of evolving variables did not affect the optimal solution when only dispersal frequencies <italic>y </italic>and <italic>z </italic>evolved, initial conditions could indeed affect the optimal solution when all three variables evolved. Closer examination showed that alternative stable states were possible, one with either all cheaters (<italic>n</italic>* = 0) or all cooperators (<italic>n</italic>* = 1), and a second with both strategies persisting (0 &lt;<italic>n* </italic>&lt; 1). Although we cannot exclude the existence of alternative interior equilibria, our numerical studies always yielded at most a single interior solution.</p>" ]

[ "<title>Results</title>", "<p>We consider two scenarios. In the first (Model 1) only dispersal in cooperators (<italic>y</italic>) and cheaters (<italic>z</italic>) evolves, but not cooperation (<italic>n</italic>). This situation would be obtained if mechanisms not explicitly included in the model (e.g., policing, [##REF##12778541##44##]) controlled the level of cooperation, or if the frequencies of cooperative behaviors were either not subject to evolution, or labile to it over much longer time scales than dispersal. More generally however, empirical study suggests that cooperative behaviors are subject to selection [##UREF##40##78##, ####REF##12955142##79##, ##REF##17055982##80####17055982##80##] and we consider the case (Model 2) in which dispersal and the frequency of cooperators (<italic>n</italic>) and cheaters (1-<italic>n</italic>) co-evolve.</p>", "<p>In addition to optimal levels of dispersal (Model 1), and of cooperation and dispersal (Model 2), we examine the effects of model parameters on dispersal specialization σ = <italic>y</italic>*/(<italic>y</italic>*+<italic>z*</italic>), and for Model 2 only, overall cooperation Φ = <italic>n</italic>*(1-<italic>y</italic>*)+(1-<italic>n</italic>*)<italic>z</italic>* (i.e., the sum of cooperators not dispersing and of cheaters dispersing). Note that when σ = 0 (or σ = 1), although all cooperators (cheats for σ = 1) are sedentary it is not necessarily true that all cheats (cooperators for σ = 1) disperse.</p>", "<title>Model 1</title>", "<p>Optimal solutions always yielded partial or complete specialization, with cooperators tending to disperse more than cheaters (i.e., σ &gt; 0.5) for high costs of cooperation (<italic>s</italic>) compared to public good's effect (<italic>P</italic>), and low cooperator frequencies (<italic>n</italic>) (Figure ##FIG##0##1##). The reverse trends promote relative cheater dispersal (σ &lt; 0.5; Fig. ##FIG##0##1##). The impact of effective group size (<italic>k</italic>) is more complex. Higher <italic>k </italic>tends to polarize dispersal to either cooperators (<italic>y* </italic>&gt; 0, <italic>z</italic>* = 0) or cheaters (<italic>y</italic>* = 0, <italic>z</italic>* &gt; 0), and increases the parameter space in which cooperators dominate dispersal (areas with σ * = 1; Fig. ##FIG##0##1##).</p>", "<p>Low effective group size (low <italic>k</italic>) should positively associate with kin competition, and in agreement with previous work [##UREF##41##81##,##REF##3626575##82##], we find that low <italic>k </italic>is associated with higher overall dispersal, <italic>d* </italic>(Figure ##FIG##1##2a##). Not surprisingly, <italic>d* </italic>increases with lower cooperator frequencies (<italic>n</italic>) and public good effects (<italic>P</italic>) (Fig. ##FIG##1##2a##). However, the effects of <italic>k </italic>and <italic>n </italic>on the separate cooperator (<italic>y*</italic>) and cheater (<italic>z*</italic>) dispersal frequencies are more complex (Figs. ##FIG##1##2b, c##). In particular, low <italic>k </italic>was always found to drive cheaters to disperse (Fig. ##FIG##1##2c##), whereas the effect on cooperators depended strongly on cooperator frequency (<italic>n</italic>) and public good productivity (<italic>P</italic>) (Fig. ##FIG##1##2b##).</p>", "<p>Cheater and cooperator dispersal can be understood as follows. When the group is dominated by cheaters (low <italic>n</italic>) and production of the public good (<italic>P</italic>) is small, increasing cooperator sedentariness (1-<italic>y*</italic>) has little beneficial effect on fitness (<italic>w</italic>), due to insufficient marginal gains via both individual exploitation (<italic>E</italic>; eqn. 3) and the group effects (<italic>G</italic>; eqn. 4). As a consequence, cooperators are selected to disperse more, relative to cheaters. Cheaters may disperse at high levels nonetheless (e.g., case of <italic>n </italic>= 0.1, <italic>k </italic>= 1.2 in Fig. ##FIG##1##2c##), because in so doing, they lessen the effects of the tragedy of the commons on individual fitness of their kin. In contrast, when the group is dominated by cooperators (high <italic>n</italic>) and public good production is high (<italic>P</italic>), marginal fitness increases with cooperator sedentariness and, due to kin competition (<italic>k</italic>), cheaters are selected to disperse more, relative to cooperators.</p>", "<title>Model 2</title>", "<p>Permitting social evolution introduces the possibility that the frequency of cooperators or cheaters fixes to zero or one, in which case associations (σ) between dispersal and social strategies are irrelevant. We find that depending on parameter combinations, either only a single global optimum is obtained, or two alternative local optima are possible. In the latter case, which state is obtained depends on initial levels of <italic>y, z </italic>and <italic>n </italic>in the numerical simulations. Figure ##FIG##2##3## shows the fraction of simulations with random initial levels of <italic>n</italic>, <italic>y </italic>and <italic>z</italic>, achieving either an internal equilibrium (0 &lt;<italic>n* </italic>&lt; 1), or one with all cooperators (<italic>n* </italic>= 1), or one with all cheaters (<italic>n</italic>* = 0) for different costs of cooperator dispersal (<italic>c</italic>; Fig. ##FIG##2##3a##) and effective group sizes (<italic>k</italic>, Fig. ##FIG##2##3b##). For simplicity in the analyses below, we employ a single arbitrary starting condition (<italic>n </italic>= <italic>y </italic>= <italic>z </italic>= 0.5).</p>", "<p>We observed four basic outcomes (Fig. ##FIG##3##4##): (1) fixation of cooperators (<italic>n* </italic>= 1), (2) fixation of cheaters (<italic>n* </italic>= 0), or coexistence of cooperators and cheaters with (3) the former only being sedentary (σ * = 0), or (4) the latter only being sedentary (σ * = 1). When σ * = 0 or σ * = 1 (i.e., all cooperators or cheaters sedentary, respectively), we further found outcomes in which all cheaters dispersed (<italic>z</italic>* = 1) or all cooperators dispersed (<italic>y* </italic>= 1), respectively. Parameter effects are generally similar to Model 1, but with some notable contrasts.</p>", "<p>Whereas in Model 1, the relative cost of cooperator (<italic>c</italic>) and cheater (<italic>e</italic>) dispersal did not yield a simple threshold condition for optimal outcomes (not shown), it did so for Model 2. We found that when cooperators and cheaters coexisted and <italic>e </italic>&gt; <italic>c</italic>, cooperators dispersed and cheaters did not (i.e., σ * = 1) (Figs. ##FIG##3##4a, b##). The reverse held when <italic>c </italic>&gt; <italic>e </italic>(Figs. ##FIG##3##4c, d##). Low effective group size (<italic>k</italic>) increases cooperator persistence (i.e., smaller areas in which <italic>n* </italic>= 0 in Fig. ##FIG##3##4##), with the effects on cheater persistence contingent on other parameters (i.e., differences in areas with <italic>n* </italic>= 0 in Fig. ##FIG##3##4##). More interestingly, whereas when <italic>e </italic>&gt; <italic>c</italic>, lower <italic>k </italic>shifts the parameter space permitting cooperators and cheaters to coexist and has little effect on the area in which all cooperators disperse (<italic>y</italic>* = 1), when <italic>c </italic>&gt; <italic>e</italic>, it expands the area of coexistence and that in which all cheaters disperse (<italic>z</italic>* = 1) (Fig. ##FIG##3##4##). Finally, relatedness (<italic>r</italic>*) generally increases with high <italic>P</italic>:<italic>s </italic>ratios, low <italic>k</italic>, and high costs to cooperator dispersal, <italic>c</italic>, with respect to cheater dispersal, <italic>e </italic>(Fig. ##FIG##4##5##). Interestingly, specialization in dispersal by cheaters and in sedentariness by cooperators tends to associate with high, but not the highest levels of relatedness (cf Figs. ##FIG##3##4c##, ##FIG##4##5c##).</p>", "<p>If we define the functional role of a cooperator as contributing to the public good, and that well functioning groups minimize the impact of cheats on the public good, then, trivially, specialization resulting in mobile cooperators and sedentary cheaters corresponds to a non-social, individualistic scenario, and cannot be considered a group related phenomenon. There are however two ways in which the impact of cheaters on the commons can be reduced: either 1-<italic>n* </italic>decreases and/or <italic>z* </italic>increases. Figure ##FIG##5##6## presents the effects of model processes on overall cooperation, defined as Φ = <italic>n</italic>* (1-<italic>y*</italic>) + (1-<italic>n*</italic>) <italic>z*</italic>. We see that although high levels of Φ are generally promoted for high <italic>P</italic>:<italic>s </italic>ratios, perfect overall cooperation (Φ = 1) is most readily obtained at low <italic>k </italic>and intermediate <italic>P</italic>:<italic>s </italic>ratios (e.g. Fig. ##FIG##5##6c##).</p>" ]

[ "<title>Discussion</title>", "<p>Our results are in broad agreement with the tenets of kin selection theory for explaining dispersal [##REF##9750181##45##,##UREF##41##81##,##REF##3626575##82##] and the maintenance of cooperative behaviors [##REF##5875341##83##, ####REF##17416674##84##, ##UREF##42##85####42##85##]. Specifically, we found that dispersal specialization leading to high levels of overall cooperation (Φ) is promoted by sufficient benefit to cost ratios (<italic>P </italic>: <italic>s</italic>) of cooperation and by kin selection (low <italic>k</italic>). The one apparent discrepancy to previous theory is, whereas higher benefit (<italic>P</italic>) to cost (<italic>s</italic>) ratios promote cooperation, higher kin selection (low <italic>k</italic>) was sometimes observed to reduce the relative frequency of cooperators (<italic>n*</italic>) (cf. Figs. ##FIG##3##4c, d##). This can be explained if we consider cheaters dispersing from the group as a type of cooperative behavior. Dispersing cheaters are effectively 'cooperative' because of the incurred individual cost of dispersal (<italic>e</italic>), and the benefits to the group in having less negative impact on the commons (<italic>Q</italic>) (cf. Figs ##FIG##3##4c, d## with Figs. ##FIG##5##6c, d##). Moreover, we found that partial or total specialization of otherwise somatic cheats as dispersing germ line occurred without the need for costly modifiers [##UREF##43##86##] or the repression of cheaters [[##UREF##44##87##,##UREF##45##88##], but see [##REF##18707497##22##]], suggesting that the mechanism identified here is applicable to a wider range of organisms where these mechanisms do not sufficiently reduce somatic cheating, or cannot evolve. Conversely, control mechanisms such as rewarding and punishment, which might be operating in many systems (see examples in Additional file ##SUPPL##0##1##), do not preclude the functionality of the mechanism demonstrated in this study (cf. Model 1).</p>", "<p>The examples presented in Table S1 (see Additional file ##SUPPL##0##1##) and our theoretical findings suggest a common conceptual and mechanistic foundation for the evolution of cooperation and individual functional specialization within groups (e.g., multicellularity). Most of the empirical examples share the feature that cooperators are less dispersive than more competitive individuals. For instance, low dispersal coincides with physical binding in bacteria that generate biofilms as a public good by polymer production [##REF##12955142##79##,##UREF##46##89##] (but see ref. [##REF##17210916##90##] for an alternative interpretation), with alloparental care of offspring in cooperative breeding [##UREF##47##91##], or with complete genetic altruism in certain eusocial insects [##UREF##48##92##]. It is worth noting that a consistent differentiation of roles regarding sedentariness and dispersal in relation to cooperation and cheating may be much more common in nature than currently believed (e.g. [##REF##17683620##93##]). Because there is no prior formal theory predicting such a relationship, empirical research on this issue is rare and suitable data are therefore scant. We stress that our theory does not elucidate the precise evolutionary pathway leading to complete multicellularity [##UREF##9##16##,##UREF##39##77##], but rather assesses the forces promoting or forestalling different levels of specialization of cooperators and cheaters as functional germ line and soma. As such, the observations of biased dispersal in Table S1 (see Additional file ##SUPPL##0##1##) have alternative explanations, including forced eviction [##UREF##49##94##] and individual-based habitat selection [##UREF##50##95##]. Experimental (e.g., [##REF##12955142##79##,##REF##17055982##80##,##REF##9770494##96##, ####REF##11140681##97##, ##UREF##51##98####51##98##]), phylogenetic (e.g., [##UREF##4##6##]), and theoretical (e.g., [##REF##12547910##13##,##REF##18707497##22##] and see discussion below) approaches are fruitful avenues to explore alternative explanations and pathways.</p>", "<p>Transitions in individuality and social complexity are generally thought to require some form of reduction in genetic variance during the reproductive process [##UREF##11##20##,##UREF##39##77##]. Genetic heterogeneity can emerge from many sources [##REF##15525396##99##], and the recursive equation 6 in our framework greatly simplifies these, only explicitly including the effects of dispersal. Our results confirm the importance of relatedness in achieving multicellularity, but also show that the highest levels of relatedness did not necessarily yield full specialization of cooperators or cheaters as dispersers, and that complete specialization could occur at relatedness levels as low as 0.7 (Fig. ##FIG##3##4d##). As such, our findings could extend to some systems in which groups are formed by the initial aggregation of non-kin [##REF##11127911##10##,##REF##15278839##74##,##UREF##44##87##,##REF##15525396##99##]. Further study is needed to explore this prediction in detail, since our model did not explicitly account for different lineages, and as such we do not know how spatial heterogeneities in relatedness might influence our results [##UREF##52##100##].</p>", "<p>Our findings have precedent, both in the study of symbiotic associations, and investigations of cooperation within species. With regard to host-parasite and symbiotic interactions, previous research has considered how parasite virulence (which is analogous to cheaters exploiting cooperative groups) may evolve spatially (e.g., [##UREF##53##101##]; for reviews see [##UREF##54##102##,##REF##8919665##103##]). In the case of horizontal transmission in parasites, which is analogous to the level of dispersal in our model (see also [##UREF##38##73##]), theory generally predicts that increased horizontal transmission (<italic>z </italic>in our model) associates with higher parasite virulence (Q (1-<italic>n</italic>) in our model) [##REF##8919665##103##]. Despite allowing for relatedness between potential cooperators and cheats we have a comparable finding, whereby an increasing tragedy of the commons pushes cheating individuals to disperse; this is both because of increased individual fitness opportunities through dispersal (<italic>z</italic>) and increased inclusive fitness through lowered group effects for those related individuals that do not disperse (Q(1-<italic>n</italic>) (1-<italic>z</italic>)).</p>", "<p>In a model investigating cooperation in spatially viscous environments, van Baalen and Rand [##REF##9750181##45##] suggested that non-altruists should disperse more readily than altruists and hypothesized that this could be viewed as a transition towards multicellularity. Koella [##UREF##55##104##] studied the independent dispersal of altruists and of cheaters in a spatially explicit setting and found that a polymorphism could arise in which altruists dispersed and interacted locally, whereas cheaters evolved longer dispersal distances and exploited altruistic clusters. Hamilton and Taborsky [##UREF##50##95##] showed that when the propensities to cooperate by generalized reciprocity and to disperse evolve independently, under a wide range of conditions either cooperation or defection is associated with dispersal, depending on the probability of finding new groups and on the costs of being alone. Over most of the range of mobility costs examined, cooperation was negatively correlated with mobility, while defection was not. Ultimately, this leads to assortment between altruists and defectors in the population (see also [##REF##18707487##105##]), which secondarily can generate group selection effects [##UREF##56##106##,##REF##15380389##107##]. Hamilton and Taborsky [##UREF##50##95##] did not check for linkage effects, however. In another study of the joint evolution of altruism and mobility, Le Galliard and coworkers [##REF##15729651##108##] found that more altruism enhancing local aggregation can select for increased mobility. The synergistic selective interaction between altruism and mobility may cause dispersal to be considerably higher than that predicted in a purely selfish population, if altruism costs accelerate slowly and mobility costs are moderate. However, their model did not reveal a polymorphism to occur between selfish-mobile and altruistic-sessile phenotypes as found so often in nature, from microbes and unicellular algae to mammals (e.g. [##UREF##10##18##,##REF##8602260##32##,##REF##21680429##109##]; Additional file ##SUPPL##0##1##). Queller [##REF##11127911##10##] argued that the resolution of within-organism conflicts could occur if an altruism allele is expressed conditional on the environment, the altruistic act being an individual removing itself from the germ line in order to perform an enhanced somatic activity. Rainey [##REF##17410161##110##] verbally proposed an idea similar in some respects to these studies, in which group selection acts to promote the functional separation of germ and soma in bacterial biofilms through the dispersal of cheats (see also [##REF##12955142##79##]). Finally, Michod [##REF##16751277##14##] showed how the specialization of lower level units into germ and soma could be associated with the transfer of fitness from lower units to the new higher individual. A critical feature of his model is the tradeoff between the viability and fecundity of lower level units, which, for convex relationships, creates disruptive selection for cooperative germ and soma. Our study, whilst generating congruent results, is to our knowledge the first to demonstrate that the evolution of lower level units based on their effects on the commons can yield dispersal specialization, one of the precursors for selection at the group level and the evolution of full multicellularity.</p>" ]

[ "<title>Conclusion</title>", "<p>Our results suggest that the establishment of trait linkage between dispersal and the propensity of within-group cheating may be a general phenomenon promoting complex social organization and multicellularity. Importantly, we cautiously suggest this should be operative regardless of whether groups ever achieve higher levels of individuality, because selection on individual components will always tend to increase exploitation, and stronger group structure will tend to increase overall cooperation through kin selected benefits [##UREF##22##42##,##REF##17416674##84##]. Partial or full reduction in the negative effects of cheaters on the commons through their specialization as dispersers offers partial solutions to two problems: the evolution of cooperation in social groups and the origin of the specialization of germ and soma in multicellular organisms. Our model is, nevertheless, a highly simplified caricature of real systems and future theoretical and empirical study is needed to explore its robustness.</p>" ]

[ "<title>Background</title>", "<p>Recent work on the complexity of life highlights the roles played by evolutionary forces at different levels of individuality. One of the central puzzles in explaining transitions in individuality for entities ranging from complex cells, to multicellular organisms and societies, is how different autonomous units relinquish control over their functions to others in the group. In addition to the necessity of reducing conflict over effecting specialized tasks, differentiating groups must control the exploitation of the commons, or else be out-competed by more fit groups.</p>", "<title>Results</title>", "<p>We propose that two forms of conflict – access to resources within groups and representation in germ line – may be resolved in tandem through individual and group-level selective effects. Specifically, we employ an optimization model to show the conditions under which different within-group social behaviors (cooperators producing a public good or cheaters exploiting the public good) may be selected to disperse, thereby not affecting the commons and functioning as germ line. We find that partial or complete dispersal specialization of cheaters is a general outcome. The propensity for cheaters to disperse is highest with intermediate benefit:cost ratios of cooperative acts and with high relatedness. An examination of a range of real biological systems tends to support our theory, although additional study is required to provide robust tests.</p>", "<title>Conclusion</title>", "<p>We suggest that trait linkage between dispersal and cheating should be operative regardless of whether groups ever achieve higher levels of individuality, because individual selection will always tend to increase exploitation, and stronger group structure will tend to increase overall cooperation through kin selected benefits. Cheater specialization as dispersers offers simultaneous solutions to the evolution of cooperation in social groups and the origin of specialization of germ and soma in multicellular organisms.</p>" ]

[ "<title>Authors' contributions</title>", "<p>MEH conceived the study, developed and analyzed the model and wrote the manuscript. DJR participated in the design of the study, developed the model and participated in writing the manuscript. MT participated in the design of the study, constructed Table S1, and participated in writing the manuscript. All authors read and approved the final manuscript.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Daniel Blumstein, Ross Crozier, Steve Frank, Toby Kiers, Barbara Taborsky, Peter Taylor and Don Waller for helpful discussions, and Steve Frank, Andy Gardner, Paul Rainey, Stuart West and two anonymous reviewers for comments on earlier drafts. MEH acknowledges the Santa Fe Institute (2005), the National Center of Ecological Analysis and Research (2006–2007), and the Centre National de la Recherche Scientifique for financial support. MT and DJR acknowledge the Swiss National Science Foundation for support (SNF-grant 3100A0-105626).</p>" ]

[ "<fig id=\"F1\" position=\"float\"><label>Figure 1</label><caption><p><bold>Globally optimal associations in dispersal and exploitation strategy for Model 1</bold>. Axes: <italic>P </italic>measures the impact of the public good on individual fitness, and <italic>s </italic>is the individual cost to cooperators in contributing to the public good. σ * = <italic>y*</italic>/(<italic>z* </italic>+ <italic>y</italic>*) indexes the tendency of cooperators to disperse (σ* &gt; 0.5) or cheats to disperse (σ* &lt; 0.5). Thick curves demarcate areas of parameter space yielding different levels of σ, whereas thin lines show areas in with either <italic>y</italic>* = 1 or <italic>z</italic>* = 1. Caption <italic>a</italic>: <italic>k </italic>= 1.2, <italic>n </italic>= 0.1; caption <italic>b</italic>: <italic>k </italic>= 10, <italic>n </italic>= 0.1; caption <italic>c</italic>: <italic>k </italic>= 1.2, <italic>n </italic>= 0.9; caption <italic>d</italic>: <italic>k </italic>= 10, <italic>n </italic>= 0.9. Note that for legibility, very thin areas parallel to thick lines are omitted, in which 0.5 &lt; σ* &lt; 1 for caption <italic>c</italic>, and 0 &lt; σ* &lt; 1 for caption <italic>d</italic>. Unless otherwise noted, dispersal rates are greater than zero and less than unity. Other parameters: <italic>c </italic>= <italic>e </italic>= 0.2, <italic>Q </italic>= 0.2. See main text for numerical methods.</p></caption></fig>", "<fig id=\"F2\" position=\"float\"><label>Figure 2</label><caption><p><bold>Effects of parameters on optimal dispersal levels for Model 1</bold>. Effects of public good production (<italic>P</italic>), frequency of cooperators (<italic>n</italic>) and effective group size (<italic>k</italic>). Caption <italic>a</italic>: overall dispersal <italic>d*</italic>; Caption <italic>b</italic>: investment in cooperator dispersal <italic>y*</italic>; Caption <italic>c </italic>investment in cheater dispersal <italic>z</italic>*. Thin line: <italic>k </italic>= 1.2, <italic>n </italic>= 0.1; dashed line: <italic>k </italic>= 10, <italic>n </italic>= 0.1; thick line: <italic>k </italic>= 1.2, <italic>n </italic>= 0.9; thick dashed line: <italic>k </italic>= 10, <italic>n </italic>= 0.9. Other parameters: <italic>c </italic>= <italic>e </italic>= 0.2, <italic>Q </italic>= 0.2, <italic>s </italic>= 0.6.</p></caption></fig>", "<fig id=\"F3\" position=\"float\"><label>Figure 3</label><caption><p><bold>The fraction of simulations in Model 2 leading to different local optima</bold>. Results based on 100 simulations in which initial levels of <italic>n</italic>, <italic>y</italic>, and <italic>z </italic>are each set to a random number between zero and one, inclusive. These simulations produced one of three equilibria: <italic>n</italic>* = 0, 0 &lt;<italic>n* </italic>&lt; 1 or <italic>n* </italic>= 1. Caption <italic>a </italic>effect of the cost of cooperator dispersal (<italic>c</italic>) with <italic>P </italic>= <italic>Q </italic>= 0.3, <italic>s </italic>= 0.5, <italic>k </italic>= 2, <italic>e </italic>= 0.2; caption <italic>b </italic>effect of effective group size (<italic>k</italic>) with <italic>P </italic>= <italic>Q </italic>= 0.2, <italic>s </italic>= 0.6, <italic>e </italic>= 0.2, <italic>c </italic>= 0.3.</p></caption></fig>", "<fig id=\"F4\" position=\"float\"><label>Figure 4</label><caption><p><bold>Locally optimal associations between dispersal and exploitation strategy</bold>. The frequency of dispersal in cooperators (<italic>y</italic>) and cheaters (<italic>z</italic>) evolves, and the frequency of cooperators (<italic>n</italic>) and cheaters (1-<italic>n</italic>) evolves. Initial frequencies in numerical studies: <italic>y </italic>= <italic>z </italic>= <italic>n </italic>= 0.5. As for Figure 1 except caption <italic>a</italic>: <italic>k </italic>= 1.2, <italic>c </italic>= 0.1; caption <italic>b</italic>: <italic>k </italic>= 10, <italic>c </italic>= 0.1; caption <italic>c</italic>: <italic>k </italic>= 1.2, <italic>c </italic>= 0.3; caption <italic>d</italic>: <italic>k </italic>= 10, <italic>c </italic>= 0.3.</p></caption></fig>", "<fig id=\"F5\" position=\"float\"><label>Figure 5</label><caption><p>Relatedness, <italic>r*</italic>, associated with simulations in Figure 4.</p></caption></fig>", "<fig id=\"F6\" position=\"float\"><label>Figure 6</label><caption><p>Overall cooperation, Φ = <italic>n </italic>*(1 - <italic>y</italic>*)+(1-<italic>n</italic>*)<italic>z</italic>*, associated with simulations in Figure 4.</p></caption></fig>" ]

[ "<table-wrap id=\"T1\" position=\"float\"><label>Table 1</label><caption><p>Parameters and variables used in this study.</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\">w</td><td align=\"left\">Individual fitness</td></tr><tr><td align=\"left\">r</td><td align=\"left\">Relatedness between any two randomly selected individuals in the group</td></tr><tr><td align=\"left\">s</td><td align=\"left\">Individual cost to cooperator growth in the group</td></tr><tr><td align=\"left\">k</td><td align=\"left\">Number of individuals in a group (an inverse measure of kin selection)</td></tr><tr><td align=\"left\">c</td><td align=\"left\">Individual cost to cooperator dispersal</td></tr><tr><td align=\"left\">e</td><td align=\"left\">Individual cost to cheater dispersal</td></tr><tr><td align=\"left\">Q</td><td align=\"left\">Impact of sedentary cheaters on the individual fitness of group members (via consumption of the public good)</td></tr><tr><td align=\"left\">P</td><td align=\"left\">Impact of sedentary cooperators on the individual fitness of group members (via production of the public good)</td></tr><tr><td align=\"left\">n</td><td align=\"left\">Relative frequency of cooperators in the group (1-n is the proportion of cheaters)</td></tr><tr><td align=\"left\">z</td><td align=\"left\">Relative frequency of cheaters dispersing</td></tr><tr><td align=\"left\">y</td><td align=\"left\">Relative frequency of cooperators dispersing</td></tr><tr><td align=\"left\">d</td><td align=\"left\">Overall investment in dispersal. d = yn + z(1-n)</td></tr><tr><td align=\"left\">Φ</td><td align=\"left\">Overall cooperation with respect to the public good. Φ = n*(1-y*)+(1-n*)z*</td></tr><tr><td align=\"left\">σ</td><td align=\"left\">Association between dispersal and cooperation. σ = y/(y+z)</td></tr></tbody></table></table-wrap>" ]

[ "<disp-formula id=\"bmcM1\"><label>(1)</label><italic>w</italic>_{<italic>i </italic>}= <italic>D</italic>(<italic>n</italic>_{<italic>i</italic>}, <italic>y</italic>_{<italic>i</italic>}, <italic>z</italic>_{<italic>i</italic>}) <italic>E</italic>(<italic>n</italic>_{<italic>i</italic>}, <italic>y</italic>_{<italic>i</italic>}, <italic>z</italic>_{<italic>i</italic>}) <italic>G</italic>(<italic>n</italic>_{<italic>i</italic>}, <italic>y</italic>_{<italic>i</italic>}, <italic>z</italic>_{<italic>i</italic>}),</disp-formula>", "<disp-formula id=\"bmcM2\"><label>(2)</label><italic>D</italic>(<italic>n</italic>_{<italic>i</italic>}, <italic>y</italic>_{<italic>i</italic>}, <italic>z</italic>_{<italic>i</italic>}) = [(1 - <italic>z</italic>_{<italic>i </italic>}(1-<italic>n</italic>_{<italic>i</italic>}) - <italic>y</italic>_{<italic>i </italic>}<italic>n</italic>_{<italic>i</italic>})/(1 - <italic>z</italic>_{<italic>j </italic>}(1-<italic>n</italic>_{<italic>j</italic>}) - <italic>y</italic>_{<italic>j </italic>}<italic>n</italic>_{<italic>j </italic>}+ (1-<italic>e</italic>) <italic>z </italic>(1-<italic>n</italic>) + (1-<italic>c</italic>)<italic>y n</italic>)] + [((1-<italic>e</italic>) <italic>z</italic>_{<italic>i </italic>}(1-<italic>n</italic>_{<italic>i</italic>}) + (1-<italic>c</italic>)<italic>y</italic>_{<italic>i </italic>}<italic>n</italic>_{<italic>i</italic>})/(1 - <italic>e z </italic>(1-<italic>n</italic>) - <italic>c y n</italic>)].</disp-formula>", "<disp-formula id=\"bmcM3\"><label>(3)</label><italic>E</italic>(<italic>n</italic>_{<italic>i</italic>}, <italic>y</italic>_{<italic>i</italic>}, <italic>z</italic>_{<italic>i</italic>}) = [(1-<italic>z</italic>_{<italic>i</italic>}) (1-<italic>n</italic>_{<italic>i</italic>}) + (1-s) (1-<italic>y</italic>_{<italic>i</italic>}) <italic>n</italic>_{<italic>i</italic>}]/[(1-<italic>z</italic>_{<italic>j</italic>}) (1-<italic>n</italic>_{<italic>j</italic>}) + (1-s) (1-<italic>y</italic>_{<italic>j</italic>}) <italic>n</italic>_{<italic>j</italic>}],</disp-formula>", "<disp-formula id=\"bmcM4\"><label>(4)</label><italic>G</italic>(<italic>n</italic>_{<italic>i</italic>}, <italic>y</italic>_{<italic>i</italic>}, <italic>z</italic>_{<italic>i</italic>}) = 1 + P (1-<italic>y</italic>_{<italic>j</italic>}) <italic>n</italic>_{<italic>j </italic>}- Q (1-<italic>z</italic>_{<italic>j</italic>}) (1-<italic>n</italic>_{<italic>j</italic>}),</disp-formula>", "<disp-formula id=\"bmcM5\"><label>(5)</label>d<italic>w</italic>_{<italic>i</italic>}/d<italic>v</italic>_{<italic>i </italic>}= ∂<italic>w</italic>_{<italic>i</italic>}/∂<italic>v</italic>_{<italic>i </italic>}+ <italic>r </italic>∂<italic>w</italic>_{<italic>i</italic>}/∂<italic>v</italic>_{<italic>j</italic>}</disp-formula>", "<disp-formula id=\"bmcM6\"><label>(6)</label><italic>r</italic>(t+1) = 1/<italic>k </italic>+ (<italic>k </italic>- 1)/<italic>k </italic>(1 - <italic>d</italic>)²<italic>r</italic>(t).</disp-formula>", "<disp-formula id=\"bmcM7\"><label>(7)</label><italic>r </italic>= 1/(<italic>k </italic>- (<italic>k </italic>- 1) (1 - <italic>d</italic>)²).</disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p><bold>(Table S1)</bold>. Examples of group formation for which there is some information on dispersal, relatedness and punishment/policing.</p></caption></supplementary-material>" ]

[]

[ "<graphic xlink:href=\"1471-2148-8-238-1\"/>", "<graphic xlink:href=\"1471-2148-8-238-2\"/>", "<graphic xlink:href=\"1471-2148-8-238-3\"/>", "<graphic xlink:href=\"1471-2148-8-238-4\"/>", "<graphic xlink:href=\"1471-2148-8-238-5\"/>", "<graphic xlink:href=\"1471-2148-8-238-6\"/>" ]

[ "<media xlink:href=\"1471-2148-8-238-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Faba bean (<italic>Vicia faba </italic>L.) is currently the third most important cool-season food legume in the world. Faba bean provides an important source of dietary protein in human diet, edible oil and animal feeds. Like other grain legumes, faba bean contributes to sustainable agriculture in the management of soil fertility and plays an essential role in crop rotation. Faba bean is a diploid with 2n = 2x = 12 chromosomes [##UREF##0##1##,##UREF##1##2##], is partially cross-pollinated ranging from 4 to 84% [##UREF##2##3##], and possesses one of the largest genomes among crop legumes (~13000 Mb). The development of saturated linkage maps as well as the identification and map-based isolation of important qualitative traits or quantitative trait loci is therefore complex and expensive.</p>", "<p>A saturated genetic linkage map provides an invaluable tool in plant genetic studies and practical breeding. One of the first genetic linkage maps of faba bean was constructed by Van de Ven <italic>et al</italic>. [##UREF##3##4##] with only 17 markers; followed by successively more detailed genetic maps by Torres <italic>et al</italic>. [##UREF##4##5##] with 51 markers, Satovic <italic>et al</italic>. [##UREF##5##6##] with 157 markers, Vaz Patto <italic>et al</italic>. [##UREF##6##7##] with 116 markers, Roman <italic>et al</italic>. [##REF##12502250##8##] with 121 markers, Roman <italic>et al</italic>. [##REF##15067394##9##] with 192 markers and Avila <italic>et al</italic>. [##REF##15067393##10##] with 103 markers. To date, faba bean genetic maps have been restricted to morphological, isozyme, RFLP, RAPD, a few seed protein genes and four SSR markers. These markers have been limited either in number, transferability or in their ability to provide syntenic information with other legume species.</p>", "<p>The recent generation of abundant genomic and genetic resources focussed around the model species <italic>Medicago truncatula </italic>and <italic>Lotus Japonicus </italic>has opened up abundant opportunities for creating gene-based molecular markers that are ideal for genetic mapping in general and comparative mapping in particular [##REF##15082563##11##, ####REF##15489274##12##, ##REF##16791689##13##, ##REF##17526914##14##, ##REF##17119911##15##, ##REF##15824281##16####15824281##16##]. In this approach, oligonucleotide primers were designed from sequences of conserved regions in gene exons that flank polymorphic regions such as introns or microsatellites. This PCR-based, codominant marker system has remarkably increased the efficiency of transferring genetic information across species. Examples include the comparison of <italic>M. truncatula </italic>with alfalfa, pea, chickpea, soybean, mung bean, lentil and lupins [##REF##15489274##12##, ####REF##16791689##13##, ##REF##17526914##14##, ##REF##17119911##15##, ##REF##15824281##16##, ##REF##16416153##17##, ##REF##16642337##18####16642337##18##].</p>", "<p>Comparative genomic studies can expose and confirm phylogenetic relationships among species and determine patterns of chromosomal evolution and syntenic relationships. More importantly, comprehensive comparative genomics can facilitate back-and-forth use of genomic resources between different legumes species, and help to reduce cost and increase efficiency in genetic research as well as crop breeding. The use of conserved genome structure to assist in transferring knowledge among related plant species is well established in grasses [##REF##10899971##19##,##REF##10810140##20##] where synteny greatly assists in gene identification among related species.</p>", "<p>In this paper we report: (1) the application of gene-based markers in faba bean; (2) the development of the first exclusively gene-based genetic and comparative map for the species; (3) analysis of syntenic relationship between faba bean and <italic>M. truncatula</italic>; and (4) the levels of homology existing between faba bean, <italic>M. truncatula </italic>and lentil, a closely related species to faba bean.</p>" ]

[ "<title>Methods</title>", "<title>Genetic mapping population</title>", "<p>A gene-based genetic map of faba bean genome was constructed using a population of 94 F₆RILs generated from a cross between faba bean line Vf6 (<italic>equina </italic>type) as the pollen recipient and line Vf27 as pollen donor (<italic>paucijuga </italic>type). These accessions have been widely used in previous genetic studies [##UREF##6##7##,##REF##15067394##9##,##UREF##7##21##] and the population was developed at IFAPA, Centro Alameda del Obispo, in Córdoba, Spain, using diploid parental individuals. Total genomic DNA was isolated from each parent and F₆individual as previously described [##REF##16402186##22##].</p>", "<title>Primer design</title>", "<p>A total of 796 intron-targeted amplified polymorphic markers (ITAPs) were used for this study. These were composed of four sets of ITAPs; 340 ML and Lup primers developed from alignment of <italic>M. truncatula </italic>and <italic>Lupinus </italic>spp. database EST sequences; 160 MLG primers based on alignments between <italic>M. truncatula</italic>, <italic>Lupinus albus</italic>, and <italic>Glycine max </italic>([##REF##17119911##15##], with a subset of these denoted AtMtL-); 143 cross-species makers (MP) developed by the Department of Plant Pathology, University of California, Davis, USA [##REF##15489274##12##]; and 140 GLIP markers created by the European Grain Legumes Integrated Project (GLIP) based on primarily on <italic>M. truncatula </italic>and pea (Andrea Seres, pers. comm.). The majority of the ITAPs markers could be positioned to a physical location in the <italic>M. truncatula </italic>psuedogenome since most primers were designed from genes in characterized chromosomal regions.</p>", "<title>Polymorphism detection</title>", "<p>Each primer pair was screened on <italic>V. faba </italic>parental DNA. PCR conditions were optimised to produce clear single amplicons, and single PCR products of the same size were purified and directly sequenced. Different detection methods were used to genotype the F₆population dependent on the type of polymorphism [##REF##17119911##15##]. Details of each marker are given in Additional file ##SUPPL##0##1##.</p>", "<title>Map construction</title>", "<p>Chi-squared analysis (P &lt; 0.05) was applied to test the segregation of the mapped markers against the expected Mendelian segregation ratio for co-dominant inheritance in the faba bean F₆RIL population. Genetic linkage mapping was conducted with MultiPoint v 1.2 software [##UREF##8##23##], with a recombination fraction (rf) of 0.29 (LOD = 9.0) using 5000 bootstraps. Map distances were calculated in cM by applying the \"Kosambi\" function. Groups of linked markers that were similarly distorted were accepted for linkage mapping. Independent markers showing significant segregation distortion and markers with missing data (&gt; 10%) were rejected for linkage to avoid bias and false linkages. Genetic maps were drawn with the software program MapChart v 2.1 [##UREF##9##24##].</p>", "<title>Establishment of macrosyntenic relationships between faba bean, lentil and <italic>M. truncatula</italic></title>", "<p>Markers mapped in faba bean were located on the <italic>M. truncatula </italic>map by aligning the ESTs originally used to design the ITAPs primers with <italic>M. truncatula </italic>BACs in the Medicago pseudogenome Mt2.0 build 8/10/2007. Alignments with a BLAST E values &lt; 1e^-20, hsp identity ≥ 60%, and hsp length &gt; 50 nt were retained. Precise positions of markers were obtained by aligning the ITAP primer sequences with <italic>M. truncatula </italic>BACs using BLASTN with an expected value &lt; 1e^-4(primer length varied from 18bp to 28bp). Approximate positions of markers that had been genetically mapped in <italic>M. truncatula </italic>but not yet positioned on the physical map were obtained from the <italic>M. truncatula </italic>genetic map [##UREF##10##25##].</p>", "<title>Loci dot plot created via Grid Map</title>", "<p>Grid Map [##UREF##11##26##] was used to compare the genetic maps of faba bean and <italic>M. truncatula</italic>. Ordered loci from faba bean and <italic>M. truncatula </italic>linkage groups were listed vertically and horizontally, respectively, and dots were positioned on the diagram at the intersection of the locations of the corresponding markers in the two genetic maps.</p>" ]

[ "<title>Results</title>", "<title>Gene-based marker development</title>", "<p>Of the 796 markers screened for amplification in genomic faba bean DNA, 19% produced two or more amplicons of different sizes. Five hundred and seven were selected and optimised for single-locus amplification. Fifty percent (254) of these produced clear single band amplicons (Table ##TAB##0##1##). Seven markers that produced two amplicons (AIGP, GLIP172, GLIP429, GLIP621, GLIP651, Lup226 and MMK1) and one that produced three amplicons (LG34) were also mapped as they exhibited convenient length polymorphisms. One hundred and sixty-five polymorphic markers were identified (Table ##TAB##0##1##), of which 151 markers were used to genotype the 94 individuals of the F₆RIL population (supplementary Table ##TAB##0##1##). Apart from the GLIP markers which were selected for their ability to amplify faba bean genomic DNA before this study, 'MLG' and 'MP' markers worked equally well in faba bean (83% and 84%, respectively) but 'MP' markers produced fewer polymorphic amplicons. Amplification rate of the 'ML' markers was the poorest (34%), less than half the rate of the 'MLG' and 'MP' groups (Table ##TAB##0##1##). Likewise, the polymorphism level of the 'ML' markers was the lowest. Both 'MLG' and 'GLIP' markers produced very high levels of polymorphism in the mapping parents (70%; Table ##TAB##0##1##). DNA sequences of markers have been submitted to Genbank under accession codes <ext-link ext-link-type=\"gen\" xlink:href=\"FH893713\">FH893713</ext-link> – <ext-link ext-link-type=\"gen\" xlink:href=\"FH937528\">FH937528</ext-link>.</p>", "<p>Nineteen mapped markers (&gt; 12%) deviated significantly (P &lt; 0.05) from the expected Mendelian inheritance ratio of 1:1. About half of these markers (9) were highly distorted (P &lt; 0.01). Fifty percent of the distorted markers segregated in favour of the Vf6 parent and fifty percent in favour of Vf27. Nine of the 12 faba bean LGs contained one to four of these distorted markers, which tended to be scattered throughout the faba bean genome. However, it is noteworthy that distorted markers grouped together in chromosomes FB-1, 7 and 8 (Figure ##FIG##0##1##).</p>", "<title>The first gene-based and comparative map of Faba bean</title>", "<p>A total of 151 genic markers were used to generate the first gene-based and comparative map of faba bean. This map was constructed with a recombination fraction of 0.29 [LOD = 9.0, 27]. The map consists of seven main linkage groups (FB-1 to FB-7) and 5 fragments (FB-8 to FB-12), which varied in length from 23.6 to 324.8 cM, and spans a total of nearly 1686 cM (Figure ##FIG##0##1## and Table ##TAB##1##2##). The number of markers per LG ranged from three to 30 markers. In addition, there was one pair and thirteen unlinked markers. Eight markers co-segregated at four loci, (three loci in FB-1, one in FB-3; Figure ##FIG##0##1## and Table ##TAB##1##2##). The maximum distances between markers ranged from 13.8 cM in FB-12 to 40.2 cM in FB-3 with an overall mean gap distance of 14.6 cM (Table ##TAB##1##2##).</p>", "<p>Of the 135 genic markers that mapped to the 12 faba bean LGs in Figure ##FIG##0##1##, 127 were assigned to the <italic>M. truncatula </italic>genetic or physical maps [##UREF##10##25##]. One hundred and four of these (82%) were in syntenic regions. Clear evidence of a simple and direct macrosyntenic relationship between the <italic>V. faba </italic>and <italic>M. truncatula </italic>is presented in the dot matrix in Figure ##FIG##1##2##. The formation of clear isoclinic diagonal lines along the linkage groups provides a strong indication of the conservation of gene order in the two legume genomes. The extensive colinearity was particularly prevailing between FB LG 1, 2, 3, 4 and Mt LG 8, 1, 4, 5 where syntenic regions accounted for 90%, 86%, 50% and 47% of the <italic>M. truncatula </italic>pseudogenome, respectively (Table ##TAB##2##3##).</p>", "<p>However, chromosomal rearrangements were also evident at a moderate level. For example, <italic>M. truncatula </italic>chromosomes 1 and 7 together with 5 and 2 merged to form the faba bean LGs 2 and 4, respectively. Similarly, <italic>M. truncatula </italic>chromosome 2 splits into FB- 4 and 8 and <italic>M. truncatula </italic>chromosome 3 into FB-5 and 9 (Table ##TAB##2##3## and Figure ##FIG##1##2##). Inversions and translocations were also notable among the orthologous markers within each syntenic pair of faba bean and <italic>M. truncatula </italic>LGs (Figure ##FIG##1##2##).</p>", "<title>Evidence of macrosynteny among faba bean, lentil and Medicago truncatula</title>", "<p>A high level of co-linearity was found between the faba bean, lentil and <italic>M. truncatula </italic>genomes based on the macro-synteny established between faba bean and <italic>M. truncatula </italic>(this study) or lentil and <italic>M. truncatula </italic>[##REF##17119911##15##] using only common orthologous markers which mapped in all three species (Figures ##FIG##2##3## and ##FIG##3##4##). The pattern of homology between faba bean and <italic>M. truncatula </italic>was similar to that between lentil and <italic>M. truncatula</italic>: for example, two linkage groups FB-1 and Len-II were exclusively syntenic to <italic>M. truncatula </italic>LG_8 and shared seven markers in common. Common markers were evenly distributed in all the three corresponding LGs suggesting that FB-1 and Len-II are essentially co-linear (Figure ##FIG##2##3A##). Likewise, FB-2 and Len-III were both syntenic to Mt-1 and orthologous to each other with nine markers in common (Figure ##FIG##2##3B##). Other examples are FB-3, Len-I and Mt-4; and FB-4, Len-V and Mt-5 (Figure ##FIG##1##2##, ##FIG##3##4A## and Phan et al., 2007 [##REF##17119911##15##]). FB-5 and FB-9 were co-linear with lentil Len-VII and LenVI, respectively, and both pairs of these linkage groups were colinear with Mt-3 (Figure ##FIG##3##4B##). This suggests shared ancestral chromosomal changes in faba bean and lentil compared to <italic>M. truncatula </italic>and confirms their phylogenetically closer relationship.</p>", "<title>Evidence of fine scale macrosynteny on individual BACs between faba bean, lentil and Medicago truncatula</title>", "<p>Several ITAP markers used in this study were designed from different genes on the same <italic>M. truncatula </italic>BACs. These markers often mapped in clusters in two or more species and in common syntenic regions of corresponding linkage groups. An example is shown in Figure ##FIG##2##3A## with markers LG89-LG98 designed from <italic>M. truncatula </italic>BAC AC140032 and LG83-LG88 from <italic>M. truncatula </italic>BAC AC138131 in <italic>M. truncatula </italic>chromosome 8. Similarly, with the exception of LG34c, markers LG31-LG43 designed from <italic>M. truncatula </italic>BAC AC152751 in Mt chromosome 1 were located in one area of FB-2 and Len-III as shown in Figure ##FIG##2##3B##; markers AnMtL6-AnMtL8 and LG102-LG112 from <italic>M. truncatula </italic>BAC AC147712 and AC135800 in Mt chromosome 3 were grouped together on corresponding pairs of LGs i.e. FB-5 and Len-VII and FB-9 and Len-VI; respectively (Figure ##FIG##3##4B##).</p>" ]

[ "<title>Discussion and conclusion</title>", "<p>The first genetic map of faba bean composed exclusively with gene-based co-dominant molecular markers was constructed using a F₆RIL population between lines Vf6 and Vf27. The map is also the first to enable the establishment of syntenic relationships between faba bean and the model legume <italic>M. truncatula</italic>, comparison with other legume species, and integration with genetic maps available in faba bean.</p>", "<p>The map is composed of 12 linkage groups and 151 genetic markers. Although the number of chromosomes in faba bean has been reported as 2n = 12 [##UREF##1##2##], the number of linkage groups in recent genetic maps in the species range from 13 to 18 [##UREF##6##7##, ####REF##12502250##8##, ##REF##15067394##9##, ##REF##15067393##10####15067393##10##] and previously as many as 48 have been reported [##UREF##5##6##]. The high number of linkage groups compared to the number of chromosomes may be due to the fact that faba bean possesses one of the largest genomes among cultivated legumes (~13000 Mb). This compares with other well-characterised species such as <italic>M. truncatula</italic>, chickpea, soybean, lentil and pea which have genomes of ~450 Mb, ~740 Mb, ~1200 Mb, ~4000 Mb and ~4000 Mb respectively [##UREF##13##28##].</p>", "<p>Of the 24 non-orthologous markers found in this study, eight were from primer pairs where more than one PCR gel band was present and where two or more such amplicons were mapped. In each case, at least one amplicon mapped syntenically. The percentages of markers sequenced in faba bean were lower compared to lentil (63%, 26% and 55% compared to 93%, 69% and 65% for MP, ML and MLG markers respectively, Table ##TAB##0##1##). This was due to higher proportion of markers amplifying multiple bands in faba bean compared to lentil (data not shown), which may imply duplication. Differences in amplification, sequencing and polymorphism rates among different types of markers used for this study reflect the mode of design of the markers. Since 'MP' and 'MLG' markers were often based on the homology of more than two phylogenetically distant species, they are more likely to work in different legume lineages. The same observation was reported for these primer sets in lentil [##REF##17119911##15##].</p>", "<p>Despite the large differences in genome sizes between <italic>M. truncatula </italic>and <italic>V. faba</italic>, a simple and direct relationship between the two genomes was identified in this study. Given the number of markers used (151), the syntenic regions cover a large proportion of <italic>M. truncatula </italic>pseudogenome with 90%, 87%, 66%, 62% and 47% for <italic>M. truncatula </italic>chromosomes 8, 1, 3, 4 and 5, respectively (Table ##TAB##2##3##). The appearance of clear isoclinic diagonal lines along the linkage groups in Figure ##FIG##1##2## also demonstrates strong evidence for the extensive co-linearity between linkage group pairs of the two species. Similar high levels of conservation have also been reported between <italic>L. culinaris </italic>ssp. <italic>culinaris </italic>and <italic>M. truncatula </italic>[##REF##17119911##15##] and other closely related legumes such as <italic>L. culinaris </italic>ssp. <italic>culinaris </italic>and <italic>P. sativum </italic>[##UREF##14##29##], <italic>M. sativa </italic>and <italic>P. sativum </italic>[##REF##15340836##30##], <italic>M. truncatula </italic>and <italic>P. sativum </italic>[##REF##16416153##17##], <italic>M. truncatula </italic>and <italic>M. sativa </italic>[##REF##15082563##11##]. This study also shows markers originally designed from genes on the same BAC clustered in corresponding syntenic areas in lentil and faba bean. The mapping populations were too small to resolve marker order in lentil and faba bean but extensive conservation of gene order,(and microsynteny) has been shown in previous studies between other legume species at similar or greater phylogenetic distances [##REF##15489274##12##,##REF##12175015##31##, ####REF##14557858##32##, ##REF##16102170##33##, ##REF##17003129##34####17003129##34##], and to some extent between <italic>M. truncatula </italic>and <italic>Arabidopsis </italic>[##REF##16102170##33##, ####REF##17003129##34##, ##REF##12644654##35####12644654##35##].</p>", "<p>A higher level of homology between <italic>V. faba </italic>and <italic>L. culinaris </italic>ssp. <italic>culinaris </italic>compared to that between <italic>V. faba </italic>and <italic>M. truncatula </italic>could be inferred from this study based on the common markers mapped in the two genomes, common homology with <italic>M. truncatula </italic>and similar pattern of rearrangements (Figures ##FIG##2##3##, ##FIG##3##4## and Phan <italic>et al</italic>., 2007 [##REF##17119911##15##]). This finding agrees with phylogenetic studies that place the genera <italic>Vicia</italic>, <italic>Lens </italic>and <italic>Pisum </italic>within the tribe Viceae while <italic>Medicago </italic>and <italic>Melilotus </italic>form a parallel tribe Trifolieae within the Galegoid or cool season legumes [##REF##12644643##36##], and is consistent with different levels of macrosynteny observed between <italic>M. truncatula</italic>, <italic>P. sativum</italic>, <italic>V. radiata</italic>, <italic>G. max</italic>, and <italic>Phaceolus vulgaris </italic>dependent on phylogenetic distance [##REF##15489274##12##]. However, chromosomal rearrangements were evident (Figures ##FIG##1##2## and ##FIG##2##3##).</p>", "<p>Rearrangements involving Mt6 and Mt3 in particular may explain the differences in chromosome number between the two species (<italic>M. truncatula</italic>: n = 8; <italic>V. faba</italic>: n = 6). Mt6 might be considered unusual and is largely composed of heterochromatic DNA [##REF##11489182##37##], contains few transcribed genes [##REF##15082563##11##] and a large proportion of resistance gene analogues [##REF##12059101##38##]. In this study no corresponding linkage group was detected in faba bean, as found previously in pea [##REF##15489274##12##] and lupin [##REF##16791689##13##,##REF##17526914##14##], together with less than five percent estimated coverage by the <italic>L. japonicus </italic>genome [##REF##17003129##34##]. In faba bean FB5 and FB9 appear to correspond to Mt3. This configuration is supported by a similar pattern in lentil (Figure ##FIG##3##4B## and Phan <italic>et al</italic>., 2007 [##REF##17119911##15##]), although a larger number of markers are needed to confirm this.</p>", "<p>The faba bean comparative map constructed here is consistent with the pattern of chromosome conservation previously observed, where different levels of conservation were found to be relatively consistent between <italic>M. truncatula </italic>and other legume species i.e. high conservation of <italic>M. truncatula </italic>chromosomes 1, 5 and 8; moderate conservation in the <italic>M. truncatula </italic>chromosomes 2, 3, 4 and 7 and lowest conservation in the <italic>M. truncatula </italic>chromosome 6 (Figures ##FIG##2##3##, ##FIG##3##4## and [##UREF##15##39##]). As described above, no homology was identified with <italic>M. truncatula </italic>chromosome 6 in this study. The alignment of this faba bean map with lentil and the current <italic>M. truncatula </italic>genome based on <italic>M. truncatula </italic>genome assembly Mt2.0 is slightly different to that based on an earlier assembly [##REF##17119911##15##]. The changes can be observed in Figure ##FIG##2##3B## where orthologous markers which were syntenic to <italic>M. truncatula </italic>chromosome 6 in lentil are now co-linear with <italic>M. truncatula </italic>chromosome 1 in common with faba bean.</p>", "<p>Genome studies have demonstrated different factors are responsible for genome size variation and speciation. These include ancient polyploidisation events in the case of the Brassicas [##REF##8978073##40##]; segmental or region-specific duplication [##REF##14615182##41##]; and genetic rearrangements, transposable element amplification, or combination of different genome modifications [##REF##17291821##42##]. Large scale rearrangements, duplications, or polyploidisation were not apparent in this study, possibly as a result of the focus on single locus markers, however differences in non-coding repetitive DNA or transposable elements provide a possible explanation for the large differences in genome size. Retroelements are known to account for substantial proportions of these Viceae genomes as shown by extensive studies in pea, for example [##REF##18031571##43##,##UREF##16##44##], and more recently <italic>Vicia </italic>[##REF##16585134##45##,##REF##8602145##46##]. Local genic rearrangements similar to that in found in the grasses (duplications, translocations, and insertions or deletions) may explain multiple PCR amplicons [##REF##10899971##19##,##REF##10810140##20##].</p>", "<p>The shared macrosynteny among the three species demonstrated here and even higher level of homology between <italic>L. culinaris </italic>and <italic>V. faba </italic>will undoubtedly facilitate the identification of markers closely linked to traits of interest in <italic>V. faba</italic>. Alignment of this map with existing faba bean maps containing important traits with polymorphic SSR markers and/or markers developed in this study, coupled with cross-reference to the abundant genetic information from the Medicago genome sequencing and extensive EST libraries available for the model legume species, will undoubtedly assist this process. As the parental line Vf6 has been used in a number of genetic and QTL mapping projects [##UREF##5##6##,##UREF##6##7##,##REF##15067394##9##], this map can serve as a central reference map. This study has provided a number of significant outcomes for faba bean genomics and legume genomics in general.</p>" ]

[ "<title>Discussion and conclusion</title>", "<p>The first genetic map of faba bean composed exclusively with gene-based co-dominant molecular markers was constructed using a F₆RIL population between lines Vf6 and Vf27. The map is also the first to enable the establishment of syntenic relationships between faba bean and the model legume <italic>M. truncatula</italic>, comparison with other legume species, and integration with genetic maps available in faba bean.</p>", "<p>The map is composed of 12 linkage groups and 151 genetic markers. Although the number of chromosomes in faba bean has been reported as 2n = 12 [##UREF##1##2##], the number of linkage groups in recent genetic maps in the species range from 13 to 18 [##UREF##6##7##, ####REF##12502250##8##, ##REF##15067394##9##, ##REF##15067393##10####15067393##10##] and previously as many as 48 have been reported [##UREF##5##6##]. The high number of linkage groups compared to the number of chromosomes may be due to the fact that faba bean possesses one of the largest genomes among cultivated legumes (~13000 Mb). This compares with other well-characterised species such as <italic>M. truncatula</italic>, chickpea, soybean, lentil and pea which have genomes of ~450 Mb, ~740 Mb, ~1200 Mb, ~4000 Mb and ~4000 Mb respectively [##UREF##13##28##].</p>", "<p>Of the 24 non-orthologous markers found in this study, eight were from primer pairs where more than one PCR gel band was present and where two or more such amplicons were mapped. In each case, at least one amplicon mapped syntenically. The percentages of markers sequenced in faba bean were lower compared to lentil (63%, 26% and 55% compared to 93%, 69% and 65% for MP, ML and MLG markers respectively, Table ##TAB##0##1##). This was due to higher proportion of markers amplifying multiple bands in faba bean compared to lentil (data not shown), which may imply duplication. Differences in amplification, sequencing and polymorphism rates among different types of markers used for this study reflect the mode of design of the markers. Since 'MP' and 'MLG' markers were often based on the homology of more than two phylogenetically distant species, they are more likely to work in different legume lineages. The same observation was reported for these primer sets in lentil [##REF##17119911##15##].</p>", "<p>Despite the large differences in genome sizes between <italic>M. truncatula </italic>and <italic>V. faba</italic>, a simple and direct relationship between the two genomes was identified in this study. Given the number of markers used (151), the syntenic regions cover a large proportion of <italic>M. truncatula </italic>pseudogenome with 90%, 87%, 66%, 62% and 47% for <italic>M. truncatula </italic>chromosomes 8, 1, 3, 4 and 5, respectively (Table ##TAB##2##3##). The appearance of clear isoclinic diagonal lines along the linkage groups in Figure ##FIG##1##2## also demonstrates strong evidence for the extensive co-linearity between linkage group pairs of the two species. Similar high levels of conservation have also been reported between <italic>L. culinaris </italic>ssp. <italic>culinaris </italic>and <italic>M. truncatula </italic>[##REF##17119911##15##] and other closely related legumes such as <italic>L. culinaris </italic>ssp. <italic>culinaris </italic>and <italic>P. sativum </italic>[##UREF##14##29##], <italic>M. sativa </italic>and <italic>P. sativum </italic>[##REF##15340836##30##], <italic>M. truncatula </italic>and <italic>P. sativum </italic>[##REF##16416153##17##], <italic>M. truncatula </italic>and <italic>M. sativa </italic>[##REF##15082563##11##]. This study also shows markers originally designed from genes on the same BAC clustered in corresponding syntenic areas in lentil and faba bean. The mapping populations were too small to resolve marker order in lentil and faba bean but extensive conservation of gene order,(and microsynteny) has been shown in previous studies between other legume species at similar or greater phylogenetic distances [##REF##15489274##12##,##REF##12175015##31##, ####REF##14557858##32##, ##REF##16102170##33##, ##REF##17003129##34####17003129##34##], and to some extent between <italic>M. truncatula </italic>and <italic>Arabidopsis </italic>[##REF##16102170##33##, ####REF##17003129##34##, ##REF##12644654##35####12644654##35##].</p>", "<p>A higher level of homology between <italic>V. faba </italic>and <italic>L. culinaris </italic>ssp. <italic>culinaris </italic>compared to that between <italic>V. faba </italic>and <italic>M. truncatula </italic>could be inferred from this study based on the common markers mapped in the two genomes, common homology with <italic>M. truncatula </italic>and similar pattern of rearrangements (Figures ##FIG##2##3##, ##FIG##3##4## and Phan <italic>et al</italic>., 2007 [##REF##17119911##15##]). This finding agrees with phylogenetic studies that place the genera <italic>Vicia</italic>, <italic>Lens </italic>and <italic>Pisum </italic>within the tribe Viceae while <italic>Medicago </italic>and <italic>Melilotus </italic>form a parallel tribe Trifolieae within the Galegoid or cool season legumes [##REF##12644643##36##], and is consistent with different levels of macrosynteny observed between <italic>M. truncatula</italic>, <italic>P. sativum</italic>, <italic>V. radiata</italic>, <italic>G. max</italic>, and <italic>Phaceolus vulgaris </italic>dependent on phylogenetic distance [##REF##15489274##12##]. However, chromosomal rearrangements were evident (Figures ##FIG##1##2## and ##FIG##2##3##).</p>", "<p>Rearrangements involving Mt6 and Mt3 in particular may explain the differences in chromosome number between the two species (<italic>M. truncatula</italic>: n = 8; <italic>V. faba</italic>: n = 6). Mt6 might be considered unusual and is largely composed of heterochromatic DNA [##REF##11489182##37##], contains few transcribed genes [##REF##15082563##11##] and a large proportion of resistance gene analogues [##REF##12059101##38##]. In this study no corresponding linkage group was detected in faba bean, as found previously in pea [##REF##15489274##12##] and lupin [##REF##16791689##13##,##REF##17526914##14##], together with less than five percent estimated coverage by the <italic>L. japonicus </italic>genome [##REF##17003129##34##]. In faba bean FB5 and FB9 appear to correspond to Mt3. This configuration is supported by a similar pattern in lentil (Figure ##FIG##3##4B## and Phan <italic>et al</italic>., 2007 [##REF##17119911##15##]), although a larger number of markers are needed to confirm this.</p>", "<p>The faba bean comparative map constructed here is consistent with the pattern of chromosome conservation previously observed, where different levels of conservation were found to be relatively consistent between <italic>M. truncatula </italic>and other legume species i.e. high conservation of <italic>M. truncatula </italic>chromosomes 1, 5 and 8; moderate conservation in the <italic>M. truncatula </italic>chromosomes 2, 3, 4 and 7 and lowest conservation in the <italic>M. truncatula </italic>chromosome 6 (Figures ##FIG##2##3##, ##FIG##3##4## and [##UREF##15##39##]). As described above, no homology was identified with <italic>M. truncatula </italic>chromosome 6 in this study. The alignment of this faba bean map with lentil and the current <italic>M. truncatula </italic>genome based on <italic>M. truncatula </italic>genome assembly Mt2.0 is slightly different to that based on an earlier assembly [##REF##17119911##15##]. The changes can be observed in Figure ##FIG##2##3B## where orthologous markers which were syntenic to <italic>M. truncatula </italic>chromosome 6 in lentil are now co-linear with <italic>M. truncatula </italic>chromosome 1 in common with faba bean.</p>", "<p>Genome studies have demonstrated different factors are responsible for genome size variation and speciation. These include ancient polyploidisation events in the case of the Brassicas [##REF##8978073##40##]; segmental or region-specific duplication [##REF##14615182##41##]; and genetic rearrangements, transposable element amplification, or combination of different genome modifications [##REF##17291821##42##]. Large scale rearrangements, duplications, or polyploidisation were not apparent in this study, possibly as a result of the focus on single locus markers, however differences in non-coding repetitive DNA or transposable elements provide a possible explanation for the large differences in genome size. Retroelements are known to account for substantial proportions of these Viceae genomes as shown by extensive studies in pea, for example [##REF##18031571##43##,##UREF##16##44##], and more recently <italic>Vicia </italic>[##REF##16585134##45##,##REF##8602145##46##]. Local genic rearrangements similar to that in found in the grasses (duplications, translocations, and insertions or deletions) may explain multiple PCR amplicons [##REF##10899971##19##,##REF##10810140##20##].</p>", "<p>The shared macrosynteny among the three species demonstrated here and even higher level of homology between <italic>L. culinaris </italic>and <italic>V. faba </italic>will undoubtedly facilitate the identification of markers closely linked to traits of interest in <italic>V. faba</italic>. Alignment of this map with existing faba bean maps containing important traits with polymorphic SSR markers and/or markers developed in this study, coupled with cross-reference to the abundant genetic information from the Medicago genome sequencing and extensive EST libraries available for the model legume species, will undoubtedly assist this process. As the parental line Vf6 has been used in a number of genetic and QTL mapping projects [##UREF##5##6##,##UREF##6##7##,##REF##15067394##9##], this map can serve as a central reference map. This study has provided a number of significant outcomes for faba bean genomics and legume genomics in general.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>The development of genetic markers is complex and costly in species with little pre-existing genomic information. Faba bean possesses one of the largest and least studied genomes among cultivated crop plants and no gene-based genetic maps exist. Gene-based orthologous markers allow chromosomal regions and levels of synteny to be characterised between species, reveal phylogenetic relationships and chromosomal evolution, and enable targeted identification of markers for crop breeding. In this study orthologous codominant cross-species markers have been deployed to produce the first exclusively gene-based genetic linkage map of faba bean (<italic>Vicia faba</italic>), using an F₆population developed from a cross between the lines Vf6 (<italic>equina </italic>type) and Vf27 (<italic>paucijuga </italic>type).</p>", "<title>Results</title>", "<p>Of 796 intron-targeted amplified polymorphic (ITAP) markers screened, 151 markers could be used to construct a comparative genetic map. Linkage analysis revealed seven major and five small linkage groups (LGs), one pair and 12 unlinked markers. Each LG was comprised of three to 30 markers and varied in length from 23.6 cM to 324.8 cM. The map spanned a total length of 1685.8 cM. A simple and direct macrosyntenic relationship between faba bean and <italic>Medicago truncatula </italic>was evident, while faba bean and lentil shared a common rearrangement relative to <italic>M. truncatula</italic>. One hundred and four of the 127 mapped markers in the 12 LGs, which were previously assigned to <italic>M. truncatula </italic>genetic and physical maps, were found in regions syntenic between the faba bean and <italic>M. truncatula </italic>genomes. However chromosomal rearrangements were observed that could explain the difference in chromosome numbers between these three legume species. These rearrangements suggested high conservation of <italic>M. truncatula </italic>chromosomes 1, 5 and 8; moderate conservation of chromosomes 2, 3, 4 and 7 and no conservation with <italic>M. truncatula </italic>chromosome 6. Multiple PCR amplicons and comparative mapping were suggestive of small-scale duplication events in faba bean. This study also provides a preliminary indication for finer scale macrosynteny between <italic>M. truncatula</italic>, lentil and faba bean. Markers originally designed from genes on the same <italic>M. truncatula </italic>BACs were found to be grouped together in corresponding syntenic areas in lentil and faba bean.</p>", "<title>Conclusion</title>", "<p>Despite the large size of the faba bean genome, comparative mapping did not reveal evidence for polyploidisation, segmental duplication, or significant rearrangements compared to <italic>M. truncatula</italic>, although a bias in the use of single locus markers may have limited the detection of duplications. Non-coding repetitive DNA or transposable element content provides a possible explanation for the difference in genome sizes. Similar patterns of rearrangements in faba bean and lentil compared to <italic>M. truncatula </italic>support phylogenetic studies dividing these species into the tribes Viceae and Trifoliae. However, substantial macrosynteny was apparent between faba bean and <italic>M. truncatula</italic>, with the exception of chromosome 6 where no orthologous markers were found, confirming previous investigations suggesting chromosome 6 is atypical. The composite map, anchored with orthologous markers mapped in <italic>M. truncatula</italic>, provides a central reference map for future use of genomic and genetic information in faba bean genetic analysis and breeding.</p>" ]

[ "<title>Authors' contributions</title>", "<p>SE and RPO designed the research. HTTP and SE wrote the manuscript. HTTP, SE and MG performed marker polymorphism discovery and population genotyping in Perth. JH assisted with informatics procedures. AMT, CMA and SC-I were involved in developing recombinant inbred lines and marker polymorphism discovery of GLIP markers in Córdoba. All authors read and approved the final manuscript.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>This research was supported by an ARC Linkage project LP0454871, the New South Wales Departments of Primary Industry and the GLIP project FP6-2002-FOOD-1-506223.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>A gene-based genetic linkage map of faba bean (<italic>Vicia faba </italic>L)</bold>. Marker distance is given in cM. *indicates markers with distorted segregation.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Matrix plot of common gene-based markers mapped in faba bean and <italic>M. truncatula</italic></bold>. The faba bean and <italic>M. truncatula </italic>loci are listed vertically and horizontally, respectively, according to their linkage group order.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Evidence of shared macrosynteny between <italic>V. faba </italic>chromosomes FB-1 and FB-2, <italic>L. culinaris </italic>and <italic>M. truncatula</italic></bold>. Common orthologous markers are depicted by dashed lines and marker distances are provided in centi-Morgans for <italic>M. truncatula </italic>only. The figures exclude markers that could not be positioned in the <italic>M. truncatula </italic>psuedogenome. A ~ indicates markers that map distally in the corresponding <italic>M. truncatula </italic>chromosome relative to faba bean, and markers suffixed a, b or c denotes derivation from primer pairs that produced multiple PCR products.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Evidence of shared macrosynteny between <italic>V. faba </italic>chromosomes FB-3 and FB-5, <italic>L. culinaris </italic>and <italic>M. truncatula</italic></bold>. Common orthologous markers are depicted by dashed lines and marker distances are provided in centi-Morgans for <italic>M. truncatula </italic>only. The figures exclude markers that could not be positioned in the <italic>M. truncatula </italic>psuedogenome. Markers highlighted * were previously mapped [##REF##16642337##18##] but not positioned in the <italic>M. truncatula </italic>psuedogenome. # indicates markers orthologous between lentil and faba bean but unmapped in <italic>M. truncatula </italic>and markers suffixed a, b or c denotes derivation from primer pairs that produced multiple PCR products.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Efficiency of gene-based markers used to construct the comparative genetic linkage map of faba bean</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Marker type</td><td align=\"left\">Screened</td><td align=\"left\">Amplification^a</td><td align=\"left\">Length polymorphism</td><td align=\"left\">Sequenced^b</td><td align=\"left\">CAPS/SNP^c</td><td align=\"left\">Mapped</td></tr></thead><tbody><tr><td align=\"left\">MP</td><td align=\"left\">143</td><td align=\"left\">120 (84%)</td><td align=\"center\">6</td><td align=\"left\">75 (63%)</td><td align=\"left\">45 (60%)</td><td align=\"left\">36</td></tr><tr><td align=\"left\">ML</td><td align=\"left\">340</td><td align=\"left\">115 (34%)</td><td align=\"center\">9</td><td align=\"left\">30 (26%)</td><td align=\"left\">16 (53%)</td><td align=\"left\">25</td></tr><tr><td align=\"left\">MLG</td><td align=\"left\">160</td><td align=\"left\">132 (83%)</td><td align=\"center\">5</td><td align=\"left\">73 (55%)</td><td align=\"left\">50 (70%)</td><td align=\"left\">42</td></tr><tr><td align=\"left\">GLIP</td><td align=\"left\">140</td><td align=\"left\">140</td><td align=\"center\">6</td><td align=\"left\">76 (54%)</td><td align=\"left\">54 (71%)</td><td align=\"left\">48</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\">Total</td><td align=\"left\">796</td><td align=\"left\">507</td><td align=\"center\">26</td><td align=\"left\">254</td><td align=\"left\">165</td><td align=\"left\">151</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Properties of the faba bean comparative map</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Linkage group</td><td align=\"left\">Length of LGs (cM)</td><td align=\"left\">No of markers</td><td align=\"left\">No of loci</td><td align=\"left\">Average marker spacing^a(cM)</td><td align=\"left\">Largest distance between markers (cM)</td></tr></thead><tbody><tr><td align=\"center\">LG-1</td><td align=\"center\">324.8</td><td align=\"center\">30</td><td align=\"center\">27</td><td align=\"center\">12.5</td><td align=\"center\">32.5</td></tr><tr><td align=\"center\">LG-2</td><td align=\"center\">313.1</td><td align=\"center\">27</td><td align=\"center\">27</td><td align=\"center\">12.0</td><td align=\"center\">31.8</td></tr><tr><td align=\"center\">LG-3</td><td align=\"center\">238.5</td><td align=\"center\">18</td><td align=\"center\">17</td><td align=\"center\">14.9</td><td align=\"center\">40.2</td></tr><tr><td align=\"center\">LG-4</td><td align=\"center\">210.7</td><td align=\"center\">13</td><td align=\"center\">13</td><td align=\"center\">17.6</td><td align=\"center\">29.8</td></tr><tr><td align=\"center\">LG-5</td><td align=\"center\">151.9</td><td align=\"center\">12</td><td align=\"center\">12</td><td align=\"center\">13.8</td><td align=\"center\">26.8</td></tr><tr><td align=\"center\">LG-6</td><td align=\"center\">120.7</td><td align=\"center\">8</td><td align=\"center\">8</td><td align=\"center\">17.2</td><td align=\"center\">36.9</td></tr><tr><td align=\"center\">LG-7</td><td align=\"center\">119.2</td><td align=\"center\">7</td><td align=\"center\">7</td><td align=\"center\">19.9</td><td align=\"center\">35.9</td></tr><tr><td align=\"center\">LG-8</td><td align=\"center\">49.9</td><td align=\"center\">6</td><td align=\"center\">6</td><td align=\"center\">10.0</td><td align=\"center\">29.5</td></tr><tr><td align=\"center\">LG-9</td><td align=\"center\">64.1</td><td align=\"center\">5</td><td align=\"center\">5</td><td align=\"center\">16.0</td><td align=\"center\">22.6</td></tr><tr><td align=\"center\">LG-10</td><td align=\"center\">41.8</td><td align=\"center\">3</td><td align=\"center\">3</td><td align=\"center\">15.4</td><td align=\"center\">30.7</td></tr><tr><td align=\"center\">LG-11</td><td align=\"center\">27.5</td><td align=\"center\">3</td><td align=\"center\">3</td><td align=\"center\">13.8</td><td align=\"center\">19.9</td></tr><tr><td align=\"center\">LG12</td><td align=\"center\">23.6</td><td align=\"center\">3</td><td align=\"center\">3</td><td align=\"center\">11.8</td><td align=\"center\">13.8</td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"center\">Total</td><td align=\"center\">1685.8</td><td align=\"center\">135</td><td align=\"center\">131</td><td/><td/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Colinearity between the faba bean and <italic>M. truncatula </italic>genomes</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">FB LG</td><td align=\"center\">No. of markers</td><td align=\"left\">No. of non-orthologous markers</td><td align=\"left\">No. of unassigned markers</td><td align=\"left\">Current Mt genetic coverage (LG: cM, No. colinear markers)^1,2</td><td align=\"left\">Current Mt pseudogenome coverage (bp)¹</td><td align=\"left\">Current faba bean genetic coverage (cM)²</td></tr></thead><tbody><tr><td align=\"left\">FB-1</td><td align=\"center\">30</td><td align=\"center\">6</td><td align=\"center\">2</td><td align=\"left\">8: 0 – 68.2, 11</td><td align=\"left\">8: 736241 – 32433863 (90%)</td><td align=\"left\">0 – 324.8 (100%)</td></tr><tr><td align=\"left\">FB-2</td><td align=\"center\">27</td><td align=\"center\">4</td><td align=\"center\">2</td><td align=\"left\">1: 0 – 58.5, 9 7: 58.4 – 60, 3</td><td align=\"left\">1: 59849 – 26963044 (86%) 7: 25725918 – 26383168 (2%)</td><td align=\"left\">152.6 – 313.1 (51.3%) 157.1 – 210.7 (25.4%)</td></tr><tr><td align=\"left\">FB-3</td><td align=\"center\">18</td><td align=\"center\">1</td><td align=\"center\">3</td><td align=\"left\">4: 48.1 – 59, 11</td><td align=\"left\">4: 14544831 – 34128326 (50%)</td><td align=\"left\">0 – 238.5 (100%)</td></tr><tr><td align=\"left\">FB-4</td><td align=\"center\">13</td><td align=\"center\">2</td><td align=\"center\">0</td><td align=\"left\">5: 0 – 32, 6 2: 12.7 – 37.9, 3</td><td align=\"left\">5: 594897 – 17960204 (47%) 2: 8491236 – 15201671 (24%)</td><td align=\"left\">0 – 142 (67%) 250.5 – 256.3 (1.9%)</td></tr><tr><td align=\"left\">FB-5</td><td align=\"center\">12</td><td align=\"center\">1</td><td align=\"center\">1</td><td align=\"left\">3: 28 – 70.3, 9</td><td align=\"left\">3: 10888141 – 28213807 (46%)</td><td align=\"left\">0 – 140.3 (92%)</td></tr><tr><td align=\"left\">FB6</td><td align=\"center\">8</td><td align=\"center\">4</td><td align=\"center\">0</td><td align=\"left\">1: 2 – 2.2, NA</td><td align=\"left\">1: 4942324 – 5151182 (0.7%)</td><td align=\"left\">-</td></tr><tr><td align=\"left\">FB-7</td><td align=\"center\">7</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"left\">7: 22.6 – 52.8, 4</td><td align=\"left\">7: 9999559 – 22150020 (38%)</td><td align=\"left\">0 – 58.9 (49%)</td></tr><tr><td align=\"left\">FB-8</td><td align=\"center\">6</td><td align=\"center\">1</td><td align=\"center\">0</td><td align=\"left\">2: 48 – 57.8, 5</td><td align=\"left\">2: 17523205 – 20646361 (11%)</td><td align=\"left\">0 – 49.9 (100%)</td></tr><tr><td align=\"left\">FB-9</td><td align=\"center\">5</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"left\">3: 62 – 72.5, 5</td><td align=\"left\">3: 19000000* – 30712150 (31%)</td><td align=\"left\">0 – 64.1 (100%)</td></tr><tr><td align=\"left\">FB-10</td><td align=\"center\">3</td><td align=\"center\">1</td><td align=\"center\">0</td><td align=\"left\">4: 0 – 7.4, NA</td><td align=\"left\">4: 537530 – 3489843 (8%)</td><td align=\"left\">-</td></tr><tr><td align=\"left\">FB-11</td><td align=\"center\">3</td><td align=\"center\">3</td><td align=\"center\">0</td><td align=\"left\">NA</td><td align=\"left\">NA</td><td align=\"left\">-</td></tr><tr><td align=\"left\">FB-12</td><td align=\"center\">3</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"left\">4: 60.4 – 61.1, NA</td><td align=\"left\">4: 36664007 – 38186081 (4%)</td><td align=\"left\">-</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>Supplementary table 1. Orthologous PCR markers developed in this study and genetically mapped in a F₆faba bean population between lines Vf6 and Vf27.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>^aFigures in parentheses are percentages of amplified markers of the total markers screened.</p><p>^bFigures in parentheses are percentages of sequenced markers of the total amplified markers.</p><p>^cFigures in parentheses are percentages of polymorphic markers of the total sequenced markers.</p></table-wrap-foot>", "<table-wrap-foot><p>^aCalculated by dividing the length of the chromosome (cM) by the number of space/distance between markers/loci.</p></table-wrap-foot>", "<table-wrap-foot><p>NA: not applicable</p><p>¹Data was either based on information from <italic>M. truncatula </italic>genome sequencing website [##UREF##10##25##] or from Choi <italic>et al</italic>. [##REF##15489274##12##].</p><p>²Data is provided for three or more colinear markers and excludes markers on the same chromosome rearranged relative to the order in <italic>M. truncatula</italic>.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2164-9-380-1\"/>", "<graphic xlink:href=\"1471-2164-9-380-2\"/>", "<graphic xlink:href=\"1471-2164-9-380-3\"/>", "<graphic xlink:href=\"1471-2164-9-380-4\"/>" ]

[ "<media xlink:href=\"1471-2164-9-380-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The ATP-Binding Cassette (ABC) transporting system is essential for uptake of specific nutrients by bacteria [##REF##8336670##1##]. ABC transporters are usually made up of a soluble-binding protein, a transmembrane protein, and an ATP-binding protein. The soluble-binding protein, also known as bacterial periplasmic-binding protein (bPBP), resides in the periplasm of Gram-negative bacteria. It selectively binds small molecules (i.e., amino acids, inorganic phosphate, or sugars), and carries them to the transmembrane component. As the loaded bPBP docks to the transmembrane unit, its conformation changes such that the substrate can be transferred to the transmembrane protein, and the bPBP is functionally recycled [##REF##9711542##2##].</p>", "<p>A large number of resolved x-ray crystal structures indicate that bPBP is composed of two domains connected by a flexible polypeptide linker that allows the protein to cycle between open and closed conformations. At equilibrium, the apo structure fluctuates between both conformations. Introduction of ligand stabilizes the closed conformation and thus concomitantly shifts the equilibrium. The ubiquitous bPBP hinge-bending motion makes them attractive targets for many practical applications, including drug delivery [##REF##10835100##3##] and biosensors [##REF##12381848##4##]. They are attractive drug delivery agents because the equilibrium between open and closed is sensitive to local environment (i.e., pH, temperature, or presence of interacting enzymes). Thus, it is feasible to engineer bPBP to bind to a drug of interest and then release it when the complex reaches a desired cell or compartment where conventional drug delivery methods fail to reach [##REF##10835100##3##]. The ligand-mediated hinge-bending motion in bPBPs can also be used to design fluorescent biosensors by covalently attaching fluorphore molecules so that fluorescence intensity and/or wavelength changes as a result of hinge-bending motion [##REF##12381848##4##]. In this way, the high specificity of binding and ligand-mediated conformational changes in bPBPs can be exploited to monitor for the presence of a specific ligand.</p>", "<p>A complete understanding of bPBP requires both mechanical and thermodynamic descriptions. Mechanical descriptions are necessary in order to detail changes in flexibility and mechanical couplings upon ligand binding, whereas thermodynamic descriptions are necessary to account for an ensemble of conformations associated with the apo and ligated states. Most computational methods focus only on one of the two phenomena. There are several common methods to explore protein thermodynamics (i.e., Monte Carlo sampling [##REF##15197271##5##], a variety of Ising-like models [##UREF##0##6##,##REF##10500173##7##] such as COREX [##REF##8876652##8##], and free energy decomposition schemes [##REF##12646369##9##, ####REF##8393940##10##, ##REF##8393941##11####8393941##11##]). Unfortunately, these methods generally fail to reproduce experimental thermodynamic response (i.e., excess heat capacity profiles). In addition, while most of these methods do consider an ensemble of conformations, they lack descriptions of mechanical couplings between sites within the protein. On the other hand, mechanical models (i.e., FIRST [##REF##11391777##12##] and Elastic Network Models (ENMs) [##REF##16280623##13##]) do provide detailed flexibility information, yet they completely lack any thermodynamic considerations. In principle, all-atom molecular dynamics (MD) simulations could be used to predict any mechanical and thermodynamic property of a protein. However, in practice, MD is much too computationally expensive to explore the thermodynamic limit [##REF##10322213##14##]. Consequently, Go-like models have been the mainstay approach that employs molecular dynamics simulation in conjunction with simplified molecular mechanics potentials tuned to the native state (or multiple states) [##REF##15740352##15##,##REF##16006532##16##].</p>", "<p>In this report, we employ a minimal Distance Constraint Model (mDCM) to investigate stability/flexibility relationships of four homologous bPBPs. We choose to use the mDCM among the available methods because; (<italic>i</italic>.) it uniquely synthesizes mechanical and thermodynamic descriptions, (<italic>ii</italic>.) its predictions compared to experiment have consistently achieved overall good agreement across a diverse population of proteins [##REF##8995351##18##,##REF##15740352##15##,##UREF##1##17##], and (<italic>iii</italic>.) it is a tractable modeling paradigm requiring just ~10 minutes of compute time for a 300 residue protein per thermodynamic condition.</p>" ]

[]

[ "<title>Results</title>", "<title>The bacterial periplasmic binding protein family</title>", "<p>Within the Structural Classification of Proteins (SCOP) database [##REF##7723011##31##], there are 29 different binding protein classes within the bPBP family, which SCOP calls the phosphate binding protein-like family. Several classes within the family include multiple species orthologs, and within each ortholog, many binding proteins have been crystallized within different states (i.e., presence/absence of ligand, wild-type vs. mutant, etc.). Fig. ##FIG##1##2a## shows the neighbor-joining phylogenetic tree built from the Probalign [##REF##17485479##32##,##REF##16954142##33##] multiple sequence alignment of 29 representative structures. Ideally, we would like to perform a comparative QSFR analysis across all 29 binding protein classes; however, pragmatic considerations make this impractical for manual manipulations. Since a large scale comparison will need automation based on prior experience, we extend our previous comparative QSFR analysis of a mesophilic/thermophilic RNase H pair [##REF##16287093##22##] to four different binding proteins. There is experimental <italic>C</italic>_{<italic>p </italic>}data for only one member of the family, the histidine binding protein (HBP). Three of the homologs, which include lysine/arginine/ornithine binding protein (LAOBP) [##REF##8496186##34##], glutamine binding protein (GBP) [##REF##9571045##35##] and HBP [##REF##8161536##36##], represent a closely related subfamily of amino acid binding proteins. Juxtaposed to this close-knit group is the distantly related phosphate binding protein (PhBP) [##REF##9228942##37##], the namesake of the SCOP family, which provides a point of reference over larger evolutionary distances. In order to circumvent parameterization issues, we apply the HBP parameterization to all four homologs, which we have previously demonstrated to be a satisfactory option [##REF##16287093##22##], especially when considering mechanical response. Two of the binding proteins have been crystallized in both the presence and absence of ligand, whereas the other two have only been crystallized in the presence of ligand. Table ##TAB##1##2## lists the PDB ID's and other relevant information of the six structures investigated herein.</p>", "<p>As expected from the phylogeny, structural superposition of all four ligated binding proteins (see Fig. ##FIG##1##2b##) confirms that the three amino acid binding proteins are much more structurally similar to each other than to PhBP. In Fig. ##FIG##1##2c##, PhBP is removed to highlight the similarity within the amino acid binding proteins. Table ##TAB##2##3## provides a more quantifiable description of (dis)similarity across the four binding proteins. Therein, pairwise root mean square deviation (RMSD) are computed using the combinatorial extension algorithm [##REF##9796821##38##], and percent sequence identities are computed from the pairwise global alignment using a dynamic programming algorithm [##UREF##5##39##]. Secondary structure topologies are mostly conserved across the four binding proteins considered here. However, across the entire bPBP family, several members have noncanonical secondary structure topologies, which indicates that a structure alignment of the complete family would be even less conserved.</p>", "<p>From the superposition of the apo and ligated LAOBP conformations (Fig. ##FIG##2##3a##), it is clear that there is pronounced conformational changes on ligand binding. In fact, the RMSD between the two LAOBP conformations is 3.5 Å. Similarly, the pairwise RMSD between the GBP pair is 4.0 Å. Although the RMSD values reported in Table ##TAB##2##3## represent overall dissimilarity between pairs, this does not properly describe local similarities because members of the bPBP family are composed of two domains connected by a flexible linker that allows the protein to transition from apo and ligated conformations. SCOP does not discriminate between the two domains; nevertheless, common domain identification algorithms (i.e., PDP [##REF##12584135##40##] and DomainParser [##REF##12560490##41##,##REF##11159328##42##]) do identify the two domains. In Table ##TAB##1##2##, the PDP domain boundaries are indicated. Note that the large domain is interrupted by insertion of the small one, resulting in two linkers across the domain boundary. The calculated pairwise RMSD between the large and small LAOBP domains individually is 0.4 Å and 0.6 Å, respectively (see Fig. ##FIG##2##3b–c##). (The corresponding RMSDs between the GBP domains are 0.6 Å and 0.8 Å.) These strong domain-specific similarities indicate that the large and small domains move more or less intact upon ligand binding, which is indicative of a hinge-like motion at the domain boundaries [##REF##17587456##43##].</p>", "<title>Differences within the H-bond networks</title>", "<p>Since the mDCM is primarily based on H-bond networks, understanding how they vary is critical to a proper understanding of the model predictions. Initially, we compared H-bond networks by simply counting pairwise topological differences within the networks of two structures. Table ##TAB##1##2## describes global H-bond statistics showing the number of H-bonds and average total energy across the four bPBP. The number of H-bonds observed within the four ligated bPBP homologs varies from 327 to 504, which is trivially explained by protein size. The correlation coefficient between H-bond numbers and protein size (number of residues) is 0.99. Surprisingly, in the two examples of ligated and apo structural pairs, the number of H-bonds within the ligated complex is smaller than the number within the apo structure. On average, there are 6.8 H-bonds between bPBP homolog and substrate. Despite the reduction in number of H-bonds on complex formation, the average H-bond strength is significantly increased. In fact, the total H-bond energies for the complex structures are much greater than their corresponding apo structures. These global differences within the H-bond networks will be used below to explain how differences within the model predictions arise.</p>", "<p>We found that the above global comparisons failed to provide accurate descriptions of the differences within other QSFR metrics. This failure is likely due to two key issues. First, the employed H-bond energy function [##REF##9194194##26##] is extremely liberal. For example, the energy function frequently identifies very weak H-bonds, such as those between residues <italic>i </italic>and <italic>i+3 </italic>within α-helices (results not shown). They have no pronounced effect on mDCM results as their energies are very small and they rarely contribute anything to the conformational entropy (since they are redundant to other constraints). Nevertheless, the large number of these feeble H-bonds inappropriately skews a distance metric that is based on global statistics. These feeble H-bonds could be filtered out using an energy cut-off, but then we would have to determine this cut-off. Second, due to the nonadditive nature of the mDCM, local topological considerations, which are lost in global metrics, have a considerable effect on output.</p>", "<p>We opted to employ a simpler but effective approach; where we compare H-bond networks by plotting a H-bond contact map (for example, see Fig. ##FIG##3##4##) to visualize essential differences. The liberal nature of the H-bond potential, which is primarily manifest within secondary structure elements, can be (visually) ignored because these feeble H-bonds are clustered within a high density of the stabilizing H-bonds. Moreover, the approach straightforwardly highlights where local hydrogen bond topology differences occur. Whether comparisons are made between any two of the four ligated bPBP homologs or between the apo/ligated pairs, key differences in H-bonding mainly occur between non-secondary structure residues. As discussed below, this result has substantial affect on the observed mechanical linkage properties.</p>", "<title>Thermodynamic descriptions</title>", "<p>Employing the HBP best-fit parameters (Table ##TAB##0##1##) for each of the four ligated structures, their predicted <italic>C</italic>_{<italic>p </italic>}curves as a function of temperature, relative to their predicted melting temperature (i.e. <italic>T – T</italic>_{<italic>m</italic>}), (see Fig. ##FIG##4##5a##) shows a high degree of diversity. For example, the maximal <italic>C</italic>_{<italic>p </italic>}ranges from 26.4 kcal/(mol·K) for GBP to 174.1 kcal/(mol·K) for PhBP. These differences are not particularly surprising, and, moreover, the underlying H-bond networks explain the variation within the heat capacities. For example, greater H-bond numbers is strongly correlated with <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}(<italic>R </italic>= 0.90). Similarly, <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}is also strongly correlated to the number of residues within the protein (<italic>R </italic>= 0.87), which occurs because the number of H-bonds and the number of residues are almost perfectly correlated (<italic>R </italic>= 0.99). Across the four structures, the average number of H-bonds per residue is 1.47 (standard deviation = 0.08). Based on Eq. (4), the effect of H-bond numbers and total H-bond energy on <italic>C</italic>_{<italic>p </italic>}can be conceptualized. Greater H-bond numbers provide more opportunities for enthalpic fluctuations to occur, thus increasing the <italic>C</italic>_{<italic>p</italic>}. In the same manner, the total H-bond energies, <italic>U</italic>_{<italic>hb</italic>}^{<italic>max</italic>}, are even more strongly correlated to <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}(<italic>R </italic>= -0.97). A large part of this relationship is explained by the fact the number of H-bonds is, of course, strongly related to the total H-bond energy (<italic>R </italic>= -0.97). However, the increased correlation to <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}is due to the greater effect upon the total enthalpy when a stronger H-bond is removed.</p>", "<p>The number of H-bonds in each structure is much less strongly correlated to <italic>T</italic>_{<italic>m </italic>}(<italic>R </italic>= 0.69); melting points are included in Table ##TAB##3##4##. The reduced correlation is primarily due to the nontrivial way entropies depend on the distribution of H-bonds, not just simple total numbers. The free energy landscapes, <italic>G</italic>(<italic>θ</italic>) = -<italic>RT</italic>ln<italic>Z</italic>(<italic>θ</italic>), of each of the four structures are plotted at their respective <italic>T</italic>_{<italic>m </italic>}in Fig. ##FIG##4##5b##. Each landscape contains two minima separated by an intervening transition state barrier, indicative of first order (2-state) kinetics. The locations of key free energy landscape features, θ_{<italic>nat</italic>}, θ_{<italic>TS</italic>}, and θ_{<italic>unf</italic>}, are also provided in Table ##TAB##3##4##. These values are within the range established previously over a structurally diverse set of globular proteins [##REF##15498582##21##]. The heights of the free energy barriers shown in Fig. ##FIG##4##5b## are strongly correlated to <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}(<italic>R </italic>= 0.92). The mDCM recapitulates commonly found correlation between barrier height and <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}when there is two-state kinetics [##UREF##6##44##]. However, this correlation is not absolutely necessary (neither theoretically nor experimentally). For example, proteins with pronounced <italic>C</italic>_{<italic>p </italic>}curves can be barrier free [##REF##15591110##45##], and need not demonstrate cooperative folding/unfolding transitions (e.g., myoglobin [##REF##9521125##46##]).</p>", "<p>In addition to the bPBP-ligand complex structures, LAOBP [##REF##8496186##34##] and GBP [##REF##8831790##47##] have been crystallized in the open (apo) forms. Differences within the apo vs. ligated LAOBP hydrogen bond networks are plotted in Fig. ##FIG##5##6##. Juxtaposed to the variability in <italic>C</italic>_{<italic>p </italic>}discussed above, comparison of the ligated and apo structures reveals several clear trends. First, in both cases the <italic>T</italic>_{<italic>m </italic>}of the apo structure is reduced compared to its ligated counterpart (Fig. ##FIG##4##5c##). This predicted down shift for <italic>T</italic>_{<italic>m </italic>}when LAOBP and GBP is ligated is fully consistent with experimental observations [##REF##10866829##27##]. The mDCM also predicts <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}to be substantially lowered upon ligand binding, which arises due to the reduced likelihood of enthalpic fluctuations. This effect is shown in Fig. ##FIG##6##7a##, where <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}is plotted versus total H-bond energy for the four bPBP-ligand complexes. The values of the apo structures are superimposed onto the plot of the four complexes, which confirms that the observed changes within <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}upon ligand removal are strongly associated with the loss of H-bonds. Of course, addition/removal of the ligand is also associated with hydrophobic interactions, free ligand entropy and its chemical potential, all of which are absent within the mDCM. Nevertheless, the importance of the H-bond network as a dominate factor has been pointed out by Cooper [##REF##10885396##48##], showing how the observed <italic>C</italic>_{<italic>p </italic>}changes between folded and unfolded protein conformations arise through fluctuations within the H-bond network. Consistent with our results, the same <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}reduction within the apo form of HBP is observed within the experimental DSC curves [##REF##10866829##27##]. The reduction of <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}is accompanied by a lower energy barrier (see Fig. ##FIG##4##5d##). Interestingly, the mDCM predicts a linear relationship between <italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}and free energy barrier height over the six cases studied (Fig. ##FIG##6##7b##). In fact, the mDCM predicts an unfolding transition with virtually no barrier for apo GBP, suggesting that unfolding/folding is a continuous transition (second order transition). It will be quite interesting to know if future experiments are consistent with this prediction.</p>", "<title>Flexibility along the backbone</title>", "<p>Compared to the large differences within the thermodynamic predictions, predictions of backbone flexibility are more conserved. Fig. ##FIG##7##8## color-codes the flexibility index onto the structural alignment of the four bPBP homologs. Regions corresponding to α-helices are predicted to be very rigid, and are the most conserved. Flexibility and rigidity predictions for β-strands are less conserved than found for α-helices. There is large variation in the flexibility index within coil regions, and, as expected, highly flexible regions uniquely occur within the coils. The observed conservation of the backbone flexibility is attributed to the similar H-bond networks within the secondary structure elements. This result is not unexpected as a strong correlation between secondary structure and H/D exchange experiments is well established [##REF##15581564##49##]. It is worth mentioning that our backbone flexibility measures are consistent with those from Elastic Network Models, which are sensitive to fold, but not sequence. Specifically, (the flexible component of the flexibility index) nearly perfectly correlates to the hinge sites predicted by Bahar et al. [##REF##10733987##50##]. These results show that the fold of the protein primarily determines backbone flexibility. Moreover, backbone flexibility is well conserved because of the underlying H-bond sub-network involving secondary structure elements, which are well conserved within a family. As described above, the H-bond network is significantly altered upon complex formation. Consistent with the hinge motion inferred from Fig. ##FIG##2##3b–c##, differences within the H-bond network between the apo/ligated pair primarily occur within non-secondary structure elements (refer back to Fig. ##FIG##5##6##). Consequently, backbone flexibility is not significantly affected outside the binding site region, which can be seen in Fig. ##FIG##8##9##.</p>", "<title>Cooperativity correlation</title>", "<p>In previous work [##REF##16287093##22##], we demonstrated that cooperativity correlation is moderately conserved across an orthologous RNase H pair. However, this is clearly not the case here; juxtaposed to the flexibility conservation along the backbone, there is immense variation within cooperativity correlation (see Fig. ##FIG##9##10##). In all cases, the large domain is primarily composed of one large rigid cluster, which results in rigidity correlation in each of the four corners (because the domain is interrupted) of the cooperativity correlation plots. Conversely, the small domain demonstrates much variability within its cooperativity correlation. The small domain generally contains flexibly and rigidly correlated regions within itself and extending into the larger domain. However, the small domain in LAOBP is mostly rigidly correlated with itself and the rest of the protein, indicating that it is primarily composed on one larger rigid cluster. These differences will be discussed again below.</p>", "<p>Differences between the apo/ligated pairs are inline with expectations. In the case of LAOBP, each domain appears as rigidly correlated with itself, whereas the couplings between the two domains are a mix of both <italic>weak </italic>flexible and rigid correlations (Fig. ##FIG##10##11##). This result is exactly consistent with the inferred hinge motion from the structural analyses discussed above and demonstrated in Fig. ##FIG##2##3##. In the case of GBP, the apo structure is significantly destabilized (<italic>Δ T</italic>_{<italic>m</italic>}<italic>= </italic>39 K). This fact, coupled with the continuous transition observed within apo GBP, results in a cooperativity correlation plot that, as might be expected, is primarily flexibly correlated.</p>" ]

[ "<title>Discussion</title>", "<p>The QSFR results presented demonstrate considerable variability in both mechanical and thermodynamic descriptions of the bPBP family. This observed variability is explained by differences within the underlying H-bond networks. Most notably, variability of thermodynamic quantities correlate well to overall statistical characteristics of H-bond networks, which is consistent with the arguments made by Cooper [##REF##10885396##48##]. Moreover, the predicted reduction of <italic>T</italic>_{<italic>m </italic>}upon ligand binding for LAOBP and GBP are qualitatively consistent with the same effect that is experimentally observed for HBP. The observed differences in the thermodynamic response is not particularly unexpected, except, we were initially surprised by the low barrier height found in the free energy landscape for GBP. However, within the context of the mDCM predictions, the total H-bond energy is an excellent indicator for predicting <italic>C</italic>_{<italic>p</italic>}^{<italic>max</italic>}, and this in return correlates well with barrier height. In the context of this simple extensive H-bond network property (i.e. total energy) over the structures studied, the dramatic deviation from two-state behavior in GBP is not an exceptional case.</p>", "<p>The predicted conservation of backbone flexibility is consistent with natural expectations. On the other hand, variability found within cooperativity correlation measures is, at first, somewhat surprising. In prior work we compared mesophilic and thermophilic RNase H orthologs, and the substantial differences in the cooperativity correlation plots (at their respective <italic>T</italic>_{<italic>m</italic>}) was contributed to different thermodynamic stability requirements of these two proteins. Nevertheless, after coarse-graining the flexibly correlated regions into contiguous stretches, we identified conserved features presumably needed to support common functional requirements of these two proteins. Our hypothesis that flexibility and rigidity properties of the native state would be conserved across proteins at their corresponding <italic>T</italic>_{<italic>m </italic>}within the same family was inconclusive to that study. Starting this work, we maintained the hypothesis that the dominant features in the cooperativity correlations important for function would be markedly conserved. Although general features indicative of the overall fold are indeed conserved, it is the variation that most stands out. Although we can expand the number of proteins to analyze within a family, this report clearly shows that cooperativity correlations are very sensitive to the H-bond network, and the variance in these measures will be the dominate feature. Moreover, other recent results (unpublished) involving nine thioredoxin homologs, also conclude that variability in cooperativity correlation is significant, and sensitive to non-secondary structure elements within the H-bond network.</p>", "<p>Clearly, the mDCM, by construction, must have some degree of sensitivity to the characteristics of the H-bond network due to the heavy emphasis it places on H-bonds and the properties of network rigidity. To address the relative importance of H-bond interactions, in ongoing work, we have modified the DCM to deemphasize the affects of native and disordered torsions, and demonstrated that the H-bond interactions within the mDCM account for the bulk of the properties characterizing the thermodynamic transition. It is quite remarkable that despite the simplicity of the mDCM, it captures the essential elements of protein thermodynamics and mechanical linkage mechanisms. However, consider what may happen when more complexity is modeled to explicitly account for solvent effects that include the hydrophobic effect (also work in progress). The likelihood that adding nonspecific interactions will decrease variance in mechanical response is remote. The hydrophobic interactions are of entropic nature related to transfer of water from the interior of a protein to aqueous solution. Based on this simple argument, and having observed no indication to the contrary in alternative model explorations, the H-bond network is expected to remain the key determinant for rigidity and flexibility properties of a protein. Keeping this in mind, we come full circle in understanding why the mDCM flexibility/rigidity predictions are robust despite a greater variance in predicted melting temperatures and other thermodynamic quantities. The nonspecific interactions will be important to modify and perturb protein thermodynamic stability, but the native H-bond network (as determined by X-ray crystallography) will be present in the exact same way as currently modeled in mDCM. We therefore take the view that the mDCM is capturing the most essential element for understanding cooperativity within proteins important for their function.</p>", "<p>The H-bond network explains the dichotomy between conservation within backbone flexibility and the lack of conservation within cooperativity correlation. For the most part, pairwise differences within the H-bond networks occur within non-secondary structure (primarily sidechain) H-bonds, meaning that the secondary structure H-bonds are mostly conserved. The conservation within these secondary structure H-bonds is what leads to conserved flexibility along the backbone. Note that backbone flexibility is not exactly conserved; in fact there are many local differences within Fig. ##FIG##7##8##. These differences presumably arise due to the observed differences elsewhere and the long-range nature of network rigidity [##REF##11391777##12##]. Furthermore, while the secondary structure H-bonds may be conserved within the contact map analysis, this does not mean their energies are equivalent, which also affects the flexibility predictions. Nevertheless, in spite of these nuanced effects, the qualitative conservation within backbone flexibility is a result of the conserved nature of the secondary structure H-bonds.</p>", "<p>Compared to local secondary structure H-bonds, differences within sidechain-sidechain H-bonds are expected to result in key differences within the topology of the H-bond networks due to their ability to span across long stretches of sequence. Meaning, a change in just a handful of critically placed non-secondary structure H-bonds can drastically alter mechanical linkage properties. These changes would be most pronounced within the cooperativity correlation plots that explicitly rely on the linkage information. As with the variability within the thermodynamic quantities, the differences between the apo/ligated pairs is used to bolster the argument that the predicted differences are real. In both cases (LAOBP and GBP), the changes upon complex formation are consistent with intuition. Moreover, as we have discussed in previous works [##REF##15542549##19##, ####REF##16542678##20##, ##REF##15498582##21##, ##REF##16287093##22####16287093##22##], cooperativity correlation can be interpreted in terms of allostery, and it is a well-known that allosteric response can vary significantly across a family [##REF##14998571##51##, ####REF##4985590##52##, ##REF##11343922##53####11343922##53##].</p>", "<p>While the arguments above do not prove that trends within the observed predictions are real, they do strongly suggest that this is the case. And, if certain predictions within the current mDCM remain suspect, the above results clearly indicate that future, more sophisticated DCMs should be able to accurately describe trends within mechanical and thermodynamic properties. The results presented here clearly demonstrate how subtle differences within the H-bond networks can lead to unexpected and pronounced complexity. These results are very exciting because it suggests that the paradigm of the DCM is effective in elucidating consequences of altered H-bond networks, for which the computation design of H-bond networks has precedence [##REF##16831445##54##]. More generally, the results presented here suggest that monitoring QSFR will be important to protein design with targeted mechanical and thermodynamic properties, and should be feasible in the near future.</p>" ]

[]

[ "<p>This is an open access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Gram-negative bacteria use periplasmic-binding proteins (bPBP) to transport nutrients through the periplasm. Despite immense diversity within the recognized substrates, all members of the family share a common fold that includes two domains that are separated by a conserved hinge. The hinge allows the protein to cycle between open (apo) and closed (ligated) conformations. Conformational changes within the proteins depend on a complex interplay of mechanical and thermodynamic response, which is manifested as an increase in thermal stability and decrease of flexibility upon ligand binding.</p>", "<title>Results</title>", "<p>We use a distance constraint model (DCM) to quantify the give and take between thermodynamic stability and mechanical flexibility across the bPBP family. Quantitative stability/flexibility relationships (QSFR) are readily evaluated because the DCM links mechanical and thermodynamic properties. We have previously demonstrated that QSFR is moderately conserved across a mesophilic/thermophilic RNase H pair, whereas the observed variance indicated that different enthalpy-entropy mechanisms allow similar mechanical response at their respective melting temperatures. Our predictions of heat capacity and free energy show marked diversity across the bPBP family. While backbone flexibility metrics are mostly conserved, cooperativity correlation (long-range couplings) also demonstrate considerable amount of variation. Upon ligand removal, heat capacity, melting point, and mechanical rigidity are, as expected, lowered. Nevertheless, significant differences are found in molecular cooperativity correlations that can be explained by the detailed nature of the hydrogen bond network.</p>", "<title>Conclusion</title>", "<p>Non-trivial mechanical and thermodynamic variation across the family is explained by differences within the underlying H-bond networks. The mechanism is simple; variation within the H-bond networks result in altered mechanical linkage properties that directly affect intrinsic flexibility. Moreover, varying numbers of H-bonds and their strengths control the likelihood for energetic fluctuations as H-bonds break and reform, thus directly affecting thermodynamic properties. Consequently, these results demonstrate how unexpected large differences, especially within cooperativity correlation, emerge from subtle differences within the underlying H-bond network. This inference is consistent with well-known results that show allosteric response within a family generally varies significantly. Identifying the hydrogen bond network as a critical determining factor for these large variances may lead to new methods that can predict such effects.</p>" ]

[ "<title>Experimental</title>", "<title>The distance constraint model</title>", "<p>The DCM is based on a free energy decomposition scheme combined with constraint theory, such that microscopic interactions are represented as mechanical distance constraints. Each distance constraint is associated with an enthalpic and entropic contribution [##UREF##1##17##]. A Gibbs ensemble of accessible microstates is defined by a set of topologically distinct mechanical frameworks. A single mechanical framework encompasses an ensemble of all accessible molecular conformations that are consistent with a specified set of distance constraints. As a result, each framework is defined by the topology of constraint placement, and the enthalpy of the framework is calculated as a linear sum of enthalpy components over all constraints present. While preserving a specified distance constraint topology, the protein will sweep over its accessible phase space as it changes shape (geometry). As the model is quite simple, it is assumed that as long as the constraint topology does not change, different geometries consistent with the specified constraints are degenerate in enthalpy. As a result, the conformational entropy of each mechanical framework is meaningful. It has been a major problem that conformational entropy is generally a nonadditive property of free energy decompositions [##REF##8995351##18##]. As explained below, we account for this nonadditive property using network rigidity to obtain a good upper bound estimate of conformational entropy by summing entropy components over a preferential set of independent constraints.</p>", "<p>Specifically, the DCM decomposes the Gibbs free energy of a single framework into sums of different types of microscopic interactions also modeled as one (or more) distance constraint(s). Each interaction type, <italic>t</italic>, is assigned enthalpic and entropic values (<italic>H</italic>_{<italic>t </italic>}and <italic>S</italic>_{<italic>t</italic>}, respectively). Enthalpic contributions are linearly summed, whereas entropic contributions are only summed over independent constraints, which are identified by a network rigidity algorithm. Key to this rigidity algorithm is to recursively build up a mechanical framework one constraint at a time and, during this process, to identify which constraints are independent. Interestingly, the set of independent constraints that this recursive process identifies is not unique, because the identification of redundancy depends on the specific ordering of constraint placement. Recalling that only independent constraints will contribute to the total conformational entropy, it is clear that any order provides an upper bound estimate to the true conformational entropy. Due to the recursive nature of the rigidity algorithm, a rigorous lowest upper bound estimate is achieved by placing the lowest entropy constraints preferentially before any other constraints with higher entropy values.</p>", "<p>Regarding the long-range nature of network rigidity as an underlying mechanical interaction captures the nonadditive nature of conformational entropy. The free energy for a given framework, <italic>F</italic>, is calculated by:</p>", "<p></p>", "<p>where <italic>N</italic>_{<italic>t </italic>}is the total number of interactions of type <italic>t</italic>, and <italic>I</italic>_{<italic>t </italic>}is the total number of <italic>independent constraints </italic>of type <italic>t</italic>, which is uniquely identified because of the preferential ordering. Subsequently, the partition function can be constructed from all possible frameworks. Detailed descriptions of how to calculate the Gibbs free energy, free energy landscapes, and partition functions using the DCM can be found in several publications [##REF##15542549##19##, ####REF##16542678##20##, ##REF##15498582##21##, ##REF##16287093##22####16287093##22##]. Solving the DCM can be done using a number of different techniques. For example, transfer matrix methods have been applied to the helix to coil transition [##REF##14754182##23##,##REF##15307195##24##]. Additionally, we recently developed an <italic>ab initio </italic>method based on analytical solutions of the configuration integral that model distance constraints as delta functions (unpublished). Here we employ a novel non-homogenous mean field approximation that we previously developed [##REF##15542549##19##,##UREF##2##25##].</p>", "<title>Free energy decomposition</title>", "<p>Within the mDCM, the microscopic interactions explicitly modeled are covalent bonds, hydrogen bonds (H-bonds), and torsional-forces; note that salt bridges represent a special case of H-bonds. The covalent bonds within a protein do not break/form due to thermal fluctuations, and are therefore quenched constraints (ever present). Quenched constraints require no parameterization for their enthalpy and entropy contributions because they only shift the free energy of a protein by a constant. The mDCM is based on three types of fluctuating interactions that need parameterization (H-bonds and two types of torsion-forces). An enthalpic and entropic value is associated with each interaction (see Table ##TAB##0##1##). Intramolecular H-bond enthalpies are described using a common multi-atom empirical potential [##REF##9194194##26##] that depends on quantum mechanical hybridization, and on the local environment defined by all atom positions. The H-bond potential energy function ensures that the enthalpy of an intramolecular H-bond ranges from -8 to 0 kcal/mol.</p>", "<p>Different interaction types are modeled by a different number of distance constraints. The two torsion-force interactions are each modeled as one distance constraint, while the multi-particle H-bond and covalent bond interactions are each modeled as five constraints [##REF##11391777##12##]. The entropic cost, <italic>γ</italic>_{<italic>env</italic>}, for each independent constraint representing the H-bond is modeled as a linear relationship to the H-bond enthalpy by <italic>γ</italic>_{<italic>env </italic>}= <italic>γ</italic>_{<italic>max </italic>}(1+E/8), where <italic>γ</italic>_{<italic>max </italic>}is a fitting parameter. It follows that 0 ≤ <italic>γ</italic>_{<italic>env </italic>}≤ <italic>γ</italic>_{<italic>max </italic>}is the allowed entropy range per H-bond distance constraint. Each of the five distance constraints modeling a single H-bond are assigned the same entropy value. Sigma covalent bonds are free to rotate and are partitioned into native-like or disordered conformational states that represent an Ising-like (binary) discrimination of torsion-forces. The (native, disordered) conformation includes all dihedral angles that are (similar to, substantially differing from) those defined in the native three-dimensional structure. Table ##TAB##0##1## lists enthalpies and entropies for the torsion-force constraints.</p>", "<title>Free energy landscape</title>", "<p>The free energy landscape is defined using two order parameters: the number of native torsion-force constraints, <italic>N</italic>_{<italic>nt</italic>}, and the number of H-bond constraints, <italic>N</italic>_{<italic>hb </italic>}(see Fig. ##FIG##0##1a##). The free energy of each macrostate, <italic>G(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>)</italic>, is calculated using a mean-field approximation [##REF##15542549##19##] by Monte Carlo sampling over frameworks satisfying the two order parameters; as few as 200 samples are needed for good statistics. Using the enthalpic and entropic parameters in Table ##TAB##0##1##, the free energy of a given macrostate is calculated by:</p>", "<p></p>", "<p>where <italic>U</italic>_{<italic>hb </italic>}is the intramolecular H-bond energy, <italic>u </italic>is an average H-bond energy to solvent, <italic>v </italic>is the energy of a native-like torsion angle, <italic>S</italic>_{<italic>c</italic>}<italic>(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>) </italic>is the conformational entropy, and <italic>S</italic>_{<italic>m</italic>}<italic>(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>) </italic>is the mixing entropy of the macrostate associated with the number of ways of distributing <italic>N</italic>_{<italic>nt </italic>}native-torsions and <italic>N</italic>_{<italic>hb </italic>}H-bonds within the protein. While not specified here, <italic>S</italic>_{<italic>c</italic>}<italic>(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>) </italic>is a linear function of <italic>γ</italic>_{<italic>env</italic>}, <italic>δ</italic>_{<italic>nat</italic>}, <italic>δ</italic>_{<italic>dis</italic>}, and <italic>I</italic>_{<italic>t</italic>}, where <italic>t </italic>denotes each of the three fluctuating constraint types (see Table ##TAB##0##1##) [##REF##15542549##19##]. In most systems investigated previously [##REF##15542549##19##, ####REF##16542678##20##, ##REF##15498582##21##, ##REF##16287093##22####16287093##22##], hysteresis is observed near the melting temperature, <italic>T</italic>_{<italic>m</italic>}, meaning there are two minima within the free energy landscape corresponding to the native and unfolded ensembles. These two free energy basins are schematically shown in Fig. ##FIG##0##1a## and will appear when the protein exhibits two-state folding characteristics.</p>", "<p>Values of the three free parameters {<italic>u</italic>, <italic>v</italic>, <italic>δ</italic>_{<italic>nat</italic>}} are determined by fitting to experimental heat capacity (<italic>C</italic>_{<italic>p</italic>}) curves from differential scanning calorimetry (DSC) [##REF##15498582##21##]. To the best of our knowledge, the DCM is the only microscopic computational modeling scheme that can quantitatively reproduce experimental protein heat capacity curves. <italic>C</italic>_{<italic>p </italic>}is directly computed from the enthalpic fluctuations, ⟨Δ<italic>H</italic>²⟩, across the ensemble using the following relationship:</p>", "<p></p>", "<p>The fluctuations are computed from the enthalpic portions of Eq. (2) using:</p>", "<p></p>", "<p>Simulated annealing is used to search through the parameter-space to find the best-fit solution(s) to the experimental <italic>C</italic>_{<italic>p </italic>}data. In previous works [##REF##15542549##19##,##REF##15498582##21##], two of the parameters (<italic>γ</italic>_{<italic>max </italic>}and <italic>δ</italic>_{<italic>dis</italic>}) listed in Table ##TAB##0##1## have been fixed, and are now treated as transferable. For a given thermodynamic condition, the entire free energy landscape and many thermodynamic response functions are calculated in a matter of ~10 minutes for a 300-residue protein on a desktop computer using one CPU. A workflow diagram for the mDCM is provided in Additional file ##SUPPL##0##1##.</p>", "<title>Ensemble averaging network rigidity</title>", "<p>In addition to the thermodynamic descriptions, the DCM also computes mechanical properties of a protein. The mechanical and thermodynamic aspects to the problem are actually intertwined such that one cannot be calculated without the other. In virtually instantaneous compute times, graph rigidity algorithms identify rigid and flexible regions, independent constraints, redundant constraints, and correlated motions. FIRST is capable of doing these calculations in a fraction of a second for a 300-residue protein. In contrast, the DCM runs in matter of minutes, because these rigidity calculations are repeated millions of times using Monte Carlo sampling throughout the various nodes in the free energy landscape. A large number of quantities that describe the mechanical nature of the protein can be derived (see below), and these properties are averaged over many accessible frameworks. Randomly perturbing the constraint topology initially defined by the native template structure generates the ensemble of these accessible frameworks.</p>", "<p>The rigidity information from distinct constraint topologies is appropriately averaged using Boltzmann weights, given as exp(-<italic>G</italic>(<italic>N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}), within the free energy landscape node of interest. It is worth noting that the program FIRST [##REF##11391777##12##], a popular implementation of network rigidity for proteins, uses an <italic>ad hoc </italic>sliding energetic cut-off threshold, <italic>E</italic>_{<italic>c</italic>}, to determine when H-bond constraints are present or not. FIRST is an athermal model where each constraint is either present or not; meaning, it does not allow for fluctuations. When a constraint is present, regardless of what it is being used to model (i.e., covalent vs. noncovalent bond), it is treated as equivalent. In the DCM, large numbers of accessible constraint topologies are sampled, their associated free energies calculated, and then the native free energy basin is identified. As such, the most probable states (free energy basins) are determined, which eliminates the arbitrariness of <italic>E</italic>_{<italic>c</italic>}.</p>", "<p>Once mDCM parameterization is achieved, a complete set of mechanical and thermodynamic quantities can be calculated. A large number of mechanical quantities (including: rigid cluster decomposition, independent and redundant constraints, rigid cluster susceptibility, global flexibility, local flexibility index, and flexibility propagation) are evaluated in a thermodynamically meaningful way. Moreover, averages of local mechanical properties can be used to identify long-range couplings between sites (i.e., allosteric communication). As such, the DCM provides a natural mechanism to evaluate Quantitative Stability/Flexibility Relationships (QSFR) [##REF##16287093##22##].</p>", "<title>Selection of DCM parameters</title>", "<p>Remarkably, backbone flexibility predicted by FIRST with suitable <italic>E</italic>_{<italic>c</italic>}, applied to a single constraint topology (i.e. derived from the native state template structure), is usually similar to the result from the DCM [##REF##15498582##21##]. This is an indication that there is a most probable state that is sharply peaked around typical backbone flexibility characteristics. Ensemble averaging makes rigidity/flexibility predictions robust, while eliminating the arbitrariness in <italic>E</italic>_{<italic>c</italic>}. Nevertheless, the DCM has three free parameters {<italic>u</italic>, <italic>v</italic>, <italic>δ</italic>_{<italic>nat</italic>}} that are obtained by fitting to heat capacity, for example. From previous [##REF##16287093##22##] and unpublished work, we find that rigidity/flexibility properties are largely insensitive to a ± 15% deviation in these three free parameters, and there is a corresponding variance of about ± 10% in the predicted <italic>T</italic>_{<italic>m </italic>}(corresponding to ± 35 K). Given these sensitivity levels, the fitting parameters determined from one reference protein structure (based on its measured <italic>C</italic>_{<italic>p</italic>}) is then applied on all similar proteins, such as those from the same family. In this work, the three fitting parameters {<italic>u</italic>, <italic>v</italic>, <italic>δ</italic>_{<italic>nat</italic>}} are taken from previous work [##REF##15542549##19##], where they were determined by fitting Histidine Binding Protein (HBP) to the <italic>C</italic>_{<italic>p </italic>}data as reported in reference [##REF##10866829##27##]. As such, we are able to characterize emergent mechanical and thermodynamic properties across a series of proteins in a consistent way. Similarities and differences in both mechanical and thermodynamic properties will therefore reflect structural differences found in these proteins.</p>", "<title>Relating thermodynamic and mechanical quantities</title>", "<p>As described above, an ensemble of mechanical frameworks is needed to obtain thermodynamic information from a statistical mechanics description of the system, and mechanical quantities are averaged over this ensemble. Moreover, correlation functions describing detailed rigidity/flexibility properties within a protein can be calculated based on probability measures over the ensemble of constraint topologies. We generically refer to all thermodynamic and mechanical information taken together as QSFR. The QSFR measures employed in this work are now described.</p>", "<p>The free energy landscape is described, as defined in Eq. 2, by the function <italic>G(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>)</italic>. The macrostate of the protein is specified by <italic>(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>)</italic>, which defines a node within Fig. ##FIG##0##1a##. Besides free energy, other physical properties are expressed as a function of the macrostate. For each node, consider a physical property given by <italic>x(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>)</italic>, which is already an average over the sub-ensemble of all constraint topologies associated with the macrostate. One such property of particular importance is the global flexibility of a protein, which is the number of independent degrees of freedom per residue. We call this quantity the global flexibility order parameter, defined by <italic>θ</italic>' (<italic>N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}) ≡ ⟨<italic>I</italic>_{<italic>dis </italic>}(<italic>N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>})⟩/<italic>n</italic>, where <italic>n </italic>is the number of residues in the protein. The quantity ⟨<italic>I</italic>_{<italic>dis</italic>}(<italic>N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>})⟩ is the average number of independent disordered torsion constraints over a sub-ensemble of constraint topologies of the macrostate.</p>", "<p>Collapsing the free energy landscape in constraint space onto the global flexibility order parameter makes a direct connection between free energy of a protein and its flexibility. The process entails calculating a partition function, <italic>Z(θ)</italic>, given by:</p>", "<p></p>", "<p>where <italic>B(θ, N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>) </italic>is a binning function. The binning function is 1 whenever <italic>-Δθ &lt; |θ-θ '(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>)| </italic>≤ <italic>Δθ</italic>, and 0 otherwise. Note that <italic>Δθ </italic>is a tolerance used for bin size, which has been set to 0.005 because we plot functions of <italic>θ </italic>in increments of 0.01. Thus, all nodes of the original two-dimensional landscape must fall into one, and only one, <italic>θ </italic>bin. The free energy of the protein as a function of its global flexibility is given as <italic>G</italic>(<italic>θ</italic>, <italic>T</italic>) = -<italic>RT</italic>ln<italic>Z</italic>(<italic>θ</italic>). Consequently, this mapping straightforwardly results in a simple one dimensional free energy landscape.</p>", "<p>At the <italic>T</italic>_{<italic>m</italic>}, proteins that fold via two-state kinetics will have a free energy landscape that looks like a \"W\". That is, <italic>G(θ, T</italic>_{<italic>m</italic>}<italic>) </italic>will have two minimum and an intervening barrier that separates the two minima. In general, it is difficult to interpret an order parameter as a reaction coordinate. We have suggested that the <italic>θ </italic>may be a good reaction coordinate, but have not studied kinetics to confirm this. However, it is instructive to explain features about <italic>θ </italic>that would be required if it did indeed serve as a reaction coordinate. Within the mDCM, there are two types of fluctuating constraints, which define the two-dimensional free energy landscape shown in Fig. ##FIG##0##1a##. Proceeding from left to right, <italic>θ </italic>will decrease since the protein rigidifies as more native torsion force constraints are added. Similarly, progressing from bottom to top, <italic>θ </italic>will decrease in the same manner as more H-bonds are added. Consequently, the locus of points along any one of the light grey lines will have nearly constant <italic>θ </italic>because a reduction of rigidity in one direction is offset by an increase in the other. The lower left-hand side of the plot coincides to small values of <italic>θ</italic>, whereas the upper-right coincides with larger values. The red line demonstrates the most direct route between the two free energy basins, over a single saddle. The saddle (Fig. ##FIG##0##1b##) reflects a state with a mixture of rigid and flexible regions coexisting, which involved a nucleation process of either forming rigid regions upon folding, or breaking apart rigid regions upon unfolding.</p>", "<p>From simple geometric considerations, and the construct of the mDCM, the path along the global flexibility order parameter must always cross the saddle. This motivates calculating other physical quantities in terms of <italic>θ</italic>. We calculate the average properties of a physical property, <italic>x(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>)</italic>, over a restricted ensemble based on the protein's global flexibility, given by:</p>", "<p></p>", "<p>For example, it is of interest to select <italic>θ </italic>to be one of the following: <italic>θ</italic>_{<italic>nat</italic>}, <italic>θ</italic>_{<italic>TS</italic>}, or <italic>θ</italic>_{<italic>unf</italic>}, which identifies the native basin, the assumed transition state maximum, and the unfolded minimum, respectively, in the function <italic>G(θ,T</italic>_{<italic>m</italic>}<italic>)</italic>. Selecting <italic>θ</italic>_{<italic>nat </italic>}allows us to compare mechanical properties that are averaged over a native state ensemble. Note, in the results presented herein, the native state ensemble is extended slightly to include a range over <italic>θ </italic>near <italic>θ</italic>_{<italic>nat </italic>}(equation not shown), which increases statistics and accounts for all variance within the entire free energy basin. Nevertheless, expanding the range of <italic>θ </italic>does not lead to any qualitative differences within QSFR.</p>", "<p>The value of <italic>θ</italic>_{<italic>nat </italic>}gives the average number of independent degrees of freedom per residue that is available to the protein in its native state. We also probe where these degrees of freedom are localized, and quantify the flexible and rigid parts of the protein. To do so, we define a flexibility index [##REF##11391777##12##,##REF##15542549##19##,##REF##15498582##21##] to describe the extent of flexibility/rigidity along the protein backbone. The flexibility index of residue <italic>i</italic>, <italic>f</italic>_{<italic>i</italic>}, is defined by Eq. 6 with X → <italic>f</italic>_{<italic>i </italic>}and . The quantities and are functions of the macrostate, <italic>(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>) </italic>and <italic>(N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}<italic>)</italic>, where they are the average densities of independent degrees of freedom and redundant constraints, respectively, at residue <italic>i</italic>. These quantities have the same definition as used in FIRST [##REF##11391777##12##], except here they are averaged uniformly over a sub-ensemble of constraint topologies that have <italic>N</italic>_{<italic>nt </italic>}native torsion constraints, and <italic>N</italic>_{<italic>hb </italic>}intramolecular H-bonds in the network. After the second averaging procedure, we obtain the flexibility index for the native sate of the protein as a function of temperature. The range of <italic>f</italic>_{<italic>i</italic>}<italic>(T) </italic>over all residues in a protein is always less than 1, and almost always greater than -1. A negative flexibility index indicates that the region is over-constrained, meaning it is rigid with more constraints than necessary to be rigid. A more negative value implies a greater density of redundant constraints. A positive flexibility index indicates that the region is flexible, meaning it is capable of motion involving a certain number of degrees of freedom. A more positive value implies a greater density of degrees of freedom to describe the motion. A flexibility index near 0 indicates a marginally rigid region having neither excess constraints nor degrees of freedom. However, due to the ensemble averaging, these marginal regions are often split between being slightly over-constrained and slightly flexible. Marginal regions are particularly interesting, as they have a high degree of susceptibility to perturbations in the network.</p>", "<p>While <italic>f</italic>_{<italic>i </italic>}is a measure of flexibility along the backbone, cooperativity correlations are defined to describe intramolecular couplings between sites within the protein. These couplings are both thermodynamic and mechanical in nature. Cooperativity correlation between two residues <italic>i </italic>and <italic>j</italic>, <italic>C</italic>_{<italic>ij</italic>}, describes both rigid and flexible couplings. Flexibility correlation quantifies the extent that two residues, across the thermodynamic ensemble, are simultaneously flexible and connected by a path in which flexibility can propagate. Conversely, rigidity correlation quantifies the extent that two residues are simultaneously within the same rigid cluster. Zero cooperatively correlation indicates that two sites are neither flexibly nor rigidly coupled. Frequently, the most flexible portions of a protein (e.g., the N- or C-termini) have no correlation to the rest of the protein due to the fact that the mechanical aspect of the definition (meaning that two sites must be connected by a path in which flexibility can propagate) is not satisfied. The correlation matrices are functions of temperature and global flexibility too. Since mechanical response will be temperature dependent, we need to address at what temperature should be used when comparing mechanical response of different proteins, but are part of the same family. In general, we work at <italic>T</italic>_{<italic>m </italic>}because, first, there is weak temperature dependence in any of the native ensemble averaged mechanical properties. Second, and more importantly, the <italic>T</italic>_{<italic>m </italic>}of each considered protein establishes a corresponding location along the folding transition, which is in contrast to a set absolute temperature, of say 320 K. Due to weak temperature dependence, the exact value of <italic>T</italic>_{<italic>m </italic>}is not that important as far as the mechanical properties are of concern. Therefore, the error in predicting <italic>T</italic>_{<italic>m </italic>}is not of much concern for making QSFR comparisons across similar proteins.</p>", "<title>Protein structure preparation</title>", "<p>The employed H-bond potential [##REF##9194194##26##] requires explicit hydrogen atoms. Most methods of adding hydrogen atoms assume pH = 7 and do not account for changes in residue-specific titration profiles based on electrostatic interactions or solvent conditions. Here, the pH of the <italic>C</italic>_{<italic>p </italic>}data used to parameterize the mDCM is 8.3. Consequently, we use the <italic>pK</italic>_{<italic>a </italic>}calculation implemented within the University of Houston Brownian Dynamics suite of programs [##UREF##3##28##] to calculate whether or not a titratable residue should be protonated or not at a specific pH (see [##UREF##4##29##,##REF##14581189##30##] for details of the calculation). Hydrogen atoms are (kept, removed) if their probability for protonation is (greater, lesser) than 50 percent.</p>", "<title>Abbreviations</title>", "<p>ABC – ATP-Binding Cassette; bPBP – bacterial Periplasmic Binding Proteins; ENM – Elastic Network Model; MD – Molecular Dynamics; DCM – Distance Constraint Model; QSFR – Quantitative Stability/Flexibility Relationships; mDCM – minimal Distance Constraint Model; DSC – Differential Scanning Calorimetry; HBP – Histidine Binding Protein; LAOBP – Lysine/Arginine/Ornithine Binding Protein; GBP – Glutamine Binding Protein; PhBP – Phosphate Binding Protein; RMSD – Root Mean Square Deviation</p>", "<title>Authors' contributions</title>", "<p>All authors contributed to execution of the presented work. DRL and DJJ planned and oversaw the executed research. DRL, SD, and DJJ wrote the paper. All authors read and approved the final manuscript.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Dr. Jim Mottonen for proofreading this paper and for stimulating discussions of the results. Key to the DCM is the use of graph-rigidity algorithms. This algorithm is claimed in U.S. Patent Number 6,014,449, which has been assigned to the Board of Trustees Michigan State University. Used with permission. This work is supported by NIH grant R01 GM073082 to DJJ and DRL.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>(a) </bold>Cartoon of the free energy landscape in two-dimensional constraint space. Each point on the two-dimensional grid defines a macrostate, <italic>(N_{<italic>nt</italic>}, N_{<italic>hb</italic>})</italic>, where the free energy, <italic>G(N_{<italic>nt</italic>}, N_{<italic>hb</italic>})</italic>, is calculated. The green shading is meant to describe the native (lower-right) and unfolded (upper-left) basins within the free energy landscape. (Notice that the axes are decreasing from bottom to top and left to right.) At times it is convenient to express the free energy as a function of a one-dimensional flexibility order parameter, <italic>θ(N_{<italic>nt</italic>}, N_{<italic>hb</italic>})</italic>. Grey dashed lines represent (approximate) fronts of constant global flexibility due to tradeoff between two constraints types. The red line denotes the shortest path crossing a single saddle from the unfolded to folded basins. <bold>(b) </bold>An example one-dimensional free energy landscape highlights the straddling barrier that must be crossed as the protein transitions between folded and unfolded.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>(a) </bold>Phylogenetic tree of representatives from the 29 protein classes within the bPBP family. The four proteins investigated herein are highlighted (Blue = LAOBP, Green = HBP, Purple = GBP, and Red = PhBP). <bold>(b) </bold>Structure superposition of the four binding proteins (color-coding the same). PhBP has been removed in <bold>(c) </bold>to highlight the conservation within the three amino acid binding proteins.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>(a) </bold>Structure superposition of the ligated (blue) and apo (red) LAOBP conformations as calculated by the combinatorial extension algorithm. The overall RMSD between the two conformations is 3.5 Å. Structure superposition of the <bold>(b) </bold>large and <bold>(c) </bold>small domains is also provided. The domain specific RMSDs are 0.6 Å and 0.4 Å, respectively.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p>H-bond contact maps of ligated LAOBP (green squares) and HBP (blue x's). Each point on the plot indicates that a H-bond is present between the corresponding residues. Most differences occur between non-secondary structure H-bonds, which primarily involve sidechain interactions and within coil regions.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>(a) </bold>Heat capacity curves and <bold>(b) </bold>free energy landscapes of the four ligated bPBP homologs. In <bold>(c) </bold>and <bold>(d) </bold>the heat capacity curves and free energy landscapes, respectively, of the apo structures (dashed lines) are compared to the ligated counterparts. Color coding is conserved throughout.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p>H-bond contact maps of ligated LAOBP (green squares) and apo LAOBP (green x's). Most differences occur between non-secondary structure H-bonds, which primarily involve sidechain interactions and within coil regions.</p></caption></fig>", "<fig position=\"float\" id=\"F7\"><label>Figure 7</label><caption><p><bold>(a) </bold>Maximal heat capacity vs. total H-bond energy and <bold>(b) </bold>maximal heat capacity vs. free energy barrier height. In both cases, the regression lines, whose equations are provided, and correlation coefficients are computed on the four ligated bPBP homologs (colored blue). The values for the two apo-structures (colored red) are superimposed onto the plot to demonstrate they fit the observed trend.</p></caption></fig>", "<fig position=\"float\" id=\"F8\"><label>Figure 8</label><caption><p>Structural alignment of the four ligated bPBP homologs color-coded by flexibility index. Flexible regions (positive flexibility index) are colored red, whereas rigid regions (negative flexibility index) are colored blue.</p></caption></fig>", "<fig position=\"float\" id=\"F9\"><label>Figure 9</label><caption><p><bold>(a) </bold>Flexibility index of the ligated (blue) and apo (blue) LAOBP. In <bold>(b) </bold>and <bold>(c)</bold>, the flexibility index of the apo and ligated structures, respectively, is mapped to structure. The most pronounced changes occur within the ligand-binding pocket at the hinge between the two domains.</p></caption></fig>", "<fig position=\"float\" id=\"F10\"><label>Figure 10</label><caption><p>Cooperativity correlation plots of the ligated <bold>(a) </bold>LAOBP, <bold>(b) </bold>HBP, <bold>(c) </bold>GBP, and <bold>(d) </bold>PhBP structures. Red indicates flexibly correlated regions, blue indicates rigidly correlated regions, and white indicates regions of no correlation.</p></caption></fig>", "<fig position=\"float\" id=\"F11\"><label>Figure 11</label><caption><p>Cooperativity correlation plots of the <bold>(a) </bold>ligated and <bold>(b) </bold>apo LAOBP structures. Note that the color-coding is the same as in Fig. 9, and that panel <bold>(a) </bold>is exactly the same as Fig. 9a.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Parameterization of the mDCM.¹</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\">Parameter</td><td align=\"center\">Value²</td><td align=\"center\">Treatment</td><td align=\"left\">Description</td></tr></thead><tbody><tr><td align=\"center\"><italic>U_{<italic>hb</italic>}</italic></td><td align=\"center\">Context dependent</td><td align=\"center\">Empirical potential [##REF##9194194##26##]</td><td align=\"left\">Intramolecular H-bond energy</td></tr><tr><td align=\"center\"><italic>γ</italic>_{<italic>max</italic>}</td><td align=\"center\">1.99</td><td align=\"center\">Constant</td><td align=\"left\">Linearly relates H-bond pure entropy to its energy</td></tr><tr><td align=\"center\"><italic>u</italic></td><td align=\"center\">-1.91</td><td align=\"center\">Fitting</td><td align=\"left\">H-bond to solvent energy upon breaking of intramolecular H-bond</td></tr><tr><td align=\"center\"><italic>v</italic></td><td align=\"center\">-0.64</td><td align=\"center\">Fitting</td><td align=\"left\">Native torsional constraint energy</td></tr><tr><td align=\"center\"><italic>δ_{<italic>nat</italic>}</italic></td><td align=\"center\">1.42</td><td align=\"center\">Fitting</td><td align=\"left\">Native torsional constraint pure entropy</td></tr><tr><td align=\"center\"><italic>δ_{<italic>dis</italic>}</italic></td><td align=\"center\">2.56</td><td align=\"center\">Constant</td><td align=\"left\">Disordered torsional constraint pure entropy</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Descriptions of the structures used in this report.¹</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\">LAOBP</td><td align=\"center\">HBP</td><td align=\"center\" colspan=\"2\">GBP</td><td align=\"center\">PhBP</td></tr></thead><tbody><tr><td align=\"center\">Source</td><td align=\"center\" colspan=\"2\"><italic>S. typhimurium</italic></td><td align=\"center\"><italic>E. coli</italic></td><td align=\"center\"><italic>E. coli</italic></td><td/><td align=\"center\"><italic>E. coli</italic></td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"center\">PDB ID</td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"1LST\">1LST</ext-link></td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"2LAO\">2LAO</ext-link>*</td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"1HSL\">1HSL</ext-link></td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"1WDN\">1WDN</ext-link></td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"1GGG\">1GGG</ext-link>*</td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"1IXH\">1IXH</ext-link></td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"center\">Bound ligand²</td><td align=\"center\">Lysine</td><td align=\"center\">n/a</td><td align=\"center\">histidine</td><td align=\"center\">glutamine</td><td align=\"center\">n/a</td><td align=\"center\">phosphate</td></tr><tr><td align=\"center\">Number of residues</td><td align=\"center\">238</td><td align=\"center\">238</td><td align=\"center\">238</td><td align=\"center\">223</td><td align=\"center\">220</td><td align=\"center\">321</td></tr><tr><td align=\"center\">Large domain³</td><td align=\"center\" colspan=\"2\">1–90, 191–240</td><td align=\"center\">1–90,191–238</td><td align=\"center\" colspan=\"2\">4–88, 182–226</td><td align=\"center\">1–75,255–321</td></tr><tr><td align=\"center\">Small domain</td><td align=\"center\" colspan=\"2\">91–190</td><td align=\"center\">91–190</td><td align=\"center\" colspan=\"2\">89–181</td><td align=\"center\">76–254</td></tr><tr><td align=\"center\">Resolution (Å)</td><td align=\"center\">1.80</td><td align=\"center\">1.90</td><td align=\"center\">1.89</td><td align=\"center\">1.94</td><td align=\"center\">2.30</td><td align=\"center\">0.98</td></tr><tr><td align=\"center\">Number of H-bonds</td><td align=\"center\">352</td><td align=\"center\">357</td><td align=\"center\">327</td><td align=\"center\">327</td><td align=\"center\">293</td><td align=\"center\">504</td></tr><tr><td align=\"center\">Number of H-bonds to substrate</td><td align=\"center\">5</td><td align=\"center\">n/a</td><td align=\"center\">6</td><td align=\"center\">9</td><td align=\"center\">n/a</td><td align=\"center\">7</td></tr><tr><td align=\"center\">Total H-bond energy (Kcal/mol)⁴</td><td align=\"center\">-942.11</td><td align=\"center\">-875.20</td><td align=\"center\">-714.00</td><td align=\"center\">-757.28</td><td align=\"center\">-633.40</td><td align=\"center\">-1274.28</td></tr><tr><td align=\"center\">Avg. H-bond energy (kcal/mol)</td><td align=\"center\">-2.68</td><td align=\"center\">-2.45</td><td align=\"center\">-2.18</td><td align=\"center\">-2.32</td><td align=\"center\">-2.16</td><td align=\"center\">-2.53</td></tr><tr><td align=\"center\">Std. dev. H-bond energy (kcal/mol)</td><td align=\"center\">2.34</td><td align=\"center\">2.27</td><td align=\"center\">2.12</td><td align=\"center\">2.22</td><td align=\"center\">2.05</td><td align=\"center\">2.30</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>All to all pairwise comparisons of the four ligated bPBP structures.¹</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">LAOBP</td><td align=\"center\">HBP</td><td align=\"center\">GBP</td><td align=\"center\">PhBP</td></tr></thead><tbody><tr><td align=\"left\">LAOBP</td><td align=\"center\">---</td><td align=\"center\">1.1 Å</td><td align=\"center\">1.5 Å</td><td align=\"center\">4.2 Å</td></tr><tr><td align=\"left\">HBP</td><td align=\"center\">69.9%</td><td align=\"center\">---</td><td align=\"center\">1.9 Å</td><td align=\"center\">4.4 Å</td></tr><tr><td align=\"left\">GBP</td><td align=\"center\">30.3%</td><td align=\"center\">29.3%</td><td align=\"center\">---</td><td align=\"center\">4.1 Å</td></tr><tr><td align=\"left\">PhBP</td><td align=\"center\">16.6%</td><td align=\"center\">17.5%</td><td align=\"center\">17.1%</td><td align=\"center\">---</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Thermodynamic descriptions of the bPBPs.¹</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\">LAOBP</td><td align=\"center\">HBP</td><td align=\"center\" colspan=\"2\">GBP</td><td align=\"center\">PhBP</td></tr></thead><tbody><tr><td align=\"center\">PDB ID</td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"1LST\">1LST</ext-link></td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"2LAO\">2LAO</ext-link>*</td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"1HSL\">1HSL</ext-link></td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"1WDN\">1WDN</ext-link></td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"1GGG\">1GGG</ext-link>*</td><td align=\"center\"><ext-link ext-link-type=\"pdb\" xlink:href=\"1IXH\">1IXH</ext-link></td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"center\"><italic>T</italic>_{<italic>m </italic>}(K)</td><td align=\"center\">385</td><td align=\"center\">376</td><td align=\"center\">336</td><td align=\"center\">336</td><td align=\"center\">297</td><td align=\"center\">386</td></tr><tr><td align=\"center\"><italic>C</italic>_{<italic>p</italic>}^{<italic>max </italic>}(kcal/mol·K)</td><td align=\"center\">113.6</td><td align=\"center\">70.8</td><td align=\"center\">38.5</td><td align=\"center\">26.4</td><td align=\"center\">8.7</td><td align=\"center\">174.1</td></tr><tr><td align=\"center\">Barrier (kcal/mol)</td><td align=\"center\">13.6</td><td align=\"center\">8.4</td><td align=\"center\">4.3</td><td align=\"center\">2.6</td><td align=\"center\">0.0</td><td align=\"center\">13.1</td></tr><tr><td align=\"center\">θ_nat</td><td align=\"center\">0.8</td><td align=\"center\">0.9</td><td align=\"center\">1.0</td><td align=\"center\">0.8</td><td align=\"center\">n/a²</td><td align=\"center\">0.9</td></tr><tr><td align=\"center\">θ_TS</td><td align=\"center\">1.2</td><td align=\"center\">1.3</td><td align=\"center\">1.3</td><td align=\"center\">1.1</td><td align=\"center\">n/a</td><td align=\"center\">1.2</td></tr><tr><td align=\"center\">θ_unf</td><td align=\"center\">2.1</td><td align=\"center\">1.9</td><td align=\"center\">1.8</td><td align=\"center\">1.6</td><td align=\"center\">n/a</td><td align=\"center\">2.0</td></tr></tbody></table></table-wrap>" ]

[ "<disp-formula id=\"bmcM1\"><label>(1)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1752-153X-2-17-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>G</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>F</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munder><mml:mo>∑</mml:mo><mml:mi>t</mml:mi></mml:munder><mml:mrow><mml:msub><mml:mi>H</mml:mi><mml:mi>t</mml:mi></mml:msub><mml:msub><mml:mi>N</mml:mi><mml:mi>t</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mi>T</mml:mi><mml:mstyle displaystyle=\"true\"><mml:munder><mml:mo>∑</mml:mo><mml:mi>t</mml:mi></mml:munder><mml:mrow><mml:msub><mml:mi>S</mml:mi><mml:mi>t</mml:mi></mml:msub><mml:msub><mml:mi>I</mml:mi><mml:mi>t</mml:mi></mml:msub></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM2\"><label>(2)</label><italic>G</italic>(<italic>N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}) = <italic>U</italic>_{<italic>hb</italic>}(<italic>N</italic>_{<italic>hb</italic>}) - <italic>uN</italic>_{<italic>hb </italic>}+ <italic>vN</italic>_{<italic>nt </italic>}- <italic>T </italic>[<italic>S</italic>_{<italic>c</italic>}(<italic>N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>}) + <italic>S</italic>_{<italic>m</italic>}(<italic>N</italic>_{<italic>nt</italic>}, <italic>N</italic>_{<italic>hb</italic>})]</disp-formula>", "<disp-formula id=\"bmcM3\"><label>(3)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M2\" name=\"1752-153X-2-17-i2\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>C</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>d</mml:mi><mml:mi>H</mml:mi></mml:mrow><mml:mrow><mml:mi>d</mml:mi><mml:mi>T</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mrow><mml:mo>〈</mml:mo><mml:mrow><mml:mi>Δ</mml:mi><mml:msup><mml:mi>H</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow><mml:mo>〉</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:mi>R</mml:mi><mml:msup><mml:mi>T</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM4\"><label>(4)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M3\" name=\"1752-153X-2-17-i3\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mrow><mml:mo>〈</mml:mo><mml:mrow><mml:mi>Δ</mml:mi><mml:msup><mml:mi>H</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow><mml:mo>〉</mml:mo></mml:mrow><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munder><mml:mo>∑</mml:mo><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>h</mml:mi><mml:mi>b</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:munder><mml:mrow><mml:msubsup><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mi>H</mml:mi><mml:mo>−</mml:mo><mml:mrow><mml:mo>〈</mml:mo><mml:mi>H</mml:mi><mml:mo>〉</mml:mo></mml:mrow></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>h</mml:mi><mml:mi>b</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:mstyle><mml:mtext> </mml:mtext><mml:msup><mml:mi>e</mml:mi><mml:mrow><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:mi>G</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>h</mml:mi><mml:mi>b</mml:mi></mml:mrow></mml:msub><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>R</mml:mi><mml:mi>T</mml:mi></mml:mrow></mml:mfrac></mml:mrow></mml:msup></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM5\"><label>(5)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M4\" name=\"1752-153X-2-17-i4\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>Z</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>θ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munder><mml:mo>∑</mml:mo><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>h</mml:mi><mml:mi>b</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:munder><mml:mrow><mml:mi>B</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mi>θ</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>h</mml:mi><mml:mi>b</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:mstyle><mml:mtext> </mml:mtext><mml:msup><mml:mi>e</mml:mi><mml:mrow><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:mi>G</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>h</mml:mi><mml:mi>b</mml:mi></mml:mrow></mml:msub><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>R</mml:mi><mml:mi>T</mml:mi></mml:mrow></mml:mfrac></mml:mrow></mml:msup></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM6\"><label>(6)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M5\" name=\"1752-153X-2-17-i5\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>X</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>θ</mml:mi><mml:mo>,</mml:mo><mml:mi>T</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munder><mml:mo>∑</mml:mo><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>h</mml:mi><mml:mi>b</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:munder><mml:mrow><mml:mi>x</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>h</mml:mi><mml:mi>b</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:mstyle><mml:mtext> </mml:mtext><mml:mfrac><mml:mrow><mml:mi>B</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>θ</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>h</mml:mi><mml:mi>b</mml:mi></mml:mrow></mml:msub><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>⋅</mml:mo><mml:msup><mml:mi>e</mml:mi><mml:mrow><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:mi>G</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mrow><mml:mi>h</mml:mi><mml:mi>b</mml:mi></mml:mrow></mml:msub><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>R</mml:mi><mml:mi>T</mml:mi></mml:mrow></mml:mfrac></mml:mrow></mml:msup></mml:mrow><mml:mrow><mml:mi>Z</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>θ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M6\" name=\"1752-153X-2-17-i6\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>x</mml:mi><mml:mo>→</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:msubsup><mml:mi>ρ</mml:mi><mml:mi>i</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>d</mml:mi><mml:mi>f</mml:mi></mml:mrow></mml:msubsup><mml:mo>−</mml:mo><mml:msubsup><mml:mi>ρ</mml:mi><mml:mi>i</mml:mi><mml:mrow><mml:mi>r</mml:mi><mml:mi>d</mml:mi><mml:mi>c</mml:mi></mml:mrow></mml:msubsup><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M7\" name=\"1752-153X-2-17-i7\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>ρ</mml:mi><mml:mi>i</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>d</mml:mi><mml:mi>f</mml:mi></mml:mrow></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M8\" name=\"1752-153X-2-17-i8\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>ρ</mml:mi><mml:mi>i</mml:mi><mml:mrow><mml:mi>r</mml:mi><mml:mi>d</mml:mi><mml:mi>c</mml:mi></mml:mrow></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M9\" name=\"1752-153X-2-17-i7\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>ρ</mml:mi><mml:mi>i</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>d</mml:mi><mml:mi>f</mml:mi></mml:mrow></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M10\" name=\"1752-153X-2-17-i8\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>ρ</mml:mi><mml:mi>i</mml:mi><mml:mrow><mml:mi>r</mml:mi><mml:mi>d</mml:mi><mml:mi>c</mml:mi></mml:mrow></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M11\" name=\"1752-153X-2-17-i7\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>ρ</mml:mi><mml:mi>i</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>d</mml:mi><mml:mi>f</mml:mi></mml:mrow></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>Supplementary Figure 1. Workflow diagram describing normal usage of the mDCM.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>¹Note that the disordered torsional constraint energy is the reference energy and is equal to 0.00. The mDCM ignores potential entropic contributions on the formation of H-bonds to solvent. ²The values of the three fitting parameters were determined by fitting HBP to experimental <italic>C</italic>_{<italic>p </italic>}curves [##REF##10866829##27##].</p></table-wrap-foot>", "<table-wrap-foot><p>¹All four bPBP homologs have been crystallized in the ligated conformation. Additionally, two bPBP homologs (marked with asterisks) have been crystallized in the open apo conformation. ²Non-biologically relevant co-crystallized solutes (i.e., metal ions) are ignored. ³Domain boundaries are identified from the ligated structures using PDP [##REF##12584135##40##]. ⁴Total H-bond energy is simply the sum of the energies for each H-bond observed within the structure.</p></table-wrap-foot>", "<table-wrap-foot><p>¹Numbers above the diagonal represent pairwise RMSD (Å) between each structure (calculated using the combinatorial extension algorithm [##REF##9796821##38##]), whereas numbers below the diagonal represent percent residue identity, as calculated from each global pairwise alignment.</p></table-wrap-foot>", "<table-wrap-foot><p>¹All four bPBP homologs have been crystallized in the ligated conformation. Additionally, two bPBP homologs (marked with asterisks) have been crystallized in the open apo conformation. ²The apo GBP structure is predicted to have a continuous transition between folded and unfolded.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1752-153X-2-17-1\"/>", "<graphic xlink:href=\"1752-153X-2-17-2\"/>", "<graphic xlink:href=\"1752-153X-2-17-3\"/>", "<graphic xlink:href=\"1752-153X-2-17-4\"/>", "<graphic xlink:href=\"1752-153X-2-17-5\"/>", "<graphic xlink:href=\"1752-153X-2-17-6\"/>", "<graphic xlink:href=\"1752-153X-2-17-7\"/>", "<graphic xlink:href=\"1752-153X-2-17-8\"/>", "<graphic xlink:href=\"1752-153X-2-17-9\"/>", "<graphic xlink:href=\"1752-153X-2-17-10\"/>", "<graphic xlink:href=\"1752-153X-2-17-11\"/>" ]

[ "<media xlink:href=\"1752-153X-2-17-S1.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Complementary therapies, such as massage, are increasingly used by health professionals to manage a wide range of symptoms, including pain [##REF##9439258##1##]. A literature search revealed a paucity of well-designed randomised controlled trials to determine the effectiveness of massage for chronic pain. This study aims to determine whether a single session of massage reduces short term pain and anxiety in patients with chronic pain.</p>", "<p>The use of complementary therapies is steadily growing, both in Europe and in the US [##REF##10812758##2##, ####REF##9820257##3##, ##UREF##0##4####0##4##]. Health care practitioners are increasingly incorporating such therapies within their care [##UREF##1##5##]. This increased used is despite lack of evidence of effectiveness [##UREF##2##6##]. Health professionals' desire to deliver best practice is coupled with government initiatives to ensure an evidence based approach to care. For example, in the UK the National Institute for Clinical Excellence and NHS Quality Improvement Scotland have been set up; in the US the Agency for Health Care Research &amp; Quality, and in Australia the National Institute for Clinical Studies all to try and ensure the health care delivered is based on what is known to be effective [##UREF##3##7##, ####UREF##4##8##, ##UREF##5##9##, ##UREF##6##10####6##10##].</p>", "<p>The holistic nature of nursing means massage is an attractive intervention for nurses to be able to offer patients. In addition to any direct therapeutic benefit, it allows nurses time with patients. It may be particularly relevant in settings where patients need immediate relief. Many nurses have undertaken courses in massage, however there is little robust evidence of its effectiveness. In today's evidence driven health care environment, if nurses are to justify the resources involved in training and delivery, they need evidence that massage is an effective treatment.</p>", "<p>A literature search of Medline, EmBase, Cinahl and the Cochrane Controlled Trials Register and Cochrane Database of Systematic Reviews was undertaken, using free text and MeSH terms. Studies were sought which were randomised controlled trials, of manual massage therapies with or without oils, and which compared massage to at least one other treatment group (including usual care) in patients with chronic non malignant pain. Studies were excluded if they used non-manual massage or where massage was administered as part of a combination of therapies or was used for acute pain. Twenty eight studies were identified, of which eight met the criteria for inclusion.</p>", "<p>After the RCT described in this paper was completed, a systematic review of massage for low back pain was published [##REF##12076429##11##]. Outcomes considered in this review were pain, return to work, subjective change in symptoms and functional status. This superseded the literature review carried out prior to the conduct of the RCT reported in this paper. The systematic review concluded that massage might be beneficial for subacute and chronic non-specific low back pain, but that more studies were needed to confirm these conclusions [##REF##12076429##11##]. Two of the RCTs included in review [##REF##12076429##11##] found patients had less anxiety after massage [##REF##10906914##12##,##REF##11264915##13##]. Studies of chronic pain other than low back pain have found less anxiety and less pain [##UREF##7##14##, ####REF##9622469##15##, ##REF##9383925##16####9383925##16##].</p>", "<p>In summary, despite the widespread and increasing use of massage for chronic pain, evidence of effectiveness is lacking, studies tend to be small and properly controlled studies are needed. An RCT to assess the effectiveness of massage for chronic pain was thus designed to add to the knowledge base in this area.</p>" ]

[ "<title>Methods</title>", "<title>Aim</title>", "<p>The aim of the study is to determine the effectiveness of a single session of nurse-administered massage for the short term relief of chronic non-malignant pain and anxiety.</p>", "<title>Design</title>", "<p>A randomised controlled trial design was utilised, in which the patients were assigned to two groups by simple block randomisation, using blocks of 10. Randomisation was carried out by use of a computerised random number generator and provided from a central office unconnected with the study. After obtaining informed consent from the patient, the allocation was obtained by telephone from the central office, thus allocation was concealed.</p>", "<title>Sample and Setting</title>", "<p>Patients aged 18 and over and attending a regional pain relief unit in England as in or out patients, who had experienced pain for three months or longer, whose pain was described as moderate or severe on the four point verbal rating pain intensity scale were eligible for the study. Patients were excluded if they did not speak English, did not consent or if they had taken any analgesics in the two hours prior to treatment. Outpatients were being followed up for the management of their chronic pain, and inpatients were admitted to the pain unit where patients stayed for one or more days in order to investigated ways of controlling their pain. The participants were a consecutive series of eligible patients and data collection took place between 1998 and 2000.</p>", "<title>Interventions</title>", "<p>The experimental group received a 15 minute manual massage of their back, neck and shoulders using sweet almond oil. The control group received a 15 minute visit to talk about pain clinic treatment. The treatment for both the experimental and control groups was carried out by two registered nurses qualified in massage (ITEC diploma). Each therapist (JM, KC) performed a sequence of the same massage techniques with the patients in the massage treatment group. This involved effleurage, petrissage and kneading techniques. For patients in the control group the therapist visited the patient for 15 minutes and encouraged them to talk about their pain and pain treatment. This is representative of usual care in this pain clinic, where health care staff talk to patients about their pain and its management, and no massage was performed for the control group. Patients in both groups were asked to try and complete at least one full hour after the intervention before requesting analgesics, but could request analgesics at any time after the intervention. If they received an analgesic, they were then excluded from any further follow-up assessments.</p>", "<title>Outcomes</title>", "<p>An independent nurse observer (DC), blind to the patients' treatment group, carried out assessments of the patient. These were carried out immediately before treatment (baseline), immediately after treatment (post treatment) and at 1, 2, 3 and 4 hours after treatment. The observer also recorded whether or not the patient had broken the blinding by inadvertently revealing their group allocation (for example, by referring to \"my massage\"). Once blinding was broken, it was considered unblinded for all subsequent assessments for that patient.</p>", "<p>Pain intensity was assessed using a 0–100 mm visual analogue scale (VAS), the four point verbal rating pain intensity scale with descriptors of none = 0, mild = 1, moderate = 2 and severe = 3, both have been shown to be reliable, valid and appropriate for clinical use [##REF##16000093##17##], and The McGill Pain Questionnaire Pain Rating Index [##REF##1235985##18##]. Pain Relief was assessed using five point verbal rating scale with descriptors of none = 0, slight = 1, moderate = 2, good = 3 and complete relief = 4. Anxiety was assessed with the six item Spielberger short form State-Trait Anxiety Inventory, [##REF##1393159##19##]. At the end of the study period patients made an overall rating of the study treatment using a five point scale (poor = 0, fair = 1, good = 2, very good = 3 or excellent = 4). They also rated whether they had achieved 50% pain relief [##REF##9483161##20##].</p>", "<title>Sample Size</title>", "<p>The sample size was calculated for comparing mean difference in pain score and for the study to have a power of 90% [##UREF##8##21##]. The results of a study of pain relief through relaxation in elderly hip fracture patients suggests that when using a pain scale rating pain from 1 to 10 that the standard deviation for patients with chronic pain on this scale is 2.5 [##REF##6374586##22##]. On this scale a mean difference of 1.5 in pain scores between the treatment groups could be considered clinically worthwhile. Since in the current study a 100 mm VAS pain scale will be used a standard deviation of 25 mm and a mean difference of 15 mm are consistent with the earlier study [##REF##6374586##22##]. With type I error set at 0.05, then the study would require 60 patients per group. If a difference as large as 20 mm between the pain scores was attained this could be demonstrated with 34 patients per group. In the current study it was decided, based on the difference found in the previous study of elderly hip fracture patients, recruit 60 to each group: a total of 120 patients.</p>", "<title>Ethical considerations</title>", "<p>The study received ethical approval from the Local Research Ethics Committee (Study Code NAPREC 97.053). All participants recruited to the study had received a verbal explanation of the study and were provided with a written information sheet. All potential participants had at least 24 hours between receiving information and deciding whether or not they wished to take part in the study. All participants gave written consent.</p>", "<title>Data analysis</title>", "<p>In the analysis baseline pain and anxiety scores are compared for the two treatment groups using t-tests. To investigate the treatment effect, for each patient their pain scores after treatment are compared to their baseline pain score (pre-treatment). The mean difference in change in pain score for the two treatment groups is compared using a 2 sample t-test. Similarly changes in anxiety between baseline and post treatment are compared for the two groups sample t-test. Box plots are used to show graphically the differences between groups. The box includes the 25^th, 50^thand 75^thpercentiles; the whiskers indicate the range of the data, with asterisks indicating outliers. Statistical significance was set at the 5% level. The data were analysed using SPSS version 12 and MINITAB version 13.</p>" ]

[ "<title>Results</title>", "<p>In total 103 patients participated in the study of whom 52 were randomly allocated to the massage group and 51 to the control group. Two patients allocated to the massage group withdrew at baseline, one because they received morphine just prior to starting the study and the other because they were unable to understand the assessment scales, thus the analysis reports results for the 101 patients who completed the baseline assessments. The patients were aged between 21 and 81 years, with mean age 53.4 years (SD 13.3 years) and 58.4% were female. The mean time that the patients had experienced chronic pain was 10.4 years (SD 8.94 years). 87% of the patients had a diagnosis which included back pain and 58% had more than one pain site. At baseline, 60.4% had moderate and 39.6% had severe pain as recorded on the verbal rating scale.</p>", "<p>Figure ##FIG##0##1## shows the number of patients still taking part in the study at each assessment point. After the first hour the drop out from both groups, but particularly from the control group was rapid. By the 2 hour post-treatment assessment only 23% of patients in the control group remained in the study and by the 3 hour assessment only 10% of the control group were available for assessment. In comparison for the massage group 72% of patients remained in the study at 2 hours post-treatment, 54% at 3 hours and 36% at 4 hours post treatment. Because of the very high withdrawal rate at 2, 3 and 4 hours after treatment, the results from these time periods are impossible to interpret and are therefore not presented.</p>", "<p>The assessment of the patients at all time points was by an observer who was blind to the treatment that the patient received (DC). Occasionally the patient would break the blinding by comments that they made to the observer. For 89 of the 100 patients in which a record of blinding was made, the observer remained blind throughout the assessment, so blinding was maintained in 89% of patients.</p>", "<p>The demographic characteristics and baseline scores for the massage and control groups are compared in Table ##TAB##0##1##. Comparisons between massage and control groups were tested using chi-squared or two sample t-tests as appropriate. There are no significant differences in the age, gender mix or duration of pain for the two treatment groups.</p>", "<p>The pain verbal rating scale scores are \"moderate\" or \"severe\" for all patients, with 40% of the massage group and 39% of the control group recording severe pain at baseline. The similarity of the distribution of pain scores between the two groups is shown in the boxplot in Figure ##FIG##1##2##.</p>", "<p>Comparisons post-treatment of pain VAS scores are shown in Figure ##FIG##2##3## and Table ##TAB##1##2##. Figure ##FIG##2##3## illustrates that the 95% confidence intervals for mean pain VAS for the massage and control groups overlap at baseline, the means are not significantly different. However, the confidence intervals are clearly distinct both immediately post treatment and at 1 hour post treatment, consistent with the massage group having significantly lower mean pain VAS. For the control group mean pain VAS for those who continue beyond 1 hour (the minority) is 36.4, but mean pain VAS for those in the control group who withdraw at 1 hour (the majority) is 74.7, a significantly higher mean (p &lt; 0.001).</p>", "<p>Table ##TAB##1##2## compares the baseline score to that immediately post treatment, the massage group experience a statistically significant (p &lt; 0.001 using paired t-test) mean reduction in pain of 16.7 mm (SD 21.2). For the control group the mean change in pain score is -0.04 mm (SD 16.0) which is not significant (p = 0.985 using paired t-test). For those in the massage group there is a significantly greater reduction in pain score both immediately post treatment and at 1 hour post treatment than in the control group. The 95% confidence interval for the difference in mean pain reduction at one hour post treatment between the massage and control groups is 5.47 mm to 24.70 mm, that is the massage group can expect on average to benefit by between 5.47 mm and 24.70 mm greater reduction than the control group up to one hour post treatment.</p>", "<p>Figure ##FIG##3##4## illustrates the reduction in pain VAS score for the massage group and the negligible change in pain VAS score for the majority of members of the control group. For both groups the median is central in the box of the boxplot, thus supporting the assumption that the data is normally distributed and that a t-test can be used to compare these changes.</p>", "<p>The results from the four point verbal rating pain, the four point pain distress scale and the McGill Pain Questionnaire produce the same findings as the pain VAS scale. That is, statistically significant differences between massage and control groups immediately and one hour post treatment.</p>", "<p>An example of these similarities is shown in comparisons of McGill Pain Questionnaire scores from baseline to immediately post treatment. The massage group experience a statistically significant (p &lt; 0.001 using paired t-test) mean reduction in McGill pain score of 11.5 (SD 9.93). For the control group the mean change in McGill pain score is 1.22 (SD 7.16) which is not statistically significant (p = 0.237 using paired t-test).</p>", "<p>Table ##TAB##2##3## showed, using a chi-squared test, a significant association between treatment group and whether the patient gets 50% pain relief, with only one person in the control group but 18/31 and 18/32 patients in the massage group reporting 50% pain relief immediately post and 1 hour post treatment.</p>", "<p>Table ##TAB##3##4## illustrates results from the Spielberger short-form anxiety scale. It shows that comparing the baseline score to that immediately post treatment, the massage group experience a statistically significant (p &lt; 0.001 using paired t-test) mean reduction in STAI score of 3.57 (SD 3.02). For the control group the mean change in STAI score is 0.00 (SD 3.39) which is not statistically significant (p = 1.00 using paired t-test). For those in the massage group there is a statistically significantly greater reduction in anxiety score both immediately post treatment and at 1 hour post treatment than in the control group. The 95% confidence interval for the difference in mean anxiety score reduction at one hour post treatment between the massage and control groups is 1.67 to 4.36 that is the massage group can expect on average to benefit by between 1.67 and 4.36 greater reduction in anxiety score than the control group up to one hour post treatment.</p>", "<p>At the final assessment for the patient, either at 4 hours post treatment or at the point that they requested an analgesic and thus withdrew from the study the patient was asked to assess whether the treatment they had received was poor, fair, good, very good or excellent. In the massage group, 34 rated the treatment as good, very good or excellent, 12 as fair and 4 as poor. In the control group, 4 rated it as good, very good or excellent, 14 as fair and 33 as poor. A chi-squared test for the patients' assessment of the treatment demonstrates that those receiving the massage rate this treatment statistically significantly better than the control group (X²= 46.6, df = 2, p &lt; 0.001).</p>", "<p>At the final assessment the patient was asked whether they would like to have the treatment again. A significantly higher proportion of those in the massage group (X²= 24.4, df = 1, p &lt; 0.001) said they would like the treatment again. Of the 49 patients in the massage group who were asked this question all 49 (100%) said yes they would like the treatment again. Of the 42 patients in the control group asked this question, 25 (60%) said yes they would like the treatment again.</p>" ]

[ "<title>Discussion</title>", "<p>This study showed that, compared to baseline, massage reduced pain and anxiety for up to one hour. There were no such differences for the control group. This suggests that massage could be a useful short term intervention, to reduce both pain and anxiety. Although an hour's pain relief after a massage may seem a short duration, this could be very worthwhile if, for example, the patient is anxious or fearful about a procedure they are about to undergo, or they are waiting for an analgesic to take effect, or to give them some short term respite from pain during the day and thus a sense of a degree of control over the pain.</p>", "<p>The results showed that these patients with moderate to severe chronic pain are used to taking analgesics, with 77% of patients in the control group dropping out to take an analgesic after one hour post intervention, and 92% control patients had dropped out by 4 hours post intervention.</p>", "<p>One of the problems of demonstrating any statistically significant effect of massage over time was the large drop out rate. Once the patient had taken an analgesic, the effects due to the massage could no longer be reliably elicited and they had to be excluded. The differential drop out rate between the groups is interesting. At 2 hours post treatment 72% of patients having massage compared to only 23% of those in the control group remained in the study. At 3 hours post treatment 53% of massage patients compared to only 10% of control patients remained. At the final assessment after 4 hours, over a third (36%) of massage patients remained compared to only 8% of those in the control group. This gives at least an indication that because patients who dropped out did so because they requested an analgesic, those in the massage group who stayed in the study did so because they had less pain.</p>", "<p>The study had aimed to recruit 120 patients, 60 in each group. In the event, 101 were included in the study. This was based on a previous study of relaxation in hip fracture patients which had a standard deviation of 25 mm and a mean difference of 15 mm [##REF##1235985##18##]. If we take the actual VAS pain scores comparing baseline to immediately post treatment, there is a mean difference of 16 mm and a standard deviation of 21 mm. For a mean difference and standard deviation of this size a sample size of 45 per group, gives power of 95%. So, looking at this outcome, the study is adequately powered.</p>", "<p>The extent of a statistically significant reduction of pain and whether it is clinically significant is important to consider. Compared to baseline, those in the massage group had a mean reduction in pain of 16.7 mm on a 0–100 scale immediately after treatment. If 50% pain relief is taken as clinically important, 36% of the massage group reported at least this level of relief immediately post treatment, compared to 0% of the control group. This provides some evidence that the massage can have quite a large effect in just over a third of patients. Although a benefit of reduced pain for one hour may seem a limited effect, it does provide something of a break from pain.</p>", "<p>It is then a clinical question whether the 15 minute investment of time in doing the massage is worth the benefits outlined in the paper. When patients either completed the study or withdrew, 68% of massage patients compared to only 8% of control patients said the intervention had been good, very good or excellent, so it would seem over two-thirds of patients did find the treatment beneficial. This could be one area where carers may be able to take on this role, although anecdotally, we are aware that some clinicians may be reluctant to make this suggestion to carers, fearing litigation if there are any adverse effects. Health professionals wishing to integrate massage into their practice need to consider professional training, and the requirements of the regulatory bodies of their own country. For example, in the UK, the Royal College of Nursing produced guidance on integrating complementary therapies into clinical care [##UREF##9##23##]. It would also be sensible to liaise with the appropriate committees within their own hospital or community setting regarding the level of training required to practice (for example, the formal structures addressing risk management, quality assurance and professional development).</p>", "<p>This study adds to the international knowledge base about the effectiveness of massage for chronic pain. It has demonstrated that massage can produce short term reductions in pain and anxiety, and improvements in pain relief which are valued by patients. These are important potential benefits for people in whom the management of pain is often a challenge.</p>" ]

[ "<title>Conclusion</title>", "<p>Massage is effective in the short term for chronic pain of moderate to severe intensity, and has a small anxiety reducing effect. It could be a useful addition to techniques offered to patients as part of their care. Nurses are well placed to be trained in and to deliver massage. More research is needed to explore the benefits of repeated treatments of massage in patients with chronic pain.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Massage is increasingly used to manage chronic pain but its benefit has not been clearly established. The aim of the study is to determine the effectiveness of a single session of nurse-administered massage for the short term relief of chronic non-malignant pain and anxiety.</p>", "<title>Methods</title>", "<p>A randomised controlled trial design was used, in which the patients were assigned to a massage or control group. The massage group received a 15 minute manual massage and the control group a 15 minute visit to talk about their pain. Adult patients attending a pain relief unit with a diagnosis of chronic pain whose pain was described as moderate or severe were eligible for the study. An observer blind to the patients' treatment group carried out assessments immediately before (baseline), after treatment and 1, 2, 3 and 4 hours later. Pain was assessed using 100 mm visual analogue scale and the McGill Pain Questionnaire. Pain Relief was assessed using a five point verbal rating scale. Anxiety was assessed with the Spielberger short form State-Trait Anxiety Inventory.</p>", "<title>Results</title>", "<p>101 patients were randomised and evaluated, 50 in the massage and 51 in the control group. There were no statistically significant differences between the groups at baseline interview. Patients in the massage but not the control group had significantly less pain compared to baseline immediately after and one hour post treatment. 95% confidence interval for the difference in mean pain reduction at one hour post treatment between the massage and control groups is 5.47 mm to 24.70 mm. Patients in the massage but not the control group had a statistically significant reduction in anxiety compared to baseline immediately after and at 1 hour post treatment.</p>", "<title>Conclusion</title>", "<p>Massage is effective in the short term for chronic pain of moderate to severe intensity.</p>", "<title>Trial Registration</title>", "<p>[ISRCTN98406653]</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>KS conceived the study and participated in its design and coordination and drafted the manuscript. NC participated in the design, conducted the statistical analysis and helped to draft the manuscript, JM and KC were involved in the data collection, delivering the intervention and helped draft the manuscript, DC participated in the design of the study and the data collection and helped draft the manuscript. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1472-6955/7/10/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>This study was funded by the Oxfordshire Health Services Research Fund. The funder was not involved in the design, conduct or interpretation of the study. Thank-you to all patients and staff involved in this study for their time and support.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Participant Flow at each stage of study.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Baseline (pre-treatment) pain VAS scores for the massage and control groups.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Mean and 95% confidence interval for the mean pain VAS scores for the massage group (- - - - -) and the control group (―) at three time points.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Change in pain VAS score between baseline and immediately post treatment for both the massage group and the control group.</bold> The change is calculated as baseline – post-treatment, so positive values indicate pain has been reduced by treatment.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Comparison of the characteristics of the massage and control groups at baseline (pre-treatment)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Characteristic</bold></td><td align=\"center\"><bold>Massage group</bold></td><td align=\"center\"><bold>Control Group</bold></td><td align=\"center\"><bold>p-value</bold></td></tr></thead><tbody><tr><td align=\"left\">Number in group</td><td align=\"center\">50</td><td align=\"center\">51</td><td/></tr><tr><td align=\"left\">Gender (male:female)</td><td align=\"center\">22:28</td><td align=\"center\">20:31</td><td align=\"center\">0.626 NS</td></tr><tr><td align=\"left\">Age in years Mean (SD)</td><td align=\"center\">51.5 (13.4)</td><td align=\"center\">55.2 (12.9)</td><td align=\"center\">0.171 NS</td></tr><tr><td align=\"left\">Duration of pain in years. Mean (SD)</td><td align=\"center\">9.86 (7.47)</td><td align=\"center\">11.0 (10.2)</td><td align=\"center\">0.540 NS</td></tr><tr><td align=\"left\">Therapist (A:B)</td><td align=\"center\">27:23</td><td align=\"center\">27:24</td><td align=\"center\">0.915 NS</td></tr><tr><td align=\"left\">Pain VAS. Range 0–100. Mean (SD)</td><td align=\"center\">57.7 (18.0)</td><td align=\"center\">62.3 (16.6)</td><td align=\"center\">0.200 NS</td></tr><tr><td align=\"left\">McGill Pain Q'aire Pain Rating Index Range 0–78 Mean (SD)</td><td align=\"center\">27.3 (11.5)</td><td align=\"center\">25.1 (14.4)</td><td align=\"center\">0.390 NS</td></tr><tr><td align=\"left\">Spielberger <italic>SF </italic>STAI Range 6–24 Mean (SD)</td><td align=\"center\">13.3 (4.16)</td><td align=\"center\">12.0 (4.15)</td><td align=\"center\">0.139 NS</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Comparison of massage group and control group for both pain VAS scores and changes in pain VAS score from baseline, at post treatment assessments*.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Time point</bold></td><td align=\"center\" colspan=\"3\"><bold>Massage group</bold></td><td align=\"center\" colspan=\"3\"><bold>Control group</bold></td><td align=\"center\"><bold>p-value (2 sample t-test)</bold></td></tr><tr><td/><td colspan=\"6\"><hr/></td><td/></tr><tr><td/><td align=\"center\">Mean</td><td align=\"center\">SD</td><td align=\"center\">N#</td><td align=\"center\">Mean</td><td align=\"center\">SD</td><td align=\"center\">N#</td><td/></tr></thead><tbody><tr><td align=\"left\">Baseline score</td><td align=\"center\">57.7</td><td align=\"center\">18.0</td><td align=\"center\">49</td><td align=\"center\">62.3</td><td align=\"center\">16.6</td><td align=\"center\">47</td><td align=\"center\">0.200 NS</td></tr><tr><td colspan=\"8\"><hr/></td></tr><tr><td align=\"left\">Immediately post treatment score</td><td align=\"center\">41.1</td><td align=\"center\">20.2</td><td align=\"center\">49</td><td align=\"center\">62.8</td><td align=\"center\">20.9</td><td align=\"center\">47</td><td/></tr><tr><td align=\"left\">Change from baseline</td><td align=\"center\">16.7</td><td align=\"center\">21.2</td><td align=\"center\">48</td><td align=\"center\">-0.04</td><td align=\"center\">16.0</td><td align=\"center\">45</td><td align=\"center\">0.000</td></tr><tr><td colspan=\"8\"><hr/></td></tr><tr><td align=\"left\">1 hour post treatment score</td><td align=\"center\">44.8</td><td align=\"center\">23.5</td><td align=\"center\">47</td><td align=\"center\">64.9</td><td align=\"center\">26.7</td><td align=\"center\">43</td><td/></tr><tr><td align=\"left\">Change from baseline</td><td align=\"center\">12.4</td><td align=\"center\">21.3</td><td align=\"center\">47</td><td align=\"center\">-2.66</td><td align=\"center\">24.0</td><td align=\"center\">41</td><td align=\"center\">0.002</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Comparison of massage group and control group for 50% pain relief at post treatment assessment.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Time point</bold></td><td align=\"center\" colspan=\"2\"><bold>Massage group 50% pain relief</bold></td><td align=\"center\" colspan=\"2\"><bold>Control group 50% pain relief</bold></td><td align=\"center\"><bold>p-value</bold></td></tr><tr><td/><td colspan=\"5\"><hr/></td></tr><tr><td/><td align=\"center\">Yes</td><td align=\"center\">No</td><td align=\"center\">Yes</td><td align=\"center\">No</td><td/></tr></thead><tbody><tr><td align=\"left\">Immediately post treatment <break/>Change from baseline</td><td align=\"center\">18</td><td align=\"center\">31</td><td align=\"center\">0</td><td align=\"center\">51</td><td align=\"center\">0.000</td></tr><tr><td align=\"left\">1 hour post treatment <break/>change from baseline</td><td align=\"center\">18</td><td align=\"center\">32</td><td align=\"center\">1</td><td align=\"center\">49</td><td align=\"center\">0.000</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Comparison of massage group and control group for both Spielberger STAI scores and changes in STAI score from baseline, at post treatment assessments.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Time point</bold></td><td align=\"center\" colspan=\"3\"><bold>Massage group</bold></td><td align=\"center\" colspan=\"3\"><bold>Control group</bold></td><td align=\"center\"><bold>p-value</bold></td></tr><tr><td/><td colspan=\"6\"><hr/></td><td/></tr><tr><td/><td align=\"center\">Mean</td><td align=\"center\">sd</td><td align=\"center\">n</td><td align=\"center\">Mean</td><td align=\"center\">Sd</td><td align=\"center\">N</td><td/></tr></thead><tbody><tr><td align=\"left\">Baseline score</td><td align=\"center\">13.3</td><td align=\"center\">4.16</td><td align=\"center\">50</td><td align=\"center\">12.0</td><td align=\"center\">4.15</td><td align=\"center\">47</td><td align=\"center\">0.139 NS</td></tr><tr><td colspan=\"8\"><hr/></td></tr><tr><td align=\"left\">Immediately post treatment score</td><td align=\"center\">9.72</td><td align=\"center\">3.43</td><td align=\"center\">47</td><td align=\"center\">12.2</td><td align=\"center\">5.15</td><td align=\"center\">47</td><td/></tr><tr><td align=\"left\">Change from baseline</td><td align=\"center\">3.57</td><td align=\"center\">3.02</td><td align=\"center\">47</td><td align=\"center\">0.00</td><td align=\"center\">3.39</td><td align=\"center\">44</td><td align=\"center\">0.000</td></tr><tr><td colspan=\"8\"><hr/></td></tr><tr><td align=\"left\">1 hour post treatment score</td><td align=\"center\">10.5</td><td align=\"center\">3.90</td><td align=\"center\">50</td><td align=\"center\">12.8</td><td align=\"center\">5.06</td><td align=\"center\">47</td><td/></tr><tr><td align=\"left\">Change from baseline</td><td align=\"center\">2.74</td><td align=\"center\">3.10</td><td align=\"center\">50</td><td align=\"center\">-0.28</td><td align=\"center\">3.43</td><td align=\"center\">43</td><td align=\"center\">0.000</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>* Note that at each time point adding the change from baseline scores to the post treatment scores will not necessarily equal the baseline scores, because some patients have dropped out after the baseline</p><p>SD = Standard Deviation</p><p>NS = Not significant</p><p># the table has some missing data on VAS scores at baseline, but data available on other outcomes. One participant in the massage group and 3 in control group did not complete VAS pain scores, and thus were not included in the later comparisons to baseline.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1472-6955-7-10-1\"/>", "<graphic xlink:href=\"1472-6955-7-10-2\"/>", "<graphic xlink:href=\"1472-6955-7-10-3\"/>", "<graphic xlink:href=\"1472-6955-7-10-4\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Molecular studies of ovarian granulosa cells have determined that the granulosa cells of various species express potassium, calcium, sodium, and chloride channels. These channels have electrical activity and generate action potentials. Porcine granulosa cells express a potassium current (I_A), a delayed rectifier K⁺current (I_K) and Ca²⁺currents [##REF##8394746##1##,##REF##11752222##2##]. Ion channels such as Kv1.1, Kv1.2, Kv1.3, Kv1.4, Kv1.5, Kv1.6, KCNQ1, KCNE1 have been identified in porcine granulosa cells [##REF##11752222##2##]. Kir6.1 and Kv4.2 are expressed in human granulosa cells [##REF##16478818##3##,##REF##15991246##4##]. Ca²⁺subunits Cav1.2 and Cav3.2 are expressed in human granulosa cells and calcium type currents are also found in human granulosa cells [##REF##15356055##5##,##REF##15671094##6##]. Human granulosa cells express a Ca²⁺activated K⁺current (BK_Ca), a transient outward K⁺current and an ATP-sensitive potassium channel [##REF##16478818##3##,##REF##15991246##4##,##REF##12466354##7##]. In hen granulosa cells, chloride channels are activated by cAMP during LH-stimulated progesterone production [##UREF##0##8##].</p>", "<p>During aging, potassium, calcium and sodium channels activities and levels are altered. For the potassium and calcium channels, the channels in the cells in the brain, heart, liver, and pancreas all change during the process of aging [##REF##12040053##9##, ####REF##10993747##10##, ##REF##16621391##11##, ##REF##15611028##12####15611028##12##]. Cumulatively, these changes include a decrease in the total number of ion channels present and alterations in the distribution and activity of the channels. For the sodium channels, the changes associated with developmental aging in retinal ganglion cells, myocardium and in kidney epithelium cells include shifts in the number and alterations in conduction activity [##REF##9607716##13##, ####REF##16476705##14##, ##REF##9950962##15####9950962##15##]. These reports suggest that there are specific age-related patterns in the expression and physiological activity of ion channels.</p>", "<p>Hyperpolarization activated cyclic nucleotide gated (HCN) channels generate a pacemaker current (I_h) that controls spontaneous pacemaker activity in the heart and brain [##REF##15975637##16##, ####REF##12471170##17##, ##REF##16460277##18##, ##REF##15963351##19####15963351##19##]. There are four members of the HCN gene family and they belong to the voltage-gated K+ superfamily. The four forms of HCN genes (HCN1-4) have highly conserved core transmembrane and cyclic nucleotide binding regions, with each of the four proteins having a six transmembrane region. The four HCN genes have different distributions in the heart and brain, suggesting that they have different functions. HCN channels have been in found in neurosecretory neurons of the hypothalamus, retinal rod photoreceptors, hair cells of the auditory system, olfactory neurons, spinal cord dorsal root ganglion neurons, and the enteric nervous system [##REF##15975637##16##, ####REF##12471170##17##, ##REF##16460277##18##, ##REF##15963351##19##, ##REF##10228147##20##, ##REF##11459060##21##, ##REF##12514127##22##, ##REF##16882011##23##, ##REF##16648453##24##, ##REF##16879992##25####16879992##25##]. The wide distribution of the HCN channels suggests that they have roles in a number of different physiological conditions. In addition to the wide distribution of these channels, it has been previously reported that HCN4 expression in the hippocampus is related to developmental age, suggesting that these channels also have aging-related changes [##REF##16882011##23##,##REF##16648453##24##].</p>", "<p>To our knowledge, no prior studies have investigated the HCN channels in the ovary. Given the important roles of HCN in other organs and given the aging-related changes found in potassium, calcium, sodium and HCN channels, it was hypothesized that HCN channels play vital roles in the ovary and that alterations of their expression would be found during reproductive aging. In this study, we analyzed the expression and localization of HCN1-4 in the rat ovary to assess this postulate.</p>" ]

[ "<title>Methods</title>", "<title>Animals and treatment</title>", "<p>Sprague-Dawley rats (Harlan, Indianapolis, IN) of three age groups were studied: 1.) \"young\", 26 days old, immature control females; 2.) \"adult\", 65–75 day old, adult control females and; 3.) \"reproductive aging\", 8–9 month old retired breeders, experimental females with declining fertility [##REF##17636211##26##]. The animals were maintained under standard housing conditions with a 12 h:12 h light cycle. They were provided access to standard rat chow (Harlan, Indianapolis, IN) and water ad libitum. The animals were euthanized by an overdose of carbon dioxide. Subsequently, both ovaries were dissected out from each animal; one ovary was snap frozen and stored at -80°C while the other one was fixed in 10% formalin and stored at 4°C for paraffin sectioning. All procedures were approved by the Institutional Animal Care and Use Committee of the University at Buffalo (GYN07042N).</p>", "<title>RNA isolation and RT-PCR</title>", "<p>Rat ovarian total RNA was isolated using Trizol (GibcoBRL, Life Technologies, Grand Island, NY). RT-PCR was performed as previously described by our laboratory [##REF##16879992##25##], using a Promega Access RT-PCR kit (Access RT-PCR System, Promega, Madison, WI). PCR primers were designed to amplify rat HCN1-4 mRNA (Table ##TAB##0##1##) and were slightly modified from mouse primers used previously [##REF##16879992##25##]. Positive control for HCN1-4 was brain RNA and the negative control was running the PCR reaction without the cDNA template. PCR conditions were as follows: 45°C for 45 min, 94°C for 2 min, and then 40 cycles of 94°C for 30 sec, 60°C for 1 min, 68°C for 2 min, and a final extension of one cycle at 68°C for 7 min. The analysis of the RT-PCR reaction products was by agarose gel electrophoresis.</p>", "<title>Western blotting</title>", "<p>Total protein was isolated by procedures used previously [##REF##16879992##25##,##REF##17636211##26##]. In brief, rat tissues were lysed in RIPA buffer, containing 50 mM Tris-HCl, 150 mM NaCl, 0.1% sodium dodecyl sulfate (SDS), 1% NP-40, 1 mM phenylmethanesulfonyl fluoride (PMSF), and 0.5% N, N'-dicyclohexylcabodiimide (DCC) [##REF##16879992##25##,##REF##17636211##26##], along with protease inhibitors (Sigma, St. Louis, MO, 1:100). The protein concentrations were determined by the Bradford method (Bio-Rad, Hercules, CA). Fifty micrograms of protein from rat tissues under reducing conditions were loaded onto a 10% (HCN2 and HCN3) or an 8% (HCN1 and HCN4) Tris-Glycine SDS-polyacrylamide gel (Invitrogen, Carlsbad, CA). After electrophoresis, the proteins were electrically transferred to a nitrocellulose membrane (VWR International, West Chester, PA), blocked with 5% skim milk in TTBS (TBS with 0.1% Tween 20), and then incubated overnight at 4°C with rabbit polyclonal antibodies against HCN1-4 (anti-HCN 1, 2, 3 or 4; product # APC-056, APC-030, APC-057, APC-052, respectively; Alomone Labs Ltd., Jerusalem, Israel) [##REF##17636211##26##] at a dilution of 1:200. Horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (1:2500; Amersham Pharmacia, Piscataway, NJ) was used to identify the protein bands and they were amplified using SuperSignal West Pico Chemiluminescent Substrate Kit (Pierce Biotechnology, Rockford, IL). Visualization of the protein bands was by CL-Xposure film (Pierce Biotechnology).</p>", "<title>Immunohistochemistry and H score semiquantitative analysis</title>", "<p>After fixation, ovaries were embedded in paraffin and cut at 4 μm thick sections that were placed on Starfrost Adhesive positively charged microscope slides (Mercedes Medical, Germany) and the procedures used were as previously described [##REF##16875679##27##]. Sections were deparaffinized using xylene and rehydrated using graded alcohol series. Sections were rinsed in distilled water then incubated for 30 minutes in 4N HCl at 37°C for antigen retrieval. Slides were cooled to room temperature, and then washed in PBS for 5 minutes. The slides were then placed in sodium borohydride diluted in PBS at a concentration of 1 mg/ml. The slides were then rinsed three times in PBS. The tissue sections were then blocked for 1 hour at room temperature using 5% goat serum and 5% BSA. The slides were transferred to a humidified chamber, and a rabbit polyclonal primary antibody against one of the HCN channels described in Western blotting. The HCN2 and HCN3 anitibodies were applied at dilutions of 1:200, and HCN1 and HCN4 were at dilutions of 1:50. As negative control, the primary antibodies were omitted. To confirm specificity of immunostaining, an additional negative control was performed for each of the channels. The anti-HCN1-4 antibodies were pre-incubated with the appropriate antibody control antigen as follows: HCN 1 _6–24Peptide, HCN 2 _147–161Peptide, HCN 3 _727–744Peptide, and HCN4-GST fusion protein (provided by Alomone labs). For the HCN1-3 preadsorption control solution, 1 μg of peptide was incubated with 1 μg of antibody and for HCN 4 preadsorption control solution, 3 μg of fusion protein was incubated with 1 μg of antibody for one hour at 37°C, centrifuged at 12,000 rpm, and then applied to the tissue sections. The slides were incubated with primary antibodies overnight at 4°C. The following morning, slides were washed 3 times in PBS for 5 minutes each. To visualize the primary antibody, Alexa Fluor goat anti-rabbit 594 (Molecular Probes, Eugene, OR) were applied to the tissue sections for 30 minutes at room temperature in the dark. To visualize HCN1, 2, and 3, the Alexa Fluor was applied at a dilution of 1:1000. To visualize HCN4, the secondary antibody was applied at a dilution of 1:500. Slides were air dried for 1 hour and cover-slipped using ProLong Antifade (Molecular Probes). The sections were then viewed on a Nikon Eclipse E400 fluorescent Microscope (Micro Video Instruments, Avon, MA) using the appropriate filters.</p>", "<p>A modified H score system was used to analyze the HCN channels 1 to 4 staining [##REF##16875679##27##,##REF##16260272##28##]. This scoring system was based on two criteria: the distribution of the staining, and the intensity of the staining. The following scale was used to determine the distribution of the stain in a structure: less than 50% of the structure stained was scored as 1, greater than or equal to 50% of the structure stained was scored as 2. To determine the intensity of the stain the following scale was used: no stain = 0, weak staining = 1, moderate staining = 2, and intense staining = 3. The numbers obtained for the distribution and intensity were then multiplied together for a combined score. A total score of 6 was the maximum for any structure. Two independent observers scored the sections for H score analysis. The results from each reviewer were compared and any discrepancy greater than 10% resulted in a reevaluation with both reviewers. Follicles were divided into classes based on the criteria of Oktay et al [##REF##7492681##29##]. Primordial, primary, preantral, and antral follicles were included in the H score analysis, as well as thecal cells, oocytes, and corpora lutea.</p>", "<title>Statistical analysis</title>", "<p>Three separate rats in each age group were used for the RT-PCR and western blot experiments. Data for the ovarian immunofluorescence H score analysis are presented as mean +/- standard error and represent results from experiments repeated in triplicate. Statistical analyses of the ovarian H scores for HCN1-4 were performed using a one-way ANOVA followed by a linear contrast (SPSS version 11). P &lt; 0.05 was considered statistically significant.</p>" ]

[ "<title>Results</title>", "<title>Gene expression and western blot analysis of HCN1-4</title>", "<p>HCN1-4 mRNA expression in the rat ovary was determined by RT-PCR (Figure ##FIG##0##1A##). For all four HCN channels, RT-PCR demonstrated the presence of these mRNA in the ovaries in all reproductive stages studied. Western blot analysis showed protein bands for HCN1-3 in rat ovaries of the three reproductive stages studied (Figure ##FIG##0##1B##). For HCN4, no protein was detected in the rat ovarian tissue by our whole ovarian extract western blot analysis. However, a 150 kDa protein band was found for rat brain and heart, two positive controls.</p>", "<title>Immunohistochemistry of ovarian HCN1-4</title>", "<p>Figure ##FIG##1##2## depicts ovarian follicles and the staining patterns found using immunofluorescence to analyze for the localization of HCN1-4 in young, adult and reproductively aged rat ovaries. By H score analysis (Figure ##FIG##1##2##), differences were detected in the spatial and temporal localization of HCN1-4 in the rat ovary. All four HCN proteins were detected in ovaries of all three reproductive ages. However, there were specific spatial protein differences in the distribution of HCN1-4. For HCN1, HCN2, and HCN3, the proteins were found in oocytes and in the granulosa cells of primary, preantral, and antral follicles. In addition, all three proteins were found in thecal cells. Furthermore, all three proteins were localized to corpora lutea. For HCN4 experimental tissue sections, HCN4 protein expression was localized only to the oocytes. In the HCN1-4 negative control experiments, both types of negative control experiments, the preabsorbed control experiments and the omission of the primary antibody experiments, were appropriately negative.</p>", "<p>In addition to the spatial findings above, there were age-related findings related to the reproductive age of the ovaries for the specific ovarian structures studied (Figure ##FIG##2##3##). For the following structures, there were differences in the H scores for the HCN proteins for the three reproductive ages studied: 1.) oocytes: HCN1 (decline in H score with increasing reproductive age; p &lt; 0.05), HCN3 (decline in H score with increasing reproductive age; p &lt; 0.01) and HCN4 (decline in H score with increasing reproductive age; p &lt; 0.01); 2.) preantral follicle granulosa cells: HCN3 (decline in H score with increasing reproductive age; p &lt; 0.05); 3.) primary follicle granulosa cells: HCN3 (decline in H score with increasing reproductive age; p &lt; 0.01); 4.) thecal cells: HCN3 (decline in H score with increasing reproductive age; p &lt; 0.01).</p>" ]

[ "<title>Discussion</title>", "<p>All four types of HCN channels are expressed in the ovary as evidenced by RT-PCR, western blot, and immunohistochemical results presented in this report. To our knowledge, this is the first description of the changes in distribution of the HCN channels in ovarian structures in the reproductive life-cycle. These channels have specific patterns in the different ovarian cell types. HCN channels 1–3 are expressed in oocytes, granulosa cells of primary, preantral, and antral follicles, the thecal cells, and in luteal cells, while HCN4 is only expressed in oocytes. This suggests that different ovarian structures use different combinations of HCN channels for normal physiological function. Furthermore, HCN4 appears to be oocyte specific and, thus, this protein may be useful to define the physiological status of an oocyte.</p>", "<p>Ion channels are involved in ovarian steroidogenesis. Potassium channels mediate gonadotropin regulated progesterone secretion in human granulosa cells [##REF##16478818##3##,##REF##15991246##4##,##REF##12466354##7##,##REF##12080003##30##]. L- and T-type Ca²⁺channels mediate hCG stimulated progesterone secretion in human granulosa cells [##REF##15356055##5##]. Sodium channels down regulate progesterone production in primate granulosa cells [##REF##10894155##31##]. Potassium channels and cAMP are involved in FSH-stimulated progesterone production in pig granulosa cells [##UREF##1##32##]. Given that HCN channels are located in all the cell types which are involved in steroidogenesis, the granulosa, theca and corpora lutea cells, it would not be unreasonable to hypothesize that the HCN channels also participate in this important ovarian activity. HCN channels have been identified in secretory cells including GnRH neurons, pancreatic β-cells, and pituitary lactotrophs [##REF##15975637##16##, ####REF##12471170##17##, ##REF##16460277##18##, ##REF##15963351##19####15963351##19##,##REF##16879992##25##]. Several studies have described membrane hyperpolarization in granulosa cells. Activation of BK_Cachannels resulting in membrane hyperpolarization is required for steroidogenesis in human luteinized granulosa cells [##REF##12466354##7##]. Thus, membrane hyperpolarization may be a mechanism controlling steroid production in the granulosa cells of the ovary. In contrast to most voltage gated channels, HCN channels are activated by membrane hyperpolarization [##REF##15975637##16##, ####REF##12471170##17##, ##REF##16460277##18##, ##REF##15963351##19####15963351##19##]. In granulosa cells it is possible that hyperpolarization of the cell membrane could activate HCN channels, thereby resulting in membrane depolarization. Depolarization could activate calcium and cAMP signaling, thus resulting in activation of steroidogenic enzymes and thereby increasing steroid production. This hypothesis would be supported by demonstration of functional HCN channels in granulosa cells.</p>", "<p>There are age related changes in potassium, calcium and sodium channel expression [##REF##12040053##9##, ####REF##10993747##10##, ##REF##16621391##11##, ##REF##15611028##12##, ##REF##9607716##13##, ##REF##16476705##14##, ##REF##9950962##15####9950962##15##]. In the data presented in this report, HCN1 and HCN2 had minimal variation through the aging process, with HCN1 only exhibiting declining levels in the oocytes during reproductive aging. This suggests that the expression of these two channels remains relatively constant throughout granulosa and thecal cell reproductive aging and may have minor or unchanging roles in ovarian physiologic functions such as steroidogenesis or peptide hormone production. HCN3 has different expression patterns in the granulosa cells and theca cells during the aging process, indicating that there is an age-dependent expression of HCN3 in ovarian structures, suggesting that the changes in steroidogenesis during aging might be modulated through this protein. HCN3 expression decreases during aging in oocytes, granulosa cells of preantral follicles, and in theca cells, suggesting a possible function in the decrease of ovarian function in advancing reproductive age. The mechanism of this decline is not yet known and understanding of the mechanism could lead to further insights into the overall aging-related reduction of ovarian function. To date, the physiologic processes in reproductive aging are not yet fully understood. HCN channels may play a role in ovarian aging and, in addition, they could serve as immunohistochemical biomarkers for reproductive aging.</p>", "<p>HCN4 is an oocyte specific channel and may be an indicator of oocyte quality. There is a linear decrease in the expression of HCN4 throughout the reproductive aging process in the female rat. Brewster et al. and Surges et al. showed a steady decrease in HCN4 expression throughout the maturation process of the rat hippocampus [##REF##16882011##23##,##REF##16648453##24##]. In the ovary, growth differentiation factor 9 (GDF-9) has been found to be oocyte specific [##REF##10379900##33##, ####REF##10067849##34##, ##REF##16384866##35####16384866##35##]. GDF-9 has been demonstrated to be essential to the growth and differentiation of early ovarian follicles. In cultured bovine granulosa cells, GDF-9 stimulated the proliferation of granulosa cells from small and large antral follicles and can disrupt the production of progesterone and estradiol [##REF##16648300##36##]. The functions of HCN4 in the oocyte have not been determined to date, but it is possible that the channel may also be necessary for growth and differentiation of ovarian follicles or for steroid production.</p>" ]

[ "<title>Conclusion</title>", "<p>In conclusion, HCN1-4 channels are expressed in the ovary and there are differential expression patterns for the channels. HCN1-3 are expressed in ovarian structures including oocytes, granulosa and thecal cells, and in luteal cells, while HCN 4 is only expressed in the oocytes. There are decreases in the expression of HCN1, 3, and 4 in the oocytes during reproductive aging, with the decrease in HCN4 being the most pronounced. HCN3 expression are also decreased in the granulosa cells of preantral follicles and theca cells in reproductive aging. Future studies need to be conducted to determine the specific roles of HCN channels in the ovaries and the physiological reasons for the changes in the expression of the channels through the aging process.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Aim of this study was to test the hypothesis that levels of hyperpolarization activated cyclic nucleotide gated channels 1 to 4 (HCN1-4) are linked to the reproductive age of the ovary.</p>", "<title>Methods</title>", "<p>Young, adult, and reproductively aged ovaries were collected from Sprague-Dawley rats. RT-PCR and western blot analysis of ovaries was performed to investigate the presence of mRNA and total protein for HCN1-4. Immunohistochemistry with semiquantitative H score analysis was performed using whole ovarian histologic sections.</p>", "<title>Results</title>", "<p>RT-PCR analysis showed the presence of mRNA for HCN1-4. Western blot analysis revealed HCN1-3 proteins in all ages of ovarian tissues. Immunohistochemistry with H score analysis demonstrated distinct age-related changes in patterns of HCN1-3 in the oocytes, granulosa cells, theca cells, and corpora lutea. HCN4 was present only in the oocytes, with declining levels during the reproduction lifespan.</p>", "<title>Conclusion</title>", "<p>The evidence presented here demonstrates cell-type and developmental age patterns of HCN1-4 channel expression in rat ovaries. Based on this, we hypothesize that HCN channels have functional significance in rat ovaries and may have changing roles in reproductive aging.</p>" ]

[ "<title>Abbreviations</title>", "<p>HCN: hyperpolarization activated cyclic nucleotide gated channel; RT-PCR: reverse transcriptase polymerase chain reaction; mRNA: messenger ribonucleic acid; Kv1.1 – 1.6: potassium voltage-gated channel, shaker-related subfamily, members 1–6; KCNQ1: potassium voltage-gated channel, subfamily Q, member 1; KCNE1: potassium voltage-gated channel, Isk-related family, member 1; Kir6.1: potassium inwardly-rectifying channel, subfamily J, member 8; Kv4.2: potassium voltage-gated channel, Shal-related family, member 2; I_A: potassium current; I_K: delayed rectifier potassium current; Cav1.2: calcium channel, voltage-dependent, L type, alpha 1C subunit; Cav3.2: calcium channel, voltage-dependent, T type, alpha 1 H subunit; Ca^2+: calcium; K^+: potassium; BK_ca: calcium activated potassium current; ATP: adenosine tri-phosphate; cAMP: cyclic adenosine mono-phosphate; LH: luteinizig hormone; I_h: pacemaker current, hyperpolarization-activated current, hyperpolarization-activated cation current; RNA: ribonucleic acid; cDNA: complimentary deoxyribonucleic acid; SDS: sodium dodecyl sulfate; PMSF: phenylmethylsulfonyl fluoride; DCC: dicyclohexlcabodiimide; TTBS: tris buffered saline with 0.1% Tween 20; HRP: horseradish peroxidase; HCl: hydrochloric acid; PBS: phosphate buffered saline; hCG: human chorionic gonadotropin; FSH: follicle stimulating hormone; GnRH: gonadotropin releasing hormone; β: beta</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>JY conceived of the study along with AA, and participated in its design and coordination and helped to draft the manuscript. BSK is responsible for the Western blots and RT-PCR, data and statistical analysis, and manuscript preparation. LG carried out the immunohistochemistry and baseline research articles for the initial research. JP carried out the immunohistochemistry, baseline research articles, data and statistical analysis, and manuscript preparation. CP carried out the immunohistochemistry as well as data analysis. AA conceived of the study along with JY, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>This work was presented, in part, at the American Society for Cell Biology Annual Meeting as Abstract #2226 in San Diego, CA in 2006.</p>", "<p>We would like to thank Charles Wu for his financial gift and the Walter Mazol fund.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>(A) </bold>RT-PCR analysis of ovarian HCN1-4 gene expression. <bold>(a) </bold>A 248 bp HCN1 RT-PCR band in all three developmental ages studied. <bold>(b) </bold>A 377 bp HCN2 RT-PCR band was evident in all three developmental ages studied. <bold>(c) </bold>A 364 bp HCN3 RT-PCR band was expressed in all three developmental stages studied. <bold>(d) </bold>A 219 HCN4 RT-PCR band was present in all three developmental stages studied. n = 3 animals studied per gene per reproductive stage. M = marker; B = brain; Y = young; A = adult; RA = reproductive aging; N = negative; H = heart. <bold>B</bold>. Western blot analysis of ovarian lysates for HCN1-4 protein expression of young, adult and reproductively aged rats. <bold>(a) </bold>Western blot analysis for HCN1 in ovaries. <bold>(b) </bold>Western blot analysis for HCN2 in ovaries. <bold>(c) </bold>Western blot analysis for HCN3 in ovaries. <bold>(d) </bold>Western blot analysis for HCN4 in ovaries. n = 3 animals studied per protein per reproductive stage.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Immunofluorescence localization of HCN1-4 in young, adult and reproductively aged rat ovaries. Experimental and negative control studies are presented from consecutive ovarian sections for each HCN protein for each reproductive state. <bold>(A) </bold>Localization of HCN1-4 channels in the young ovary. <bold>(B) </bold>Localization of HCN1-4 channels in the adult ovary. <bold>(C) </bold>Localization of HCN1-4 channels in the reproductive aging ovary. n = 3–9 animals studied per protein per reproductive stage.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>H Score analysis of HCN1-4 channels in the different ovarian structures throughout the aging process. <bold>(A) </bold>The H scores of HCN1 (a), HCN2 (b), HCN3 (c), and HCN4 (d) channels in the oocytes during the reproductive aging process. <bold>(B) </bold>H score analysis of granulosa cells in different follicle classes. (a-d) H scores of HCN1-4 channels in primary follicles. (e-h) Results of H score analysis of HCN1-4 channels in preantral follicle granulosa cells. (i-l) H scores for HCN1-4 channels in the granulosa cells of antral follicles. <bold>(C) </bold>H score analysis of HCN1 (a), HCN2 (b), HCN3 (c), and HCN4 (d) expression in theca cells. <bold>(D) </bold>H score analysis of HCN1-4 channels (a-d respectively) in the corpora lutea. N/A – not applicable, as the young rats do not yet have corpora lutea. n = 3–9 animals studied per protein per reproductive stage.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Primers for RT-PCR reactions</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Gene</td><td align=\"left\">GenBank Accession No.</td><td align=\"left\">Forward primer (5'-3')</td><td align=\"left\">Reverse primer (3'-5')</td><td align=\"left\">Size of product (bp)</td></tr></thead><tbody><tr><td align=\"left\">HCN1</td><td align=\"left\"><ext-link ext-link-type=\"gen\" xlink:href=\"NM_053375\">NM_053375</ext-link></td><td align=\"left\">TTCATGCAGAGGCAGTTCAC</td><td align=\"left\">CACGGTGTTGTTGTTTGCTC</td><td align=\"left\">248</td></tr><tr><td align=\"left\">HCN2</td><td align=\"left\"><ext-link ext-link-type=\"gen\" xlink:href=\"NM_053684\">NM_053684</ext-link></td><td align=\"left\">CCATGCTGACAAAGCTCAAA</td><td align=\"left\">CGAGCTGAGATCATGCTGAA</td><td align=\"left\">377</td></tr><tr><td align=\"left\">HCN3</td><td align=\"left\"><ext-link ext-link-type=\"gen\" xlink:href=\"NM_053685\">NM_053685</ext-link></td><td align=\"left\">TCGGACACTTTCTTCCTGCT</td><td align=\"left\">GGTTGAAGATGCGAACCACT</td><td align=\"left\">364</td></tr><tr><td align=\"left\">HCN4</td><td align=\"left\"><ext-link ext-link-type=\"gen\" xlink:href=\"NM_021658\">NM_021658</ext-link></td><td align=\"left\">GGGCTTCTCCTGTAGCCTTT</td><td align=\"left\">TGAGCTTCAGGTCCTGTGTG</td><td align=\"left\">219</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1477-7827-6-35-1\"/>", "<graphic xlink:href=\"1477-7827-6-35-2\"/>", "<graphic xlink:href=\"1477-7827-6-35-3\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Malaria is a major public health problem in sub-Saharan Africa and <italic>Plasmodium falciparum </italic>infection is a leading cause of morbidity and mortality inflicting a huge economic burden in countries where the disease is endemic [##REF##15759000##1##]. It is estimated that death toll of malaria exceeds one million people each year, and the victims are primarily children under the age of five [##REF##11832956##2##]. Until the early 1980s, the African highlands (generally referred to areas of &gt;1,500 m above sea level) were either free of malaria or had very low incidences of the disease; however, since the late 1980s a series of malaria epidemics has occurred [##UREF##0##3##, ####UREF##1##4##, ##REF##2894515##5##, ##REF##8055757##6##, ##REF##9615495##7##, ##REF##9866748##8##, ##REF##11859368##9####11859368##9##]. Among the many factors that may contribute to the highland malaria epidemics, resistance of the parasites to multiple antimalarials has not been extensively investigated. Resistance to antimalarial drugs is one of the major obstacles for effective malaria control. The first case of chloroquine (CQ) resistance in Kenya was reported in 1977 [##REF##382468##10##]. In 1993, resistance levels had reached 70% [##REF##8698011##11##]. In 1998, the Ministry of Health of Kenya changed the first line of treatment from chloroquine to sulfadoxine – pyrimethamine (SP; Fansidar^®) [##REF##11123822##12##]. In 2004, the Ministry of Health of Kenya officially changed the first-line drug to artemether/lumefantrin (Coartem™) [##UREF##2##13##].</p>", "<p>Drug resistance in malaria parasites is associated with genetic mutations in target genes and can be monitored using molecular methods. CQ resistance is determined by the major point mutation at codon 76 of the <italic>P. falciparum </italic>CQ resistance transporter (<italic>pfcrt</italic>) gene [##REF##11517439##14##], which is highly correlated with increased clinical CQ tolerance and treatment failure [##REF##11517439##14##, ####REF##15085184##15##, ##REF##11937421##16####11937421##16##]. In addition, point mutations in <italic>P. falciparum </italic>multi-drug resistance gene 1 (<italic>pfmdr1</italic>) (e.g., N86Y, Y183F, S1034C, N1042D, and D1246Y) have been shown to modulate CQ resistance [##REF##15032630##17##] and possibly lumefantrine resistance [##REF##15717281##18##]. Resistance to antifolates is associated with point mutations in the dihydrofolate reductase (<italic>pfdhfr</italic>) and dihydropteroate synthetase (<italic>pfdhps</italic>) genes [##REF##11703841##19##]. Pyrimethamine (PY) resistance is conferred by the key mutation at codon 108 in the <italic>pfdhfr </italic>gene, while additional mutations at positions 51 and 59 increase the levels of resistance [##REF##2183222##20##]. The 164L mutation common in Southeast Asia has been shown to confer PY resistance. Although this mutation is not common in Africa, a recent study by McCollin et al detected the 164L mutation in western Kenya [##REF##16779725##21##]. Similarly, sulfadoxine (SD) resistance is conferred by a key mutation at codon 437 in the <italic>pfdhps </italic>gene and modulated by additional mutations at codons 436, 540, 581, and 613 [##UREF##3##22##]. Multiple mutations in these two genes result in antifolate treatment failure [##REF##9364963##23##]. Molecular methods have been developed to detect point mutations in genes responsible for drug resistance [##REF##9015515##24##, ####REF##11172152##25##, ##REF##11807721##26####11807721##26##], and have been widely used to monitor drug resistance in epidemiological surveys [##REF##11937421##16##]. In this study, we have analyzed the mutations of antimalarial drug resistance genes for CQ (<italic>pfcrt </italic>and <italic>pfmdr1</italic>) and antifolates (<italic>pfdhps </italic>and <italic>pfdhfr</italic>) in samples from patients with acute malaria infections and school children with asymptomatic infections at one lowland and two highland locations in western Kenya.</p>" ]

[ "<title>Methods</title>", "<title>Sample collection</title>", "<p>Sampling in two highland sites (Kisii, elevation ~1600 m above sea level; and Kakamega, elevation 1,480–1,560 m) and one neighboring lowland site (Kombewa, elevation ~1200 m) from June through August 2005 were conducted as a part of malaria surveillance activities. Average annual malaria prevalence among primary school children during the sampling period was 10.3%, 42.7% and 75% in Kisii, Kakamega and Kombewa, respectively. Malaria transmission intensity, measured by entomological inoculation rate (EIR), was 0.4, 16.6 and 31.1 infectious bites per person per year in Kisii, Kakamega and Kombewa, respectively [##REF##16619599##27##]. At each site, 100 samples were collected from patients admitted with acute <italic>P. falciparum</italic> infection at a local hospital and 100 samples from primary school children (age 6–14 years old) with asymptomatic <italic>P. falciparum</italic> infections diagnosed by microscopy. Symptomatic malaria patients were treated by clinicians in the hospital with antimalarial drug, Coartem which achieved cure rates of up to 95%, even in areas of multi-drug resistance. Each sample consisted of ~200 μl of finger-prick blood spotted on filter papers. Filter papers were dried and stored at -20°C until parasite DNA extraction. The human subject protocol involved in this study has been approved by University of California, Irvine (UCI) and Kenya Medical Research Institute (KEMRI). Informed consent was provided by the parents/guardians of the children and assent from the children was obtained prior to the sample collection.</p>", "<title>Parasite DNA extraction and species identification</title>", "<p>DNA was extracted from the blood filters using the Saponin/Chelex method [##REF##15463789##28##]. Parasite DNA was extracted from one quarter of a blood spot of about 1 cm in diameter and dissolved in 200 μl of distilled water. Three microliters of the parasite DNA were used as the template for polymerase chain reaction (PCR). To avoid complications from mixed parasite species infections, a nested PCR method was used to verify <italic>P. falciparum </italic>infections and exclude the presence of other <italic>Plasmodium </italic>species in each sample [##REF##10348249##29##]. Approximately 3–5% of samples containing mixed parasite species or other parasite species were identified in all the study sites. Samples containing only <italic>P. falciparum </italic>DNA were used for genotyping analysis.</p>", "<title>Molecular detection of mutations in drug targets</title>", "<p>Mutations in the <italic>pfcrt </italic>(K76T) and <italic>pfmdr1 </italic>genes were detected using a PCR-restriction fragment length polymorphism (RFLP) method. The fragment encompassing <italic>pfcrt </italic>codon 76 was amplified and digested with <italic>Apo</italic>I, which cleaves the wild type into 111 and 34 bp fragments [##REF##12798014##30##]. Similarly, fragments containing codons 86, 184, 1034, 1042, and 1246 of the <italic>pfmdr1 </italic>gene were separately amplified by PCR and digested with respective restrictive enzymes as described previously [##REF##12798014##30##, ####REF##10747279##31##, ##REF##10802315##32##, ##REF##17550470##33####17550470##33##]. Screening for mutations associated with antifolate resistance at codons 16, 50, 51, 59, 108, and 164 of the pfdhfr gene and codons 436, 437, 540, 581, and 613 of the pfdhps gene was performed by nested PCR and mutation-specific restriction enzyme digestions [##REF##7611566##34##, ####REF##9603482##35##, ##REF##14628939##36####14628939##36##].</p>", "<title>Statistical analysis</title>", "<p>The difference in frequencies of point mutations in the four aforementioned genes between highland and lowland parasite populations was determined by using the χ²tests or Fisher's exact test for datasets with sample size less than 5. Yates' correction was applied for the chi-square value, resulting in corrected <italic>P </italic>values. Statistical significance was taken at the <italic>P </italic>= 0.05 level. Association between the different mutations was tested using Fisher's exact test.</p>" ]

[ "<title>Results</title>", "<p>A total of 600 <italic>P. falciparum </italic>samples were analyzed for <italic>pfcrt, pfmdr1, pfdhfr</italic>, and <italic>pfdhps </italic>genes. Over 90% of samples were successfully amplified at the 17 test codons, and polymorphisms were detected at 10 of 17 codons screened (Table ##TAB##0##1##). Consistent with the hyperendemic settings in the study areas, we detected infections by mixed strains at seven of the 17 studied codons, which included codon 76 of <italic>pfcrt</italic>, codons 86, 184 and 1246 of <italic>pfmdr1</italic>, codon 59 of <italic>pfdhfr</italic>, and codons 437 and 540 of <italic>pfdhps </italic>(Table ##TAB##0##1##). The frequencies of mutant codons at the four genes varied slightly between the study sites, but the differences were not significant (Fig. ##FIG##0##1##). Mixed-genotype infections at <italic>pfdhfr </italic>codon 59, 437, and 540 of the antifolate target genes were low (&lt;10%) in all samples. Similarly, mixed-genotype infections at codon 76 of the <italic>pfcrt </italic>gene were low (&lt;15%). In contrast, mixed-genotype infections at codons 86, 184 and 1246 of the <italic>pfmdr1 </italic>gene were more frequent (18 – 45%) (Table ##TAB##0##1##).</p>", "<title>Mutations in CQ resistance genes</title>", "<p>Consistent with the past extensive use of CQ in this area, the <italic>pfcrt </italic>76T mutant allele was present in = 80% (including mixed infections with wild type and mutant alleles) of all samples from the three locations (Fig. ##FIG##0##1A## and ##FIG##0##1B##). In comparison, mutations at alleles 86Y, 184F and 1246Y of <italic>pfmdr1 </italic>gene were present at lower frequencies, ranging from 30 to 76% (including mixed infections) (Fig. ##FIG##0##1C## and ##FIG##0##1D##), whereas no mutation was found at codons 1034 and 1042, which was associated with CQ resistance in South America. The frequencies of mutations were generally not significantly different between the three study sites except <italic>pfcrt </italic>76T and <italic>pfmdr1 </italic>86Y between Kombewa and Kisii (<italic>P </italic>&lt; 0.05), and <italic>pfmdr1 </italic>184F between Kombewa and Kakamega (<italic>P </italic>&lt; 0.01) (Fig. ##FIG##0##1E## and ##FIG##0##1F##). Consequently, the genotype 86Y/184Y/1246Y (wildtype at codon 184) was present at a significantly higher frequency in highland than in lowland parasites (<italic>P </italic>&lt; 0.05), whereas the genotype 86Y/184F/1246Y (triple mutations) was significantly more frequent in lowland than in highland parasites (Table ##TAB##1##2##). Since certain <italic>pfmdr1 </italic>alleles were found associated with CQ resistance, we compared the association between <italic>pfmdr1 </italic>mutations and the major CQ-resistance determinant <italic>pfcrt </italic>76T. When all parasite samples were taken into account, significant associations were found between the <italic>pfcrt </italic>76T and two <italic>pfmdr1 </italic>alleles (184F and 1246Y) (<italic>P </italic>&lt; 0.01). However, <italic>pfcrt </italic>76T was not significantly associated with <italic>pfmdr1 </italic>86Y (<italic>P </italic>= 0.174), the mutation found associated with CQ-resistant field isolates in Asia and Africa.</p>", "<title>Mutations in antifolate drug resistance genes</title>", "<p>Despite relatively recent deployment of antifolate drugs for controlling malaria in western Kenya, parasites with mutations in the drug target genes, <italic>pfdhfr </italic>and <italic>pfdhps</italic>, were highly prevalent. The mutation at codon 108 (allele 108N) of the <italic>pfdhfr </italic>gene, a major determinant of PY resistance, was ubiquitously present in all samples (Table ##TAB##0##1##). In addition, mutant alleles at codons 51 and 59 were also detected at high frequencies (85% – 98%). No mutation was detected at codons 16, 50, and 164 of <italic>pfdhfr</italic>. Most samples (90% – 100%) contained the key SD-resistance determinant mutation A437G and K540E in <italic>pfdhps</italic>. In contrast, the mutant 436F allele was rare (&lt; 5%), and mutant 436A allele was not detected in any locations. Further, no mutation was detected at codons 581 and 613 of <italic>pfdhps</italic>.</p>", "<p>Increased numbers of mutations at the two antifolate target genes are associated with increased resistance to SD-PY. In the studied samples, single or double mutations were rarely encountered, whereas triple mutations in the <italic>pfdhfr </italic>gene were the most common. When the two genes were combined, quintuple mutations (51I/59R/108N of <italic>pfdhfr </italic>and 437G/540E of <italic>pfdhps</italic>) were found with a high prevalence (80–87%) at the three sites (Fig. ##FIG##0##1E## and ##FIG##0##1F##), followed by quadruple mutations (51I/59C/108N/437G/540E, 51N/59R/108N/437G/540E, 51I/59R/108N/437G/540K, 51I/59R/108N/437A/540E) (8% – 16%) (Table ##TAB##1##2##).</p>", "<title>Mutation frequency difference between asymptomatic and symptomatic samples</title>", "<p>The <italic>pfcrt </italic>K76T mutation frequencies were not significantly different in the asymptomatic and symptomatic samples in the lowland Kombewa site (96% vs. 94%; Fisher's exact test, <italic>P </italic>= 0.537) and two highland sites (86% vs. 92% in Kakamega, χ²= 1.41, df = 1, <italic>P </italic>= 0.235; 80% vs. 83% in Kisii, χ²0.03, df = 1, <italic>P </italic>= 0.863). Similarly, there was no significantly difference in mutation frequencies of <italic>pdmdr1 </italic>gene at test codons between asymptomatic and symptomatic samples in the lowland (Kombewa: χ²= 1.44, df = 1, <italic>P </italic>= 0.230 at codon N86Y; χ²= 0.33, df = 1, <italic>P </italic>= 0.566 at codon Y184F; χ²= 0, df = 1, <italic>P </italic>= 1.000 at codon D1246Y) and two highland sites (Kakamega: χ²= 0.07, df = 1, <italic>P </italic>= 0.791 at codon N86Y; χ²= 0.01, df = 1, <italic>P </italic>= 0.920 at codon Y184F; χ²= 0.26, df = 1, <italic>P </italic>= 0.610 at codon D1246Y. Kisii: χ²= 0.87, df = 1, <italic>P </italic>= 0.351 at codon N86Y; χ²= 0, df = 1, <italic>P </italic>= 1.000 at codon Y184F; χ²= 1.40, df = 1, <italic>P </italic>= 0.237 at codon D1246Y). Similarly, for the <italic>pfdhfr</italic>/<italic>pfdhps </italic>quintuple mutation frequencies, no significantly difference was found in the asymptomatic and symptomatic samples in both lowland site (Kombewa: χ²= 0.16, df = 1, <italic>P </italic>= 0.689) and highland sites (Kakamega: χ²= 0.20, df = 1, <italic>P </italic>= 0.655. Kisii: χ²= 0.32, df = 1, <italic>P </italic>= 0.572).</p>" ]

[ "<title>Discussion</title>", "<p>In the present study, we analyzed the mutations of four known drug resistance genes in both highland and lowland parasite populations of western Kenya. We showed that the frequencies of key mutations in the <italic>pfcrt</italic>, <italic>pfmdr1</italic>, <italic>pfdhps</italic>, and <italic>pfdhfr </italic>genes that were implicated in resistance to CQ and SP were very high. We further demonstrated that there was no difference in the frequencies of key mutations between symptomatic and asymptomatic malaria volunteers. Our experiment design did not allow us to test clinical or parasitological efficacy in symptomatic infections after treatment; nor was the mutation prevalence/in vivo resistance before 1998 when the national policies of treating uncomplicated malaria were changed to SP was examined. Nevertheless, the big sample size in this study coupled to lack of differences in the frequency of resistant parasite genotypes in the two highland areas and the low land area clearly demonstrates a high frequency of drug resistant mutants circulating in the study areas.</p>", "<p>Resistance to CQ is largely determined by the K76T mutation in the <italic>pfcrt </italic>gene, and enhanced by mutations at other sites of this gene and mutations in the <italic>pfmdr1 </italic>gene. Since resistant phenotypes often have fitness costs [##UREF##4##37##], their prevalence is expected to decline after removal of the selective pressure. For example, the prevalence of mutant alleles of <italic>pfcrt </italic>76T decreased from 64.5% in 2002 to 16% in 2004 and that of mutant <italic>pfmdr1 </italic>86Y alleles decreased from 46.6% to 2.7% two and half year after CQ withdrawal in coastal Tanzania [##REF##17172390##38##]. Kublin et al. reported that the prevalence of the chloroquine-resistant <italic>pfcrt </italic>76T genotype decreased from 85% in 1992 to 13% in 2000. In 2001, chloroquine cleared 100% of 63 asymptomatic <italic>P. falciparum </italic>infections, no isolates were resistant to chloroquine in vitro, and no infections with the chloroquine-resistant <italic>pfcrt </italic>76T genotype were detected [##REF##12792863##39##]. Similarly, Laufer et al. demonstrated that chloroquine was again an efficacious treatment for malaria, 12 years after it was withdrawn from use in Malawi [##REF##17093247##40##]. Although Kenya had a similar drug policy, the <italic>pfcrt </italic>K76T mutation in western Kenya was still predominant. Consistent with the observation of high <italic>pfcrt </italic>K76T mutation frequency in the study sites, <italic>pfmdr1 </italic>mutations that enhance CQ resistance (e.g., 86Y) were also present at relatively high frequencies (37.2 – 45.2%) in western Kenya. This may be partly caused by wide availability and consumption of CQ after it was replaced as first-line antimalarial drug by antifolate drug (SP) in 1998 [##REF##11123822##12##,##UREF##2##13##]. Our results indicated that parasite populations in western Kenya were still highly resistant to 4-aminoquinoline drugs.</p>", "<p>Several years after introduction of SP as the first-line antimalarial drug in Kenya, mutant genotypes have already become common. We found that mutations at codon 108 of the <italic>pfdhfr </italic>gene and codon 437 of the <italic>pfdhps </italic>gene, which are the major determinants for sulfadoxine and pyrimethamine resistance, respectively, were highly prevalent in our study sites. Furthermore, increased resistance to antifolate drugs was correlated with mutations at additional sites, which were also detected in our present study. In Kenya, Nzila et al. [##REF##10722502##41##] reported that the <italic>pfdhfr </italic>triple mutant (codons 108, 51 and 59) was associated with seven-day treatment failure using Fansidar, and this association was strengthened by the presence of double mutations of the <italic>pfdhps </italic>gene at codons 437 and 540. Such \"quintuple\" mutants were also strongly correlated with Fansidar treatment failure in Malawi [##REF##11807721##26##]. These quintuple mutants have already reached 80–87% prevalence in the parasite populations in western Kenya. This is consistent with the observation in another highland site in Chogoria, near Mount Kenya [##REF##11784419##42##]. Since mutations at different sites of the <italic>pfdhfr </italic>and <italic>pfdhps </italic>genes might be a stepwise process, two mutations at codon 59 of <italic>pfdhfr </italic>and codon 540 of <italic>pfdhps </italic>can accurately predict the presence of this quintuple mutant [##REF##11807721##26##]. The molecular method for monitoring antifolate resistance in <italic>P. falciparum </italic>can thus be simplified to detect the presence of these two mutations. This method has been shown to provide the best means of predicting clinical treatment outcomes in the patient population, which consists primarily of children from endemic areas of Africa [##REF##14628939##36##]. In our study, either of these two mutations has exceeded 75% prevalence in the parasite populations, suggesting that at least 50% of the parasites carry these double mutations. Consistent with other genotype studies of <italic>pfdhfr </italic>from Kenyan isolates, we did not observe mutations at codon 164 [##REF##9364963##23##,##REF##10722502##41##,##UREF##5##43##], although McCollum et al. identified 164L mutation in several samples in western Kenya [##REF##16779725##21##].</p>", "<p>The causes of the malaria epidemics in the East African highlands are complex and could be the result of interactions between environmental changes, vectors and parasites. Antimalarial drug resistance has been invoked as a major factor [##REF##11859368##9##], however the molecular epidemiology of drug resistance in these highlands is unknown. Mbaisi et al. [##UREF##5##43##] reported a significantly lower prevalence of mutations at codon 437 of the <italic>pfdhps </italic>gene and codons 86 and 1246 of the <italic>pfmdr </italic>gene in lowland (Kisumu) samples than in highland (Kericho, Magadi, and Entosopia) samples in western Kenya, whereas this trend was the opposite for the mutation at codon 436 of the <italic>pfdhps </italic>gene. In contrast, our study did not detect significant differences in frequency of mutations at 14 codons in four drug resistant genes between samples from lowland and highland areas in western Kenya, although the prevalence of mutations at codons 86 and 1246 of the <italic>pfmdr </italic>gene was higher in samples from the lowlands (Kombewa) than in those from the highlands (Kakamega and Kisii). This result suggests either a similar drug selection pressure or significant gene flow between lowland and highland parasite populations aided by human travel [##REF##16229773##44##], as detected by microsatellite and merozoite surface protein 1 and 2 (<italic>msp</italic>-1 and <italic>msp</italic>-2) [##REF##18165519##45##].</p>" ]

[ "<title>Conclusion</title>", "<p>Together with the evidence that there was no significance difference in the frequencies of key resistance-conferring mutations in <italic>pfcrt</italic>, <italic>pfmdr1</italic>, <italic>pfdhps</italic>, and <italic>pfdhfr </italic>genes between symptomatic and asymptomatic malaria infections in western Kenya highland, high frequencies of these mutations in symptomatic and asymptomatic infections suggest that drug resistance of malaria parasites may be an important contributor to malaria-induced morbidity and mortality. However, the role of drug resistance as a driving force for malaria outbreaks in the highlands has not been established. If drug resistance were the main driving force, the lowland and highland sites should expect similar epidemic pattern of malaria incidence. Moreover, drug resistance level should increase over time until a new antimalarial drug is adopted. Thus, the number of malaria patients would be expected to increase gradually over time, but malaria incidence should not exhibit a large inter-annual variation. In contrast, dramatic fluctuations in malaria incidence in the African highlands were observed [##REF##11859368##9##,##REF##10516785##46##,##REF##14983017##47##]. Significant association between climate variability and malaria incidence suggests that climate factors may play an important role in the East African highlands [##REF##14983017##47##]. Further studies are required to examine the interactions between climate factors and drug resistance on malaria incidence in African highlands.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Since the late 1980s a series of malaria epidemics has occurred in western Kenya highlands. Among the possible factors that may contribute to the highland malaria epidemics, parasite resistance to antimalarials has not been well investigated.</p>", "<title>Methods</title>", "<p>Using parasites from highland and lowland areas of western Kenya, we examined key mutations associated with <italic>Plasmodium falciparum </italic>resistance to sulfadoxine – pyrimethamine and chloroquine, including dihydrofolate reductase (<italic>pfdhfr</italic>) and dihydropteroate synthetase (<italic>pfdhps</italic>), chloroquine resistance transporter gene (<italic>pfcrt</italic>), and multi-drug resistance gene 1 (<italic>pfmdr1</italic>).</p>", "<title>Results</title>", "<p>We found that &gt;70% of samples harbored 76T <italic>pfcrt </italic>mutations and over 80% of samples harbored quintuple mutations (51I/59R/108N <italic>pfdhfr </italic>and 437G/540E <italic>pfdhps</italic>) in both highland and lowland samples. Further, we did not detect significant difference in the frequencies of these mutations between symptomatic and asymptomatic malaria volunteers, and between highland and lowland samples.</p>", "<title>Conclusion</title>", "<p>These findings suggest that drug resistance of malaria parasites in the highlands could be contributed by the mutations and their high frequencies as found in the lowland. The results are discussed in terms of the role of drug resistance as a driving force for malaria outbreaks in the highlands.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>DZ: Participated in the design of the study, conducted data collection, statistical analysis and drafting of the manuscript. YA: Conducted sample collection and helped with writing the manuscript. AG: Faciliated and conducted field sample collection. LC: Participated in the design of the study, interpretation of data and revising the manuscript. DM: Participated in experimental work and revising of the manuscript. GY: Conceived the study, participated in collection of samples, and participated in manuscript preparation. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2334/8/105/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>The authors thank Hong Chen, Emmanuel A. Temu, Thomas Loescher, David Warhurst and two anonymous reviewers for constructive comments on the manuscript. The work was supported by NIH grants R01 A150243, D43 TW01505, and R03 TW007360.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Frequency of mutations in <italic>Plasmodium falciparum </italic>chloroquine resistance transporter gene (<italic>pfcrt</italic>) (A and B), multi-drug resistance gene (<italic>pfmdr1</italic>) (C and D) and quintuple mutations in dihydrofolate reductase (<italic>pfdhfr</italic>)/dihydropteroate synthetase (<italic>pfdhps</italic>) (E and F) genes in symptomatic and asymptomatic volunteers in highland (Kakamega and Kisii) and lowland (Kombewa) sites of western Kenya. The <italic>pfdhfr/pfdhps </italic>quintuple mutation refers to 51<underline>I</underline>/59<underline>R</underline>/108<underline>N</underline>/437<underline>G</underline>/540<underline>E</underline>. No mutations were detected at the following codons: <italic>pfdhfr </italic>(A16V, C50R, I164L and S436A), <italic>pfdhps </italic>(A581G and A613S), and <italic>pfmdr1 </italic>(S1034 and N1042D).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Frequencies of mutations in genes associated with resistance to chloroquine, sulfadoxine-pyrimethamine in <italic>P. falciparum </italic>parasites from symptomatic and asymptomatic volunteers in western Kenya.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\">Gene</td><td align=\"center\">Mutation</td><td align=\"center\">Polymorphism</td><td align=\"center\" colspan=\"2\">Kombewa</td><td align=\"center\" colspan=\"2\">Kakamega</td><td align=\"center\" colspan=\"2\">Kisii</td></tr><tr><td/><td/><td/><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td></tr><tr><td/><td/><td/><td align=\"center\">S*</td><td align=\"center\">A*</td><td align=\"center\">S</td><td align=\"center\">A</td><td align=\"center\">S</td><td align=\"center\">A</td></tr></thead><tbody><tr><td align=\"center\"><italic>pfcrt</italic></td><td align=\"center\">K76T</td><td align=\"center\">Mutant</td><td align=\"center\">87</td><td align=\"center\">81</td><td align=\"center\">86</td><td align=\"center\">77</td><td align=\"center\">75</td><td align=\"center\">70</td></tr><tr><td/><td/><td align=\"center\">Mixed</td><td align=\"center\">7</td><td align=\"center\">15</td><td align=\"center\">6</td><td align=\"center\">9</td><td align=\"center\">8</td><td align=\"center\">10</td></tr><tr><td/><td/><td align=\"center\">Wild-type</td><td align=\"center\">6</td><td align=\"center\">4</td><td align=\"center\">8</td><td align=\"center\">14</td><td align=\"center\">17</td><td align=\"center\">20</td></tr><tr><td align=\"center\"><italic>Pfmdr1</italic></td><td align=\"center\">N86Y</td><td align=\"center\">Mutant</td><td align=\"center\">28</td><td align=\"center\">41</td><td align=\"center\">28</td><td align=\"center\">39</td><td align=\"center\">10</td><td align=\"center\">23</td></tr><tr><td/><td/><td align=\"center\">Mixed</td><td align=\"center\">44</td><td align=\"center\">22</td><td align=\"center\">45</td><td align=\"center\">37</td><td align=\"center\">40</td><td align=\"center\">18</td></tr><tr><td/><td/><td align=\"center\">Wild-type</td><td align=\"center\">28</td><td align=\"center\">37</td><td align=\"center\">27</td><td align=\"center\">24</td><td align=\"center\">50</td><td align=\"center\">59</td></tr><tr><td/><td align=\"center\">Y184F</td><td align=\"center\">Mutant</td><td align=\"center\">24</td><td align=\"center\">15</td><td align=\"center\">7</td><td align=\"center\">5</td><td align=\"center\">11</td><td align=\"center\">26</td></tr><tr><td/><td/><td align=\"center\">Mixed</td><td align=\"center\">26</td><td align=\"center\">30</td><td align=\"center\">23</td><td align=\"center\">26</td><td align=\"center\">33</td><td align=\"center\">19</td></tr><tr><td/><td/><td align=\"center\">Wild-type</td><td align=\"center\">50</td><td align=\"center\">55</td><td align=\"center\">70</td><td align=\"center\">69</td><td align=\"center\">56</td><td align=\"center\">55</td></tr><tr><td/><td align=\"center\">D1246Y</td><td align=\"center\">Mutant</td><td align=\"center\">24</td><td align=\"center\">37</td><td align=\"center\">40</td><td align=\"center\">28</td><td align=\"center\">42</td><td align=\"center\">24</td></tr><tr><td/><td/><td align=\"center\">Mixed</td><td align=\"center\">42</td><td align=\"center\">30</td><td align=\"center\">36</td><td align=\"center\">44</td><td align=\"center\">21</td><td align=\"center\">29</td></tr><tr><td/><td/><td align=\"center\">Wild-type</td><td align=\"center\">34</td><td align=\"center\">33</td><td align=\"center\">24</td><td align=\"center\">28</td><td align=\"center\">37</td><td align=\"center\">47</td></tr><tr><td align=\"center\"><italic>pfdhfr</italic></td><td align=\"center\">N51I</td><td align=\"center\">Mutant</td><td align=\"center\">98</td><td align=\"center\">96</td><td align=\"center\">98</td><td align=\"center\">98</td><td align=\"center\">96</td><td align=\"center\">95</td></tr><tr><td/><td/><td align=\"center\">Mixed</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td></tr><tr><td/><td/><td align=\"center\">Wild-type</td><td align=\"center\">2</td><td align=\"center\">4</td><td align=\"center\">2</td><td align=\"center\">2</td><td align=\"center\">4</td><td align=\"center\">5</td></tr><tr><td/><td align=\"center\">C59R</td><td align=\"center\">Mutant</td><td align=\"center\">81</td><td align=\"center\">75</td><td align=\"center\">88</td><td align=\"center\">79</td><td align=\"center\">90</td><td align=\"center\">86</td></tr><tr><td/><td/><td align=\"center\">Mixed</td><td align=\"center\">4</td><td align=\"center\">10</td><td align=\"center\">2</td><td align=\"center\">7</td><td align=\"center\">2</td><td align=\"center\">0</td></tr><tr><td/><td/><td align=\"center\">Wild-type</td><td align=\"center\">15</td><td align=\"center\">15</td><td align=\"center\">10</td><td align=\"center\">14</td><td align=\"center\">8</td><td align=\"center\">14</td></tr><tr><td/><td align=\"center\">S108N</td><td align=\"center\">Mutant</td><td align=\"center\">100</td><td align=\"center\">100</td><td align=\"center\">100</td><td align=\"center\">100</td><td align=\"center\">100</td><td align=\"center\">100</td></tr><tr><td/><td/><td align=\"center\">Mixed</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td></tr><tr><td/><td/><td align=\"center\">Wild-type</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td></tr><tr><td align=\"center\"><italic>pfdhps</italic></td><td align=\"center\">S436F</td><td align=\"center\">Mutant</td><td align=\"center\">4</td><td align=\"center\">2</td><td align=\"center\">4</td><td align=\"center\">4</td><td align=\"center\">2</td><td align=\"center\">2</td></tr><tr><td/><td/><td align=\"center\">Mixed</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0</td></tr><tr><td/><td/><td align=\"center\">Wild-type</td><td align=\"center\">96</td><td align=\"center\">98</td><td align=\"center\">96</td><td align=\"center\">96</td><td align=\"center\">98</td><td align=\"center\">98</td></tr><tr><td/><td align=\"center\">A437G</td><td align=\"center\">Mutant</td><td align=\"center\">94</td><td align=\"center\">95</td><td align=\"center\">92</td><td align=\"center\">93</td><td align=\"center\">88</td><td align=\"center\">93</td></tr><tr><td/><td/><td align=\"center\">Mixed</td><td align=\"center\">4</td><td align=\"center\">5</td><td align=\"center\">4</td><td align=\"center\">7</td><td align=\"center\">2</td><td align=\"center\">0</td></tr><tr><td/><td/><td align=\"center\">Wild-type</td><td align=\"center\">2</td><td align=\"center\">0</td><td align=\"center\">4</td><td align=\"center\">0</td><td align=\"center\">10</td><td align=\"center\">7</td></tr><tr><td/><td align=\"center\">K540E</td><td align=\"center\">Mutant</td><td align=\"center\">88</td><td align=\"center\">95</td><td align=\"center\">92</td><td align=\"center\">91</td><td align=\"center\">88</td><td align=\"center\">93</td></tr><tr><td/><td/><td align=\"center\">Mixed</td><td align=\"center\">10</td><td align=\"center\">5</td><td align=\"center\">4</td><td align=\"center\">9</td><td align=\"center\">2</td><td align=\"center\">0</td></tr><tr><td/><td/><td align=\"center\">Wild-type</td><td align=\"center\">2</td><td align=\"center\">0</td><td align=\"center\">4</td><td align=\"center\">0</td><td align=\"center\">10</td><td align=\"center\">7</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Percentage of haplotypes of <italic>P. falciparum </italic>parasites collected from symptomatic and asymptomatic volunteers in western Kenya.</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\">Haplotypes</td><td align=\"center\" colspan=\"2\">Kombewa</td><td align=\"center\" colspan=\"2\">Kakamega</td><td align=\"center\" colspan=\"2\">Kisis</td></tr><tr><td/><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td></tr><tr><td/><td align=\"center\">S*</td><td align=\"center\">A*</td><td align=\"center\">S</td><td align=\"center\">A</td><td align=\"center\">S</td><td align=\"center\">A</td></tr><tr><td align=\"left\"><italic>Pfmdr1</italic></td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  86<underline>Y</underline>/184Y/1246<underline>Y</underline></td><td align=\"center\">24</td><td align=\"center\">26</td><td align=\"center\">44</td><td align=\"center\">46</td><td align=\"center\">38</td><td align=\"center\">36</td></tr><tr><td align=\"left\">  86<underline>Y</underline>/184<underline>F</underline>/1246<underline>Y</underline></td><td align=\"center\">32</td><td align=\"center\">34</td><td align=\"center\">24</td><td align=\"center\">23</td><td align=\"center\">16</td><td align=\"center\">13</td></tr><tr><td align=\"left\">  86N/184<underline>F</underline>/1246D</td><td align=\"center\">12</td><td align=\"center\">15</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">25</td><td align=\"center\">27</td></tr><tr><td align=\"left\">  86<underline>Y</underline>/184<underline>F</underline>/1246D</td><td align=\"center\">18</td><td align=\"center\">10</td><td align=\"center\">0</td><td align=\"center\">5</td><td align=\"center\">1</td><td align=\"center\">0</td></tr><tr><td align=\"left\">  86<underline>Y</underline>/184Y/1246D</td><td align=\"center\">2</td><td align=\"center\">0</td><td align=\"center\">16</td><td align=\"center\">10</td><td align=\"center\">1</td><td align=\"center\">3</td></tr><tr><td align=\"left\">  86N/184Y/1246D</td><td align=\"center\">6</td><td align=\"center\">5</td><td align=\"center\">8</td><td align=\"center\">10</td><td align=\"center\">11</td><td align=\"center\">12</td></tr><tr><td align=\"left\">  86N/184<underline>F</underline>/1246<underline>Y</underline></td><td align=\"center\">3</td><td align=\"center\">5</td><td align=\"center\">4</td><td align=\"center\">3</td><td align=\"center\">8</td><td align=\"center\">5</td></tr><tr><td align=\"left\">  86N/184Y/1246<underline>Y</underline></td><td align=\"center\">3</td><td align=\"center\">5</td><td align=\"center\">4</td><td align=\"center\">3</td><td align=\"center\">0</td><td align=\"center\">4</td></tr><tr><td/><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><italic>Pfdhfr/pfdhps</italic></td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Quintuple</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  51<underline>I</underline>/59<underline>R</underline>/108<underline>N</underline>/437<underline>G</underline>/540<underline>E</underline></td><td align=\"center\">87</td><td align=\"center\">84</td><td align=\"center\">86</td><td align=\"center\">83</td><td align=\"center\">85</td><td align=\"center\">80</td></tr><tr><td align=\"left\"> Quadruple</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  51<underline>I</underline>/59C/108<underline>N</underline>/437<underline>G</underline>/540<underline>E</underline></td><td align=\"center\">7</td><td align=\"center\">5</td><td align=\"center\">3</td><td align=\"center\">8</td><td align=\"center\">4</td><td align=\"center\">4</td></tr><tr><td align=\"left\">  51N/59<underline>R</underline>/108<underline>N</underline>/437<underline>G</underline>/540<underline>E</underline></td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">4</td><td align=\"center\">3</td><td align=\"center\">2</td><td align=\"center\">6</td></tr><tr><td align=\"left\">  51<underline>I</underline>/59<underline>R</underline>/108<underline>N</underline>/437<underline>G</underline>/540K</td><td align=\"center\">1</td><td align=\"center\">5</td><td align=\"center\">1</td><td align=\"center\">2</td><td align=\"center\">1</td><td align=\"center\">2</td></tr><tr><td align=\"left\">  51<underline>I</underline>/59<underline>R</underline>/108<underline>N</underline>/437A/540<underline>E</underline></td><td align=\"center\">1</td><td align=\"center\">1</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">3</td><td align=\"center\">4</td></tr><tr><td align=\"left\"> Triple</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  51<underline>I</underline>/59<underline>R</underline>/108<underline>N</underline>/437A/540K</td><td align=\"center\">2</td><td align=\"center\">3</td><td align=\"center\">2</td><td align=\"center\">2</td><td align=\"center\">5</td><td align=\"center\">4</td></tr><tr><td align=\"left\"> Double</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  51<underline>I</underline>/59C/108<underline>N</underline>/437A/540K</td><td align=\"center\">2</td><td align=\"center\">2</td><td align=\"center\">4</td><td align=\"center\">2</td><td align=\"center\">0</td><td align=\"center\">0</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>* S – symptomatic; A – asymptomatic.</p><p>Note: No mutations were detected at the following codons: <italic>pfdhfr </italic>(A16V, C50R and I164L); <italic>pfdhps </italic>(S436A, A581G and A613S); and <italic>pfmdr1 </italic>(S1034 and N1042D).</p></table-wrap-foot>", "<table-wrap-foot><p>* S – symptomatic; A – asymptomatic; The mutation amino acids are underlined.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2334-8-105-1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>In the USA, more people die from lung cancer than any other type of cancer [##UREF##0##1##]. This is true for both men and women. In 2004, lung cancer accounted for more deaths than breast cancer, prostate cancer, and colon cancer combined [##UREF##1##2##].</p>", "<p>Lung cancer can metastasize to virtually any bone, although the axial skeleton and proximal long bones are most commonly involved [##REF##12527569##3##]. The primary symptom resulting from bone involvement is pain, which may have a pleuritic component when the ribs are involved. Bone pain is present in up to 25% of all patients at presentation [##REF##12527569##3##].</p>", "<p>Patients commonly seek chiropractic care with musculoskeletal complaints [##REF##10714540##4##,##REF##12391720##5##]. Through history and examination, chiropractic physicians have an opportunity to assess patients and determine whether serious conditions are present that may necessitate medical referrals.</p>", "<p>Patients with previously identified or yet to be identified cancer may seek care with chiropractic physicians. This case report demonstrates previously undiagnosed lung cancer with widespread metastatic foci.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<title>Chiropractic considerations</title>", "<p>The identification of primary or secondary metastatic cancer requires careful consideration with regard to history and physical examination. A key objective for the chiropractic physician is to identify \"red flags\" as quickly as possible. This is especially true for any disease process that may weaken bone.</p>", "<p>The application of directed force into spinal or osseous structures inherent to the chiropractic adjustment mandate careful evaluative procedure. Janse defined the adjustment as a specific form of articular manipulation using long or short lever techniques with specific contacts and is characterized by a dynamic thrust of controlled velocity, amplitude and direction [##UREF##2##6##].</p>", "<p>While chiropractic physicians are challenged with the responsibility of attempting to identify relative and absolute contraindications to spinal adjustments, sometimes early onset, insidious and seemingly innocuous symptoms may delay early identification [##REF##1402413##7##,##UREF##3##8##].</p>", "<title>Clinical considerations</title>", "<p>When primary cancer is not yet identified, metastatic extension to skeletal structures can at times be difficult to detect [##REF##1402413##7##,##UREF##3##8##]. As was illustrated in this case, clinical considerations that may assist or delay the identification of metastatic bone disease include:</p>", "<p>1. Early in the course of the disease progression, important red flag identifiers may not initially be present and can delay early identification.</p>", "<p>2. Initial pain presentations may be suggestive of common clinical conditions that are less aggressive.</p>", "<p>3. Patients may or not be aware of, or report, the existence of a primary cancer.</p>", "<p>4. Pain can be initially mild to severe and is often progressive in nature and unremitting despite therapeutic interventions.</p>", "<p>5. It is sometimes extremely difficult to positively identify metastatic disease due to complex clinical factors [##REF##1402413##7##,##UREF##3##8##].</p>", "<p>Red flag indicators for metastatic bone disease include: age over 50 or under 20 years, a history of cancer, constitutional symptoms including unexplained weight loss, pain worse at night or in atypical areas, no significant improvement after &gt; 1 month of conservative (non-invasive) care, pain that has no mechanical exacerbating or remitting factors, and severe disabling pain affecting a child or adolescent [##UREF##4##9##].</p>", "<title>Diagnostic imaging considerations</title>", "<p>Humphrey reported that about 25% of people with lung cancer do not have symptoms from advanced cancer when their lung cancer is found [##REF##15126259##10##]. Maghfoor reported that 7–10% of patients with lung cancer are asymptomatic and their cancers are diagnosed incidentally after a CXR was performed for other reasons [##UREF##5##11##]. Numerous studies have shown that the chest radiograph lacks sensitivity in detecting mediastinal lymph node metastases and in detecting chest wall and mediastinal invasion [##REF##8284352##12##].</p>", "<p>CT has become the major imaging modality of choice in the evaluation of patients with bronchogenic carcinoma [##UREF##6##13##]. Traditionally, chest CT for staging of lung cancer is extended into the abdomen to include the adrenal glands. Whether this requires intravenous contrast material is debatable [##UREF##6##13##]. Patz et al. [##REF##10405720##14##] concluded that contrast-enhanced CT extended to include the liver rarely adds to routine nonenhanced CT through the adrenal glands and does not influence management decisions.</p>", "<p>The evaluation of the mediastinum with magnetic resonance imaging (MRI) is approximately equal to that of CT with regard to the staging of bronchogenic carcinoma and MRI is significantly more accurate for detecting direct mediastinal invasion [##REF##1847239##15##]. Other studies have confirmed the usefulness of MRI, particularly in the evaluation of chest wall invasion and the local staging of superior sulcus tumors [##REF##2916014##16##,##REF##8451432##17##]. The general conclusion of these studies is that MRI has advantages in the assessment of both chest wall and mediastinal invasion [##UREF##6##13##].</p>", "<p>Indications for the use of whole body positron emission tomography imaging in lung cancer using 18-fluorodeoxyglucose (FDG-PET) in patients with non-small cell lung cancer include high clinical index of suspicion of high grade malignancy and radiographic evidence of nodal enlargement [##UREF##6##13##]. In addition, PET scans may be helpful in centers where mediastinoscopy is not readily available and in patients with significant comorbid conditions who are borderline candidates for surgery, with locally advanced disease, solitary brain metastasis, and cases of local recurrence that might qualify for reoperation [##REF##9768946##18##,##REF##15189955##19##].</p>", "<p>Bone scintigraphy in the detection of metastatic disease has significant limitations. Although it has high sensitivity, it is noted for having very low specificity that ranges from 50%–60% [##UREF##6##13##]. Bone scintigraphy should probably be limited to cases in which patients have specified clinical indicators of bone metastasis [##REF##1652164##20##].</p>", "<p>When evaluating suspected pulmonary metastasis, CXR and CT of the chest are rated by the American College of Radiology (ACR) scale as: \"9 – most appropriate\" (Rating Scale: 1-Least appropriate, 9-Most appropriate) [##UREF##7##21##]. It is generally accepted that chest radiography, with posteroanterior (PA) and lateral views, should be the initial imaging test in patients without known or suspected thoracic metastatic disease [##REF##2052672##22##, ####REF##9515171##23##, ##REF##8934121##24####8934121##24##]. Compared with chest radiography, CT is much more sensitive for detecting pulmonary nodules, because of its lack of superimposition and its high contrast resolution [##REF##2052672##22##, ####REF##9515171##23##, ##REF##8934121##24####8934121##24##].</p>" ]

[ "<title>Conclusion</title>", "<p>Lung cancer is a significant and aggressive primary cancer with a predilection for skeletal metastasis. When primary lung cancer is not previously identified, metastatic disease to skeletal structures may initially manifest as musculoskeletal complaints. Careful diagnostic evaluation and decision making may allow for earlier diagnosis.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>The objective of this case report is to describe the clinical presentation of a patient who complained of shoulder pain and was diagnosed with carcinoma of the scapula and spine that metastasized from the lung.</p>", "<title>Case presentation</title>", "<p>A 76-year-old man without a history of cancer sought chiropractic care for right shoulder pain. Careful evaluation, radiographs, and subsequent imaging revealed primary and metastatic lung cancer. The patient was referred to his primary care physician for immediate medical care. Diagnostic images are included in this case to provide a comprehensive depiction of the scope of the patient's disease.</p>", "<title>Conclusion</title>", "<p>Musculoskeletal symptoms are commonly encountered in chiropractic practice. It is important to recognize that primary lung cancer may be unidentified, and musculoskeletal symptoms may reflect the first sign of primary or metastatic pulmonary disease. Thoughtful evaluative procedure and clinical decision making, combined with the use of appropriate diagnostic tests may allow timely identification of primary or metastatic disease.</p>" ]

[ "<title>Case presentation</title>", "<title>Case report</title>", "<p>A 76-year-old male sought chiropractic care for complaints of right shoulder pain and mild right arm weakness. The onset of pain was insidious and of one week's duration. Pain was rated 8/10 on a visual analogue scale (0 = no pain, 10 = the worst pain of one's life). The pain was described as severe and worsened with movement. Additional symptoms included mild shortness of breath and posterior thoracic pain on respiration.</p>", "<p>The patient's past medical history included headache, degenerative joint disease affecting the cervical spine, and a benign thyroid nodule. The patient reportedly smoked tobacco products for 50 years. He was a retired electrician.</p>", "<p>The patient was afebrile. Vital signs were normal. Respirations were 18 cycles per minute. The lungs were clear to auscultation. The patient reported upper thoracic pain on inspiration.</p>", "<p>A non-tender, mild decrease in active range of motion of the cervical spine was noted in all planes. No tenderness was elicited on palpation of the cervical spine. Cervical compression and Soto-Hall tests were negative. Valsalva maneuver was negative. Neurologic examination revealed no focal deficits.</p>", "<p>Examination of the right shoulder revealed exquisite tenderness on palpation of the lateral border of the scapula with muscle spasm affecting the ipsilateral infraspinatus, teres major, and teres minor muscles. Active ranges of shoulder motion were restricted and painful in abduction, internal, and external rotation.</p>", "<p>Plain film radiographs of the right shoulder (AP with internal and external rotation views) and thoracic spine (AP and lateral views) were performed. Disruption of the cortical margin of the lateral border of the right scapula was noted as evidenced by an indistinct lucency (see Figure ##FIG##0##1##). In addition, a suspicious mass was noted in the hilar region of the right lung. Complete loss of the right hilar vascular detail secondary to the tumor mass effect were noted with visualized subsegmental infiltrate densities. No evidence of pleural effusion was noted.</p>", "<p>The initial diagnostic impression included: suspicious right lung pathology and apparent lytic process affecting the scapula of an unknown origin. The patient was referred for imaging evaluations that included chest x-ray (CXR) and computed tomographic (CT) evaluation of the chest. He was referred to his primary care medical physician.</p>", "<p>The CXR and CT examination of the chest, abdomen and pelvis revealed:</p>", "<p>1. A large mass in the right upper lobe of the lung with associated mediastinal and hilar adenopathy (see Figures ##FIG##1##2## and ##FIG##2##3##).</p>", "<p>2. Metastatic disease of the scapula (see Figure ##FIG##3##4##).</p>", "<p>3. Metastatic liver disease.</p>", "<p>Subsequent bone scintigraphy revealed abnormal increased accumulation of radiopharmaceutical along the lateral aspect of the right scapula (see Figure ##FIG##4##5##). MRI evaluation revealed additional metastatic foci including the cervical, thoracic and lumbar spinal regions as evidenced by multiple regions of decreased signal intensity are visualized on T1 weighted images (see Figures ##FIG##5##6## and ##FIG##6##7##). Biopsy confirmed a primary lung carcinoma origin. Unfortunately, the patient succumbed to the disease within 3 months of its diagnosis.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>JD conceived the study and drafted the manuscript. GJD participated in the care of the patient and provided data related to the case. Both authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>Written informed consent was obtained from the decedent's wife for publication of this case report and accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.</p>", "<p>The authors wish to thank Anthony V. D'Antoni, DC, MS, PhD(c) and Steven Yeomans, DC, FACO for their thorough editorial assistance.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>AP radiograph of the right scapula reveals a focal indistinct lucency and lytic destruction of the lateral scapular cortical margin</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>PA chest radiograph reveals a right hilar mass</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>CT of the chest reveals a large mass in the right upper lobe of the lung with associated mediastinal and hilar adenopathy.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>CT of the chest reveals cortical lucency, expansile destruction, and medullary invasion due to metastatic lung carcinoma affecting the right scapula</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>Bone scintigraphy of the right scapula reveals increased uptake where metastatic lung carcinoma is present</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p>MRI sagittal T1WI reveals scattered foci of decreased signal intensity reflective of metastatic disease affecting the cervical and thoracic spine regions.</p></caption></fig>", "<fig position=\"float\" id=\"F7\"><label>Figure 7</label><caption><p><bold>MRI sagittal T1WI reveals scattered foci of decreased signal intensity reflective of metastatic disease affecting the thoraco-lumbar spine</bold>.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1746-1340-16-8-1\"/>", "<graphic xlink:href=\"1746-1340-16-8-2\"/>", "<graphic xlink:href=\"1746-1340-16-8-3\"/>", "<graphic xlink:href=\"1746-1340-16-8-4\"/>", "<graphic xlink:href=\"1746-1340-16-8-5\"/>", "<graphic xlink:href=\"1746-1340-16-8-6\"/>", "<graphic xlink:href=\"1746-1340-16-8-7\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[]

[]

[]

[]

[ "<p>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.</p>", "<title>Background</title>", "<p>Chromosome location is often used as a scaffold to organize genomic information in both the living cell and molecular biological research. Thus, ever-increasing amounts of data about genomic features are stored in public databases and can be readily visualized by genome browsers. To perform <italic>in silico </italic>experimentation conveniently with this genomics data, biologists need tools to process and compare datasets routinely and explore the obtained results interactively. The complexity of such experimentation requires these tools to be based on an e-Science approach, hence generic, modular, and reusable. A virtual laboratory environment with workflows, workflow management systems, and Grid computation are therefore essential.</p>", "<title>Findings</title>", "<p>Here we apply an e-Science approach to develop SigWin-detector, a workflow-based tool that can detect significantly enriched windows of (genomic) features in a (DNA) sequence in a fast and reproducible way. For proof-of-principle, we utilize a biological use case to detect regions of increased and decreased gene expression (RIDGEs and anti-RIDGEs) in human transcriptome maps. We improved the original method for RIDGE detection by replacing the costly step of estimation by random sampling with a faster analytical formula for computing the distribution of the null hypothesis being tested and by developing a new algorithm for computing moving medians. SigWin-detector was developed using the WS-VLAM workflow management system and consists of several reusable modules that are linked together in a basic workflow. The configuration of this basic workflow can be adapted to satisfy the requirements of the specific <italic>in silico </italic>experiment.</p>", "<title>Conclusion</title>", "<p>As we show with the results from analyses in the biological use case on RIDGEs, SigWin-detector is an efficient and reusable Grid-based tool for discovering windows enriched for features of a particular type in any sequence of values. Thus, SigWin-detector provides the proof-of-principle for the modular e-Science based concept of integrative bioinformatics experimentation.</p>" ]

[ "<title>Findings</title>", "<p>Genomic information is encoded in DNA and as such retained in a fairly steady configuration. In contrast to RNA, proteins, and metabolites, DNA is organized by a limited number of large chromosomes with relatively stable DNA sequences. Therefore, position in the DNA sequence, i.e., chromosome location, provides a convenient and essential scaffold for both the living cell and molecular biological research. In cells, for example, chromosomal organization is important for gene-transcription processes. Expression-profiling studies showed that gene expression is not only controlled at the level of individual genes, but also via autonomous regulation of chromosomal domains [##REF##12144710##1##, ####REF##12478293##2##, ##REF##12214599##3##, ##REF##12915492##4##, ##REF##16791208##5####16791208##5##]. This suggests the existence of higher-order transcriptional regulatory mechanisms related to DNA organization or structures. The use of chromosomal organization in the life sciences is exemplified by the popularity of genome browsers that use chromosome location to map many genomic features, such as genes and their products, regulatory elements, gene expression, and epigenetic markers. The search for connections between genomic features is important in unraveling cellular mechanisms.</p>", "<p>The pace at which omics experiments continuously keep producing large amounts of data about genomic features for an increasing number of sequenced genomes, creates a need for new high-throughput methods for identification of correlations between DNA related features [##REF##17571346##6##, ####REF##17072317##7##, ##REF##15920525##8##, ##REF##10471496##9##, ##REF##11904358##10##, ##REF##15388519##11##, ##REF##16859498##12####16859498##12##]. Therefore, biologists would benefit from tools that could quickly identify enriched regions of genomic features. This would allow extensive, yet convenient <italic>in silico </italic>experimentation based on routinely processing and comparing multiple datasets. However, this requires these tools to be implemented in such a way that they deal with the many steps involved in this kind of experimentation. These include: acquiring the data from local or remote data repositories, converting it to the desired format, using it with the actual application that searches for the desired enrichment (possibly using Grid computation), visualizing the results, and comparing and/or integrating multiple datasets. Therefore, such a tool should be developed applying an e-Science approach [##REF##16580600##13##, ####REF##18628864##14##, ##UREF##0##15##, ##UREF##1##16##, ##REF##17881406##17####17881406##17##]: it should be generic with respect to which data it can analyze, easy to adapt, and its parts should be reusable.</p>", "<p>In an e-Science approach, a computational environment that provides transparent access to distributed data, adequate computational resources, as well as the necessary interfacing tools, is called a <italic>virtual laboratory </italic>(<italic>VL</italic>). <italic>Workflow management systems </italic>(<italic>WMSs</italic>, [##UREF##2##18##, ####REF##16845108##19##, ##UREF##3##20##, ##UREF##4##21####4##21##]) are an example of interfacing tooling that takes care of scheduling, keeps track of task executions, and provides the management framework necessary to develop applications inside a VL. WMSs can be used to design scientific workflows that automate <italic>in silico </italic>experimentation by providing a pipeline for streaming large quantities of data through various algorithms, applications and services.</p>", "<p>This paper describes an e-Science based data integration and analysis tool: SigWin-detector. This application can detect clusters with increased (or decreased) density of a genomic feature in a DNA-related sequence in a fast and reproducible way. In the context of the development of a VL, our tool was implemented as a workflow running under WS-VLAM[##UREF##3##20##,##UREF##4##21##], a Grid-enabled WMS. A biological use case shows its relevance for biological research. SigWin-detector is based on a method previously used by Versteeg and coworkers [##REF##12915492##4##] to detect <italic>regions of increased and decreased gene expression </italic>(<italic>RIDGEs and anti-RIDGES</italic>) in human transcriptome maps (HTM). We improved the original method by i) deriving an analytical formula for computing the new hypothesis probability distribution, which replaces the costly step of estimation by random sampling and ii) developing a new algorithm for computing moving medians. While these improvements radically increase the intrinsic efficiency of the method, implementing SigWin-detector using a generic e-Science approach with access to Grid resources broadens its applicability and makes it amenable to a wide spectrum of experiments on genomic features or in fact on any sequence of values.</p>", "<title>Significant windows and the mmFDR procedure</title>", "<p>Versteeg et al. [##REF##12915492##4##] identified clusters where the median expression level of the genes involved is significantly higher than expected (RIDGEs), using a <italic>moving median false discovery rate (mmFDR) </italic>procedure (Figure ##FIG##0##1##). The mmFDR procedure identifies RIDGEs by testing the input gene-expression against the null hypothesis that the position of the genes on the chromosomes does not affect their expression levels. This same procedure can be used to identify <italic>significant windows </italic>(i.e., windows in the input sequence that have a median value that deviates significantly from expected, if assumed that the ordering of the numbers in the input sequence is random) related to any genomic feature mapped to DNA sequences. In an even wider scope, it can also be used to identify significant windows in any sequence of numbers.</p>", "<title>Avoiding permutations in the mmFDR procedure</title>", "<p>Computationally, the most expensive step in the original mmFDR procedure is the repeated determination of medians over sliding windows of permutations of the input data to estimate the probability function corresponding to the null hypothesis. Our first improvement to the original method was to derive an exact formula for this distribution (see definitions and derivation in Additional file ##SUPPL##0##1##):</p>", "<p></p>", "<p>This exact formula reduces the number of cycles of computing moving medians of an input sequence of approximately 25,000 entries from at least 5,000 to 1, giving SigWin-detector the efficiency it needs to be used routinely and for processing and comparing multiple datasets within minutes to hours, instead of days. This efficiency could not be if <italic>f</italic>(<italic>m</italic>) was estimated by sampling the permutation space <bold>E</bold>_{<italic>π</italic>}, and counting the number of times <italic>m </italic>was the median value in any sliding window of size <italic>S</italic>.</p>", "<title>Speeding up the computation of moving medians</title>", "<p>Although we removed the need for computing moving medians over permutations of the input sequence, we still need to compute medians of windows sliding over the input sequence. We developed a new algorithm to compute those moving medians efficiently by exploiting the fact that moving medians for many window sizes must be computed simultaneously (Figure ##FIG##1##2##). This new algorithm is also suitable for computing any other order-statistics.</p>", "<p>Additional Figure A1 (Additional file ##SUPPL##1##2##) shows a graph comparing our moving medians algorithm with the commonly used Hardle and Steiger's algorithm [##UREF##5##22##]. While the execution time of their algorithm increases with window size (for a fixed sequence size), the execution time of our algorithm decreases with window size (Figure A1, upper panel). Because SigWin-detector needs to compute moving medians for many window sizes, our algorithm has a clear advantage over Hardle and Steiger's algorithm. In Figure A1, the break-even point of the cumulative computation is for <italic>S</italic>_{<italic>max </italic>}around 400. The efficiency of our method can be further improved by using a mixed algorithm that uses Hardle and Steiger's algorithm for small window sizes and our algorithm for large window sizes, or by employing a divide-and-conquer approach. For example, a two-phase algorithm would start by dividing the input sequence into chunks of size 2<italic>M</italic>, with <italic>M </italic>≥ 2<italic>S</italic>_{<italic>max</italic>}, and applying the original algorithm to each chunk separately. Similarly, the second phase computes the medians for the missing sliding windows by dividing the sequence into chunks of the same size, but now using an offset <italic>M</italic>. This two-phase algorithm is also suitable for parallelization.</p>", "<title>Designing a Grid-enabled generic workflow</title>", "<p>To broaden the applicability of the mmFDR procedure, we implemented SigWin-detector using an e-Science approach by implementing a general, reusable, and adaptable tool with access to Grid resources using the WS-VLAM workflow management system[##UREF##3##20##,##UREF##4##21##].</p>", "<p>First we split the procedure into a collection of workflow components (called modules), each module performing a specific task that may be fine-tuned using parameters. The modules exchange data with each other by means of input and output ports. We then can choose the appropriate modules and compose a workflow suited to our specific needs [##UREF##1##16##]. Figure ##FIG##2##3## describes a basic workflow configuration of SigWin-detector.</p>", "<p>The SigWin-detector Config-Basic1 workflow was tested on a Grid computer cluster composed of geographically distributed computational nodes: <italic>Distributed ASCI Supercomputer 3 </italic>(<italic>DAS-3</italic>, [##UREF##6##23##]). Additional Figure A2 (Additional file ##SUPPL##1##2##) presents wall clock execution times of the SigWin-detector Config-Basic1 workflow (Figure ##FIG##2##3##) for input sequences of various sizes.</p>", "<p>The basic workflow can be altered by substituting, deleting, or adding modules. For example, we can extend the workflow to get the input sequence from a remote <italic>uniform resource identifier </italic>(<italic>URI</italic>)and then put the resulting SigWin-map back into it. We can modify the workflow to generate one SigWin-map per logical subsequence of the input sequence, instead of a single SigWin-map for the complete sequence [##UREF##1##16##]. We can also expand our workflow by computing significant windows for high median values (e.g., RIDGEs) and significant windows for low median values (e.g., anti-RIDGEs) simultaneously. The SigWin-detector workflow itself can be made into a \"composite module\" for more complex workflows. Furthermore, interconnection of WS-VLAM with the TAVERNA workbench [##REF##16845108##19##] will permit the use of the existing TAVERNA components in connection with SigWin-detector. At the moment, Grid authentication prevents WS-VLAM workflows being used outside the Grid without the extra step of Grid certification. However, we are working on a Taverna workflow that encapsulates the SigWin detector, to be made available through the myExperiment webpage [##UREF##7##24##].</p>", "<title>Biological application: finding RIDGES in a human transcriptome map</title>", "<p>Once we finished our basic SigWin-detector, we modified it (Additional file ##SUPPL##2##3##) for application in our biological use case that aims to find (anti-)RIDGES in transcriptome maps. Figures ##FIG##3##4## and ##FIG##4##5## show a series of RIDGEOGRAMS for gene expression data for a recent version of the human transcriptome map (HTM) based on the UCSC release hg18 [##REF##12915492##4##], and Table ##TAB##0##1## summarizes some RIDGE statistics. Each RIDGEOGRAM displays both RIDGEs (red-shades) and anti-RIDGEs (blue-shades), the different color shades representing different mmFDR threshold levels. The size of the resulting RIDGEOGRAMS is proportional to the number of genes on a chromosome. We determined i) genome-wide (anti-)RIDGEs, i.e., windows for which the median expression is significantly higher (lower) than expected by considering the whole genome gene expression profile in the mmFDR procedure (Figure ##FIG##3##4##), and ii) chromosome specific (anti-)RIDGEs, i.e., the same analysis, but considering only the specific chromosome gene expression profile (Figure ##FIG##4##5##). This distinction has a major effect on the outcome. If the expression values of the genes on a certain chromosome are typically significantly higher than the genome-wide values, then there are less chromosome specific than genome-wide RIDGEs (e.g., chromosome 19 in Figures ##FIG##3##4## and ##FIG##4##5## and Table ##TAB##0##1##). Conversely, if the expression values of the genes on a chromosome are typically significantly smaller than the genome-wide values, then there are more chromosome specific RIDGEs (e.g., chromosome 6 in Table ##TAB##0##1## and Figures ##FIG##3##4## and ##FIG##4##5##). In the case of anti-RIDGEs the opposite holds (e.g., chromosomes 17 in Table ##TAB##0##1## and Figures ##FIG##3##4## and ##FIG##4##5##). This example shows the importance of choosing the right sequence to compute the null hypothesis distribution. Based on the fact that chromosomes are separate molecules in a cell, one may favor the results from the individual chromosome SigWin-detector analysis to investigate potential higher-order gene expression regulatory mechanisms.</p>", "<p>The RIDGEOGRAMS shown in Figures ##FIG##3##4## and ##FIG##4##5## only take the ordering of the genes into account, and not their actual physical position in the chromosome. However, from a biological perspective it is likely that the higher order gene-expression mechanisms that underlie RIDGEs relate to an actual section of the chromosome rather than a cluster of genes just ordered by their chromosome location. So we used our SigWin-detector to take the physical gene position into account by subdividing the chromosomes in stretches of constant value (250 kb). If a stretch contains the beginning of one or more genes, their average expression value is assigned to that stretch of DNA. For this analysis we used the SigWin-detector Config-Sub2 with preprocessed HTM data and adapted parameters. The resulting RIDGEOGRAMS are proportional to the chromosome's size (Additional Figure A3, Additional file ##SUPPL##1##2##). The anti-RIDGEs show a lower cut-off caused by the many 0 values in the HTM. The results from the SigWin-detector analysis using chromosome position are substantially different to those using chromosome ordering. This application demonstrated that SigWin-detector is an e-Science tool that allows convenient in-silico experimentation. To prove that this tool is generic, we used our workflow to examine a simple sequential data set: an extended time series of hourly ground level ozone concentration measurements (Additional file ##SUPPL##3##4##).</p>", "<title>Availability and requirements</title>", "<p>• <bold>Project name</bold>: SigWin-detector</p>", "<p>• <bold>Project home page: </bold><ext-link ext-link-type=\"uri\" xlink:href=\"http://mad-db.science.uva.nl/projects/sigwin/\"/></p>", "<p>• <bold>Programming language: </bold>C++</p>", "<p>• <bold>Other requirements: </bold>SigWin-detector needs the WS-VLAM workflow management system. WS-VLAM has a client distribution and site distribution.</p>", "<p><italic>i. WS-VLAM client distribution: </italic>The WS-VLAM composer, a graphical interface used for creating, modifying, and submitting workflows. Needs Java virtual machine (version1.5 or higher).</p>", "<p><italic>ii. WS-VLAM site distribution: </italic>The WS-VLAM engine, which is needed for running the workflows in a Grid. The WS-VLAM engine needs a GLOBUS GT4 (4.0.3) installation.</p>", "<p>To download these WS-VLAM distributions (Additional file ##SUPPL##4##5##) go to <ext-link ext-link-type=\"uri\" xlink:href=\"http://staff.science.uva.nl/~gvlam/wsvlam/\"/>, click the \"Distributions\" tab and follow the instructions in it.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>MAI carried out the entire research project and wrote the manuscript. MFvB participated in development of the statistical methods. MR was involved in the conceptualization of the analytical formula, in the e-Science approach, and in the coordination of the project. ASZB, DV, and WA worked on the development and support of WS-VLAM. AHvK developed the methods for the genomic mapping of expression data and was involved in the development of the statistical methods. TMB conceived the study, participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>We thank R. Monajemi for assistance with the HTM data sets, R. H. Bisseling for checking the mathematics, L. O. Hertzberger for his constant support, and J. Batson for proofreading the paper. This work was carried out in the context of the Virtual Laboratory e-Science project <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.vl-e.nl\"/> and BioRange program of the Netherlands Bioinformatics Centre (NBIC). VL-e is supported by a BSIK grant from the Dutch Ministry of Education, Culture and Science (OC&amp;W) and the ICT innovation program of the Ministry of Economic Affairs (EZ). BioRange is supported by a BSIK grant through the Netherlands Genomics Initiative (NGI).</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Using a mmFDR method to detect RIDGEs in a human transcriptome map</bold>. Schematic representation of the moving median false discovery rate (mmFDR) procedure identifying regions of high and low density of gene expression (RIDGEs and anti-RIDGEs, respectively) [##REF##12915492##4##]. (A) Input sequence, a human transcriptome map (HTM), i.e., expression values of genes ordered by their chromosome location (cyan; chromosome 6). (B) <italic>mm</italic>(<italic>w</italic>), moving medians of the HTM for a given window size <italic>S</italic>. (C) Determination of the high and low mmFDR thresholds at a given level <italic>α</italic>: The high threshold <italic>m</italic>_{<italic>k </italic>}is the smallest gene expression value for which the , here <italic>f</italic>(<italic>m</italic>) is the theoretical probability distribution of <italic>mm</italic>(<italic>w</italic>), and <italic>g</italic>(<italic>m</italic>) is the observed distribution of <italic>mm</italic>(<italic>w</italic>). (In [##REF##12915492##4##], <italic>f</italic>(<italic>m</italic>) is estimated by simple sampling). Similarly, the low threshold <italic>m</italic>_{<italic>j </italic>}is the largest gene expression value for which . (D) Selection of significant windows in chromosome 6: RIDGEs (in red) all windows for which the median gene expression is higher than or equal to <italic>m</italic>_{<italic>k</italic>}; anti-RIDGEs (in blue) all windows for which the median gene expression is lower than or equal to <italic>m</italic>_{<italic>j</italic>}. (E) Output RIDGEOGRAM of chromosome 6. Each row (y-axis) in the RIDGEOGRAM represents a window size, ranging from <italic>S </italic>= 3 to <italic>S </italic>= <italic>M </italic>(the number of genes on the chromosome). Each column (x-axis) represents a sliding window number, ranging from <italic>w </italic>= S/2 to <italic>w </italic>= <italic>M</italic>-<italic>S</italic>/2 (hence the triangular form). Color is used to mark window medians significantly above (red) or below (blue) the genome-wide median. The scheme shows median expression data for window size <italic>S </italic>= 69 and FDR thresholds level <italic>α </italic>= 5%.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Computing moving medians for many window sizes</bold>. Description of our moving medians algorithm and data structures used. The figure illustrates a computation with input sequence size <italic>N </italic>= 7, and window sizes <italic>S </italic>= 3, 5, 7. (A) <italic>Rank </italic>data structure: used to store the input sequence. The Rank data structure gives access to the input sequence in its original and ranked order. It also allows fetching elements according to their rank. (B) <italic>Marker </italic>data structure: helps navigation through the sliding windows while keeping track of the median (or any other desired order-statistics). The Marker data structure is a Boolean array used to keep track of the elements that are inside a sliding window by means of crossing out the elements that are outside it. It also has a pointer that keeps track of the <italic>i</italic>th remaining element. This pointer is used to track the median. The Marker structure assumes the sequence is in ranked order. For example, if a sliding window of size 3 of a sequence of size 7 contains elements ranked 5, 1, and 6, the corresponding Marker structure has elements ranked 2, 3, 4, and 7 crossed out, and its median pointer points to element ranked 5. (C) Moving median algorithm for window size S. Our algorithm computes the moving medians for window sizes <italic>S </italic>= <italic>Smin</italic>, <italic>Smin+</italic>d<italic>S,..., Smin+n</italic>·d<italic>S</italic>, starting at <italic>S </italic>= <italic>S</italic>_{<italic>min</italic>}. When the last sliding window of size <italic>S </italic>is reached, the algorithm proceeds to the next window size (<italic>S</italic>+d<italic>S</italic>) by inserting the elements that are in the first sliding window of size <italic>S</italic>+d<italic>S </italic>and crossing out the elements that were in the last sliding window of size <italic>S </italic>and setting the new position for the median pointer (which is element <italic>mm</italic>(<italic>S+</italic>d<italic>S</italic>) = (<italic>S</italic>+d<italic>S</italic>+1)/2). The algorithm stops after computing the medians for the largest window size.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>SigWin-Detector basic workflow using the WS-VLAM workflow composer</bold>. Upper: A snapshot of the workflow. Lower: Short description of the functionality of each module, port connections, and output ports. The ports are named by an abbreviation of the module name followed by 'i' or 'o' (input or output respectively) and the port number. Input ports are colored in blue and output ports in red. The ports are numbered in the same order they appear in the workflow.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Genome-wide RIDGES in a human transcriptome map (HTM)</bold>. Genome-wide RIDGEOGRAMS per chromosome for the HTM based on the UCSC release hg18 [##REF##12915492##4##]. The expression levels are mapped to gene number. Each RIDGEOGRAM displays a composite of both RIDGES (red-shades) and anti-RIDGEs (blue-shades) for different mmFDR rate levels: 10% (lighter shade), 5%, 1%, and 0.5% (darker shade). All the different window sizes are depicted because they give different specific results. In general, small windows suffer from noise and large windows suffer from lack of detail.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>Chromosome-specific RIDGES in a human transcriptome map (HTM)</bold>. Chromosome-specific RIDGEOGRAMS per chromosome for the HTM based on the UCSC release hg18 [##REF##12915492##4##]. The expression levels are mapped to gene number. Each RIDGEOGRAM displays a composite of both RIDGES (red-shades) and anti-RIDGEs (blue-shades) for different mmFDR rate levels: 10% (lighter shade), 5%, 1%, and 0.5% (darker shade).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>HTM statistical data</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td/><td/><td align=\"center\" colspan=\"4\"><bold>RIDGEs</bold></td><td align=\"center\" colspan=\"4\"><bold>anti-RIDGEs</bold></td></tr></thead><tbody><tr><td/><td/><td/><td/><td align=\"right\" colspan=\"2\"><bold>all window sizes</bold></td><td align=\"right\" colspan=\"2\"><bold>window sizes 19–59</bold></td><td align=\"right\" colspan=\"2\"><bold>all window sizes</bold></td><td align=\"right\" colspan=\"2\"><bold>window sizes 19–59</bold></td></tr><tr><td colspan=\"12\"><hr/></td></tr><tr><td align=\"right\"><bold>chr</bold></td><td align=\"right\"><bold>median</bold></td><td align=\"right\"><bold>size</bold></td><td align=\"right\"><bold>N</bold></td><td align=\"right\"><bold>gw-R</bold></td><td align=\"right\"><bold>chr-R</bold></td><td align=\"right\"><bold>gw-R</bold></td><td align=\"right\"><bold>chr-R</bold></td><td align=\"right\"><bold>gw-aR</bold></td><td align=\"right\"><bold>chr-aR</bold></td><td align=\"right\"><bold>gw-aR</bold></td><td align=\"right\"><bold>chr-aR</bold></td></tr><tr><td colspan=\"12\"><hr/></td></tr><tr><td align=\"right\">Y</td><td align=\"right\">11</td><td align=\"right\">57772954</td><td align=\"right\">96</td><td align=\"right\">0</td><td align=\"right\">28</td><td align=\"right\">0</td><td align=\"right\">9</td><td align=\"right\">212</td><td align=\"right\">0</td><td align=\"right\">54</td><td align=\"right\">0</td></tr><tr><td align=\"right\">21</td><td align=\"right\">15</td><td align=\"right\">46944323</td><td align=\"right\">318</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"right\">6957</td><td align=\"right\">0</td><td align=\"right\">266</td><td align=\"right\">0</td></tr><tr><td align=\"right\">18</td><td align=\"right\">16</td><td align=\"right\">76117153</td><td align=\"right\">488</td><td align=\"right\">0</td><td align=\"right\">8</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"right\">23329</td><td align=\"right\">27</td><td align=\"right\">521</td><td align=\"right\">0</td></tr><tr><td align=\"right\">13</td><td align=\"right\">19</td><td align=\"right\">114142980</td><td align=\"right\">553</td><td align=\"right\">0</td><td align=\"right\">2</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"right\">32667</td><td align=\"right\">10123</td><td align=\"right\">853</td><td align=\"right\">190</td></tr><tr><td align=\"right\">4</td><td align=\"right\">23</td><td align=\"right\">191273063</td><td align=\"right\">1172</td><td align=\"right\">0</td><td align=\"right\">323</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"right\">121113</td><td align=\"right\">0</td><td align=\"right\">5</td><td align=\"right\">0</td></tr><tr><td align=\"right\">6</td><td align=\"right\">26</td><td align=\"right\">170899992</td><td align=\"right\">1406</td><td align=\"right\">28327</td><td align=\"right\">175351</td><td align=\"right\">873</td><td align=\"right\">1803</td><td align=\"right\">73404</td><td align=\"right\">12884</td><td align=\"right\">223</td><td align=\"right\">0</td></tr><tr><td align=\"right\">8</td><td align=\"right\">26</td><td align=\"right\">146274826</td><td align=\"right\">1067</td><td align=\"right\">213</td><td align=\"right\">61</td><td align=\"right\">32</td><td align=\"right\">0</td><td align=\"right\">83720</td><td align=\"right\">3110</td><td align=\"right\">176</td><td align=\"right\">20</td></tr><tr><td align=\"right\">10</td><td align=\"right\">26</td><td align=\"right\">135374737</td><td align=\"right\">1123</td><td align=\"right\">165</td><td align=\"right\">36813</td><td align=\"right\">9</td><td align=\"right\">0</td><td align=\"right\">9239</td><td align=\"right\">611</td><td align=\"right\">453</td><td align=\"right\">379</td></tr><tr><td align=\"right\">20</td><td align=\"right\">26.5</td><td align=\"right\">62435964</td><td align=\"right\">738</td><td align=\"right\">978</td><td align=\"right\">1350</td><td align=\"right\">292</td><td align=\"right\">538</td><td align=\"right\">4171</td><td align=\"right\">0</td><td align=\"right\">7</td><td align=\"right\">0</td></tr><tr><td align=\"right\">2</td><td align=\"right\">29</td><td align=\"right\">242951149</td><td align=\"right\">1908</td><td align=\"right\">1801</td><td align=\"right\">22546</td><td align=\"right\">247</td><td align=\"right\">52</td><td align=\"right\">2871</td><td align=\"right\">376</td><td align=\"right\">34</td><td align=\"right\">2</td></tr><tr><td align=\"right\">5</td><td align=\"right\">29</td><td align=\"right\">180857866</td><td align=\"right\">1276</td><td align=\"right\">722</td><td align=\"right\">8875</td><td align=\"right\">146</td><td align=\"right\">303</td><td align=\"right\">35406</td><td align=\"right\">10231</td><td align=\"right\">298</td><td align=\"right\">234</td></tr><tr><td align=\"right\">3</td><td align=\"right\">30</td><td align=\"right\">199501827</td><td align=\"right\">1581</td><td align=\"right\">47644</td><td align=\"right\">97097</td><td align=\"right\">806</td><td align=\"right\">1491</td><td align=\"right\">77920</td><td align=\"right\">85262</td><td align=\"right\">123</td><td align=\"right\">89</td></tr><tr><td align=\"right\">X</td><td align=\"right\">30</td><td align=\"right\">154913754</td><td align=\"right\">893</td><td align=\"right\">141</td><td align=\"right\">946</td><td align=\"right\">106</td><td align=\"right\">694</td><td align=\"right\">0</td><td align=\"right\">3667</td><td align=\"right\">0</td><td align=\"right\">0</td></tr><tr><td colspan=\"12\"><hr/></td></tr><tr><td align=\"right\"><italic>genome</italic></td><td align=\"right\"><italic>33</italic></td><td align=\"right\"><italic>3080419480</italic></td><td align=\"right\"><italic>26740</italic></td><td align=\"right\"><italic>1115947</italic></td><td align=\"right\"><italic>767239</italic></td><td align=\"right\"><italic>2511</italic></td><td align=\"right\"><italic>9406</italic></td><td align=\"right\"><italic>545680</italic></td><td align=\"right\"><italic>554438</italic></td><td align=\"right\"><italic>4832</italic></td><td align=\"right\"><italic>6844</italic></td></tr><tr><td colspan=\"12\"><hr/></td></tr><tr><td align=\"right\">1</td><td align=\"right\">34</td><td align=\"right\">247249719</td><td align=\"right\">2659</td><td align=\"right\">165349</td><td align=\"right\">154517</td><td align=\"right\">1611</td><td align=\"right\">1734</td><td align=\"right\">68461</td><td align=\"right\">271099</td><td align=\"right\">75</td><td align=\"right\">88</td></tr><tr><td align=\"right\">12</td><td align=\"right\">34</td><td align=\"right\">132349534</td><td align=\"right\">1382</td><td align=\"right\">1161</td><td align=\"right\">541</td><td align=\"right\">348</td><td align=\"right\">492</td><td align=\"right\">853</td><td align=\"right\">1262</td><td align=\"right\">530</td><td align=\"right\">661</td></tr><tr><td align=\"right\">7</td><td align=\"right\">35</td><td align=\"right\">158821424</td><td align=\"right\">1273</td><td align=\"right\">10549</td><td align=\"right\">19690</td><td align=\"right\">615</td><td align=\"right\">693</td><td align=\"right\">3614</td><td align=\"right\">5125</td><td align=\"right\">388</td><td align=\"right\">479</td></tr><tr><td align=\"right\">15</td><td align=\"right\">36</td><td align=\"right\">100338915</td><td align=\"right\">859</td><td align=\"right\">232</td><td align=\"right\">29110</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"right\">0</td></tr><tr><td align=\"right\">11</td><td align=\"right\">38</td><td align=\"right\">134452384</td><td align=\"right\">1472</td><td align=\"right\">225772</td><td align=\"right\">150844</td><td align=\"right\">1508</td><td align=\"right\">814</td><td align=\"right\">0</td><td align=\"right\">14777</td><td align=\"right\">0</td><td align=\"right\">0</td></tr><tr><td align=\"right\">9</td><td align=\"right\">39</td><td align=\"right\">140273252</td><td align=\"right\">1103</td><td align=\"right\">62730</td><td align=\"right\">50105</td><td align=\"right\">214</td><td align=\"right\">571</td><td align=\"right\">7</td><td align=\"right\">27055</td><td align=\"right\">7</td><td align=\"right\">3</td></tr><tr><td align=\"right\">14</td><td align=\"right\">39</td><td align=\"right\">106368585</td><td align=\"right\">834</td><td align=\"right\">817</td><td align=\"right\">102</td><td align=\"right\">170</td><td align=\"right\">75</td><td align=\"right\">210</td><td align=\"right\">5968</td><td align=\"right\">186</td><td align=\"right\">489</td></tr><tr><td align=\"right\">17</td><td align=\"right\">44</td><td align=\"right\">78774742</td><td align=\"right\">1439</td><td align=\"right\">67267</td><td align=\"right\">0</td><td align=\"right\">1236</td><td align=\"right\">0</td><td align=\"right\">841</td><td align=\"right\">72325</td><td align=\"right\">434</td><td align=\"right\">1405</td></tr><tr><td align=\"right\">16</td><td align=\"right\">47</td><td align=\"right\">88827254</td><td align=\"right\">1075</td><td align=\"right\">107253</td><td align=\"right\">1388</td><td align=\"right\">1293</td><td align=\"right\">82</td><td align=\"right\">504</td><td align=\"right\">15676</td><td align=\"right\">63</td><td align=\"right\">1248</td></tr><tr><td align=\"right\">22</td><td align=\"right\">48</td><td align=\"right\">49691432</td><td align=\"right\">580</td><td align=\"right\">34220</td><td align=\"right\">0</td><td align=\"right\">255</td><td align=\"right\">0</td><td align=\"right\">12</td><td align=\"right\">242</td><td align=\"right\">12</td><td align=\"right\">219</td></tr><tr><td align=\"right\">19</td><td align=\"right\">52</td><td align=\"right\">63811651</td><td align=\"right\">1445</td><td align=\"right\">360606</td><td align=\"right\">17542</td><td align=\"right\">2748</td><td align=\"right\">55</td><td align=\"right\">169</td><td align=\"right\">14618</td><td align=\"right\">124</td><td align=\"right\">1338</td></tr></tbody></table></table-wrap>" ]

[ "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1756-0500-1-63-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mrow><mml:mi>m</mml:mi><mml:mo>≥</mml:mo><mml:msub><mml:mi>m</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:msub><mml:mrow><mml:mi>f</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>m</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>/</mml:mo><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mrow><mml:mi>m</mml:mi><mml:mo>≥</mml:mo><mml:msub><mml:mi>m</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:msub><mml:mrow><mml:mi>g</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>m</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>≤</mml:mo><mml:mi>α</mml:mi></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M2\" name=\"1756-0500-1-63-i2\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mrow><mml:mi>m</mml:mi><mml:mo>≤</mml:mo><mml:msub><mml:mi>m</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:msub><mml:mrow><mml:mi>f</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>m</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>/</mml:mo><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mrow><mml:mi>m</mml:mi><mml:mo>≤</mml:mo><mml:msub><mml:mi>m</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:msub><mml:mrow><mml:mi>g</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>m</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>≤</mml:mo><mml:mi>α</mml:mi></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M3\" name=\"1756-0500-1-63-i3\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>f</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>r</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mi>r</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mi>K</mml:mi></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mi>N</mml:mi><mml:mo>−</mml:mo><mml:mi>r</mml:mi></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mi>S</mml:mi><mml:mo>−</mml:mo><mml:mi>K</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mi>N</mml:mi></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mi>S</mml:mi></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mi>r</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>S</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mi>N</mml:mi><mml:mo>−</mml:mo><mml:mi>r</mml:mi></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>S</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mi>N</mml:mi></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mi>S</mml:mi></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mi>S</mml:mi><mml:mi>N</mml:mi></mml:mfrac><mml:mfrac><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mi>r</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>S</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mi>N</mml:mi><mml:mo>−</mml:mo><mml:mi>r</mml:mi></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>S</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mi>N</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mi>S</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:mfrac><mml:mo>.</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>Derivation of the exact formula for the probability function <italic>f</italic>(<italic>m</italic>), and detailed description of the mmFDR-procedure.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional file 2</title><p>Additional Figures.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S3\"><caption><title>Additional file 3</title><p>Description of alternative SigWin-detector workflow configurations.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S4\"><caption><title>Additional file 4</title><p>Applicability of SigWin-detector: periodic time series of air quality data.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S5\"><caption><title>Additional file 5</title><p>This tar file contains the source files of the WS-VLAM modules needed to run the SigWin-detector workflow, and some examples. To uncompress use. ▪ tar -xvzf SigWin-VLAM.v1.1.tar.gz (Linux users). ▪ WinZip or a similar tool.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>chr: chromosome, median: median expression of all genes on a chromosome, size: chromosome size in base pairs, N: number of genes in chromosome, gw-R: number of genome-wide RIDGEs in a chromosome, chr-R: number of chromosome-specific RIDGEs in a chromosome, gw-aR: number of genome-wide anti-RIDGEs in a chromosome, chr-aR: number of chromosome-specific anti-RIDGEs in a chromosome.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1756-0500-1-63-1\"/>", "<graphic xlink:href=\"1756-0500-1-63-2\"/>", "<graphic xlink:href=\"1756-0500-1-63-3\"/>", "<graphic xlink:href=\"1756-0500-1-63-4\"/>", "<graphic xlink:href=\"1756-0500-1-63-5\"/>" ]

[ "<media xlink:href=\"1756-0500-1-63-S1.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1756-0500-1-63-S2.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1756-0500-1-63-S3.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1756-0500-1-63-S4.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1756-0500-1-63-S5.gz\" mimetype=\"application\" mime-subtype=\"x-zip-compressed\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[]

[]

[]

[]

[]

[ "<p>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.</p>", "<title>Background</title>", "<p>The introduction of PSA testing in Austria led to a steep increase of the incidence of prostate cancer. We want to present the course of the number of newly diagnosed cases of prostate cancer in Austria since 1997, and set these numbers in relation to the total of radical prostatectomies (with resection of lymph nodes) in the same time period. All numbers were retrieved from health statistics of Statistics Austria. The report period of cancer cases and of RPE comprises the years 1997–2004. All calculations were performed for totals as well as for 5-year age groups (40–89 years of age).</p>", "<title>Findings</title>", "<p>The number of prostate cancer cases rose from 1997 to 2004 by 35%, while the number of RPE rose by 94% in the same time period. The proportion of RPE in relation to new cases rose from 41% in 1997 to 59% in 2004.</p>", "<title>Conclusion</title>", "<p>A slight decrease of prostate cancer mortality can already be observed in Austria, but the question of over-treatment still awaits analysis.</p>" ]

[ "<title>Findings</title>", "<p>Opportunistic Prostate-Specific-Antigen testing (PSA) of healthy men started at the beginning of the 1990s and led to a steep increase of prostate cancer incidence [##REF##11711757##1##, ####REF##16091872##2##, ##UREF##0##3####0##3##]. In Austria data on surgical treatment of prostate cancer are available since 1997. The purpose of this paper is to present the correlation between the number of newly detected cases and the number of radical prostatectomies with resection of lymph nodes (RPE) by 5 year age group.</p>", "<p>Data on the number of newly detected cases and number of prostatectomies where obtained from Statistics Austria. The data on radical prostatectomies by age group are available since 1997 only. The data set covers the age group 40 to 89 years comprising a population of 1.738,655 men in the year 2000. Since cases are anonymized it is indeterminable whether newly detected cases are operated in the year of diagnosis or in the following year, if ever.</p>", "<p>Cases treated by perineal cryosurgery were excluded, due to the small total number (0 to 2 cases per year).</p>", "<p>The number of newly detected cases and of RPE (1997 through 2004) as well as the calculated percentage per year by 5 year age group are given in table ##TAB##0##1##. The absolute number of prostate cancer cases rose from 3999 in 1997 to 5416 in 2004 (+35.4%, all age groups). The absolute number of RPE rose from 1648 in 1997 to 3200 in 2004 (+94.2%, all age groups). The percentage of RPE in relation to the number of newly detected cases rises from 41% in 1997 to 59% in 2004 (all age groups). The increase of this percentage is observed in all age groups younger than 70 years but not in older age groups, the linear trend over time is significant for all age groups and for totals, except for age group 40–44 years.</p>", "<p>The increase of the incidence of prostate cancer is well documented for Austria [##REF##11711757##1##,##REF##16091872##2##], however the absolute numbers are for the first time set in relation to the number of RPE nationwide. Within 8 years, the total number of RPE increased by 94.2%, thus nearly doubled, with the increase mainly occurring in the age group 55–69 years. Starting at age 70 years the percentage of RPE remains more or less stable on a substantially lower level.</p>", "<p>The number of RPE we used in our calculations represents the official hospital discharge statistics of Austria comprising of the collected data of all hospitals (including private hospitals), but excludes a very small number of self-pay patients. Thus our results negligibly may underestimate the relation of RPE to the number of newly detected cases. A further limitation of our analyses is caused by legal constrains: due to data protection laws we are limited to the use of depersonalized data.</p>", "<p>The increase of RPE in Austria can be explained by at least three factors: a) operation technique meliorated significantly, now is a standard procedure, perioperative mortality is low and the risk of postoperative morbidity such as incontinence and erectile dysfunction decreased in Austria [##REF##16688458##4##,##REF##16835007##5##], consequently acceptance of RPE in patients increased. b) By the introduction of PSA testing the target group of prostate cancer screening shifted towards younger age groups, meaning that carcinoma more frequently is sought in younger men [##UREF##0##3##]. c) Because of the slow progression of most cases, RPE only makes sense when health status permits RPE and patients have a life expectancy of at least 10 years [##REF##15888698##6##].</p>", "<p>The high percentage of RPE in men younger than 70 years surely also is influenced by the fact that counselling (towards RPE or radiation therapy) in Austria is performed by urologists. In this context, it is worth mentioning, that the primary intention of this manuscript is to provide the data which can serve as a basis for a constructive discussion about counselling and the provision of the most adequate therapeutic regimes.</p>", "<p>The reason for the more or less stable percentage in men aged 70 years or older may lie in the fact that therapy of prostate cancer in these patients more often relies on watchful waiting, active surveillance, external beam radiation, brachytherapy or hormonal treatment. Detailed data for these therapies are not available.</p>", "<p>In conclusion, currently 59% of prostate cancer cases detected are treated by RPE, a percentage steadily rising from 1997 on. On one hand, the dramatic increase of RPE already leads to a slight reduction of prostate cancer mortality in Austria [##UREF##0##3##,##REF##11549491##7##], on the other hand the question of over-testing and over-treatment, and added morbidity remains unanswered and analyses are still pending for Austria.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>GH and CV planned the study and drafted the manuscript. CV obtained the data, GH and CV performed the statistical analysis. SM and GS contributed in preparing the manuscript. All authors read and approved the final manuscript.</p>" ]

[]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Number of newly detected cases of prostate cancer in Austria, and number of radical prostatectomies 1997–2004, as well as proportion of RPE/incidence, and P value for linear time trend, totals and 5-year age groups of men aged 40–89.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\">Age</td><td align=\"center\" colspan=\"3\">1997</td><td align=\"center\" colspan=\"3\">1998</td><td align=\"center\" colspan=\"3\">1999</td><td align=\"center\" colspan=\"3\">2000</td><td align=\"center\" colspan=\"3\">2001</td><td align=\"center\" colspan=\"3\">2002</td><td align=\"center\" colspan=\"3\">2003</td><td align=\"center\" colspan=\"3\">2004</td><td/></tr></thead><tbody><tr><td/><td align=\"center\">Nndc</td><td align=\"center\" colspan=\"2\">RPE</td><td align=\"center\">Nndc</td><td align=\"center\" colspan=\"2\">RPE</td><td align=\"center\">Nndc</td><td align=\"center\" colspan=\"2\">RPE</td><td align=\"center\">Nndc</td><td align=\"center\" colspan=\"2\">RPE</td><td align=\"center\">Nndc</td><td align=\"center\" colspan=\"2\">RPE</td><td align=\"center\">Nndc</td><td align=\"center\" colspan=\"2\">RPE</td><td align=\"center\">Nndc</td><td align=\"center\" colspan=\"2\">RPE</td><td align=\"center\">Nndc</td><td align=\"center\" colspan=\"2\">RPE</td><td/></tr><tr><td colspan=\"26\"><hr/></td></tr><tr><td/><td align=\"center\">n</td><td align=\"center\">n</td><td align=\"center\">%</td><td align=\"center\">n</td><td align=\"center\">n</td><td align=\"center\">%</td><td align=\"center\">n</td><td align=\"center\">n</td><td align=\"center\">%</td><td align=\"center\">n</td><td align=\"center\">n</td><td align=\"center\">%</td><td align=\"center\">n</td><td align=\"center\">n</td><td align=\"center\">%</td><td align=\"center\">n</td><td align=\"center\">n</td><td align=\"center\">%</td><td align=\"center\">n</td><td align=\"center\">n</td><td align=\"center\">%</td><td align=\"center\">n</td><td align=\"center\">n</td><td align=\"center\">%</td><td align=\"center\">P-value</td></tr><tr><td colspan=\"26\"><hr/></td></tr><tr><td align=\"center\">40–44</td><td align=\"center\">3</td><td align=\"center\">4</td><td align=\"center\">&gt;100</td><td align=\"center\">5</td><td align=\"center\">4</td><td align=\"center\">80</td><td align=\"center\">5</td><td align=\"center\">4</td><td align=\"center\">80</td><td align=\"center\">16</td><td align=\"center\">14</td><td align=\"center\">87</td><td align=\"center\">18</td><td align=\"center\">15</td><td align=\"center\">83</td><td align=\"center\">17</td><td align=\"center\">18</td><td align=\"center\">&gt;100</td><td align=\"center\">22</td><td align=\"center\">17</td><td align=\"center\">77</td><td align=\"center\">25</td><td align=\"center\">25</td><td align=\"center\">100</td><td align=\"center\">0.338</td></tr><tr><td align=\"center\">45–49</td><td align=\"center\">26</td><td align=\"center\">21</td><td align=\"center\">81</td><td align=\"center\">40</td><td align=\"center\">37</td><td align=\"center\">92</td><td align=\"center\">44</td><td align=\"center\">35</td><td align=\"center\">80</td><td align=\"center\">74</td><td align=\"center\">71</td><td align=\"center\">96</td><td align=\"center\">85</td><td align=\"center\">81</td><td align=\"center\">95</td><td align=\"center\">95</td><td align=\"center\">87</td><td align=\"center\">92</td><td align=\"center\">125</td><td align=\"center\">123</td><td align=\"center\">98</td><td align=\"center\">109</td><td align=\"center\">118</td><td align=\"center\">&gt;100</td><td align=\"center\">&lt; 0.001</td></tr><tr><td align=\"center\">50–54</td><td align=\"center\">127</td><td align=\"center\">99</td><td align=\"center\">78</td><td align=\"center\">149</td><td align=\"center\">118</td><td align=\"center\">79</td><td align=\"center\">166</td><td align=\"center\">157</td><td align=\"center\">95</td><td align=\"center\">199</td><td align=\"center\">171</td><td align=\"center\">86</td><td align=\"center\">256</td><td align=\"center\">240</td><td align=\"center\">94</td><td align=\"center\">266</td><td align=\"center\">259</td><td align=\"center\">97</td><td align=\"center\">267</td><td align=\"center\">279</td><td align=\"center\">&gt;100</td><td align=\"center\">254</td><td align=\"center\">258</td><td align=\"center\">&gt;100</td><td align=\"center\">&lt; 0.001</td></tr><tr><td align=\"center\">55–59</td><td align=\"center\">371</td><td align=\"center\">300</td><td align=\"center\">61</td><td align=\"center\">440</td><td align=\"center\">382</td><td align=\"center\">87</td><td align=\"center\">496</td><td align=\"center\">433</td><td align=\"center\">87</td><td align=\"center\">522</td><td align=\"center\">482</td><td align=\"center\">92</td><td align=\"center\">541</td><td align=\"center\">518</td><td align=\"center\">96</td><td align=\"center\">559</td><td align=\"center\">492</td><td align=\"center\">88</td><td align=\"center\">597</td><td align=\"center\">552</td><td align=\"center\">92</td><td align=\"center\">585</td><td align=\"center\">612</td><td align=\"center\">&gt;100</td><td align=\"center\">&lt; 0.001</td></tr><tr><td align=\"center\">60–64</td><td align=\"center\">490</td><td align=\"center\">369</td><td align=\"center\">75</td><td align=\"center\">528</td><td align=\"center\">359</td><td align=\"center\">68</td><td align=\"center\">589</td><td align=\"center\">454</td><td align=\"center\">77</td><td align=\"center\">784</td><td align=\"center\">688</td><td align=\"center\">88</td><td align=\"center\">907</td><td align=\"center\">798</td><td align=\"center\">88</td><td align=\"center\">995</td><td align=\"center\">884</td><td align=\"center\">89</td><td align=\"center\">1158</td><td align=\"center\">1067</td><td align=\"center\">92</td><td align=\"center\">1105</td><td align=\"center\">980</td><td align=\"center\">89</td><td align=\"center\">&lt; 0.001</td></tr><tr><td align=\"center\">65–69</td><td align=\"center\">841</td><td align=\"center\">515</td><td align=\"center\">61</td><td align=\"center\">880</td><td align=\"center\">568</td><td align=\"center\">65</td><td align=\"center\">924</td><td align=\"center\">560</td><td align=\"center\">61</td><td align=\"center\">938</td><td align=\"center\">666</td><td align=\"center\">71</td><td align=\"center\">962</td><td align=\"center\">685</td><td align=\"center\">71</td><td align=\"center\">910</td><td align=\"center\">651</td><td align=\"center\">72</td><td align=\"center\">1065</td><td align=\"center\">842</td><td align=\"center\">79</td><td align=\"center\">1011</td><td align=\"center\">789</td><td align=\"center\">78</td><td align=\"center\">&lt; 0.001</td></tr><tr><td align=\"center\">70–74</td><td align=\"center\">837</td><td align=\"center\">299</td><td align=\"center\">36</td><td align=\"center\">873</td><td align=\"center\">288</td><td align=\"center\">33</td><td align=\"center\">976</td><td align=\"center\">288</td><td align=\"center\">30</td><td align=\"center\">1042</td><td align=\"center\">355</td><td align=\"center\">34</td><td align=\"center\">1035</td><td align=\"center\">390</td><td align=\"center\">38</td><td align=\"center\">998</td><td align=\"center\">370</td><td align=\"center\">37</td><td align=\"center\">1074</td><td align=\"center\">500</td><td align=\"center\">47</td><td align=\"center\">1001</td><td align=\"center\">377</td><td align=\"center\">38</td><td align=\"center\">&lt; 0.001</td></tr><tr><td align=\"center\">75–79</td><td align=\"center\">554</td><td align=\"center\">34</td><td align=\"center\">6</td><td align=\"center\">655</td><td align=\"center\">31</td><td align=\"center\">5</td><td align=\"center\">737</td><td align=\"center\">38</td><td align=\"center\">5</td><td align=\"center\">698</td><td align=\"center\">40</td><td align=\"center\">6</td><td align=\"center\">674</td><td align=\"center\">42</td><td align=\"center\">6</td><td align=\"center\">644</td><td align=\"center\">26</td><td align=\"center\">4</td><td align=\"center\">732</td><td align=\"center\">119</td><td align=\"center\">16</td><td align=\"center\">718</td><td align=\"center\">36</td><td align=\"center\">5</td><td align=\"center\">&lt; 0.001</td></tr><tr><td align=\"center\">80–84</td><td align=\"center\">383</td><td align=\"center\">4</td><td align=\"center\">1</td><td align=\"center\">300</td><td align=\"center\">2</td><td align=\"center\">&lt;1</td><td align=\"center\">316</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">333</td><td align=\"center\">1</td><td align=\"center\">&lt;1</td><td align=\"center\">328</td><td align=\"center\">2</td><td align=\"center\">&lt;1</td><td align=\"center\">339</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">403</td><td align=\"center\">38</td><td align=\"center\">9</td><td align=\"center\">367</td><td align=\"center\">2</td><td align=\"center\">&lt;1</td><td align=\"center\">&lt; 0.001</td></tr><tr><td align=\"center\">85–89</td><td align=\"center\">367</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">348</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">340</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">317</td><td align=\"center\">1</td><td align=\"center\">&lt;1</td><td align=\"center\">322</td><td align=\"center\">1</td><td align=\"center\">&lt;1</td><td align=\"center\">258</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">284</td><td align=\"center\">15</td><td align=\"center\">5</td><td align=\"center\">325</td><td align=\"center\">1</td><td align=\"center\">&lt;1</td><td align=\"center\">&lt; 0.001</td></tr><tr><td colspan=\"26\"><hr/></td></tr><tr><td align=\"center\">Total</td><td align=\"center\">3999</td><td align=\"center\">1648</td><td align=\"center\">41</td><td align=\"center\">4218</td><td align=\"center\">1826</td><td align=\"center\">43</td><td align=\"center\">4593</td><td align=\"center\">1969</td><td align=\"center\">43</td><td align=\"center\">4923</td><td align=\"center\">2489</td><td align=\"center\">51</td><td align=\"center\">5128</td><td align=\"center\">2774</td><td align=\"center\">54</td><td align=\"center\">5081</td><td align=\"center\">2791</td><td align=\"center\">55</td><td align=\"center\">5727</td><td align=\"center\">3167</td><td align=\"center\">55</td><td align=\"center\">5416</td><td align=\"center\">3200</td><td align=\"center\">59</td><td align=\"center\">&lt; 0.001</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>Nndc ... Number of newly detected cases.</p><p>RPE ... Radical Prostatectomy (with resection of lymph nodes).</p><p>P-value ... P value for linear time trend.</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Bladder cancer represents a global health problem. It ranks ninth in worldwide cancer incidence. It is the 4th commonest cancer in men and the 12^thin women in the USA. It is estimated that about 67,160 Americans were diagnosed with bladder cancer in 2007 and 13,750 died of the disease [##UREF##0##1##].</p>", "<p>In Egypt, carcinoma of the bladder is the most prevalent cancer, accounting for as many as 31% of all cancer cases [##REF##8313484##2##]. Currently, it ranks first in males representing 16.2% of male cancer [##REF##16892094##3##]. The estimated incidence in males in rural areas in Egypt is about 32 per 100.000 [##UREF##1##4##].</p>", "<p>The exact etiology of bladder cancer is still unknown. Several risk factors have been accused as being involved in its pathogenesis such as cigarette smoking [##UREF##2##5##], synthetic nitrogen fertilizers [##REF##14579546##6##], organophosphate-based pesticides [##REF##11943336##7##], aromatic amines [##REF##17971596##8##], pelvic irradiation, A cyclophosphamide, chronic cystitis, schistosomiasis [##UREF##2##5##], human papilloma virus [##REF##9816001##9##], genetic predisposition, and some occupations [##UREF##2##5##]. The relative importance of such risk factors in the pathogenesis of the disease differs in different populations.</p>", "<p>The aim of this study is to identify and rank the risk factors of muscle invasive bladder cancer (MIBC) in Upper Egypt and to describe the peculiarities of the disease presentation and histopathology in this specific population.</p>" ]

[ "<title>Methods</title>", "<p>This study is an analytical, hospital based, case control study comparing MIBC cases with matched control group in age, sex and residence for the presence of risk factors of bladder cancer.</p>", "<p>The study was carried out in Upper Egypt which is a narrow strip of land that extends from the cataract boundaries of modern-day Aswan to the area between El-Aiyat and Zawyet Dahshur, south of modern-day Cairo, Figure ##FIG##0##1##.</p>", "<p>The study group were residents of upper Egypt with newly diagnosed histologically proven MIBC admitted to South Egypt Cancer Institute in 2005 (n = 130).</p>", "<p>Controls were chosen from the healthy visitors of the institute. They were matched to cases for age, sex and residence and were compared to them as regards risk factors. Two controls were matched with each case (n = 260). Full medical history, clinical examination, urinalysis and abdominal ultrasonography were done for controls to exclude the presence of any bladder lesion.</p>", "<p>Personal interview was conducted to collect socio-demographic data (age, sex, occupation and residence), history suggestive of risk factors (bilharziasis, smoking, chronic cystitis, bladder stones, family history of cancer, parents consanguinity, exposure to chemicals, pelvic radiation and cyclophosphamide chemotherapy), and mode of presentation. History of bilharziasis is defined as finding of ovae on previous urinalysis and history of medical treatment for it. Histopathological pattern of the tumor after cystectomy was recorded.</p>", "<p>SPSS program (version 13) was used for data analysis, which included descriptive analysis and logistic regression for calculation of risk factors.</p>", "<p>Approval was obtained from the ethical committee of Faculty of Medicine, Assiut University. An informed written consent was obtained from all the participants, security and confidentiality of all the information obtained was guaranteed.</p>" ]

[ "<title>Results</title>", "<p>None of the controls had any suspicious symptom or sign of MIBC, also no suspicious lesions were found during investigating the controls.</p>", "<p>The mean age of our patients was 58.34 ± 12.13. Males constituted 83.8% of the patients and 83.1% of controls. Residents of the rural areas were 93.8% of both patients and controls.</p>", "<p>There was a highly significant statistical difference between cases and controls as regards the exposure to bilharziasis, fertilizer, pesticides, recurrent cystitis, bladder stones, smoking, and positive family history of bladder cancer (p &lt; 0.001). Yet, the type of fertilizer, mode of exposure to it, the type of pesticide, mode of exposure to it, the type of smoking and the degree of relative with bladder cancer were not statistically significantly different between cases and controls.</p>", "<p>Patients' characteristics are shown in table ##TAB##0##1##. 73.8% of cases were farmers, 86.2% were married and 91.5% were illiterates.</p>", "<p>Risk factors of MIBC as calculated by risk estimate analysis are shown in table ##TAB##1##2##. The adjusted odds ratio (AOR) as estimated by stepwise logistic regression is shown in table ##TAB##2##3##. The most important risk factor was the positive family history of the disease (AOR 7.7, CI = 2.1–28.4, p &lt; 0.01) followed by exposure to pesticides (AOR 6.2, CI = 3.5–11.3, p &lt; 0.001).</p>", "<p>The results of imaging (ultrasound, IVU and CT) and cystoscopy are shown in table ##TAB##3##4##; 77% of the patients had an obstructed kidney, 100% of them had a filling defect in IVU. CT showed single bladder lesion in 90% and multiple lesions in 10% of cases.</p>", "<p>Table ##TAB##3##4## also shows some important clinical and histopathological criteria of our patients. Burning micturition was the first complaint in 73.8% of cases while hematuria was the presenting symptom in only 20.8%. Digital rectal examination revealed a palpable mass in 95.4% of the cases. Squamous cell carcinoma (SCC) constituted 67.6% of cases; 47.7% of cases had a T2b tumor at the time of first presentation. Positive lymph node affection was found in 13.3% of the patients.</p>" ]

[ "<title>Discussion</title>", "<p>The mean age of cases in this study was 58.34 ± 12.13 years which agrees with another recent report from Egypt that found that the mean age of bladder cancer cases was 56.24 ± 11 [##UREF##3##10##]. This age is less than reported in the literature for other parts of the world; Lynch and Cohen, (1995) reported that the median ages at diagnosis for urothelial carcinoma is 69 years in males and 71 years in females [##REF##8001003##11##].</p>", "<p>The male to female ratio in this study was 5.5: 1. Residents of rural areas constituted 93.8% of cases while only 6.2% of cases lived in urban areas. This difference in male to female ratio than the international ratio of 3:1 might be explained by the fact that women in Upper Egypt aren't equally involved in farming activities with men, hence they are less exposed to the risk factors of the disease that are linked to this occupation (pesticides, fertilizers and bilharziasis) [##REF##11301380##12##].</p>", "<p>Positive family history of bladder cancer was confirmed in 13.8% of cases. Many studies reported that family history plays a major rule in developing bladder cancer and familial clusters of bladder cancer have been reported in transitional cell carcinoma (TCC) [##REF##2294255##13##]. Moreover, we found that parents' consanguinity is an important risk factor for the development of bladder cancer in offspring, history of consanguinity between parents was found in 50.8% of cases (AOR = 3.9, 95%CI = 2.2–6.9, P &lt; 0.001).</p>", "<p>Among our patients 73.8% were farmers. In Upper Egypt, cancer risk in this occupational group is considered an important public health problem. Farmers are exposed to several hazardous substances such as fertilizers and pesticides. Moreover, the prevalence and severity of schistosomiasis tend to rise sharply with opportunities for exposure. In Egypt, the disease prevalence increased dramatically after installation of the High Dam, which created perennial irrigation instead of the basin one with subsequent higher exposure to bilharzial infestations [##REF##9880476##14##]. A positive past history of bilharzial infestations was obtained from 87.7% of our cases (OR = 5.8, 95% CI = 3.3–10.4, P &lt; 0.001). Due to the nature of the patients' work as farmers (frequently in contact with water), it is practically very difficult, if not impossible, to define the number of episodes of infestation or the time lag before each treatment course is given, also, the number of treatment courses given doesn't frequently reflect the real number of infestation episodes. Moreover, the heaviness of infestation which is believed to be an important factor in development of bladder cancer can never be practically measured. So an absolute history of exposure to bilharzial infestation was used in this study.</p>", "<p>There is a plethora of literature incriminating <italic>Schistosoma haematobium </italic>infestation as a risk factor for bladder cancer, but explanation for this association remains speculative [##REF##209198##15##]. Evidence that supports the association between schistosomiasis and bladder cancer includes the geographical correlation between the 2 conditions, the distinctive patterns of sex and age at diagnosis, the clinicopathological identity of schistosome-associated bladder cancer, and extensive evidence in experimentally infected animals [##REF##7728097##16##]. The relatively high frequency of bladder cancer in Egypt supports the etiological relationship to urinary schistosomiasis. Despite the marked decrease in prevalence of endemic schistosomiasis over the last 2 decades (decreased from 35% in 1983 to 1.7% in 2003, with complete eradication in certain districts.), Egypt is still paying the toll of the previously high prevalence of the disease. Comparison of the frequency of active urinary schistosomiasis previously reported during the era of high prevalence of the disease and the age-specific incidence rate indicates a strong cohort effect, figure ##FIG##1##2##. It could be anticipated that in the near future, there will be a marked decrease in bilharziasis associated bladder cancer in Egypt as a sequel to schistosomiasis control. The potential risk is the rise in incidence of bladder cancer related to other risk factors [##UREF##4##17##].</p>", "<p>History of exposure to pesticides was obtained in 82.3% of our patients (AOR 6.2, CI = 3.5–11.3, p &lt; 0.001). It is well known that persons exposed to pesticides are at greater risk of developing bladder cancer than persons with no exposure to them [##REF##2294255##13##].</p>", "<p>El-Mawla et al (2001) stated that urinary bladder stones and chronic cystitis increase the risk of developing bladder cancer and particularly SCC [##REF##11301380##12##]. History of bladder stones was found in 25.4% of our cases (AOR = 5, 95% CI = 2.2–11.4, P &lt; 0.001), while 33.1% of cases had history of recurrent cystitis (AOR = 3.1, 95% CI = 1.5–6.1, P &lt; 0.01).</p>", "<p>Some authors have claimed that bladder carcinogenesis is related to bacterial infections, which are commonly associated with bilharzial infestation, rather than the parasite itself. Urinary bacteria have a double action: (i) the production of carcinogenic nitrosamines from their precursors in urine, e.g., nitrates and secondary amines, and (ii) the secretion of the enzyme β-glucuronidase, which may clear conjugated carcinogens, yielding free carcinogenic products [##REF##36278##18##].</p>", "<p>Among the patients of this study, 80% were current or ex smokers (OR = 5.3, 95% CI = 3.2–8.7, P &lt; 0.001). Cigarette smokers are reported to have up to a fourfold higher incidence of bladder cancer than do people who have never smoked [##REF##2793235##19##]. Radosavljevic et al., (2003) stated that although smoking is still recognized as a major risk factor of cancers including bladder cancer, the increasing incidence of bladder cancer despite the reduction in smoking in the United States suggests that other environmental factors may be playing an increasing role in the development of bladder cancer [##REF##12852163##20##]. This is in accordance with the results of this study where smoking failed to be an independent risk factor for MIBC when adjusted to other environmental factors (pesticides, fertilizers and bilharziasis).</p>", "<p>The present data contradicts the common belief in Egypt about the major role bilharziasis plays in bladder cancer development. A larger role for the exposure to pesticides and fertilizers is evident in this study. Also the role of family history and consanguinity between parents seems to be higher than ever recognized. The risk factor profile of Egyptian bladder cancer has changed over the last 26 years as exposure to chemical carcinogens play a major role for the development of bladder cancer in Egypt [##REF##18188671##21##].</p>", "<p>Histopathological examination showed that 67.6% of cases had SCC, 15.4% TCC, 8.5% anaplastic carcinoma and 8.5% had other pathological types (table ##TAB##3##4##). This seams like a shift from what was previously reported regarding the percentage of SCC which exceeded 75% in Bilharzial bladders [##UREF##5##22##]. According to many Egyptian authors, the pattern of histopathology of bladder cancer showed a marked change over the previous years, where SCC constituted less than 60% of bladder cancer [##REF##11301380##12##,##REF##18188671##21##,##REF##18485392##23##]. This change might be explained by the introduction of mass treatment of bilharziasis in recent years in contrast to the increased exposure to pesticides and fertilizers.</p>", "<p>As regards the mode of presentation, globally, the most common presenting symptom of bladder cancer is painless hematuria, which occurs in about 90% of cases [##UREF##6##24##]. In this study the main presenting complaint was burning micturition, which was the presenting symptom in 73.8% of cases while hematuria was the 1st complaint in only 20.8%. This different mode of presentation might be due to the different tumor configuration which was found to be solid in 91.5% of patients. Unlike the papillary configuration which can bleed easily on shedding of the tumor cells, the solid tumor configuration delays the occurrence of hematuria. This mode of presentation complicates the picture of bladder cancer in Egypt as the patients who are usually accustomed to some kind of burning micturition due to bilharziasis, bladder stones and cystitis don't ask for medical advice until their tumors are already invasive [##REF##15729596##25##]. This was reflected on the result of digital rectal examination where bladder mass was detected in 95.4% of cases.</p>", "<p>Lymph node affection was found in only 13.3% of the patients. This might be explained by the fibrosis that affects the lymphatics in bilharzial patients. This is in agreement with the findings of other reports from Egypt [##REF##9224310##26##].</p>" ]

[ "<title>Conclusion</title>", "<p>MIBC in Upper Egypt is peculiar in that it is usually of the SCC type (although its percentage is decreasing), occurs at a younger age and presents with burning micturition rather than hematuria.</p>", "<p>Unlike the common belief, risk factors such as positive family history, parents' consanguinity, exposure to pesticides and chronic cystitis seem to play now more important roles than bilharziasis and smoking in the development of this disease in this area, yet reports on larger numbers of patients are needed to support this conclusion.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>In Egypt, where bilharziasis is endemic, bladder cancer is the commonest cancer in males and the 2^ndin females; squamous cell carcinoma (SCC) is the commonest type found, with a peculiar mode of presentation. The aim of this study is to identify and rank the risk factors of muscle invasive bladder cancer (MIBC) in Upper Egypt and describe its specific criteria of presentation and histopathology.</p>", "<title>Methods</title>", "<p>This is an analytical, hospital based, case controlled study conducted in south Egypt cancer institute through comparing MIBC cases (n = 130) with age, sex and residence matched controls (n = 260) for the presence of risk factors of MIBC. Data was collected by personal interview using a well designed questionnaire. Patients' records were reviewed for histopathology and Radiologic findings.</p>", "<title>Results</title>", "<p>The risk factors of MIBC were positive family history [Adjusted odds ratio (AOR) = 7.7], exposure to pesticides [AOR = 6.2], bladder stones [AOR = 5], consanguinity [AOR = 3.9], recurrent cystitis [AOR = 3.1], bilharziasis [odds ratio (OR) = 5.8] and smoking [OR = 5.3]. SCC represented 67.6% of cases with burning micturition being the presenting symptom in 73.8%.</p>", "<title>Conclusion</title>", "<p>MIBC in Upper Egypt is usually of the SCC type (although its percentage is decreasing), occurs at a younger age and presents with burning micturition rather than hematuria. Unlike the common belief, positive family history, parents' consanguinity, exposure to pesticides and chronic cystitis seem to play now more important roles than bilharziasis and smoking in the development of this disease in this area.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>AHZ supervised the work and gave important suggestions, MS supervised the work, participated in data analysis and manuscript writing and directed the clinical assessment, AAA-E carried out the field work, interviewing cases and controls, clinical assessment of cases and controls, data management and analysis and writing the final manuscript, DAH participated in writing the drafts, writing the final manuscript and gave very important clinical suggestions, MAA supervised the work, participated in the clinical assessment of cases and controls, directed the clinical investigations, gave very important suggestions and participated in the manuscript writing.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2407/8/250/prepub\"/></p>" ]

[]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Map of Egypt.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Bladder Cancer: Prevalence of Active Schistosomiasis (AS) by Age and Age Distribution of Bladder Cancer (BC) in Egypt. Ibrahim and Khaled (2006) ¹⁷.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Patients' characteristics:</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Patients' Characteristics</td><td/></tr></thead><tbody><tr><td align=\"left\">Occupation</td><td/></tr><tr><td align=\"left\"> • Farmer</td><td align=\"center\">96 (73.8%)</td></tr><tr><td align=\"left\"> • Trader</td><td align=\"center\">4 (3.1%)</td></tr><tr><td align=\"left\"> • Employee</td><td align=\"center\">4(3.1%)</td></tr><tr><td align=\"left\"> • House wife</td><td align=\"center\">22 (16.9%)</td></tr><tr><td align=\"left\"> • Manual worker</td><td align=\"center\">4 (3.1%)</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\">Marital status</td><td/></tr><tr><td align=\"left\"> • Single</td><td align=\"center\">2 (1.5%)</td></tr><tr><td align=\"left\"> • Married</td><td align=\"center\">112 (86.2%)</td></tr><tr><td align=\"left\"> • Divorced</td><td align=\"center\">1 (0.8%)</td></tr><tr><td align=\"left\"> • Widowed</td><td align=\"center\">15 (11.5%)</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\">Education</td><td/></tr><tr><td align=\"left\"> • Illiterate</td><td align=\"center\">119 (91.5%)</td></tr><tr><td align=\"left\"> • Read and write</td><td align=\"center\">5 (3.8%)</td></tr><tr><td align=\"left\"> • Primary school graduate</td><td align=\"center\">2 (1.6%)</td></tr><tr><td align=\"left\"> • Preparatory school graduate</td><td align=\"center\">1 (0.8%)</td></tr><tr><td align=\"left\"> • Secondary school graduate</td><td align=\"center\">1 (0.8%)</td></tr><tr><td align=\"left\"> • Intermediate education graduate</td><td align=\"center\">2(1.5%)</td></tr><tr><td align=\"left\"> • University graduate</td><td align=\"center\">0 (0%)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Risk factors of MIBC as calculated by risk estimate analysis.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"left\">p-value</td><td align=\"center\">OR</td><td align=\"center\">(95%CI)</td></tr></thead><tbody><tr><td align=\"left\">Positive family history of bladder cancer</td><td align=\"left\">p &lt; 0.001</td><td align=\"center\">13.8</td><td align=\"center\">(4–47.7)</td></tr><tr><td align=\"left\">Exposure to pesticides</td><td align=\"left\">p &lt; 0.001</td><td align=\"center\">9.4</td><td align=\"center\">(5.6–15.8)</td></tr><tr><td align=\"left\">Exposure to fertilizers</td><td align=\"left\">p &lt; 0.001</td><td align=\"center\">7.5</td><td align=\"center\">(4.4–12.8)</td></tr><tr><td align=\"left\">Bladder stones</td><td align=\"left\">p &lt; 0.001</td><td align=\"center\">7.0</td><td align=\"center\">(3.5–14.2)</td></tr><tr><td align=\"left\">Parents' consanguinity</td><td align=\"left\">p &lt; 0.001</td><td align=\"center\">6.3</td><td align=\"center\">(3.9–10.4)</td></tr><tr><td align=\"left\">Recurrent cystitis</td><td align=\"left\">p &lt; 0.001</td><td align=\"center\">6.1</td><td align=\"center\">(3.4–11.1)</td></tr><tr><td align=\"left\">Bilharziasis</td><td align=\"left\">p &lt; 0.001</td><td align=\"center\">5.8</td><td align=\"center\">(3.3–10.4)</td></tr><tr><td align=\"left\">Smoking</td><td align=\"left\">p &lt; 0.001</td><td align=\"center\">5.3</td><td align=\"center\">(3.2–8.7)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Risk factors of MIBC calculated by stepwise logistic regression:</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"left\">P-value</td><td align=\"center\">AOR</td><td align=\"center\">(95% CI)</td></tr></thead><tbody><tr><td align=\"left\">Positive family history of bladder cancer</td><td align=\"left\">P &lt; 0.01</td><td align=\"center\">7.7</td><td align=\"center\">(2.1–28.4)</td></tr><tr><td align=\"left\">Exposure to pesticides</td><td align=\"left\">P &lt; 0.001</td><td align=\"center\">6.2</td><td align=\"center\">(3.5–11.3)</td></tr><tr><td align=\"left\">Bladder stones</td><td align=\"left\">P &lt; 0.001</td><td align=\"center\">5.0</td><td align=\"center\">(2.2–11.4)</td></tr><tr><td align=\"left\">Parents' consanguinity</td><td align=\"left\">P &lt; 0.001</td><td align=\"center\">3.9</td><td align=\"center\">(2.2–6.7)</td></tr><tr><td align=\"left\">Chronic cystitis</td><td align=\"left\">P &lt; 0.01</td><td align=\"center\">3.1</td><td align=\"center\">(1.5–6.1)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Patients' clinical, radiological, cystoscopic, and histopathological criteria.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"/><td align=\"center\">(n = 130)</td></tr></thead><tbody><tr><td><bold>First complaint</bold></td><td align=\"center\"/></tr><tr><td align=\"left\"> Burning micturition</td><td align=\"center\">73.8%</td></tr><tr><td align=\"left\"> Haematuria</td><td align=\"center\">20.8%</td></tr><tr><td align=\"left\"> Loin pain</td><td align=\"center\">3.8%</td></tr><tr><td align=\"left\"> Frequency</td><td align=\"center\">1.6%</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>Ultrasound findings</bold></td><td/></tr><tr><td align=\"left\"> Visible growth</td><td align=\"center\">99.1</td></tr><tr><td align=\"left\"> Bladder stones</td><td align=\"center\">8.9%</td></tr><tr><td align=\"left\"> Obstructed kidney</td><td align=\"center\">77%</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>IVU findings</bold></td><td/></tr><tr><td align=\"left\"> Visible filling defect</td><td align=\"center\">100%</td></tr><tr><td align=\"left\"> Obstructed kidney</td><td align=\"center\">77.3%</td></tr><tr><td align=\"left\"> Normal contrast secretion</td><td align=\"center\">82.7%</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>CT</bold></td><td/></tr><tr><td align=\"left\"> Liver cirrhosisBack pressure on kidneys</td><td align=\"center\">20%</td></tr><tr><td align=\"left\"> Back pressure on kidneys</td><td align=\"center\">80%</td></tr><tr><td align=\"left\">Bladder</td><td/></tr><tr><td align=\"left\"> Single lesion</td><td align=\"center\">90%</td></tr><tr><td align=\"left\"> Multiple lesion</td><td align=\"center\">10%</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>Cystoscopy findings</bold></td><td/></tr><tr><td align=\"left\"> Involved urethra</td><td align=\"center\">0%</td></tr><tr><td align=\"left\"> Involved bladder neck</td><td align=\"center\">6.6%</td></tr><tr><td align=\"left\"> Involved ureteral orifice Tumor configuration</td><td align=\"center\">13.1%</td></tr><tr><td align=\"left\">  Solid</td><td align=\"center\">91.5%</td></tr><tr><td align=\"left\">  Papillary</td><td align=\"center\">8.5%</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>Pathological type</bold></td><td/></tr><tr><td align=\"left\"> Well differentiated SCC</td><td align=\"center\">43.8%</td></tr><tr><td align=\"left\"> Moderately differentiated SCC</td><td align=\"center\">20%</td></tr><tr><td align=\"left\"> Transitional cell carcinoma</td><td align=\"center\">15.4%</td></tr><tr><td align=\"left\"> Anaplastic carcinoma</td><td align=\"center\">8.5%</td></tr><tr><td align=\"left\"> Spindle cell carcinoma</td><td align=\"center\">5.4%</td></tr><tr><td align=\"left\"> Poorly differentiated SCC</td><td align=\"center\">3.8%</td></tr><tr><td align=\"left\"> Adenocarcinoma</td><td align=\"center\">3.1%</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>Stage of the tumor</bold></td><td/></tr><tr><td align=\"left\"> T2a</td><td align=\"center\">8.6%</td></tr><tr><td align=\"left\"> T2b</td><td align=\"center\">47.7%</td></tr><tr><td align=\"left\"> T3a</td><td align=\"center\">7.8%</td></tr><tr><td align=\"left\"> T3b</td><td align=\"center\">3.1%</td></tr><tr><td align=\"left\"> T4a</td><td align=\"center\">28.1%</td></tr><tr><td align=\"left\"> T4b</td><td align=\"center\">4.7%</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>Lymph node affection</bold></td><td/></tr><tr><td align=\"left\"> Positive</td><td align=\"center\">13.3%</td></tr><tr><td align=\"left\"> Negative</td><td align=\"center\">86.7%</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>OR = odds ratio, CI = confidence interval.</p></table-wrap-foot>", "<table-wrap-foot><p>AOR = adjusted odds ratio, CI = confidence interval.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2407-8-250-1\"/>", "<graphic xlink:href=\"1471-2407-8-250-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The latest advancements in translational medicine directed the path towards highly demanded biomarker validation studies that focused on the identification of new cancer biomarkers for cancer research community. These studies lead toward the development of biobanks that are capable of providing well annotated high quality tissue samples that can meet the current demands of cancer research community. This objective is achieved successfully by employing a well designed informatics architecture that balances and supports the routine activities of a biobank [##UREF##0##1##]. Tissue Banking Informatics is relatively novel area of biomedical informatics that deals with collecting and storage of biospecimens with associated clinicopathology annotation. The importance of the development of tissue banks has been realized for the last decade with the advent of advanced computer technology that is utilized in the development of tissue bank informatics infrastructure. These systems have the capability to include the managing aspects of biobank that can function as central role in the development of translational research initiatives [##UREF##1##2##, ####UREF##2##3##, ##UREF##3##4##, ##UREF##4##5##, ##UREF##5##6####5##6##].</p>", "<p>National Mesothelioma Virtual Bank (NMVB), funded by The Centers for Disease Control and Prevention (CDC) in association with National Institutes of Occupational Health and Safety (NIOSH), is an example of an exceptional and progressive tissue banking effort. The main objective of this resource is to provide mesothelioma pleural, pericardial and peritoneal tissue samples along with blood and DNA samples with associated cancer registry, clinicopathology, recurrence and therapy related data on a web based interface to meet the growing need of cancer research community. The multimodal data is integrated from various hospital systems into a web-based architecture which is built upon a set of common data elements (CDEs) that provide semantic and syntactic interoperability across multiple institutions to facilitate translation research. During the development of this resource we applied the lessons learnt from the earlier collaborative projects: fundamental components and standards were employed from Cooperative Breast Cancer Tissue Resource (CBCTR), Pennsylvania Cancer Alliance Bioinformatics Consortium (PCABC), Cooperative Prostate Cancer Tissue Resource (CPCTR), and the cancer Data Standard Repository (caDSR) model of caBIG [##UREF##5##6##, ####UREF##6##7##, ##UREF##7##8##, ##UREF##8##9##, ##UREF##9##10##, ##UREF##10##11####10##11##].</p>", "<p>The manuscript highlights the infrastructure components that were used to build this virtual biobank allowing it to collect clinically annotated tissues from various clinical settings which meet with statutory and best practice guidelines. This resource provides a web-based publicly accessible database that organizes the data by CDE standards and allows the end users to query on de-identified information within the warehouse through a controlled \"point and click\" interface.</p>", "<p>The resource also allows web-based requests for tissues, provides a fair review process, delivers tissues to scientists, permits growth and continual updates and integration with other ongoing efforts, and markets the whole process. Here we describe all the components that make up this comprehensive NMVB informatics architecture, discuss the use of standards, and outline the previous collaborative biorepository experiences of the team at the University of Pittsburgh, Department of Biomedical Informatics [##REF##16677389##12##].</p>" ]

[ "<title>Methods</title>", "<title>Collaborative Institutions</title>", "<p>The resource currently has three participating institutions: University of Pennsylvania (U. Penn), University of Pittsburgh (U. Pitt) and New York University (NYU). The other participants supporting this effort are the Mesothelioma Applied Research Foundation (MARF), University of Hawaii, University of Vermont, and Fox Chase Cancer Center. The collaborative efforts make this resource capable of providing diverse biospecimens including fresh frozen tissues, paraffin blocks, tissue microarrays (TMAs), blood and DNA samples with clinicopathologic and follow-up data from a wide variety of medical care settings spread across rural and urban regions.</p>", "<title>Inclusions/Exclusion Criteria of Cases into the Resource</title>", "<title>Prospective Collection Criteria</title>", "<p>The criteria for participation in the NMVB prospective study is that the individuals are receiving or seeking medical care for mesothelioma are 18 years old or older and can provide informed consent. This is usually obtained at either a doctor's office, clinic visits or upon hospital admission.</p>", "<p>The only exclusion criteria are patients younger than 18 years old and prisoner-patients, who are excluded based on federal limitations. As very few patients are likely to not be eligible for inclusion based on these criteria, the racial, gender and ethnic characteristics of the individuals approached for participation in this NMVB Registry reflect the demographics of patients receiving or seeking medical care at the mesothelioma clinics at the member institutions of the consortium. No individuals are excluded from participation in the NMVB Registry on the basis of race, ethnicity, gender or HIV status.</p>", "<title>Retrospective collection Criteria</title>", "<p>The NMVB registry utilizes an honest broker system for identifying and enrolling eligible retrospective cases into the NMVB database. The honest broker system is also utilized by the Tissue bank with Institutional Review Board (IRB) approval for collecting excess tissue and biological materials. The retrieval of samples from archived residual biospecimens that have previously been removed for clinical treatment and pathology diagnosis as well as retrospective review of associated medical records for annotation purposes fall within the IRB's exempt category [##UREF##11##13##].</p>", "<title>Patient Health Information Protection and De-Identification Process</title>", "<p>The NMVB has designed its procedures to protect the confidentiality and privacy of human subjects and IRB approval has been obtained for all activities. NMVB uses a decentralized system for specimen and data collection and storage. Each case is assigned a de-identified NMVB number. The only link to patient identity is held in reserve locally within the institution and there are no links directly connecting specimens or data to patients.</p>", "<p>The NMVB database is available to the research community on a public web site but it has been designed to produce only de-identified datasets upon query. This is possible because only de-identified data sets exist which are compliant with the 'safe harbour' approach to HIPPA (Health Insurance Portability and Accountability Act) [##UREF##12##14##]. The \"safe-harbor\" approach entails removal of all 18 identifiers enumerated at section 164.514(b) [##UREF##1##2##] of the regulations. Thus for example, a participant's age is presented as age rage, rather than the date of birth, an approach which protects the identity while still providing sufficient information for research purposes. All data requests are tracked in the secure NMVB web-based data query tool regardless of whether the purpose is clinical or research related.</p>", "<p>The Honest Broker acts as a barrier between fully identified confidential clinical patient information and the completely de-identified data made available to the research community. An honest broker is an individual, organization or system acting for or on behalf of the covered entity to collect and provide health information to the investigators in such a manner whereby it would not be reasonably possible for the investigators or others to identify the corresponding patients-subjects directly or indirectly. The honest broker cannot be one of the investigators or researcher. A researcher may use the services of an honest broker service to obtain the Protected Health Informations in a de-identified manner. De-identification means that the patient-subjects cannot be identified by researchers or others directly or indirectly through identifiers link to the patient-subject. This honest broker service will de-identify medical record information by automated or manual methods. All honest broker services are approved in advance by both the IRB of record and UPMC. If an honest broker service is not part of the UPMC covered entity, a valid business associate agreement with UPMC is executed with UPMC in order to access UPMC-held Protected Health Informations for de-identification. If an honest broker system/service is to be used to obtain de-identified Protected Health Informations, this fact must be identified in the study's IRB submission. The honest brokers are individuals who have clinical responsibilities as tissue bankers in the Health Sciences Tissue Bank (HSTB), postdoctoral fellows to manage the pathology data or as cancer registry specialists in the UPMC Network Cancer Registry. Based on their clinical job duties, their educational backgrounds and experiences vary. Depending on the nature of the projects, these bankers can work autonomously or collaboratively to meet biospecimen and/or data needs (please refer to additional file ##SUPPL##0##1##) [##UREF##11##13##].</p>", "<title>Development of the Common Data Elements</title>", "<p>Common Data Elements were developed to allow annotation of cases and accomplish characterization of biospecimens collected from different collaborating sites. The development of CDEs for the resource has recently been described in detail in a related publication from our group [##REF##18397527##15##]. In summary, the CDE development was by joint consensus of participants (including domain experts from various subcommittees) under the leadership supervision of the NMVB Coordinating Committee. The CDEs include cancer registry data (demographic, epidemiology and follow-up) at the patient level, pathology data at the specimen level including data elements to elaborate TNM staging and tumor grading, along with block level annotation and genotype data [##REF##16111498##16##]. In this process the CDE sub-committee used the experiences gained from previous projects like the PCABC [##UREF##7##8##], CPCTR [##UREF##8##9##], caBIG [##UREF##10##11##], Early Detection Research Network (EDRN) [##UREF##13##17##] and Specialized Programs of Research Excellence (SPOREs) [##UREF##14##18##]. The major standards used to develop the CDEs include the NAACCR Data Standards for Cancer Registries [##UREF##15##19##], CAP Cancer Protocol and Checklist [##UREF##16##20##], the ADASP – Association of Directors of Anatomic and Surgical Pathology (ADASP) [##REF##17145632##21##] and the American Joint Committee on Cancer (AJCC) Cancer Staging Manual [##UREF##17##22##]. The utilization of all the above mentioned standards facilitates the development of CDEs that provides an informatics support which is quite ample to achieve both syntactic and semantic interoperability across various systems [##REF##18397527##15##].</p>", "<title>Data Collection and Method of Data Transmission</title>", "<title>Retrospective Collection</title>", "<p>Each collaborative site collects data and tissue samples. Once a case becomes available the pathologists at each site review the surgical pathology report and all histological sections of the tissue facilitating the selection of the mesothelioma case. Then the pathologist selects key slides, referred to as matrix slides, according to a standardized protocol with specific features of the case which is likely to be of interest for scientific investigators. Afterwards the detailed data elements are collected on these slides, and the slides are used to obtain the corresponding matrix blocks which represent the core specimen components of the Resource. This results in block-level annotation of the case with recording of important information about the case such as the size and type of tumor present in each block of interest. Overall, this process increases the utility of the resource many fold as the block level annotation leads to finding cases queried by specific criteria at a later point. This review up front also facilitates the retrieval of such blocks when a request is received at a later point. After the review of pathology data, the data managers and certified tumor registrars retrieve the clinical and follow-up data for cases enrolled in the study. The data are derived in part from the tumor registries of the various hospitals and institutions, which include the Anatomic Pathology Laboratory Information system (APLIS) (b) Cancer Registry Information System (CRIS), and (c) Tissue Banking Inventory System (TBIS).</p>", "<title>Prospective Collection</title>", "<p>To collect mesothelioma cases prospectively, patient are being consented at the physician office or hospitals at each collaborative site. At the time of patient enrollment in the study prospectively a NMVB Health Assessment Questionnaire is handed over to the patient. The purpose of this questionnaire is to collect patient demographic, epidemiology and past medical history data. Additionally, detailed clinical information is obtained by direct review of and extraction of information from patient charts, from consultation with outpatient referring physicians, and from direct patient interviews performed by cancer registrars and clinical nurses. Data is collected and annotated using common data entry paper forms that are correlated with the CDEs developed by the Resource [##REF##16677389##12##]. Finally, collaborative sites transfer de-identified data to the central database either manually using the web-based data entry tool or electronically in spreadsheets.</p>", "<title>Quality Assurance and Check of NMVB Datasets</title>", "<p>The transmitted data from collaborative sites is reviewed at the NMVB coordinating (database) site, which processes the data according to quality assurance (QA) and quality check (QC) protocols every four months that has been established by the resource. The process of QA/QC is performed to locate any missing datasets and errors performed during data recording. The value of each data set is matched with the CDE dictionary to make certain the standards are being followed during data recording. Only the standardized data is uploaded into the database at regular basis, and data with errors is sent back to the corresponding sites along with explanations of rejection. The correction of the data is responsibility of each collaborative site. To ensure the quality of data an audit review process has been established.</p>", "<p>The resource selects 20% of the newly entered cases to be reevaluated by honest brokers, cancer registrars and data managers. After completion of their review, the audit reviewers submit a report of their findings and recommendations to the resource. The resource members discuss their findings in the next general meeting of the NMVB Coordinating Committee and make plans to implement the proposed recommendations [##REF##16677389##12##].</p>", "<p>The NMVB resource has periodic QA/QC of pathology data by assessing inter-observer concordance for the resource pathologists. These consist of two methods which include 1) the joint review of cases at meetings and 2) independent review of cases circulated among sites. For joint QA review during meetings, resource pathologists review up to 5 cases from each site, with emphasis on the 5 \"matrix slides\" selected by each pathologist at each site to include areas of difficulty or likely diagnostic differences. Joint review of cases is performed on a multi-headed microscope that permits pathologists to discuss diagnostic differences and set thresholds.</p>", "<p>Independent review of cases is performed by individual pathologist at each site on cases sent from the other collaborative sites at regular intervals. The data mangers at each site will randomly select the cases for QA review from those added to the Resource within certain cut-off dates. The review will be documented by completing the \"matrix\" fields and select \"histology\" fields of the NMVB database. Any areas that show a high level of discrepancy will be determined by data mangers and communicated to resource pathologists. The Pathology Subcommittee will then discuss their findings in the subsequent general meeting of the Coordinating Committee and provide a report with recommendations as indicated by their findings [##REF##16677389##12##].</p>", "<title>Mesothelioma Virtual Bank Central Data Warehouse</title>", "<p>The NMVB database is based on an informatics model that aids in achieving semantic and syntectic interoperability by describing the common data elements in the form of metadata or data descriptors and by using a controlled vocabulary in order to make the data understandable and sharable for end-users. The system architecture is designed to provide query speed and high security as well as expansion capabilities for incorporating new data elements or integrating existing systems at participating institutions. Emphasis has been placed to provide user access at three levels (a) NMVB statistical data for public view (b) approved investigator database query that allows seeing individual patient de-identified clinical data and (c) data manger access to query and edit the stored data. Patient privacy is of utmost importance at all levels of user's access.</p>", "<p>The NMVB data warehouse is constructed using information models captured as UML class diagrams. Unified Modeling Language (UML) is a non-proprietary language for constructing, visualizing, and documenting the artifacts of software engineering. The information model for CDEs was created to establish a structurally aligned generalized relationship [##UREF##18##23##]. Enterprise Architect (EA) software has been utilized to develop the UML class diagrams [##UREF##19##24##].</p>", "<p>The NMVB web based query tool is based on the caTISSUE Clinical Annotation Engine (CAE) application. CAE was originally developed by the University of Pittsburgh as part of the National Cancer Institute's (NCI) Cancer Biomedical Informatics Grid (caBIG) program [##UREF##20##25##]. The NMVB database allows researches to search clinically annotated Mesothelioma biospecimens via a web interface in real time. The database is made available through a publicly available website [##UREF##21##26##]. The database warehouse facilitates standardized clinical annotation structure and incorporates variety of data sets from different data sources. Two methods of annotation have been adopted manually using web based data entry tool and data imported electronically through XML files [##UREF##22##27##].</p>", "<title>Marketing of Resource Specimens</title>", "<p>Various types of media resources are being utilized for the advertisement of the NMVB biospecimens and services to the research community as follow:</p>", "<p><bold>1</bold>. <italic>The NMVB web site </italic>has been developed and maintained at Department of Biomedical Informatics at University of Pittsburgh [##UREF##23##28##]. The web site includes general information about the resource and different type of specimens available to the research community.</p>", "<p><bold>2</bold>. <italic>Mass E- mailings to Investigator</italic>: Announcement letters of invitation to utilize the resource are sent via e-mail to investigators that have published articles in tissue- based mesothelioma research. Additional names were provided by investigators who visited the NMVB booth at scientific meetings. A mechanism is provided for any email subscriber to \"opt out\" from the mailing list at the time any mass mailing is distributed.</p>", "<p><bold>3</bold>. <italic>Advertisements </italic>are placed in specialty scientific journals, research society newsletters, fliers at research meetings, and through free listings in journals and websites.</p>", "<p><bold>4</bold>. <italic>Posters and podium presentations </italic>regarding the practical use and the Resource specimens at research meetings.</p>", "<p><bold>5</bold>. <italic>Marketing booths </italic>at scientific research meetings, in collaboration to CDC and NIOSH or as individual standalone booths. Marketing surveys were conducted at scientific research meetings as well as with well-known mesothelioma researchers to estimate what resources the NMVB should focus on providing to the research investigators.</p>" ]

[ "<title>Results</title>", "<title>NMVB Biospecimens Collection</title>", "<p>The NMVB resources hold over 650 archived mesothelioma cases and prospective cases. The resource provides more than 775 biospecimes that are accrued from surgical resections and biopsies and also includes whole blood and DNA samples.</p>", "<p>This collection is made possible by the collaborative efforts of University of Pennsylvania, University of Pittsburgh and New York University. At the end of second quarter of 2008 we project the resource to have over 700 annotated cases of pleural, peritoneal and pericardial mesothelioma specimens along with blood and DNA samples available to the mesothelioma research community. The majority of these cases consist of archival paraffin blocks from surgically treated patients. In addition the first Tissue Microarray (TMA) has been made available to investigators along with de-identified clinicopathology and follow-up data. IATA Dangerous Goods Regulations [##UREF##24##29##] are implemented for the special handling and safe shipping of biospecimen to investigators. All major courier companies acknowledge IATA measures of shipping biohazardous supplies by air.</p>", "<p>The NMVB common data elements (CDE) are developed on NAACCR Data Standards for Cancer Registries [##UREF##15##19##], CAP Cancer Protocol and Checklist [##UREF##16##20##], the ADASP [##REF##17145632##21##] and the American Joint Committee on Cancer (AJCC) Cancer Staging Manual [##UREF##17##22##] standards to build mesothelioma CDE datasets. The NMVB web based Query Tool is based upon CAE version 2.0. The query tool is secure password protected and only investigator by approved IRB and Scientific Review Committee are capable to access the database. It permits end-users to develop their own case lists for their applications and quarry the data related to the specimen cases they have received from the resource for their approved studies. The query results shows only de-identified datasets associated with each approved case through disease and specimen pre-defined views of the data set (Figure ##FIG##0##1##).</p>", "<p>The NMVB Statistical Database for Public View is open to the general public. The statistical data base for public query provides summary information on all the mesothelioma cases and their associated biospecimens stored into biorespoitory. The result page shows the number of cases, specimens and blocks in the database that match with query criteria of the investigator and general statistics on a limited number of data elements. By utilizing this public view database the investigator could be able to gain enough information to decide if the resource has sufficient biospecimens to fulfill his experimental requirements (Figure ##FIG##1##2##).</p>" ]

[ "<title>Discussion</title>", "<p>The growing importance of cross-institutional translational research and advancement in the development of informatics tools that are capable of analyzing clinical tissue sample has increased the need of research community for high quality and well annotated biospecimens. To fulfill research community's requirements the NMVB has established an integrated mesothelioma biobank and web base database query tool. This system is built on an underlying architect of common data elements for characterization of tissue samples and clinical follow-up data, supported by an essential quality assurance process. By the development and implementation of NMVB database, the resource developed a web based query tool to make investigators knowledgeable of the resource biospecimens. Finally, the NMVB resource has developed efforts to market the specimens, and a process for requesting samples and data by investigators that involves an independent research evaluation panel (REP) of professionals on mesothelioma, biostatistics, and pathology that are not part of resource.</p>", "<title>Clinical Annotation and development of federated Model NMVB datasets</title>", "<p>The data mangers at each collaborative site are responsible for clinical annotation of biospecimen and collect related information manually from variety of data sources. The collected data is then integrated, de-identified and standardized according to approved study protocols. The integration is a process in which a selected patient data is collected from multiple clinical sources which includes pathology data from the APLIS, tumor marker data from the Clinical Pathology Lab Information System (CPLIS), and clinical staging and treatment data from CRIS.</p>", "<p>The challenges pertained to this process are identifying the same patient across various hospital sources and collecting the accurate data in precise context. These barriers are overcome by common linking patient identifiers and information and this helps in tracking the same patient information across multiple sources. This also reduces the chances of errors in data recording. The resource employs this solution to address the challenges that are largely due to a manual data collection and recording by the data manger. This helps in keeping the focus only on data quality. The next challenge is to collect data that requires to be monitored for context before it can be incorporated into the database. This is important for the data that has temporal relationship. The potential solutions to these challenges are automated data retrieval through electronic queries of existing systems for example electronic medical record (EMR) and cancer registry.</p>", "<p>De-identification is the process which includes isolation of patient clinical identified data sets that are kept in the medical records from the research de-identified data sets. This process is done by honest brokers who act as a barricade between clinical identified data sets and totally de-identified research data sets. The honest broker collects clinical data, identifies the patient identifiers, separates them by assigning the research de-identified number and finally made de-identified data set available to research community. The NMVB resources assign each case a 4-digit de-identified case identification number and linkage to the patient health information is kept by the honest brokers. The NMVB database at no point in time present any linkages to the patient identifiers (medical record number, data of birth, social security number) to end users.</p>", "<p>To facilitate the standardized clinical annotation process of biospecimens and to automate the process of annotation, the NMVB project team incorporated the NAACCR standard [##UREF##15##19##], CAP cancer protocol [##UREF##16##20##] and ADASP [##REF##17145632##21##] guidelines for mesothelioma as a well recognized, significant and standardized sets of data elements in their respective domain. These standards include a series of reporting guidelines for diagnostic pathology reports and outcome related descriptors for the vast majority of human malignancies. Each CAP guideline is comprised of checklists with the data elements to describe the gross as well as microscopic attributes of the neoplasm including pathological staging and perifocal reactions such as margins, angio- as well as perineural invasion with specified valid values for each data element that are important for clinical decision-making and prognostication of individual cases.</p>", "<p>The process of developing the CDEs for the NMVB has authenticated this initiative that can successfully direct to the implementation of strong human mesothelioma-related CDEs that assist in the collection of high quality data for the research community. The CDE are developed by NMVB CDE subcommittee that included experts from various clinical and research fields e.g. pathologists, thoracic surgeons, epidemiologists, bioinformaticians, biostatisticians, data managers, cancer registrars, and research scientists.</p>", "<p>Finally the purpose was to create datasets that should provide value to the end users in future. Furthermore, the definitions of the CDEs and their associated descriptors need to be clearly understandable to all those who collects data. Through the use of ISO compliance and accepted data standards the goal of collecting annotation data of high quality was achievable. The only way in which information from multiple databases can truly be shared and made useful is through the careful use of unequivocal, well defined, consistent and structured metadata. Informaticians and database developers provided the structural link that brought the CDEs together in the database, addressed technical issues, and provided guidance related to implementation of the CDEs at local institutions.</p>", "<title>Presentation of NMVB Web Base Query Tool</title>", "<p>The CAE model enables standards-based manual annotation of biospecimens with associated well-characterized clinical information. It supports importing structured data from various clinical information systems such as APLIS, CPLIS and cancer tumor registries allowing the integration of multimodal annotations within the cancer centers, providing a complete picture of a patient's disease [##REF##16677389##12##]. Additionally, the CAE is based on such a model that aids in developing and conveying the semantic interoperability of the data system by describing the common data elements in the form of metadata or data descriptors (about the content, quality, condition, and other characteristics of the data) and by using controlled vocabulary and ontology, in order to make the data understandable and sharable for end-users and flexible for the system. Hence, the overall advantage of this suite over the legacy biospecimen annotation systems includes the shared responsibilities of individual institutions for services and implementation of the required standards and vocabulary that fosters data sharing effortlessly.</p>" ]

[ "<title>Conclusion</title>", "<p>The NMVB stands as a nation wide largest resource for mesothelioma archived and prospective biosepcimens with associated clinical, pathology, recurrence, follow-up and treatment data and act as a central resource for investigators using experimental methods in translational, histopathology, validation and outcomes measures. This resource is built upon a robust and efficient informatics architecture that facilitates well organized management, standardized collection and detailed clinical annotation of cases across multiple collaborative sites. This set-up provides a manual of operation, a histopathology guide, and a database with common data elements for characterization of mesothelioma biospecimens, multimodal datasets and a quality assurance and control process.</p>", "<p>The NMVB web based query tool acts as a central source that provides a mechanism for researchers to efficiently search clinically annotated datasets and biospecimens that are pertinent to their research areas. The web-based query tool ensures patient health information protection by disclosing only de-identified data with Institutional Review Board (IRB) and scientific review committee approved standards and protocols. Additionally NMVB web site facilitates an online process of requesting mesothelioma biospecimen, acces statistical query tool for public view and access to approved investigator query tool to potential investigator. The biospecimens disbursement and database access to patient de-identified clinical data is granted after Research Evaluation Panel (REP) approval that includes experts on mesothelioma, biostatics and pathology. Different methods are adopted to market the resource that includes brochures, a website, and a booth that is being used to market the resource at scientific meetings. In fact NMVB presents as a fundamental platform for the development and implementation of an integrated tissue banking program and facilitates other tissue banking efforts through the members of the collaborative resource and their associated publications.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Advances in translational research have led to the need for well characterized biospecimens for research. The National Mesothelioma Virtual Bank is an initiative which collects annotated datasets relevant to human mesothelioma to develop an enterprising biospecimen resource to fulfill researchers' need.</p>", "<title>Methods</title>", "<p>The National Mesothelioma Virtual Bank architecture is based on three major components: (a) common data elements (based on College of American Pathologists protocol and National North American Association of Central Cancer Registries standards), (b) clinical and epidemiologic data annotation, and (c) data query tools. These tools work interoperably to standardize the entire process of annotation. The National Mesothelioma Virtual Bank tool is based upon the caTISSUE Clinical Annotation Engine, developed by the University of Pittsburgh in cooperation with the Cancer Biomedical Informatics Grid™ (caBIG™, see <ext-link ext-link-type=\"uri\" xlink:href=\"http://cabig.nci.nih.gov\"/>). This application provides a web-based system for annotating, importing and searching mesothelioma cases. The underlying information model is constructed utilizing Unified Modeling Language class diagrams, hierarchical relationships and Enterprise Architect software.</p>", "<title>Result</title>", "<p>The database provides researchers real-time access to richly annotated specimens and integral information related to mesothelioma. The data disclosed is tightly regulated depending upon users' authorization and depending on the participating institute that is amenable to the local Institutional Review Board and regulation committee reviews.</p>", "<title>Conclusion</title>", "<p>The National Mesothelioma Virtual Bank currently has over 600 annotated cases available for researchers that include paraffin embedded tissues, tissue microarrays, serum and genomic DNA. The National Mesothelioma Virtual Bank is a virtual biospecimen registry with robust translational biomedical informatics support to facilitate basic science, clinical, and translational research. Furthermore, it protects patient privacy by disclosing only de-identified datasets to assure that biospecimens can be made accessible to researchers.</p>" ]

[ "<title>Abbreviations</title>", "<p>ADASP: Association of Directors of Anatomic and Surgical Pathology; AJCC: American Joint Committee on Cancer; APLIS: Anatomic Pathology Lab Information System; caBIG: cancer Bioinformatics Grid; caDSR: cancer Data Standards Repository; CAE: Clinical Annotation Engine; CRIS: Cancer Registry Information System; CAP: College of American Pathologist; CBCTR: Cooperative Breast Cancer Tissue Resource; CDC: Center for Disease Control and Prevention; CDE: Common data element; CPCTR: Cooperative Prostate Cancer Tissue Resource; CPLIS: Clinical Pathology Lab Information System; EA: Enterprise Architect; EDRN: Early Detection Research Network; EMR: Electronic Medical Record; HIPAA: Health Insurance Portability and Accountability Act; IATA: International Air Transport Association; IRB: Institutional Review Board; ISO: International Organization for Standardization; MARF: Mesothelioma Allied Research Foundation; NAACR: North American Association of Central Cancer Registries; NCI: National Cancer Institutes; NIOSH: National Institutes of Occupational health and Safety; NMVB: National Mesothelioma Virtual bank; NYU: New York University; PCABC: Pennsylvania Cancer Alliance Bioinformatics Consortium; PHI: Protected Health Information; QA: Quality assurance; QC: Quality control; REP: Research Evaluation Panel; SPORE: Specialized Programs of Research Excellence; TBIS: Tissue Bank Information System; TMA: Tissue microarray; TNM: Tumor/Node/Metastasis; UML: Unified Medical Language; UPENN: University of Pennsylvania; UPITT: University of Pittsburgh; UPMC: University of Pittsburgh Medical Center.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>MJB, WA, SKM, LS, GF, JM and AVP contributed equally to the first draft of this manuscript. MJB, the chair for the Coordinating Committee, is responsible for leading the efforts of developing the requirements for the central database. MJB, WA, SKM, AVP, RD, JM, MF and HIP have contributed in study design, implementation and quality assurance of the database and tool. LS and AKP have contributed in the development and implementation of software tools for the data annotation and query engine in the web based interface, and incorporation of other existing standards. SBW, AS and GF have played an important role in implementation of Cancer registry data standards into the database and collection and quality assurance of follow-up and epidemiological data. NBW is the overall Project coordinator for this project. All authors have reviewed and commented on successive drafts of the manuscript and have provided the first author with approval of the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2407/8/236/prepub\"/></p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>This work is funded and supported by the Centers for Disease Control and Prevention (CDC) and the National Institute for Occupational Safety and Health (NIOSH) Grant 1U19OH009077-01 (IRB #0608194).</p>", "<p>We acknowledge all the following contributors to the development of the resource (in alphabetical order by last name):</p>", "<p><bold>Research Evaluation Panel: </bold>Dr. James Luketich, M.D, Dr. Michele Carbone, Dr. Margaret R. Karagas, Dr. Rodney Landreneau, MD, Dr. Brooke T. Mossman, Dr. Joseph R. Testa and Dr. Paolo Toniolo.</p>", "<p><bold>University of Pennsylvania School of Medicine, Department of Pathology: </bold>Lisa Miranda.</p>", "<p><bold>New York University School of Medicine: </bold>Amanda Beck, Kenneth Choi, Samantha Lincoln, Jacinta Small. Audrey Sorensen.</p>", "<p><bold>University of Pittsburgh School of Medicine: </bold>Michelle Bisceglia, Amelia M. Hensler, Linda Mignogna, Amita T. Mistry, Thomas Harper and William Shirey.</p>", "<p>Translating Research Across Community: Paula Kim</p>", "<p>We are thankful to the Department of Pathology at the University of Pittsburgh, School of Medicine, Pittsburgh, PA. for their help.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Presents the NMVB approved investigator query tool.</bold> This view provides de-identified data sets upon end-user query.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Presents the NMVB statistical summary for public view.</bold> This view provides overall statistics on all the cases and biospecimens stored into the resource.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional File 1</title><p>Honest Broker services for Tissue Banks and Clinical data.</p></caption></supplementary-material>" ]

[]

[ "<graphic xlink:href=\"1471-2407-8-236-1\"/>", "<graphic xlink:href=\"1471-2407-8-236-2\"/>" ]

[ "<media xlink:href=\"1471-2407-8-236-S1.docx\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Gastrointestinal cancers are major causes of cancer death throughout the world [##REF##15661684##1##,##REF##16368871##2##]. Recent advances of multimodal treatments have been improving the prognosis. Chemotherapy with combinations of new drugs including fluoropyrimidine, irrinotecan, and oxaliplatin, has greatly contributed to the prolonged prognosis [##REF##11006366##3##, ####REF##10744089##4##, ##REF##14665611##5##, ##REF##10944126##6##, ##UREF##0##7##, ##REF##15864716##8####15864716##8##].</p>", "<p>5-Fluorouracil (5-FU) is a key drug in combination chemotherapy and an evaluation of the predictability of 5-FU sensitivity is important to exclude those patients who would experience adverse effects. Among the molecular markers of 5-FU activity, thymidylate synthase (TS), dihydropyrimidine dehydrogenase (DPD), orotate phosphoribosyltransferase (OPRT), and thymidine phosphorylase (TP) are reported to be highly predictive of 5-FU sensitivity [##REF##145246##9##, ####REF##7577040##10##, ##REF##10778957##11##, ##REF##12576451##12##, ##UREF##1##13##, ##REF##10202277##14##, ##REF##9563897##15##, ##REF##2522792##16##, ##REF##14716816##17##, ##REF##15330198##18##, ##REF##10744051##19##, ##REF##14562021##20##, ##REF##16425285##21##, ##REF##11902527##22##, ##REF##12494248##23####12494248##23##]. 5-FU is catabolized to dihydrofluorouracil and inactivated by DPD. Thymidylate synthase is an essential DNA synthetic enzyme that is suppressed by 5-fluoro-deoxyuridine-monophosphate (FdUMP), an active metabolite of fluorouracil [##REF##145246##9##]. FdUMP and TS form covalent ternary complexes with 5, 10-methylene-tetrahydrofolate that subsequently inhibit DNA synthases [##REF##145246##9##,##REF##7577040##10##]. Colorectal cancer with both low DPD and low TS mRNA expressions has been reported to show greater antitumor effects in 5-FU-based chemotherapy [##REF##10778957##11##,##REF##12576451##12##]. Fluorouracil is converted to active metabolites by phosphorylation through three different pathways, and TP and OPRT are the key enzymes in two of these pathways [##UREF##1##13##]. Thymidine phosphorylase is an enzyme that activates 5'-deoxy-5-fluorouridine to 5-FU and then 5-FU to 5-fluoro-2'-deoxyuridine. Orotate phosphoribosyltransferase is an enzyme that converts 5-FU to 5-fluorouridine-5'-monophosphate (FUMP) and is considered to predominantly inhibit RNA synthesis. High expression of TP in a tumor is correlated with a high response rate to 5'-deoxy-5-fluorouridine [##REF##10202277##14##,##REF##9563897##15##] and high expression of OPRT in a tumor is correlated with sensitivity to 5-FU [##UREF##1##13##,##REF##2522792##16##]. For these reasons, many studies have reported that the activities of these enzymes have been associated with sensitivity to 5-FU-based chemotherapy in gastric cancer [##REF##14716816##17##, ####REF##15330198##18##, ##REF##10744051##19####10744051##19##] and colorectal cancer [##REF##10778957##11##,##REF##12576451##12##,##REF##14562021##20##, ####REF##16425285##21##, ##REF##11902527##22##, ##REF##12494248##23####12494248##23##].</p>", "<p>These observations were based on gene expressions evaluated by using fresh frozen materials, which were composed of cancer cells, stromal cells in the cancer tissues, and even normal tissues. Because gastric and breast cancers contain large amounts of stromal cells in the cancer tissues, gene expression evaluated by ordinary methods reflects that of the cancer tissues, but not cancer cells. To evaluate gene expression of the cancer cells alone, it is essential to isolate the cancer cells from the stromal cells. To achieve this, we used a laser capture microdissection plus the real time reverse transcription-polymerase chain reaction (RT-PCR) method (LCM+RT-PCR) on formalin-fixed paraffin-embedded (FFPE) samples. In our previous report, in which TS, DPD, and TP gene expressions in breast cancers were evaluated by this method, we disclosed that the gene expressions in cancer cells were significantly different from those in stromal cells [##REF##15890242##24##]. Gastric cancer, which contains large amounts of stromal cells in the cancer tissues, may show different gene expression between cancer cells and stromal cells as in breast cancers. In this study, gene expression levels of TS, DPD, TP, and OPRT in cancer cells of gastric cancer tissues were separately quantified from those in stromal cells by using the LCM+RT-PCR method. We also investigated those genes in colorectal cancer, which contains small amounts of stromal cells.</p>" ]

[ "<title>Methods</title>", "<title>Patient and samples</title>", "<p>Formalin-fixed paraffin-embedded samples from 47 patients with gastric cancers and 43 with colorectal cancers who underwent surgery were studied. This study was approved by the Institutional Review Board of the Tokyo Medical Dental University and all patients gave written consent.</p>", "<title>Laser capture microdissection</title>", "<p>A representative FFPE tumor was selected by a pathologist after examination of the hematoxylin and eosin-stained slides. 10 μm thick sections were stained with nuclear fast red to enable visualization of histology for LCM (P.A.L.M. Microlaser Technologies AG, Munich, Germany). Caner cells (Ca), cancerous stroma (Str) of the sample were dissected using the LCM technique.</p>", "<p>Formalin-fixed, paraffin-embedded tumor specimens and adjacent normal tissues were cut into serial sections with a thickness of 10 μm. For the pathological diagnosis, one slide was stained with H&amp;E and evaluated by pathologist (Figure ##FIG##0##1A##). Other sections were stained with nuclear fast red (NFR, American MasterTech Scientific Inc., Lodi, CA) to enable visualization of histology (Figure ##FIG##0##1B##). Laser capture microdissection (P.A.L.M. Microlaser Technologies AG, Munich, Germany) was performed in all the tumor samples to ensure that only tumor cells were dissected (Figure ##FIG##0##1C##). At least 25 mm²of tumor tissue or stromal tissue are collected from each FFPE block. Depending on the percentage of tumor tissue in the specimen, this generally can require from one to four 10 μm sections.</p>", "<title>Real-time quantitative RT-PCR</title>", "<p>RNA was isolated from these using a novel, proprietary procedure (Response Genetics, Los Angels, CA: United States Patent Number 6,248,535). After RNA isolation, cDNA was derived from each sample according to a previously described procedure [##REF##10675236##25##].</p>", "<p>Target cDNA sequence were amplified by quantitative PCR using a fluorescence-based real-time detection method(ABI PRISM 7900 Sequence Detection System [Taqman]; Applied Biosystems, Foster City, CA)as previously described [##UREF##2##26##,##UREF##3##27##]. The 25 μl PCR reaction mixture contained 600 nmol/l of each primer (Table ##TAB##0##1##), 200 nmol/l each of dATP, dCTP, and 1× Taqman buffer A containing a reference dye (all reagents were supplied by Applied Biosystems, Foster City, CA). The PCR conditions were 50°C for 10 s and 95°C for 10 min, followed by 42 cycles at 95°C for 15 s and 60°C for 1 min. TS, DPD, TP and OPRT gene expressions in each part of the tumors were quantified as ratios between two absolute measurements (gene of interest/beta-actin).</p>", "<title>Statistical analysis</title>", "<p>Comparison of mRNA levels between matched cancer cells and cancerous stroma was made with the Wilcoxon's rank test.</p>" ]

[ "<title>Results</title>", "<title>Gene expressions in cancer cells (Ca) and cancerous stroma (Str) in gastric cancer</title>", "<p>As shown in Figure ##FIG##1##2##, the median TS mRNA level was 2.4 in Ca and 0.47 in Str. The TS gene expression levels were significantly higher in Ca than in Str (p &lt; 0.0001). The median DPD mRNA level was 0.92 in Ca and 1.1 in Str. The DPD gene expression levels were significantly lower in Ca than in Str (p = 0.0136). The median TP mRNA level was 8.0 in Ca and 4.3 in Str. The TP gene expression levels were significantly higher in Ca than in Str (p &lt; 0.0001). The median OPRT mRNA level was 1.1 in Ca and 0.37 in Str. The OPRT gene expression levels were significantly higher in Ca than in Str (p &lt; 0.0001).</p>", "<title>Gene expressions in cancer cells (Ca) and cancerous stroma (Str) in colon cancer</title>", "<p>As shown in Figure ##FIG##2##3##, the median TS mRNA level was 1.4 in Ca and 0.44 in Str. The TS gene expression levels were significantly higher in Ca than in Str (p = 0.0002). The median DPD mRNA level was 0.30 in Ca and 0.93 in Str. The DPD gene expression levels were significantly lower in Ca than in Str (p &lt; 0.0001). The median TP mRNA was 2.4 in Ca and 3.8 in Str. The TP gene expression levels were significantly lower in Ca than in Str (p = 0.0055). The median OPRT mRNA level was 1.0 in Ca and 0.37 in Str. The OPRT gene expression levels were significantly higher in Ca than in Str (p &lt; 0.0001).</p>" ]

[ "<title>Discussion</title>", "<p>In this study, all the gene expressions were successfully estimated separately in cancer cells and cancerous stroma of human gastric and colon cancers using the LCM+RT-PCR technique. All of them showed significantly different expression levels between cancer cells and cancerous stroma.</p>", "<p>In both gastric and colon cancers, higher gene expression levels of TS and OPRT were observed in cancer cells than in cancerous stroma and a lower gene expression level of DPD was observed in cancer cells than in cancerous stroma. These gene expressions might be controlled by a common regulatory mechanism in gastric and colon cancers. In our previous study, these genes showed the same tendency in breast cancer [##REF##15890242##24##]. TS is the enzyme for DNA synthesis and cell proliferation and because cancer cells grow more rapidly than normal cells, TS gene expression is thought to be up-regulated in cancer cells. Several studies have revealed that the expression of OPRT is increased in several types of carcinoma, including gastric and colorectal carcinomas [##REF##16012756##28##, ####REF##15993511##29##, ##REF##16211289##30####16211289##30##]. OPRT is a nucleotide metabolic enzyme that is essential for cell proliferation. Thus, OPRT gene expression is thought to be up-regulated in cancer cells, like TS gene expression [##REF##1717130##31##,##REF##2162241##32##]. Evaluations of comparisons of the expression levels of DPD in cancer cells and in normal tissues are controversial. Differing results might be affected by the use of tumor tissue samples with various amounts of stromal tissue. However, in the recent study, the amount of stromal cells was taken into consideration, and the results agreed with those of previous reports that DPD expression in cancer cells was lower than in normal tissues and in stromal cells [##REF##15316940##33##,##REF##16132996##34##]. McLead <italic>et al. </italic>reported that the down-regulation of DPD expression may create a favorable environment for tumor growth. Low expression levels of DPD and decreased catabolism of uracil in cancer cells suggest that pyrimidine nucleotide pools increase [##REF##10499634##35##]. On the other hand, regarding the gene expression level of TP, the opposite result was observed. Although the gene expression level of TP was higher in cancer cells than in cancerous stroma in gastric cancer, it was lower in cancer cells than in cancerous stroma in colon cancer. It has been reported that certain solid tumors, including gastric and breast cancers, expressed elevated levels of TP as compared with stromal tissues [##REF##8758258##36##, ####REF##9355979##37##, ##REF##10353748##38##, ##REF##8757190##39####8757190##39##]. In colon cancer, several immunohistochemical studies reported that cancer cells had TP expression [##REF##8757190##39##,##REF##9570376##40##], while other studies reported that most cells expressing TP were stromal cells, especially macrophages and lymphocytes [##REF##9464495##41##,##REF##9652793##42##]. It was shown by analyzing the expression level of TP in cancer cells and cancerous stroma separately that cancer cells had TP expression, although stromal cells had higher TP expression than cancer cells. These current results regarding the expression of TS, DPD, OPRT, and TP agree with the previous data.</p>", "<p>In previous studies, biochemical assays, immunohistochemistry, enzyme-linked immunosorbent assay (ELISA), and reverse transcription-polymerase chain reaction (RT-PCR) have been used to evaluate the protein expression related to enzymes catabolizing 5-FU. Biochemical assays are often impossible to perform with minimal clinical samples. Many institutions have stored tissue samples as paraffin-embedded specimens after surgical resection, so immunohistochemical assays are convenient and inexpensive when performed on paraffin-embedded specimens. However, this method does not yield quantified results. Amplification by PCR may be performed on fresh frozen samples from resected cancer specimens to study gene expressions. However, the results are dependent on how promptly the samples were collected and stored. This method is not practical for the purpose of retrospective studies. In fresh frozen samples of cancer tissues, contamination with cancerous stroma and even normal tissue cannot be avoided. It is known that gastric and breast cancers contain large amounts of stromal tissue. We thought that the contaminations might influence the results of PCR, so we adopted LCM+RT-PCR. Laser capture microdissection provides selective isolation of defined cell populations from heterogeneous tissue sections [##REF##8875945##43##]. Moreover, the availability of real-time RT-PCR technology combined with the extraction of RNA from paraffin-embedded specimens allows quantitative and accurate measurement of gene expressions [##REF##10675236##25##,##REF##12576452##44##]. This novel method made it possible to analyze only cancer cells, so in spite of amounts of cancerous stroma, we could analyze intratumoral gene expressions equally. Furthermore, this method has another advantage. Because cancer cells and cancerous stroma are extracted separately within the same paraffin section, we can perform the analysis from a small amount of resected specimen.</p>", "<p>The time when various chemotherapies, including monoclonal antibody therapy has come, but the chemotherapy with 5-FU-based regimen has still played an important role as the treatment for many cancers. Accordingly it is thought to be useful to estimate gene expressions of enzymes, that are related to catabolism of 5-FU, in Ca and Str separately by LCM+RT-PCR. Though we need further investigate the relationship between those gene expressions and 5-FU sensitivity in clinical setting, we believe that with this technique it may be possible to predict the sensitivity of the agents before treatment using a small amount of a biopsy specimen and it leads to establish a tailor-made treatment for cancer patients.</p>" ]

[ "<title>Conclusion</title>", "<p>In this study, by using LCM+RT-PCR, we could analyze gene expressions in cancer cells and stromal tissues separately in FFPE gastric and colon cancer specimens. This method may make it possible to accurately analyze gene expressions in cancer cells from a small amount of a biopsy specimen in spite of the amount of stromal tissue and intratumoral heterogeneity.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Thymidylate synthase, dihydropyrimidine dehydrogenase, thymidine phosphorylase, and orotate phosphoribosyltransferase gene expressions are reported to be valid predictive markers for 5-fluorouracil sensitivity to gastrointestinal cancer. For more reliable predictability, their expressions in cancer cells and stromal cells in the cancerous tissue (cancerous stroma) have been separately investigated using laser capture microdissection.</p>", "<title>Methods</title>", "<p>Thymidylate synthase, dihydropyrimidine dehydrogenase, thymidine phosphorylase, and orotate phosphoribosyltransferase mRNA in cancer cells and cancerous stroma from samples of 47 gastric and 43 colon cancers were separately quantified by reverse transcription polymerase chain reaction after laser capture microdissection.</p>", "<title>Results</title>", "<p>In both gastric and colon cancers, thymidylate synthase and orotate phosphoribosyltransferase mRNA expressions were higher (p &lt; 0.0001, p &lt;0.0001 respectively in gastric cancer and P = 0.0002, p &lt; 0.0001 respectively in colon cancer) and dihydropyrimidine dehydrogenase mRNA expressions were lower in cancer cells than in cancerous stroma (P = 0.0136 in gastric cancer and p &lt; 0.0001 in colon cancer). In contrast, thymidine phosphorylase mRNA was higher in cancer cells than in cancerous stroma in gastric cancer (p &lt; 0.0001) and lower in cancer cells than in cancerous stroma in colon cancer (P = 0.0055).</p>", "<title>Conclusion</title>", "<p>By using this method, we could estimate gene expressions separately in cancer cells and stromal cells from colon and gastric cancers, in spite of the amount of stromal tissue. Our method is thought to be useful for accurately evaluating intratumoral gene expressions.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>HM prepared the formalin-fixed paraffin-embedded (FFPE) samples and drafted the manuscript. KD and PVD performed the LCM+RT-PCR. HU performed the statistical analysis. KS oversaw this study. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2407/8/210/prepub\"/></p>" ]

[]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>1A: Formalin-fixed, paraffin-embedded samples were stained with H &amp; E.</bold> 1B: Formalin-fixed, paraffin-embedded samples were stained with nuclear fast red. 1C: By laser capture microdissection , only tumor cells were dissected.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Gene expressions in cancer cells and cancerous stroma in gastric cancer</bold>. Expression levels of thymidylate synthase, orotate phosphoribosyltransferase, dihydropyrimidine dehydrogenase, and thymidine phosphorylase mRNA in cancer cells (Ca) and cancerous stroma (Str) were evaluated by laser capture microdissection plus reverse transcription-polymerase chain reaction. Thymidylate synthase, orotate phosphoribosyltransferase, and thymidine phosphorylase expression levels were higher in Ca than in Str. The dihydropyrimidine dehydrogenase expression level was higher in Str than in Ca.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Gene expressions in cancer cells and cancerous stroma in colon cancer</bold>. Expression levels of thymidylate synthase, orotate phosphoribosyltransferase, dihydropyrimidine dehydrogenase, and thymidine phosphorylase mRNA in cancer cells (Ca) and cancerous stroma (Str)were evaluated by laser capture microdissection plus reverse transcription-polymerase chain reaction. Thymidylate synthase and orotate phosphoribosyltransferase expression levels were higher in Ca than in Str. Dihydropyrimidine dehydrogenase and thymidine phosphorylase expression levels were higher in Str than in Ca.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Polymerase chain reaction primers and probes</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Primers/probes</td><td align=\"left\">Sequence</td></tr></thead><tbody><tr><td align=\"left\">TS2-764F(18 bp)</td><td align=\"left\">GCCTCGGTGTGCCTTTCA</td></tr><tr><td align=\"left\">TS2-830R(17 bp)</td><td align=\"left\">CCCGTGATGTGCGCAAT</td></tr><tr><td align=\"left\">Probe TS2-785T(21 bp)</td><td align=\"left\">6FAM-TCGCCAGC-</td></tr><tr><td align=\"left\">DPD3a-51F(19 bp)</td><td align=\"left\">AGGACGCAAGGAGGGTTTG</td></tr><tr><td align=\"left\">DPD3a-134R(20 bp)</td><td align=\"left\">GTCCGCCGAGTCCTTACTGA</td></tr><tr><td align=\"left\">Probe DPD3a-71T(29 bp)</td><td align=\"left\">6FAM-CAGTGCCTACAGTCTC-</td></tr><tr><td align=\"left\">TP3-700F(17 bp)</td><td align=\"left\">CCTGCGGACGGAATCCT</td></tr><tr><td align=\"left\">TP3-770ZR(20 bp)</td><td align=\"left\">GCTGTGATGAGTGGCAGGCT</td></tr><tr><td align=\"left\">Probe TP3-722T(25 bp)</td><td align=\"left\">6FAM-CAGCCAGAGATGTGA-</td></tr><tr><td align=\"left\">β-actin-592F(18 bp)</td><td align=\"left\">TGAGCGCGGCTACAGCTT</td></tr><tr><td align=\"left\">β-actin-651R(22 bp)</td><td align=\"left\">TCCTTAATGTCACGCACGATTT</td></tr><tr><td align=\"left\">Probe β-actin-611T(18 bp)</td><td align=\"left\">6FAM-ACCACCACGGCCGAGCGG</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>bp, base pairs</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2407-8-210-1\"/>", "<graphic xlink:href=\"1471-2407-8-210-2\"/>", "<graphic xlink:href=\"1471-2407-8-210-3\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>HIV/AIDS continues to be major public health threat with new infections on the rise. Current therapies do not completely cure the disease and there is no effective vaccine available [##REF##12835707##1##,##REF##12835704##2##]. A potentially rewarding approach is intracellular immunization using gene therapy strategies that protect viral susceptible cells from the infecting virus [##REF##15922953##3##]. Thus far, a number of promising intracellular immunization strategies have been employed using different anti-HIV molecules that act by a variety of mechanisms. Among these, nucleic acid-based approaches using ribozymes, antisense constructs, and siRNAs have received considerable attention due to their ease of expression and their non-immunological nature [##REF##15922953##3##,##REF##18066041##4##]. Some of these have entered clinical trials and safety testing with encouraging results [##REF##15922953##3##,##REF##18066041##4##]. In these studies either conventional retroviral vectors or lentiviral vectors were used for gene transfer. Although highly efficient for stable gene transfer, use of retroviral derived vectors poses a degree of risk in terms of viral mediated oncogenesis [##REF##16009045##5##]. Because of this potential risk, non-retroviral mediated gene delivery systems are being currently investigated. In this regard, Sleeping Beauty (SB) transposon system shows considerable promise [##REF##17164769##6##]. This system consists of a synthetic transposon and an associated transposase which functions by a cut and paste mechanism. Gene transposition is mediated by the transposase in a two step process in which the enzyme first recognizes the short inverted/direct (IR/DR) sequences in the transposon followed by the excision of the transposon and later integration of the transposon sequences into a target DNA region with a TA-dinucleotide sequence. The SB system can be deployed either as trans-delivery system in which the transposon and transposase are delivered by independent plasmids or a cis-delivery system in which both the components are incorporated into the same plasmid [##REF##16189271##7##]. Continued progress in this area has resulted in the derivation of more efficient transposases and more efficient gene delivery [##REF##17344320##8##]. Many mammalian cell types have been shown to be substrates for efficient SB mediated gene transfer including mouse embryonic stem cells [##REF##16084771##9##]. Thus, SB system offers a novel way of gene delivery for HIV gene therapy purposes.</p>", "<p>With regard to effective anti-HIV genes for gene therapy, siRNAs constitute highly effective gene silencing molecules due to their target specificity and improved potency [##REF##16716105##10##]. The siRNAs trigger an innate endogenous RNAi pathway for target recognition and gene silencing. Thus far, siRNAs targeted to a number of HIV genes have shown impressive gene down regulation and consequent viral inhibition both in vitro and in vivo [##REF##17726454##11##, ####REF##17145937##12##, ##REF##12907142##13##, ##REF##17406343##14####17406343##14##]. Due to their high target specificity however, a high possibility exists for siRNA viral escape mutants to arise during prolonged treatment. Indeed, such generation of viral escape mutants against specific siRNAs has already been documented [##REF##15687388##15##]. This possibility can be much reduced by targeting essential cellular molecules that aid in viral replication. Among the many cellular molecules shown to be involved in HIV infection and replication, the cell surface coreceptors CCR5 and CXCR4 are essential for viral entry by macrophage tropic R5 and T-cell tropic-X4 HIV respectively [##REF##16570858##16##,##REF##10358771##17##]. The primary HIV infection is established by R5 virus and during the later stages of disease, T-cell tropic X4 virus predominates [##REF##10358771##17##,##REF##15516706##18##]. In nature, a segment of the human population containing a 32-base pair deletion in the CCR5 gene, but apparently physiologically normal, was found to be resistant to infection by R5 tropic HIV-1 [##REF##10358771##17##,##REF##8756719##19##]. Therefore, CCR5 coreceptor is an ideal cellular target to suppress HIV infection. A number of previous studies including ours have successfully targeted both the HIV coreceptors by siRNA mediated gene silencing [##REF##17145937##12##,##REF##17670939##20##, ####REF##15813990##21##, ##REF##15306840##22####15306840##22##]. Down regulation of either of these coreceptors resulted in effective viral inhibition. However, retroviral derived vectors were used in these studies.</p>", "<p>With a long range goal of developing a non-viral gene delivery of anti-HIV genes for gene therapy, here we evaluated the utility of SB transposon system to deliver siRNA genes for stable gene transfer. Two previously well characterized siRNAs against CCR5 and CXCR4 coreceptors were introduced into SB transposon. Our results show that stable cell lines can be derived that harbor and express siRNA genes with concommittent HIV resistance.</p>" ]

[ "<title>Methods</title>", "<title>Construction of CCR5 and CXCR4 shRNA expressing SB constructs</title>", "<p>The Sleeping Beauty transposon vector pT/BH plasmid was obtained from Dr. Perry Hackett (University of Minnesota). The vector plasmid contains a multiple cloning site (MCS) flanked by a left and right inverted/direct repeat (IR/DR) elements [##REF##12842434##27##,##REF##16096013##28##]. Based on our previous data two well characterized and effective CCR5 and CXCR4 shRNAs were chosen for incorporating into the SB system plasmid [##REF##17597795##29##]. The CCR5 siRNA target sequence is 5'-GUGUCAAGUCCAAUCUAUG-3' whereas the CXCR4 siRNA target sequence is 5'-GAGUCUGAGUCUUCAAGUU-3'. The CXCR4 or CCR5 shRNA DNA cassette was generated by PCR using published protocol [##REF##12458798##30##]. In brief, PCR was done using U6 or H1 forward primer and a reverse primer containing 3'end homologous region of U6 or H1 promoter fused with CXCR4 or CCR5 shRNA sequence. The resulting PCR product was cloned into a Topo vector pCR8GW (Invitrogen, CA). A BglII site was engineered at the 5'end of forward and reverse primers. pT/BH was the transposon vector plasmid used into which a CMV driven RFP, IRES driven neomycin resistance gene and a SV40 polyadenylation signal containing cassette was cloned at the EcoRV site to derive the control RFP SB plasmid. To generate this pIRESneoRFP cassette, RFP gene was cloned as a BamHI – NotI fragment from pDsRed-N2 (Clontech, CA) into pIRESneo3 (Clontech, CA). A U6 promoter driven CXCR4 shRNA or H1 promoter driven CCR5 shRNA DNA cassette was cloned in parallel as a BglII-BglII fragment in the pT/BH plasmid at BamHI site to get pT/BH-U6CXCR4 or pT/BH-H1CCR5. The CMV-RFP-IRES-neo-SV40pA cassette was released as NruI-BstZ17I fragment and cloned at EcoRV site of pT/BH-U6CXCR4 or pT/BH-H1CCR5 plasmid to get pT/BH-U6CXCR4-CMV-RFP-IRES-neo or pT/BH-H1CCR5-CMV-RFP-IRES-neo. A hyperactive transposase expressing plasmid pHSB5, obtained from Dr. Mark Kay (Stanford University) was used to transpose the SB constructs [##REF##17344320##8##]. A schematic representation of SB constructs and transposase plasmid are shown in Figure ##FIG##0##1##.</p>", "<title>Cell culture and transfection</title>", "<p>Respective coreceptor expressing MAGI-CCR5 and MAGI-CXCR4 cell lines were obtained from the NIH AIDS Reagent Program and maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% FBS, 500 μg/ml G418, 100 μg/ml hygromycin and 1 μg/ml puromycin. Similar culturing conditions were used for GHOST-R3/X4/R5 cells with G418 concentration being 200 μg/ml [##REF##9971817##23##, ####REF##9201229##24##, ##REF##8649511##25####8649511##25##]. Cells were transfected with respective SB plasmids using Lipofectamine 2000 (Invitrogen, CA) as we previously described [##REF##15000821##31##].</p>", "<title>FACS analysis and sorting</title>", "<p>To enrich for transgenic cells, the SB transfected cells were subjected to FACS sorting based on RFP expression. The sorted cells were cultured for 4 weeks and analyzed by FACS to determine the cell surface down regulation by the respective siRNAs as described [##REF##16109172##32##]. Briefly, the transfected or untransfected control cells were washed in PBS and resuspended in FACS buffer. FITC conjugated anti-CXCR4 or anti-CCR5 antibody was added to the cells and incubated for 30 minutes at 4°C. Cells were then washed and resuspended in PBS for FACS which was done using a Coulter EPICS-XL MCL (Coulter Corporation, FL) machine and analysed with EXPO32 ADC software.</p>", "<title>HIV-1 challenge of siRNA transposed cells</title>", "<p>To determine viral resistance conferred by the down regulation of CCR5 and CXCR4 coreceptors, siRNA transposed or non-transposed cells were subjected to viral challenge with HIV-1 BaL (CCR5-tropic), HIV-1 NL4.3 (CXCR4-tropic) or HIV-1 89.6 (Dual-tropic) viral strains. The HIV-1 viral strains were obtained from the AIDS Research and Reference Reagent program, Division of AIDS, National Institute of Allergy and Infectious Diseases. Briefly, 0.5 × 10⁶transgenic GHOST-X4/R3/R5, MAGI-CXCR4 or MAGI-CCR5 cells in 6 well plates were washed and exposed to virus at an MOI of 0.01 in the presence of polybrene (4 μg/ml). Virus was allowed to adsorb for 2 hours at 37°C. Cells were then washed twice with PBS and 2 ml of complete DMEM was added [##REF##15813990##21##,##REF##15813986##33##]. Culture supernatants collected at different days post-challenge were assayed for p24 antigen by ELISA (Beckman-Coulter, CA).</p>", "<title>Transposed gene integration analysis</title>", "<p>To verify the stable transposition of the siRNA containing genes in the RFP expressing cell lines, the genomic DNA was isolated and subjected to Splinkerette PCR using a published protocol [##REF##11416868##26##]. Transposed cell genomic DNA was digested with Sau3AI (for left IR/DR junctional analysis) or NlaIII (for right IR/DR junction analysis). Splinkerretes were generated by heating equimolar amounts of long primerette (5'-CCTCCACTACGACTCACTGAAGGGCAAGCAGTCCTAACAACCATG-3') with the respective splink to 80°C and cooling it to room temperature. Splink BglII (5'-GATCCATGGTTGTTAGGACCTGGAGGGGAAATCAATCCCCT-3', 5'-phosphate) was used for left IR/DR and splink SphI (5'-GTTGTTAGGACTGCTTGGAGGGGAAAATCAATCAATCCCCT-3', 5'-phosphate) was used for right IR/DR. The splinkerretes were then ligated to the respective digested genomic DNA ends. Ligation was performed with 7.5 μM of splinkerette and 25 ng/μl of genomic DNA with T4 DNA ligase (Fermentas Inc, MD). Primary PCR was done using the ligation reaction as template with primerette short (5'-CCTCCACTACGACTCACTGAAGGGC-3') in conjunction with either long IR/DR (L2) (5'-CTGGAATTTTCCCAAGCTGTTTAAAGGCACAGTCAAC-3') for IR/DR (L) or long IR/DR (R) (5'-GCTTGTGGAGGCTACTCGAAATGTTTGACC-3') for IR/DR (R). Primary PCR was done with 10 cycles of 95°C for 5 sec and 70°C (-0.5°C per cycle) for 2 min followed by 20 cycles of 95°C for 5 sec and 65°C for 2 min. Nested PCR was done by using 1/250 dilution of primary PCR product within the secondary PCR reaction. The second PCR was done using primerette-nested (5'-GGGCAAGCAGTCCTAACAACCATG-3') in conjunction with new L1 (5'-GACTTGTGTCATGCACAAAGTAGATGTCC-3') for IR/DR (L) or IR/DR (R) KJC1 (5'-CCACTGGGAATGTGATGAAAGAAATAAAAGC-3') for IR/DR (R). Nested PCR was done with 30 cycles of 95°C for 5 sec, 61°C for 30 sec and 70°C for 90 sec. Both primary and nested PCR included a hot-start at 95°C for 1 min and a final extension of 70°C for 10 min. Oligonucleotides used for this assay were obtained from IDT (San Jose, CA). The PCR products were cloned using a Topo cloning kit (Invitrogen, CA) and sequenced for the junctional region. The sequencing was done by Laragen (Los Angeles).</p>" ]

[ "<title>Results</title>", "<title>Stable gene transfer of CXCR4 and CCR5 shRNAs by SB transposon system</title>", "<p>To investigate the utility of SB mediated gene transfer of anti-HIV-1 coreceptor siRNAs against CCR5 and CXCR4 we used the cell lines MAGI-CCR5 and MAGI-CXCR4 that constitutively express the respective individual coreceptors in addition to a GHOST-R3/X4/R5 cell line that constitutively expresses both [##REF##9971817##23##, ####REF##9201229##24##, ##REF##8649511##25####8649511##25##]. As described in the methods, the cells were transfected with the respective plasmid SB constructs. Expression of the transposed constructs was monitored by the presence of RFP fluorescence. The gene transposed cells were enriched by FACS sorting and were maintained in culture for six months to confirm stable expression of the transgenes. Expression of RFP was observed throughout the time of culture. We also evaluated cells transfected with SB constructs alone in the absence of the transposase. The RFP expression in these cells was lost within a week post transfection. In a separate set of drug selection experiments to determine the levels of gene transfer using SB system in HeLa cells, it was found that the levels of transposition were 19.5% for the RFP control (above the background 0.6% gene transfer without the transposase). The gene transfer levels for the CXCR4 siRNA and the CCR5 siRNA constructs were 10.5% and 12% respectively. To further confirm transposition mediated transgene integration in stably gene transposed cells, we analysed the genomic DNA for the presence of the respective constructs. This was achieved by PCR amplifying and sequencing the junctional region of transposon and chromosomal DNA [##REF##11416868##26##]. The typical hallmark of transposition is indicated by the presence of the dinucleotide 'TA' which was found at every insertion site analysed. To determine the transposed gene location, both left and right invert/direct repeats were sequenced at the chromosomal junctions. Sequences obtained were analysed using BLASTn software. Multiple integration events were recorded which spanned a range of chromosomal regions. The integration of representative individual SB transposons into the chromosomal DNA is summarized in Table ##TAB##0##1##. GHOST-R3/X4/R5 cells transposed with the control RFP transposon showed integration in Ch 5 and 17. Cells containing CCR5 siRNA showed Ch 5 and 20 regions at the transposon integration junction, while those transgenic for CXCR4 siRNA were found in Ch 17. In case of MAGI-CCR5 cells, control RFP transposon integrated into Ch 10 and 15. The CCR5 siRNA transposed cells showed integration in Ch 12 and 20. The integration sites for MAGI-CXCR4 cells were in Ch 6 and 12 for control RFP while those for CXCR4 siRNA transposon were in Ch 5 and 7. We also analyzed the copy numbers of integrated genes in GHOST-R3/X4/R5 cells using real time PCR. Our results showed 14.3, 6.5 and 10.8 copies per cell of the RFP control, CXCR4 siRNA and CCR5 siRNA constructs (data not shown).</p>", "<title>Down regulation of HIV-1 coreceptors CXCR4 and CCR5 in SB transposed siRNA transgenic cells</title>", "<p>The above data showed that SB transposed siRNAs are stably integrated into respective cells. We next evaluated if the stably gene modified cells show the effect of siRNA mediated gene silencing. Accordingly, the transposed cells were analysed for CXCR4 or CCR5 surface expression by FACS (Figure ##FIG##1##2##). Our results showed about 94% down-regulation of CXCR4 expression and a 97% down-regulation of CCR5 in GHOST-R3/X4/R5 cells transposed with CXCR4 or CCR5 siRNAs respectively. In the MAGI-CXCR4 cell line, the CXCR4 expression was reduced by 98% by the respective siRNA, while MAGI-CCR5 cells showed a 99% reduction in CCR5 levels as a result of respective transposon mediated siRNA expression (data not shown). Cells transposed with control SB construct without siRNA insert showed no decrease in coreceptor expression with levels similar to that shown by control unmanipulated cells. The levels of coreceptor down regulation obtained with these siRNAs in SB system are similar to that seen with that delivered via lentiviral vectors (data not shown). These results confirmed the efficacy of the respective siRNAs in mediating gene silencing of the HIV-1 coreceptors.</p>", "<title>SB transposed anti-CCR5 and CXCR4 siRNAs confer HIV-1 resistance</title>", "<p>To determine if down regulation of CCR5 and CXCR4 coreceptors conferred viral resistance, siRNA transgenic GHOST-R3/X4/R5 cells were challenged with X4-tropic (NL4-3), R5-tropic (BaL-1) and dual coreceptor tropic HIV-1 89.6 strain. Antigen ELISAs to detect viral p24 in culture supernatants were performed on various days post-infection up to three weeks (Figure ##FIG##2##3##). When challenged with X4-tropic HIV-1 NL4.3, GHOST-R3/X4/R5 cells expressing CXCR4 siRNA showed a 10 fold decrease in virus production as compared to control non-transgenic cells on day 10 post-infection. The level of viral inhibition reached upto 14 fold through day 21 post-infection. In contrast CCR5 siRNA expressing GHOST-R3/X4/R5 cells failed to show any inhibition of virus production against X4 tropic HIV-1 NL4.3. Viral challenge of GHOST-R3/X4/R5 cells expressing CCR5 siRNA with the R5-tropic HIV-1 BaL resulted in an 8 fold reduction in virus production on day 10 post-infection, which doubled to 16 fold on days 14 and 21 post-infection. GHOST-R3/X4/R5 cells expressing CXCR4 siRNA served as a negative control as they showed similar levels of infection seen in control non-transgenic cells with the R5-tropic virus challenge. In dual-tropic HIV-1 89.6 viral challenges, neither of the individual CXCR4 siRNA or CCR5 siRNA expressing GHOST-R3/X4/R5 cells showed significant protection as expected since the challenge virus could use either of the coreceptors. However there was a moderate decrease in the virus production on day 21 as compared to unmanipulated cells. Cells transposed with SB control construct without anti-HIV transgenes showed similar levels of infection as the unmanipulated cells for all three HIV-1 strains. We also challenged SB transposed MAGI-CCR5 and MAGI-CXCR4 cells with R5 or X4 tropic viral strains respectively and found similar levels of resistance (data not shown). These data collectively showed that the respective SB system delivered siRNAs are functional and mediate viral resistance.</p>" ]

[ "<title>Discussion</title>", "<p>As a first step towards exploiting a non-viral gene transfer system for HIV gene therapy, here we have shown that SB transposon system can be utilized for deriving stably gene modified cells that display HIV resistance. To achieve this goal, we employed siRNAs with proven efficacy to down regulate expression of the essential HIV-1 coreceptors CCR5 and CXCR4 with a consequent viral resistance phenotype. To our knowledge this is the first report describing gene transfer for viral resistance using a transposon system.</p>", "<p>GHOST-R3/X4/R5 cells constitutively expressing both CCR5 and CXCR4 coreceptors were used for SB mediated siRNA gene transfer in these proofs of concept studies. Since the general gene transfer efficiency is low relative to that typically obtained with lentiviral vectors [##REF##16189271##7##,##REF##15813986##33##,##REF##17005721##34##], transfected cells were enriched by FACS sorting to evaluate the effectiveness of the stably integrated siRNA transgenes. Our results have shown that transgenic cells could be cultured indefinitely with stable expression of the transposed genes. FACS analysis of the siRNA modified cells showed consistent down regulation of the respective receptors CCR5 and CXCR4 amounting up to a 94% down regulation whereas cells transposed with control SB construct lacking siRNA transgenes showed normal levels of coreceptor expression similar to unmanipulated cells. Thus, down regulation of the respective targeted coreceptors established that siRNA transgenes are functional in a SB transposon system. As determined in the viral challenge experiments, siRNA transgenic cells also showed HIV resistance. With regard to individual siRNAs, GHOST-R3/X4/R5 cells transposed with CCR5 siRNA were found to be resistant to R5 HIV-1 viral challenge, whereas the cells transposed with CXCR4 siRNA were resistant to X4 HIV-1 viral challenge thus confirming the specificity of the respective siRNAs in mediating viral resistance. As expected, no significant protection could be seen from a dual tropic viral challenge of either of the individual siRNA gene modified cells since this viral strain could use either of the coreceptors for cellular entry.</p>", "<p>To further confirm stable gene transposition of the siRNA genes, we also mapped the integration sites of the SB transposon in respective transfected cells and found that these representative cell clones harbored the transgenes in different chromosomes namely 5, 6, 7, 10, 12, 15, 17 and 20. Previous studies mapped numerous SB-mediated integration sites in cultured and primary cells and found no chromosomal preference for insertion [##REF##15743807##35##,##REF##18047689##36##]. Consistent with this observation, the above clones transposed with siRNAs also represent random transposition events.</p>", "<p>The non-viral nature of the SB system offers some advantages over the more common retro and lentiviral mediated gene transfer [##REF##17073604##37##]. Among these are that no viral sequences are involved thus minimizing insertion transcriptional activation of cellular genes and risk of generation of replication competent viruses during vector production. However, the gene transfer efficiency with the SB system remains sub-optimal compared to the viral vector systems [##REF##16084771##9##]. Future improvements in the SB system are necessary to achieve higher gene transfer efficiency to be clinically practical [##REF##17164769##6##,##REF##16084771##9##].</p>", "<p>Although shown to be effective in conferring HIV resistance to cultured cells, the present SB system needs to be further evaluated in a hematopoietic stem cell setting using CD34 progenitor cells with a high efficiency of gene transfer to be clinically useful as shown with lentivirus vectors [##REF##15922953##3##,##REF##18066041##4##]. Even if high enough efficiency gene transfer is not achievable with this system in the near future, other innovative approaches are possible that may show clinical utility. For example, currently human embryonic stem (hESC) cells show great promise in developing novel therapies [##REF##18024604##38##,##REF##17016468##39##]. The hESC have already been shown to be amenable to gene transfer with SB transposon system, and it is now routine to derive hematopoietic CD34 cells from hESC as shown by us and others [##REF##17673526##40##, ####REF##16623949##41##, ##REF##18215326##42##, ##REF##16844782##43##, ##REF##18032700##44####18032700##44##]. One can envisage that hESC can be transposed with anti-HIV siRNAs using SB system and high expressing cell clones could be derived. From these transgenic hESC clones, unlimited numbers of siRNA expressing CD34 cells could be derived for HIV gene and cell therapies. Such experiments are currently underway in our laboratory.</p>" ]

[ "<title>Conclusion</title>", "<p>SB gene transposon system can be used to deliver siRNA genes against HIV-1 coreceptors CCR5 and CXCR4 for stable gene expression. The siRNA genes are able to downregulate the respective coreceptor expression on the cell surface and thus confer resistance against HIV-1 infection by restricting viral entry. These studies have demonstrated for the first time the utility of the non-viral SB system to derive viral resistant cells and paved the way for the use of this system for HIV gene therapy studies.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Thus far gene therapy strategies for HIV/AIDS have used either conventional retroviral vectors or lentiviral vectors for gene transfer. Although highly efficient, their use poses a certain degree of risk in terms of viral mediated oncogenesis. Sleeping Beauty (SB) transposon system offers a non-viral method of gene transfer to avoid this possible risk. With respect to conferring HIV resistance, stable knock down of HIV-1 coreceptors CCR5 and CXCR4 by the use of lentiviral vector delivered siRNAs has proved to be a promising strategy to protect cells from HIV-1 infection. In the current studies our aim is to evaluate the utility of SB system for stable gene transfer of CCR5 and CXCR4 siRNA genes to derive HIV resistant cells as a first step towards using this system for gene therapy.</p>", "<title>Results</title>", "<p>Two well characterized siRNAs against the HIV-1 coreceptors CCR5 and CXCR4 were chosen based on their previous efficacy for the SB transposon gene delivery. The siRNA transgenes were incorporated individually into a modified SB transfer plasmid containing a FACS sortable red fluorescence protein (RFP) reporter and a drug selectable neomycin resistance gene. Gene transfer was achieved by co-delivery with a construct expressing a hyperactive transposase (HSB5) into the GHOST-R3/X4/R5 cell line, which expresses the major HIV receptor CD4 and and the co-receptors CCR5 and CXCR4. SB constructs expressing CCR5 or CXCR4 siRNAs were also transfected into MAGI-CCR5 or MAGI-CXCR4 cell lines, respectively. Near complete downregulation of CCR5 and CXCR4 surface expression was observed in transfected cells. During viral challenge with X4-tropic (NL4.3) or R5-tropic (BaL) HIV-1 strains, the respective transposed cells showed marked viral resistance.</p>", "<title>Conclusion</title>", "<p>SB transposon system can be used to deliver siRNA genes for stable gene transfer. The siRNA genes against HIV-1 coreceptors CCR5 and CXCR4 are able to downregulate the respective cell surface proteins and thus confer resistance against viral infection by restricting viral entry. These studies have demonstrated for the first time the utility of the non-viral SB system in conferring stable resistance against HIV infection and paved the way for the use of this system for HIV gene therapy studies.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>MT derived the experimental data and RA was responsible for the conception and overall implementation of the project. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>Work reported here was supported by NIH RO1 grants AI50492 and AI057066 to R.A. We thank Perry Hackett for the SB transposon plasmid, Mark Kay for the hyperactive SB transposase, Karen Helms and Leslie Armstrong for help with FACS sorting. We thank the NIH AIDS Research and Reference Reagents Program for HIV-1 related reagents used in this work.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Schematic representation of siRNA SB constructs</bold>. A) Control SB transposon plasmid construct with Neo resistance and RFP reporter genes. RFP is driven by a CMV promoter whereas the Neo resistance is expressed via IRES. B) SB transposon construct incorporating anti-CXCR4 siRNA driven by Pol III U6 promoter. C) SB transposon construct incorporating anti-CCR5 siRNA driven by Pol III H1 promoter. D) Plasmid construct encoding the hyperactive transposase under CMV promoter.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Cell surface down regulation of CCR5 or CXCR4 coreceptors in siRNA transfected GHOST-R3/X4/R5 cells</bold>. GHOST-R3/X4/R5 cells that constitutively express CCR5 and CXCR4 coreceptors were transfected with control RFP, CCR5 or CXCR4 siRNA constructs. RFP expressing transgenic cells were FACS sorted and cultured. To determine the down regulation of respective coreceptors, the cells were stained with respective FITC tagged antibodies and FACS analyzed. The down regulation of CCR5 coreceptor (Panel A) <italic>was determined </italic>by comparing CCR5 levels in untransfected (A1), control RFP transfected (A2) and CCR5 siRNA transfected (A3) cells. The CXCR4 coreceptor down regulation is shown by comparing CXCR4 levels in untransfected (B1), control RFP transfected (B2) and CXCR4 siRNA transfected (B3) cells. The percent down regulation of CCR5 (A4) or CXCR4 (B4) coreceptors is also indicated.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>HIV-1 challenge of siRNA transposed GHOST-R3/X4/R5 cells</bold>. To determine viral resistance, siRNA transposed transgenic cells were challenged with HIV-1 NL4.3 (CXCR4 tropic virus), HIV-1 BaL (CCR5 tropic virus) or HIV-1 89.6 (dual tropic virus) viruses at an MOI of 0.01. On various days post-infection, cell culture supernatants were collected and analyzed for p24 antigen levels by ELISA to determine the levels of viral inhibition. Untransposed (◆), control RFP transposed (■), CXCR4 siRNA transposed (×) or CCR5 siRNA transposed (○). Panel A – NL4.3, Panel B – BaL, Panel C – 89.6.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Chromosomal integration of different SB constructs.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Cell Line</td><td align=\"left\">SB Construct</td><td align=\"left\">Chromosomal Location</td></tr></thead><tbody><tr><td align=\"left\">GHOST-R3/X4/R5</td><td align=\"left\">RFP control</td><td align=\"left\">Ch 5q34-q35, Ch 17q25.1</td></tr><tr><td/><td align=\"left\">CXCR4 siRNA</td><td align=\"left\">Ch 17q23.3</td></tr><tr><td/><td align=\"left\">CCR5 siRNA</td><td align=\"left\">Ch 5q34-q35.1, Ch 20q13.2</td></tr><tr><td align=\"left\">MAGI-CXCR4</td><td align=\"left\">RFP control</td><td align=\"left\">Ch 6p22.3, Ch 12q14.2</td></tr><tr><td/><td align=\"left\">CXCR4 siRNA</td><td align=\"left\">Ch 5q33.1, Ch 7q31.1</td></tr><tr><td align=\"left\">MAGI-CCR5</td><td align=\"left\">RFP control</td><td align=\"left\">Ch 10p12.31, Ch 15q11</td></tr><tr><td/><td align=\"left\">CCR5 siRNA</td><td align=\"left\">Ch 12p11.2, Ch 20q13.3</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1742-6405-5-16-1\"/>", "<graphic xlink:href=\"1742-6405-5-16-2\"/>", "<graphic xlink:href=\"1742-6405-5-16-3\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Cutaneous malignant melanomas are highly aggressive tumors with unpredictable biological behavior [##REF##16405570##1##]. Metastases in brain, bones and viscera with subsequent ascites development, are frequent [##REF##16405570##1##]. The progression of a transformed melanocyte to malignant melanoma is accompanied by gradual acquisition of multiple genetic alterations that lead to losses of onco-suppressor genes and increased tumor hypermutability [##REF##15280398##2##]. Malignant melanomas display both types of known genomic instability in neoplasia; chromosomal instability (CIN) and microsatellite instability (MIN) [##REF##15280398##2##,##REF##9872311##3##]. MIN has been observed in 30% of cutaneous malignant melanomas [##REF##11437939##4##]. However, the great majority of malignant melanomas examined by various cytogenetic methods, exhibit increased karyotypic complexity, extended aneuploidy and heteroploidy [##REF##7553595##5##, ####REF##11943340##6##, ##REF##14618616##7####14618616##7##]. Recurrent chromosomal imbalances in skin melanomas include losses of chromosomes 1p, 6q and 9p [##REF##15280398##2##,##REF##14578177##8##]. Tumor progression and aggressive behavior have been associated with imbalances of chromosomes 7, 10 and 17 [##REF##15280398##2##,##REF##14618616##7##].</p>", "<p>Most human tumors including melanomas maintain sufficient telomere length for continuous growth by expressing telomerase [##REF##7605428##9##,##REF##16920589##10##], the remainder are thought to utilize a variety of telomere recombination mechanisms termed alternative lengthening of telomeres (ALT) [##REF##15548547##11##]. Observations on transformed and tumor cell lines that lack telomerase, linked ALT phenotype to highly increased structural chromosomal instability and extreme telomeric length deviation ranging from very long to extremely short telomeres [##REF##15548547##11##]. We report a MIN, CIN, and TRAP negative (Telomere Repeat Amplification Protocol), highly aggressive melanoma metastasis to the peritoneal cavity, with unusually stable abnormal pseudodiploid karyotype, and relatively short but not dysfunctional telomeres.</p>" ]

[ "<title>Methods</title>", "<title>Immunochemistry-Cytopathology</title>", "<p>Cell material from the peritoneal aspirations was subjected to routine diagnostic cytopathology protocols including Giemsa, Papanicolaou, Hematoxylin-Eosin (BDH-Chemicals) stains and immunocytochemistry using the melanocyte specific antibody S-100 (Dako). Immunocytochemical staining against S-100 was performed using Horse Radish Peroxidase (HRP) (Dako). Cell smears were re-hydrated, treated with 3% hydrogen peroxide for 15 minutes and rinsed with Tris Buffered Saline with 0.05% Tween 20 (Dako). After cooling for 20 min, sections were incubated with the primary antibody (rabbit antihuman monoclonal S-100 antibody, 1:400 dilution, Dako) for 1 hour at room temperature and then incubated for 45 min with an anti-mouse HRP labelled polymer (EnVision+System-HRP, Dako). Finally slides were treated with a diaminobenzidine (DAB) chromogenic substrate (Dako) for 10 min, counterstained with hematoxylin, dehydrated and coverslipped.</p>", "<title>Short term cultures/Cytogenetic analysis</title>", "<p>Malignant cells from two peritoneal aspirations were collected by centrifugation (10 min/1500 rpm/25°C). They were subsequently cultured in eight 25 cm²T-flasks at 37°C and 5% CO₂, in Dulbecco's Minimum Essential Medium supplemented with 10% fetal bovine serum, 0.08 mg/ml amphotericin, 25 units/ml penicillin, and 25 pg/ml of streptomycin (Invitrogen). When high mitotic index was reached, cells were exposed to colcemid (0.1 μg/ml) (Invitrogen) for 30 min, in 37°C, and harvested using trypsine (Invitrogen), after 0.075 KCL hypotonic treatment and Methanol/Acetic acid (BDH-Chemicals) fixation. For the construction of the representative karyotype, we combined G-Banding after Trypsine and Giemsa (GTG-Banding), inverted 4',6-diamidino-2-phenylindole (DAPI)-banding, subtelomeric FISH (TeloVysion-Vysis) and telomeric FISH (Dako), in a total of 400 metaphases from 8 short term monolayer cell cultures. For dual-color interphase or metaphase FISH we used satellite probes specific for chromosomes 7 and 17 (Cytocell). In brief, our general FISH protocol was based on pepsin pre-treatment, formamide or NaOH target denaturation, over-night hybridization and high stringency post hybridization washes. Telomere-specific Peptide Nucleic acid Analog (PNA) hybridizations were performed using a Cy3-(Indocarbocyanine)- conjugated (CCCTAA)₃probe (Dako), according to manufacturer's instructions. All FISH preparations were mounted and counterstained with VectaShield antifade medium (Vector), containing 0.1 μg/ml DAPI (Sigma). GTG-Banding was performed after trypsine denaturation (Invitrogen) and Giemsa (BDH-Chemicals) staining. Digital images were captured in a Perceptive Systems Imaging, a Metasystems or an Applied Imaging molecular cytogenetics workstations equipped with fluorescent Zeiss, or Nikon microscopes. Quantification of telomeric PNA fluorescence was performed in 500 chromatids on DAPI counterstained metaphase preparations in a single hybridization experiment using the Isis software (Metasystems).</p>", "<title>Microsatellite instability assay</title>", "<p>Two mononucleotide markers, BAT-25 and BAT-26 were tested for microsatellite instability by radioactive PCR after Polyacrylamide Gel Electrophoresis (PCR-PAGE) assay, using the following primers: BAT25.1 (5'-TCGCCTCCAAGAATGTAAGT-3'), BAT25.2 (5'-TCTGCATTTTAACTATGGCTC-3'), BAT26.1 (5'-TGACTACTTTTGACTTCAGCC-3') and BAT26.2 (5'-AACCATTCAACATTTTTAACCC-3'). Experiment was monitored by controls for human microsatellite stability (normal genomic and MIN DNA from a patient with human Hereditary Non-Polyposis Colon Cancer – HNPCC-).</p>", "<title>TRAP assay</title>", "<p>Telomerase activity of cell lysates was analyzed by the telomeric repeat amplification protocol (TRAP) assay with a TRAPeze Telomerase Detection kit (Intergen) according to manufacturer's instructions. Approximately 10⁶cells were harvested and lysed in 400 μl of 1× CHAPS (3-[(3-Cholamidopropyl)dimethylammonio]propanesulfonic acid, 3-[(3-Cholamidopropyl)-dimethylammonio]-1-propanesulfonate) lysis buffer [Tris-HCl 10 mM, pH 7.5; 1 mM EGTA (ethylene glycol tetraacetic acid), 1 mM MgCl₂, 0.5% CHAPS 10% glycerol, DEPC (Diethylpyrocarbonate) treated water on ice for 30 min. Cell debris were spun down for 20 minutes at 12,000 r.p.m at 4°C. Each reaction was carried out by using 2 μl of supernatant, 1 μl of each primer, 0.5 μl of Taq-Polymerase (TAKARA), 10 μl of solution-Q (Qiagen), 5 μl of 10× buffer, 2 μl of dNTPs, in DEPC treated water in final volume of 50 μl. The primers used for the TRAP-assay PCR, were TS-5'-AATCCGTCGAGCAGAGTT-3' and Cxa-5'-GTGTAACCCTAACCCTAACCC-3'. The PCR program consisted first of an incubation at 30°C for 30 min and then in a thermocycler, 94°C for 2 min; 94°C for 30 s, 50°C for 25 s, 72°C for 30 s (33×); 72°C for 1 min. PCR products were electrophoresed in a 10% 19:1 acrylamide gel (Sigma)/0.5× TBE (Tris/Borate/EDTA) buffer using the mini protean II gel system (Biorad). Gels were stained with 2 μl of SYBR Green (Sigma) for 15 min at room temperature in 50 ml of TBE 0.5× buffer, and then exposed to UV light and visualized by a Kodak image acquisition station.</p>", "<title>Flow FISH</title>", "<p>To measure cellular telomere length, short term cultured cells were hybridized in situ with a fluorescent telomere-specific peptide nucleic acid probe, according to manufacturer's protocol. Briefly, cells were washed in PBS, and re-suspended to 10⁵cells/100 μl of a hybridization mixture (Dako) containing 70% formamide and a telomere-specific FITC (Fluorescein isothiocyanate)-conjugated PNA probe. Control samples were re-suspended in hybridization solution without probe to obtain background fluorescence values. After hybridization, cells were spun down and washed twice with 4 ml PBS (Phosphate Buffered Saline) at 40°C for 10 min and finally re-suspended in PBS containing 0.1% Bovine Serum Albumin, 10 μg/ml RNase A (Roche) and 0.1 μg/ml propidium iodide (Calbiochem-Novabiochem). Cells were analyzed on a FACScan flow cytometer (Becton Dickinson) or stored at 4°C before analysis.</p>" ]

[ "<title>Results</title>", "<title>Patient history and ascitic fluid samples</title>", "<p>Peritoneal fluid samples were obtained by two subsequent paracenteses (within a 12-day interval) of a 38-year-old woman, presented at the Department of Gynecology, Laikon Hospital, with ascites and solid structures at her ovaries as revealed by CT-scan. Two years ago the patient had a less than 1.5 cm large, cutaneous nevus excised from the anterior surface of her left hip. The primary tumor was characterized as a nodular melanoma, Clark's level 3, Breslow's depth 2.0 mm. One out of 14 inguinal nodes, excised in a subsequent operation, was found to be invaded. She received 6 cycles of chemotherapy (cis-platin-dacarbazine) and remained disease-free for 15 months. The cytologic examination of the ascitic aspiration confirmed the presence of malignant cells positive for the melanocyte specific antibody S-100 (Figure. ##FIG##0##1##). The patient refused to be operated, gave her written consent for further research on the specimens obtained, and expired 40 days after presentation.</p>", "<title>Cytogenetic analysis</title>", "<p>G-Banding analysis (according to ISCN 1995) [##UREF##0##12##] from 8 short-term cell cultures of two peritoneal aspirations taken in an interval of 12 days, showed a 46,XX,del(6)(q23?qter),del(9)(p10pter),der(10)t(7;10)(q31.3qter::p13)del(10)(p14?pter),der(11)t(5;11)(q22.3qter;q23)del(11)(q24?qter),i(17q) pseudodiploid karyotype, in 94–96% of 200 mitoses examined (Figure. ##FIG##1##2A##). Endoreduplication was observed in 4–6% of the malignant cells leading to a 92,XXXX,idemx2 karyotype. Subtelomeric FISH specific for all human telomeres except for chromosomes 16, 19, 20 and the short arms of acrocentric chromosomes, was used to assist in the description of marker chromosomes identified by G-Banding (Figure. ##FIG##1##2A##), and to verify deletions spanning up to the end of rearranged chromosomes. To examine if this remarkable karyotypic stability was not confined only to dividing mitotic cells, we performed dual color interphase FISH with probes specific for centromeres 7 and 17, in 200 interphase nuclei obtained from 2 short-term cell cultures from both aspirations. Centromeres 7 and 17 showed notable numerical stability in these populations. The rates of whole genome endoreduplication were similar to those of the karyotyped mitotic cells (Figure. ##FIG##1##2B##).</p>", "<title>Examination of factors related to chromosome stability</title>", "<p>In an attempt to attribute the karyotypic stability of this metastatic melanoma to measurable parameters related to chromosome stability in the context of neoplastic continuous growth, we examined microsatellite instability (MIN) and telomerase activity. Microsatellite unstable tumors show a significantly lower rate of chromosomal instability as compared to the MIN negative [##REF##9872311##3##]. To rule out underlying microsatellite genomic instability in this metastatic melanoma, we tested by PCR-PAGE the robust mononucleotide repeat markers BAT-25 and BAT-26. Both loci have been shown to be sensitive markers of MIN [##REF##7604266##13##]. Compared to positive and negative controls, this metastatic melanoma displayed no micro-satellite instability (Figure. ##FIG##2##3A##). Ectopic expression of telomerase in normal fibroblasts has been connected to karyotypic stability [##REF##9916803##14##]. We conducted a TRAP assay to test telomerase activity in cultured cells from 2 sub-cultures from both peritoneal aspirations. In both samples this assay was negative (Figure. ##FIG##2##3B##). To examine if these melanoma cells followed the ALT-pathway of telomere maintenance [##REF##15548547##11##] we compared the relative telomeric length of our specimen by Quantitative-PNA-Flow-FISH [##REF##11813198##15##] with the pseudodiploid human acute T cell leukemia JURKAT cell line, normal human fibroblasts and an ALT-positive cell line [##REF##11439347##16##]. Cell material for this test was obtained from a short-term subculture that was previously karyotyped and found to be composed exclusively from chromosomally abnormal mitotic cells. This comparison revealed that the melanoma cells had relatively short telomeres (Figure. ##FIG##2##3C##). PNA-telomeric FISH on 500 chromatids from 10 randomly picked metaphase spreads showed that most of the 46 chromosomes of this metastatic melanoma were uniformly capped with telomeric repeats (Figure. ##FIG##2##3D##) and no signs of structural chromosome instability attributed to telomere dysfunction such as end-to-end fusions and dicentric chromosomes were evident.</p>" ]

[ "<title>Discussion</title>", "<p>Metastatic transition in most human tumors is accompanied by a series of complex recurrent and stochastic chromosomal anomalies. These changes reflect the evolutionary pressure held by the cancer cells to bypass natural barriers and re-establish continuous growth into unrelated histopathologic environments [##REF##959840##17##,##REF##15743675##18##]. In this report, the karyotype of the primary tumor is not available, therefore the relative simplicity of genomic imbalances encountered in metastasis, permits only a hypothetical reconstruction of the chromosomal evolution of the disease. It has been proposed that melanomas develop through a mode of karyotypic evolution, common to both low and high complexity karyotypes [##REF##15280398##2##]. To become malignant, an apparently normal melanocyte of this patient underwent multiple karyotypic alterations involving breakpoints in at least 7 different chromosomes as well as chromosomal losses and non-disjunctions. Although we cannot define the temporal order of the recorded rearrangements, we postulate that the hemizygous deletions 6q23qter and 9p- might be early events in the chromosomal evolution of this melanoma. Translocations and deletions involving the q-arm of human chromosome 6 have been found in more than 80% of melanomas [##REF##7553595##5##]. According to Hoglund et al (2005) [##REF##15645488##19##], deletions of the distal 6q should be considered early chromosomal lesions in melanomas. Moreover, the short arm of chromosome 9 is the site of several cell cycle regulators that have been linked with familial disease, or associated to melanoma progression and aggressive behavior [##REF##15280398##2##]. The gains of genomic material and the additional deletions involving 10p, 11q and 17p, were by-products of unbalanced chromatid separation of balanced translocations and the isochromosome formation. These more complex alterations might represent later events in the process of the karyotypic evolution of the disease. Chromosomes 10 and 11 are frequently lost in metastatic melanomas whereas chromosome 7 is frequently gained [##REF##15280398##2##,##REF##7553595##5##,##REF##14618616##7##,##REF##12112321##20##,##REF##17363583##21##].</p>", "<p>Rearrangements affecting the short arm of chromosome 17, where the p53 gene is located, have been implicated in the pathogenesis of malignant melanoma [##REF##15280398##2##]. It is interesting that although p53 deficiency has been related to increased rates of numerical chromosome instability or polyploidy [##REF##14702042##22##], in this melanoma hemizygosity of p53 was not associated with continuous genomic instability. MIN tumors display extremely low rates of CIN [##REF##9872311##3##]. We ruled-out the possibility that this melanoma belonged to this type of tumors. We also ruled-out CIN in our specimens, since this metastatic pseudodiploid tumor was highly cytogenetically stable by all means examined. These results are compatible with those of Abdel-Rahman et al. 2001 and Fabarius et al. 2003, who observed that chromosomes of near-diploid cells are structurally much more stable than those of highly aneuploid counterparts [##REF##11226274##23##,##REF##12742157##24##]. Perhaps, the rare, melanoma described here, is unusually stable, because it is near-diploid, in contrast to the majority of highly aneuploid genomically unstable melanomas.</p>", "<p>The majority of human malignant melanomas and melanoma cell lines studied with the TRAP assay were found to express telomerase activity [##REF##16920589##10##,##REF##10408692##25##]. Furthermore, telomerase activity has been connected to aggressiveness of melanomas [##REF##10195394##26##]. In continuous neoplastic growth, insufficiently protected telomeres tend to undergo end-to-end fusions and to produce numerous complex chromosome rearrangements such as dicentric chromosomes and inverted duplications [##REF##7957062##27##, ####REF##10805796##28##, ##REF##16166375##29####16166375##29##]. No evidence of such lesions was found in our specimens. The transient stage of structural chromosomal instability in this case, equally involved subtelomeric, centromeric and genomic regions, and gave rise to translocations with canonical orientation. Surprisingly, this metastatic tumor was negative for telomerase activity. Moreover, no signs of recombinatorial telomere elongation were present [##REF##15548547##11##] since flow FISH showed relatively short telomeres and PNA FISH displayed a uniform terminal capping of virtually all chromosomes of this melanoma with TTAGGG repeats.</p>", "<p>The remarkable stability, and telomeric integrity of the metastatic tumor presented here, can be attributed either to transient telomerase activation, or the action of an unknown but efficient telomere restoration mechanism. However, we can not exclude the possibility that adequate telomeric length for clonal expansion and metastasis was already acquired by the cancer progenitor melanocyte. This assumption might correlate with the relatively young age of the patient. A thorough examination of a series of human osteosarcomas revealed a category of tumors that do not express telomerase activity and do not display any ALT-pathway characteristics [##REF##17909028##30##]. Interestingly these tumors showed low rates of CIN [##REF##17909028##30##]. A similar sub-category might be also encountered in melanomas. The exceptional case reported here, suggests that metastatic progression in this melanoma, was not accompanied by genomic instability, telomerase activity, or the engagement of the classical alternative recombinatorial telomere lengthening (ALT) pathway.</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<p>Malignant melanomas are characterized by increased karyotypic complexity, extended aneuploidy and heteroploidy. We report a melanoma metastasis to the peritoneal cavity with an exceptionally stable, abnormal pseudodiploid karyotype as verified by G-Banding, subtelomeric, centromeric and quantitative Fluorescence in Situ Hybridization (FISH). Interestingly this tumor had no detectable telomerase activity as indicated by the Telomere Repeat Amplification Protocol. Telomeric Flow-FISH and quantitative telomeric FISH on mitotic preparations showed that malignant cells had relatively short telomeres. Microsatellite instability was ruled out by the allelic pattern of two major mononucleotide repeats. Our data suggest that a combination of melanoma specific genomic imbalances were sufficient and enough for this fatal tumor progression, that was not accompanied by genomic instability, telomerase activity, or the engagement of the alternative recombinatorial telomere lengthening pathway.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>SG conceived and coordinated the study, carried out the analysis of the results and wrote the manuscript. GP collected the samples, acquired informed consent and took part in the analysis of results and manuscript preparation. MC carried out conventional and FISH cytogenetics. SM-L carried out subtelomeric FISH. C-EJ performed the TRAP and Flow-FISH assays. PM and AL carried out and analyzed the cytopathology assays. II-F participated in the design of the study. J-LB performed and analyzed the microsatellite instability assays. SD participated in coordination of the study and analysis of results. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>We thank M. Morris for providing reagents. This work was supported by the Biomedical Research Foundation of the Academy of Athens Greece (BRFAA), in a starting intramural funding and the Grant#05NON-EU-449 of the Greek Secretariat of Research of the Greek Ministry of Development to SG.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>The cytologic examination of the ascitic fluid showed malignant cells with high mitotic index (Giemsa × 400) (A). </bold>Immunocytochemistry against the melanocyte specific antibody S-100 confirmed the presence of malignant melanocytes (Hematoxylin and DAB × 400).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>A GTG-Banding and sub-telomere specific FISH composite representative karyotype of the reported melanoma.</bold> Subtelomeric FISH verified structural integrity of most chromosomes, canonical orientation of both translocations, the deletions 6q, 10p, and 11q, as well as the isochromosome i(17q). The depicted partial dual or triple color subtelomeric FISH karyotypes derive from 23 independent pseudodiploid metaphases; each black box represents a single mitotic nucleus (Red = Spectrum Orange, Green = FITC, Purple = Spectrum Aqua ×1000) (A). Dual color interphase FISH for centromeres 7 (yellow), and 17 (green), shows remarkable numerical stability in 200 nuclei (error-bars represent the standard error of the mean) (B).</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Microsatellite instability in neoplasia (MIN) was excluded in this tumor since the microsatellite markers BAT-25 and BAT-26, showed no instability as compared to MIN positive and negative controls (A). The TRAP assay was negative for telomerase activity in cell culture material obtained from both peritoneal aspirations as compared to two well known telomerase positive human cancer cell lines (MCF-7 and HeLa) (B). Telomere length in this melanoma is relatively low as compared to the ALT U2-OS cell line, leukemic JURKAT cells and human embryonic fibroblasts (error bars represent standard error of the mean between 3 independent experiments) (C). Telomeric PNA FISH indicated uniform terminal capping with TTAGGG repeats on virtually all chromosomes in both pseudodiploid and endoreduplicated clones and low deviation of telomeric length in 500 chromatids as compared to the ALT U2-OS cell line (inverted DAPIx1000) (error bars represent standard deviation)(D).</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1755-8166-1-20-1\"/>", "<graphic xlink:href=\"1755-8166-1-20-2\"/>", "<graphic xlink:href=\"1755-8166-1-20-3\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Osteoid osteoma is a benign bone tumor of the growing skeleton representing approximately 10% of all benign bone neoplasias [##UREF##0##1##]. It usually affects children and young adults [##UREF##0##1##]. Heine in 1927 [##UREF##1##2##], Bergstrand in 1930 [##UREF##2##3##], and Jaffe in 1935 [##UREF##3##4##] identified osteoid osteoma as a clinical entity. Pain is often the only symptom of the disease and is typically described as mild and intermittent at first, becoming more constant and severe at night [##REF##8272884##5##]. When the lesions appear in the hand, diagnosis is challenging for three reasons: first, the typical pain pattern may be absent; second, lesions in the hand may have unusual clinical signs and radiographic presentations; and third, histologic features may differ from classic osteoid osteomas, which occur in the long bones [##REF##15552158##6##]. The metacarpals in particular are not a common site for osteoid osteoma and the diagnosis is often missed in the initial examination. We report a case of an osteoid osteoma in the third metacarpal, and describe the clinical presentation, radiological findings and successful outcome after surgical excision of the lesion.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<p>Osteoid osteoma is a benign bone tumor of the growing skeleton representing approximately 10% of all benign bone neoplasias. It usually affects children and young adults. Normally the tumor does not exceed 1 cm in diameter [##REF##7597885##7##]. The radiographic characteristic of osteoid osteoma is the central nidus, a 2 to 10 mm focus of osteoid nested in a more radiolucent fibrous stroma, surrounded by marginal sclerosis.</p>", "<p>Osteoid osteoma usually occurs in the second and third decade of life. Male patients are more often affected than female patients by a ratio of 2:1, and the tumor is rare in the African-American population. It has a predilection for the lower extremity, with half or more of the lesions occurring in the femur and tibia, near the end of the shaft. Of the remaining lesions, approximately 30% are equally distributed among the spine, hand and foot [##REF##1887121##8##].</p>", "<p>Localization in the hand occurs with an incidence of only about 8% of all reported cases. Nevertheless, osteoid osteoma of the hand is well described in the literature. Allieu and Lussiez [##REF##3233040##9##] and Ambrosia <italic>et al</italic>. [##REF##3655246##10##] reported the largest series of hand osteoid osteomas. The phalanges are the most frequent sites for osteoid osteoma in the hand [##REF##13108890##11##, ####REF##2012734##12##, ##REF##7887006##13####7887006##13##], followed by the carpal bones. The metacarpals are the least common sites for osteoid osteoma [##REF##12427001##14##, ####REF##10190628##15##, ##REF##14089277##16####14089277##16##].</p>", "<p>Trauma has been considered to be a contributing factor, although for others the correlation between injury and the onset of osteoid osteoma remains unclear [##REF##13108890##11##]. Carroll [##REF##13108890##11##] asserted that there is no direct correlation between them, but many cases have been reported in which an injury precedes the onset of the lesion. Kendrick and Evarts [##REF##5589607##17##] reported that 15 out of their 36 cases had had an episode of initial trauma, and the incidence reported by Bednar <italic>et al</italic>. [##REF##8294734##18##] was 11 out of 46 cases. Baron <italic>et al</italic>. [##REF##1496275##19##] described 15 patients with post-traumatic osteoid osteoma. Uda <italic>et al</italic>. [##REF##12427001##14##] reported a case of an osteoid osteoma of the metacarpal bone presenting after an injury.</p>", "<p>Clinically, patients usually present with pain and swelling. The pain, which occurs in about 80% of patients, is more severe at night and is often relieved with salicylates or other non-steroidal anti-inflammatory agents that inhibit the production of prostaglandins by the lesion [##REF##3956019##20##]. Several hypotheses have been proposed to explain the intensity of pain. Nerve endings might be stimulated by the high pressure owing to the increased blood flow within the tumor [##REF##13163104##21##]. Nerve fibers, which are presumed to be components of the autonomic nervous system, are identified in the fibrous zone around the nidus [##REF##14246937##22##]. Prostaglandins may directly stimulate free nerve endings inside or close to the tumor by lowering the nociceptive threshold [##REF##2071279##23##]. A painless osteoid osteoma in a metacarpal has been reported by Basu <italic>et al</italic>. [##REF##10190628##15##], nevertheless, all other metacarpal osteoid osteomas reported to date have presented with pain [##REF##7597885##7##,##REF##3233040##9##,##REF##3655246##10##,##REF##2012734##12##,##REF##7887006##13##,##REF##2071279##23##], as in our patient.</p>", "<p>The diagnosis of an osteoid osteoma in the metacarpals may be difficult and is usually based on clinical and radiographic findings. Conventional radiographs can show the nidus as a small lytic spot surrounded by a radiolucent ring. However, about a quarter of osteoid osteomas are not detected on plain radiographs alone. In such cases, CT, bone scintigraphy, magnetic resonance imaging and angiography are useful in making the correct diagnosis [##REF##3956019##20##]. Surgical treatment including excision of the nidus is usually curative [##REF##7597885##7##], and is the treatment of choice. Recently, minimally invasive techniques, such as percutaneous trephine or drill resection [##REF##7726054##24##,##REF##8504605##25##], with or without the subsequent injection of ethanol [##REF##11387077##26##,##REF##9162624##27##] and thermal destruction with laser photocoagulation [##REF##9874556##28##] or radiofrequency ablation [##REF##7623534##29##], have been used for the removal or destruction of the nidus.</p>", "<p>Recurrence of an osteoid osteoma is likely due to incomplete excision [##REF##639399##30##,##REF##2358492##31##]. Usually, such recurrences have been recorded after curettage or drilling and rarely after an en bloc excision. Carroll [##REF##13108890##11##] has stressed the need for careful radiological and microscopic control at the time of operation. Patients may experience a symptom-free interval after unsuccessful surgery. Recurrence of symptoms may indicate the presence of a second osteoid osteoma. Although such cases are rare, lesions with as many as three distinct nidi have been reported [##REF##4283763##32##]. Most recurrences occur in the first 7 months after primary treatment [##REF##12091665##33##] and have been associated with a nidus diameter of 1.0 to 1.5 cm [##REF##11341426##34##].</p>" ]

[ "<title>Conclusion</title>", "<p>Osteoid osteomas of the hand are challenging to diagnose for several reasons. First, the typical pain pattern may be absent. Second, lesions in the hand may have unusual clinical signs and radiographic presentations. Third, histologic features may differ from classic osteoid osteomas, which occur in the long bones.</p>", "<p>Osteoid osteomas of the metacarpal bones, although unusual, should be considered in the differential diagnosis of chronic pain in the hand of a young patient, presenting with or without a history of previous injury.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Introduction</title>", "<p>Osteoid osteoma is a benign tumor of the growing skeleton. It presents with pain, which is usually worse at night. The radiographic features consist of a central oval or round nidus surrounded first by a radiolucent area followed by another area of sclerotic bone. In the hand, osteoid osteoma is more commonly located in the phalanges and carpal bones. The metacarpals are the least common sites for osteoid osteoma.</p>", "<title>Case presentation</title>", "<p>We present a case of an osteoid osteoma of the left third metacarpal bone in a 36-year-old woman. The clinical and radiographic findings along with the surgical management of the lesion are presented. The pain disappeared immediately after the operation. At the 2-year follow-up, the patient was pain-free and there was no evidence of recurrence.</p>", "<title>Conclusion</title>", "<p>Physicians should be aware of the unusual presence and the atypical clinical presentation of this benign lesion in the metacarpal bones of the hand.</p>" ]

[ "<title>Case presentation</title>", "<p>A 36-year-old woman was referred to our clinic in May 2005 with a 1-year history of pain in her left hand. The pain was constant but increased at night and after manual labor, and was reduced by non-steroidal anti-inflammatory agents. There was no history of injury.</p>", "<p>There was a tender swelling of the head of the third metacarpal bone in the dorsum of the left hand at physical examination. The range of motion was not limited and there were no sensory disturbances. The grip strength of the left hand was slightly reduced, mainly due to pain.</p>", "<p>Blood count and biochemical profile were within the reference ranges. The radiograph showed an oval nidus surrounded by a radiolucent ring (Fig. ##FIG##0##1##).</p>", "<p>Computed tomography (CT) of the left hand clearly showed an oval radiolucent zone at the head of the third metacarpal bone and marked sclerosis around the lesion (Fig. ##FIG##1##2##). The history and clinical and radiographic findings pointed to the diagnosis of an osteoid osteoma of the head of the third metacarpal bone in the left hand. The patient was operated on 30 days later, by a dorsal approach (Fig. ##FIG##2##3a##), under a brachial plexus block. An en bloc excision of the nidus was performed using a small curette. A high-speed burr was also used to remove the sclerotic bone inside the lesion (Fig. ##FIG##2##3b##). The defect was filled with an autogenous cancellous bone graft (Fig. ##FIG##2##3c##). The hand was immobilized postoperatively with a splint.</p>", "<p>Histological examination confirmed the diagnosis of osteoid osteoma. The pain disappeared immediately after the operation. At the 2-year follow-up, the patient was pain-free and there was no evidence of recurrence (Fig. ##FIG##0##1##).</p>", "<title>Abbreviations</title>", "<p>CT: Computed tomography.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>EC carried out the operation and conceived of the idea of presenting the case report. FNX assisted at the operation and in the preparation and drafting of the manuscript. VSN and NE assisted in the drafting of the manuscript. DK made the final check and approval of the submitted manuscript. All authors read and approved the final manuscript.</p>", "<title>Consent</title>", "<p>Written informed consent was obtained from the patient for publication of this case report and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.</p>" ]

[ "<title/>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Plain radiography of the left hand</bold>. A small, oval, radiolucent lesion partially surrounded by sclerotic bone (left). No signs of recurrence at the 2-year follow-up (right).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Computed tomography showing the radiolucent zone and the marked sclerosis around the lesion (arrow).</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Surgical procedure</bold>. (a) Dorsal approach at the third metacarpal head. (b) Resection of the dorsal sclerotic bone. (c) The defect filled with an autogenous cancellous bone graft.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1752-1947-2-285-1\"/>", "<graphic xlink:href=\"1752-1947-2-285-2\"/>", "<graphic xlink:href=\"1752-1947-2-285-3\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Acquired partial lipodystrophy (APL) or Barraquer-Simons syndrome is a rare form of progressive lipodystrophy (OMIM 60879). Patients show a progressive and symmetrical lipoatrophy of subcutaneous adipose tissue starting from the face and spreading to the upper part of the body. Other alterations, such as nephropathy or central nervous system abnormalities, are often, but not always, present [##REF##14170857##1##]. Patients without any associated anomalies or medical complications have also been reported [##REF##15098243##2##]. Women are more often affected than men and most of the reported cases appear to be sporadic [##REF##122289##3##]. However, the appearance of similarly affected relatives in the families of several patients has suggested autosomal dominant inheritance [##REF##14170857##1##,##REF##2385743##4##]. Recently, through a study of 35 cases and extensive review of the literature, Garg and collaborators established a major diagnostic criterion for APL as being gradual onset of bilaterally symmetrical loss of subcutaneous fat from the face, neck, upper extremities and abdomen sparing the lower extremities [##REF##14747765##5##]. It has been reported recently that some cases of APL are associated with mutations in the <italic>lamin B </italic>gene [##REF##16826530##6##], which is reminiscent of other lipodystrophies, such as Dunnigan-type familial partial lipodystrophy type 2, which are associated with <italic>lamin A </italic>mutations [##REF##16364671##7##]. However, the precise mechanisms leading to adipose tissue atrophy in the syndrome remain unknown. Here we present a patient with typical APL and, to obtain insight into the potential etiopathogenic events leading to lipoatrophy in the disease, a gene expression analysis of subcutaneous adipose tissue is reported. The profile of gene expression is compared to the gene expression pattern in subcutaneous fat from healthy control individuals. This study focused on the analysis of the expression of genes characteristic of the adipocyte phenotype, as well as of genes involved in mitochondrial function and local inflammation status in adipose tissue, whose expression is known to be disturbed in experimental models of lipoatrophy as well as in lipodystrophy associated with antiretroviral treatment in patients with HIV.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<p>The present study constitutes the first gene expression analysis in adipose tissue from a patient with APL lipodystrophy syndrome and, to our knowledge, the first gene expression analysis of any lipodystrophy disease other than that in highly active antiretroviral-treated (HAART) patients infected with HIV. A clear limitation of the present type of study is that cause-and-effect relationships cannot be established and the observed changes in gene expression could be either a cause or consequence of the alteration in adipose tissue. However, identification of the genes showing altered expression may provide clues for further research into the etiopathogenesis of the disease.</p>", "<p>The results indicated that adipose tissue of the patient with APL showed impaired expression of genes associated with the adipocyte differentiation process and this involves genes related to adipose tissue metabolism and <italic>PPARγ</italic>, a master regulatory gene of adipogenesis. The down-regulation of <italic>PPARγ </italic>may be the main causative event of the coordinate impairment in the expression of genes encoding components of adipocyte metabolism or adipokines, as genes such as <italic>GLUT4</italic>, <italic>LPL</italic>, and the adiponectin gene are known targets of <italic>PPARγ</italic>-dependent regulation [##REF##8463205##11##]. PPARγ deficiency due to either haploinsufficiency or to dominant negative activity elicits familial partial lipodystrophy (Dunnigan) type 3 [##REF##15711581##12##] whereas reduced <italic>PPARγ </italic>expression has been observed in fat from patients with HAART-associated lipodystrophy [##REF##11937183##13##,##REF##17310817##14##]. It appears therefore that lowered PPARγ activity, regardless of its origin, may be common to multiple types of lipodystrophy. However, it cannot be excluded that PPARγ down-regulation is an effect, and not a primary cause of the APL syndrome, just as lowered PPARγ gene expression is commonly found in adipose tissue from patients with distinct pathologies leading to lipoatrophy. On the other hand, the absence of a reduction in mRNA for C/EBPα, another master regulator of adipogenesis, does not support an overall impairment of adipogenesis in patients with APL. Unaltered leptin gene expression is consistent with previous observations in patients with APL showing that most of these patients have unaltered circulating levels of leptin [##REF##14747765##5##].</p>", "<p>A major difference in the alterations observed in adipose tissue from our patient with respect to subcutaneous fat from lipodystrophic HAART patients concerns marker genes of inflammation. Whereas in HAART patients, the expression of genes related to inflammatory processes are systematically up-regulated [##REF##11937183##13##, ####REF##17310817##14##, ##REF##14532165##15####14532165##15##], gene expression for the four markers of inflammation analyzed here is completely normal in our patient. This result does not indicate a major involvement of an inflammatory environment in adipose tissue as causative of lipoatrophy in APL syndrome. However, a wider analysis of markers of inflammation covering the distinct manifestations of the inflammatory process would be required for unequivocal evaluation of the role of inflammation in adipose tissue of patients with APL.</p>", "<p>A remarkable finding in this study is the identification of reduced expression of genes for mitochondrial respiratory chain components in adipose tissue from our patient with APL. Mitochondrial impairment has been previously identified in HAART-associated lipodystrophy and it was first attributed to the toxicity of some antiretroviral drugs which cause mitochondrial DNA depletion [##REF##15681026##16##]. However, recent data have established that mitochondrial impairment is a more general phenomenon associated with HAART lipoatrophy, which also involves causative events other than mtDNA depletion [##REF##17310817##14##,##REF##16284533##17##]. The present findings identify for the first time mitochondrial disturbances in adipose tissue from a form of lipoatrophy unrelated to viral infection or antiretroviral treatment, and not associated with mtDNA depletion. This indicates that altered mitochondrial function might be a potential common disturbance in lipoatrophy regardless of its origin, a possibility which deserves further research. Recent findings have indicated a previously unrecognized role of mitochondrial biogenesis in the adipocyte differentiation process [##REF##15520860##18##], and the present observations would fit with mitochondrial impairment as being causative of lipoatrophic phenomena. However, other events related to mitochondrial disturbances, such as mutations in the tRNALys gene of mitochondrial DNA, cause lipomatosis rather than peripheral lipoatrophy [##REF##16603396##8##] and therefore a simple defective mitochondrial activity model cannot solely account for eliciting atrophy in adipose tissue.</p>" ]

[ "<title>Conclusion</title>", "<p>In summary, this is the first gene expression analysis in adipose tissue from a patient with APL. It reveals impaired gene expression for marker genes of adipogenesis, including the master regulator <italic>PPARγ</italic>, and mitochondrial function without signs of altered expression of inflammation-related genes.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Introduction</title>", "<p>Acquired partial lipodystrophy or Barraquer-Simons syndrome is a rare form of progressive lipodystrophy. The etiopathogenesis of adipose tissue atrophy in these patients is unknown.</p>", "<title>Case presentation</title>", "<p>This is a case report of a 44-year-old woman with acquired partial lipodystrophy. To obtain insight into the molecular basis of lipoatrophy in acquired partial lipodystrophy, we examined gene expression in adipose tissue from this patient newly diagnosed with acquired partial lipodystrophy. A biopsy of subcutaneous adipose tissue was obtained from the patient, and DNA and RNA were extracted in order to evaluate mitochondrial DNA abundance and mRNA expression levels.</p>", "<title>Conclusion</title>", "<p>The expression of marker genes of adipogenesis and adipocyte metabolism, including the master regulator <italic>PPARγ</italic>, was down-regulated in subcutaneous adipose tissue from this patient. Adiponectin mRNA expression was also reduced but leptin mRNA levels were unaltered. Markers of local inflammatory status were unaltered. Expression of genes related to mitochondrial function was reduced despite unaltered levels of mitochondrial DNA. It is concluded that adipogenic and mitochondrial gene expression is impaired in adipose tissue in this patient with acquired partial lipodystrophy.</p>" ]

[ "<title>Case presentation</title>", "<p>The patient was a Caucasian woman from Spain, and who was the first child of non-consanguineous, healthy parents. The neonatal period and her psychomotor development were normal. She had her first menstruation at the age of 13 years, and regularly since then. She has one child following a normal pregnancy history. At age 8 years, she noted that her subcutaneous adipose facial tissue gradually began to decrease and she complained of generalized muscle pain, predominantly in her lower legs, after exercise. The patient was first seen at age 44 years, this being the time at which adipose tissue biopsy was conducted (see below). A physical examination revealed generalized and symmetrical loss of subcutaneous fatty tissue, predominantly in her face and the upper part of her body. The facial lipoatrophy gave an impression of ageing, and a male aspect was noted. However, testosterone levels were normal (0.34 ng/ml, normal range from 0.3 to 1.2 ng/ml). Blood examination showed normal levels of glucose (98 mg/dl, normal range 45 to 135 mg/dl), whereas levels of triglycerides were higher than normal (252 mg/dl, normal range 35 to 150 mg/dl) and levels of cholesterol slightly higher than normal (234 mg/dl, normal range 100 to 220 mg/dl). However, unaltered levels of HDL-cholesterol (47 mg/dl, normal range 35 to 80 mg/dl) and LDL-cholesterol (137 mg/dl, normal range 60 to 150 mg/dl) were found. The blood examination also revealed normal muscular enzyme levels. A neurological examination indicated that deep tendon reflexes were normal and no myotonic phenomena were observed. Nerve conduction studies showed normal values in all tested nerves. A concentric needle examination showed complex repetitive discharges in all tested muscles with no spontaneous activity. Renal and liver function were, as inferred from serum enzyme levels, also normal (ALT/GPT 34 U/liter, normal range 2 to 41 U/liter; AST 27 U/liter, normal range 1 to 40 U/liter; GGT 24 U/liter, normal range 5 to 49 U/liter). Ultrasonic examination of the abdomen indicated hepatic steatosis with normal liver size and morphology, and the kidneys and spleen were normal. Cytogenetic studies revealed a normal karyotype (46XX) without evidence of chromosome breakage. Serum C3 levels were 45 mg/dL, which was abnormally low with respect to normal values (range from 85 to 180 mg/dl). Results were positive for the presence of serum complement 3 nephritic factor. Thus, the overall clinical and biochemical features of the patient led to the diagnosis of APL. The pattern of progressive loss of subcutaneous adipose tissue in the face and the upper part of the body was in accordance with the major criterion established by Misra <italic>et al. </italic>(2004) [##REF##14747765##5##] and supportive criteria were also met: onset during childhood, low serum levels of complement 3 and the presence of serum complement 3 nephritic factor.</p>", "<p>A biopsy sample of subcutaneous adipose tissue was taken from the arm. Control values of gene expression in adipose tissue were obtained from the analysis of biopsies of subcutaneous adipose tissue taken from the arms of 10 healthy control individuals (mean age 38.5 ± 9.0 years, 4/6 female/male). The patient and the individual controls gave their consent to participate in the study and the protocol was approved by the Ethics Committee of the Hospital de la Santa Creu i Sant Pau, Barcelona, Spain. Separate analysis of gene expression markers in men and women did not show any significant difference and therefore reference values of gene expression in healthy men and women were shown together. After homogenization in RLT buffer (Qiagen, Hilden. Germany), an aliquot was used for isolation of DNA, which was performed using a standard phenol/chloroform extraction methodology. Another aliquot of the homogenate was used for RNA extraction using a column-affinity based methodology (RNEasy, Qiagen, Hilden, Germany). For mRNA analysis TaqMan Reverse Transcription and -RT-PCR reagents were used (Applied Biosystems, Foster City, CA, USA). One microgram of RNA was transcribed into cDNA using random-hexamer primers and real-time reverse transcriptase-polymerase chain reaction (RT-PCR) was performed on an ABI PRISM 7700HT sequence detection system. The TaqMan RT-PCR was performed in a final volume of 25 μl using TaqMan Universal PCR Master Mix, NoAmpErase UNG reagent and the specific gene expression primer probes. The TaqManGene Expression assays used were: COX4I1 (subunit IV of cytochrome c oxidase, <italic>COIV</italic>), Hs00266371_m1; ATPase5J (subunit F6 of F0-ATPsynthase) Hs0036588_m1; PPARGC1 (<italic>PGC-1α</italic>), Hs0013304_m1; CEBPA (CCAAT/enhancer binding protein-α, <italic>C/EBBP-α</italic>), Hs00269972_s1; PPARG (peroxisome-proliferator activated receptor-γ, <italic>PPAR-γ</italic>), Hs00234592_m1; RB1 (retinoblastoma protein, <italic>pRB</italic>), Hs00153108_m1; DLK1 (<italic>Pref-1</italic>), Hs00171584_m1; UCP2 (uncoupling protein-2, <italic>UCP-2</italic>), Hs00163349_m1; UCP3 (uncoupling protein-3, <italic>UCP-3</italic>), Hs00243297_m1; NRF1 (nuclear respiratory factor-1, <italic>NRF-1</italic>) Hs00192316_m1; LPL (lipoprotein lipase, <italic>LPL</italic>), Hs00173425_m1; LEP (leptin, <italic>LEP</italic>), Hs00174877_m1; Complement precursor-3, Hs00163811_m1; TNF (tumor necrosis factor-α, <italic>TNF-α</italic>), Hs00174128_m1; SLC2A4 (glucose transporter, <italic>GLUT4</italic>), Hs00168966_m1; APM1 (adiponectin), Hs00605917_m1; β2-microglobulin, Hs9999907_m1; MCP-1 (monocyte chemoattractant protein-1, <italic>MCP-1</italic>), Hs00234140_m1; CD68 antigen, Hs00154355. Primers and probe for the detection of cytochrome c oxidase subunit II (<italic>COII</italic>) mRNA and mtDNA abundance were designed as previously reported [##REF##16603396##8##]. Controls with no RNA, primers, or reverse transcriptase were included in each set of experiments. Each sample was run in duplicate and the amount of mRNA for the gene of interest in each sample was normalized to that of the reference control using the comparative (2^-ΔCT) method. Data for gene transcripts are expressed as the ratio of relative abundance of the mRNA of the gene of interest respect to 18S rRNA (Hs99999901_s1).</p>", "<p>Examination of gene expression for master regulatory factors associated with the promotion of adipogenesis indicated that peroxisome proliferator-activated-γ (<italic>PPARγ</italic>) mRNA was the only mRNA significantly down-regulated in the patient with APL with respect to controls (Table ##TAB##0##1##). The expression of mRNA for another positive regulator of adipogenesis, CCAAT/enhancer binding protein-α (<italic>C/EBPα</italic>) mRNA was not significantly altered. The mRNA levels of retinoblastoma protein pRb, a protein that may have negative effects on adipogenesis [##REF##12479814##9##], and of <italic>Pref-1</italic>, a known negative regulator of adipogenesis [##REF##17116701##10##] were also unchanged. Among adipokines, leptin mRNA levels were unaltered in the patient whereas adiponectin mRNA was down-regulated. The mRNA for two marker genes of adipose tissue metabolism, insulin-sensitive glucose transporter-4 (<italic>GLUT4</italic>) and lipoprotein lipase (<italic>LPL</italic>), were also down-regulated in the patient with respect to controls. In contrast, the mRNA levels for three marker genes of inflammation (tumor necrosis factor-α, <italic>TNFα</italic>; <italic>MCP-1</italic>; and β2-microglobulin) as well as the marker of macrophage infiltration CD68 were unaltered in the patient, with values almost identical to those of controls. Complement component-3 gene expression was also unchanged.</p>", "<p>Levels of transcripts corresponding to components of the respiratory chain system (OXPHOS), either mtDNA-encoded (<italic>COII</italic>) or nuclear DNA-encoded (<italic>COIV </italic>and <italic>ATP synthase F0 subunit 6</italic>) were reduced in adipose tissue from the patient with respect to healthy controls. No significant changes were observed for mtDNA abundance in adipose tissue from the patient with respect to reference control values (1.08-fold change with respect to the mean control values). Neither <italic>UCP2 </italic>mRNA levels nor <italic>UCP3 </italic>mRNA levels were altered in the patient relative to controls. Likewise, transcript levels for PPARγ-coactivator-1α <italic>(PGC-1α</italic>) and nuclear respiratory factor-1 (<italic>NRF-1</italic>), regulatory factors of mitochondrial biogenesis, were also unaltered in the patient with respect to controls.</p>", "<title>Abbreviations</title>", "<p>18S rRNA: 18S ribosomal RNA; APL: Acquired partial lipodystrophy (Barraquer-Simons syndrome); C/EBP-α: CCAAT/enhancer binding protein-α; COIV: subunit IV of cytochrome c oxidase; COII: subunit II of cytochrome c oxidase; <italic>GLUT4</italic>: Glucose transporter-4; HAART: Highly active antiretroviral-treatment; LPL: Lipoprotein lipase; MCP-1: Monocyte chemoattractant protein-1; mtDNA: Mitochondrial DNA; NRF1: Nuclear respiratory factor-1; PGC-1α: peroxisome-proliferator activated receptor-γ co-activator-1α; <italic>PPAR-γ</italic>: Peroxisome-proliferator activated receptor-γ; pRB: retinoblastoma protein; Pref-1: Preadipocyte factor-1; TNF-α: tumor necrosis factor-α; UCP2: Uncoupling protein-2; UCP3: Uncoupling protein-3.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>PD and II analyzed and interpreted the data from the physical examination of the patient and PD was a major contributor in writing the manuscript. RRG, EG and II performed the analysis and interpretation of data in relation to renal and liver function, cytogenetic analysis and blood analysis. EGA and ALA carried out muscle analysis. JCD performed the analysis and interpretation of mitochondrial DNA data and JG and MG analyzed the mRNA. FV designed and coordinated the study and was responsible for final writing of the manuscript. All authors read and approved the final manuscript.</p>", "<title>Consent</title>", "<p>Written informed consent was obtained from the patient for publication of this case report. A copy of the written consent is available for review by the Editor-in-Chief of this journal.</p>" ]

[ "<title>Acknowledgements</title>", "<p>Supported by Ministerio de Educación y Ciencia (grant SAF2005-01722) and Fondo de Investigaciones Sanitarias Ministerio de Sanidad y Consumo (grant PI052336), Spain.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>mRNA expression of genes involved in adipogenesis, metabolism, inflammation and mitochondrial function in adipose tissue from our patient with APL</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"left\"><bold>Healthy controls</bold></td><td align=\"left\"><bold>APL</bold></td><td align=\"left\"><bold>Change</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Adipogenesis (+)</bold></td><td/><td/><td/></tr><tr><td align=\"left\"><italic>PPARγ </italic>mRNA</td><td align=\"left\">2.78 (1.81–3.74) × 10^-5</td><td align=\"left\">1.35 × 10^-5</td><td align=\"left\">↓</td></tr><tr><td align=\"left\"><italic>C/EBPα </italic>mRNA</td><td align=\"left\">1.55 (0.93–2.03) × 10^-4</td><td align=\"left\">1.66 × 10^-4</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><bold>Adipogenesis (-)</bold></td><td/><td/><td/></tr><tr><td align=\"left\"><italic>pRb </italic>mRNA</td><td align=\"left\">7.52 (3.47–11.57) × 10^-6</td><td align=\"left\">3.84 × 10^-6</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><italic>Pref-1 </italic>mRNA</td><td align=\"left\">0.44 (0.03–0.84) × 10^-9</td><td align=\"left\">0.04 × 10^-9</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><bold>Adipokines</bold></td><td/><td/><td/></tr><tr><td align=\"left\"><italic>Leptin </italic>mRNA</td><td align=\"left\">0.73 (0.41–1.02) × 10^-4</td><td align=\"left\">0.85 × 10^-4</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><italic>Adiponectin </italic>mRNA</td><td align=\"left\">1.05 (0.69–1.39) × 10^-3</td><td align=\"left\">0.49 × 10^-3</td><td align=\"left\">↓</td></tr><tr><td align=\"left\"><bold>Metabolism</bold></td><td/><td/><td/></tr><tr><td align=\"left\"><italic>GLUT4 </italic>mRNA</td><td align=\"left\">0.96 (0.43–1.47) × 10^-5</td><td align=\"left\">0.13 × 10^-6</td><td align=\"left\">↓</td></tr><tr><td align=\"left\"><italic>LPL </italic>mRNA</td><td align=\"left\">2.88 (2.15–3.47) × 10^-4</td><td align=\"left\">0.73 × 10^-4</td><td align=\"left\">↓</td></tr><tr><td align=\"left\"><bold>Inflammation</bold></td><td/><td/><td/></tr><tr><td align=\"left\"><italic>TNFα </italic>mRNA</td><td align=\"left\">5.18 (2.74–7.63) × 10^-7</td><td align=\"left\">4.56 × 10^-7</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><italic>MCP-1 </italic>mRNA</td><td align=\"left\">0.24 (0.07–0.40) × 10^-5</td><td align=\"left\">0.19 × 10^-5</td><td align=\"left\">=</td></tr><tr><td align=\"left\">β<italic>2-microglobulin </italic>mRNA</td><td align=\"left\">2.85 (2.06–3.49) × 10^-4</td><td align=\"left\">3.12 × 10^-4</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><italic>CD68 </italic>mRNA</td><td align=\"left\">0.75 (0.26–1.24) × 10^-4</td><td align=\"left\">1.01 × 10^-4</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><bold>Complement system</bold></td><td/><td/><td/></tr><tr><td align=\"left\"><italic>Component 3 mRNA</italic></td><td align=\"left\">1.99 (0.91–3.07) × 10^-4</td><td align=\"left\">2.08 × 10^-4</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><bold>OXPHOS</bold></td><td/><td/><td/></tr><tr><td align=\"left\"><italic>COII </italic>mRNA</td><td align=\"left\">1.17 (0.74–1.51) × 10^-2</td><td align=\"left\">0.35 × 10^-2</td><td align=\"left\">↓</td></tr><tr><td align=\"left\"><italic>COIV </italic>mRNA</td><td align=\"left\">4.24 (3.42–4.81) × 10^-5</td><td align=\"left\">1.55 × 10^-5</td><td align=\"left\">↓</td></tr><tr><td align=\"left\"><italic>ATPsynthase F0-6 mRNA</italic></td><td align=\"left\">6.77 (4.85–8.69) × 10^-5</td><td align=\"left\">2.10 × 10^-5</td><td align=\"left\">↓</td></tr><tr><td align=\"left\"><bold>UCPs</bold></td><td/><td/><td/></tr><tr><td align=\"left\"><italic>UCP2 </italic>mRNA</td><td align=\"left\">4.05 (2.74–5.77) × 10^-5</td><td align=\"left\">5.72 × 10^-5</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><italic>UCP3 </italic>mRNA</td><td align=\"left\">4.13 (1.67–6.50) × 10^-7</td><td align=\"left\">2.56 × 10^-7</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><bold>Mitochondriogenesis regulators</bold></td><td/><td/><td/></tr><tr><td align=\"left\"><italic>PGC-1α </italic>mRNA</td><td align=\"left\">1.12 (0.71–1.51) × 10^-6</td><td align=\"left\">0.82 × 10^-6</td><td align=\"left\">=</td></tr><tr><td align=\"left\"><italic>NRF-1 </italic>mRNA</td><td align=\"left\">4.06 (2.96–5.26) × 10^-6</td><td align=\"left\">3.37 × 10^-6</td><td align=\"left\">=</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>APL, Acquired partial lipodystrophy. Values of mRNA expression in adipose tissue from healthy controls are shown as means (95% confidence interval in parentheses), expressed as the ratio of relative abundance of the mRNA of the gene of interest relative to 18S rRNA. Reduced mRNA expression values in APL below confidence intervals are shown as ↓.</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>It is known that vitamin D and especially its activated metabolite 1,25-dihydroxyvitamin D₃(<italic>1,25D</italic>₃), are involved in controlling the normal function of the endocrine pancreas, and particularly insulin secretion [##REF##6323527##1##,##REF##16563471##2##]. Vitamin D deficiency inhibits rat pancreatic secretion and turnover of insulin, leading to impaired glucose tolerance, while replacement therapy with <italic>1,25D</italic>₃, is able to reverse these abnormalities [##REF##6250216##3##,##REF##9854180##4##]. <italic>1,25D</italic>₃also affects the insulin receptor (IR), the protein to which insulin must bind to carry out its multiple biological actions in the cells. In this respect, our group has reported the first demonstration that <italic>1,25D</italic>₃increased human IR mRNA levels, the IR number, and the insulin response in U-937 human promonocytic cells through mechanisms that involve direct transcriptional activation of the human IR gene [##REF##7698218##5##, ####REF##11075718##6##, ##REF##12125099##7####12125099##7##].</p>", "<p>In non-obese diabetic mice, vitamin D deficiency accelerates type 1 diabetes [##REF##14758446##8##], while chronic administration of <italic>1,25D</italic>₃reduces the incidence of diabetes in these mice, principally by modulating immune mechanisms [##REF##1397723##9##, ####REF##7926338##10##, ##REF##15971062##11####15971062##11##]. Streptozotocin-induced diabetes is another animal model of diabetes that comprises both toxic and inflammatory mechanisms [##REF##9231661##12##]. Mononuclear infiltration and the altered morphology of islets coupled with the disappearance of beta cells are among the histological changes reported in the pancreas of these animals [##REF##9231661##12##,##REF##4926377##13##]. Indeed, hyperglycemia and hypoinsulinemia, have also been described in this experimental model [##REF##137248##14##,##REF##150797##15##]. This hypoinsulinemia was associated with increased insulin binding in the kidney [##REF##6749631##16##,##REF##3537445##17##] and liver [##REF##137248##14##,##REF##6389544##18##] and with somewhat controversial insulin binding results in adipose tissue [##REF##150797##15##,##REF##154427##19##,##REF##6802695##20##]. The increases in renal and hepatic IRs, were accompanied by elevated IR mRNA expression in both tissues, and it could be reversed by treatment with insulin [##REF##1280238##21##, ####REF##1386819##22##, ##REF##7556957##23####7556957##23##]. Despite the different alterations in IR mRNA levels and insulin binding, streptozotocin-induced diabetic rats characteristically display insulin resistance [##REF##150797##15##,##REF##6389544##18##,##REF##154427##19##]. In this diabetic model, the administration of 1,25D₃for 8 weeks was reported to improve diabetes attenuating pancreatic islet damage and decreasing the insulin requirements [##REF##9231661##12##].</p>", "<p><italic>1,25D</italic>₃acts in its genomic effects as a ligand for the vitamin D receptor (VDR, NR1I1) [##REF##10219237##24##]. This receptor is a member of the superfamily of nuclear receptors that regulates gene expression as a vitamin D-dependent transcription factor. It exerts this action by binding, preferentially as a heterodimer with the retinoid × receptor (RXR), to vitamin D response elements (VDREs) in the promoter regions of target genes [##REF##16946007##25##]. A VDRE generally consists of two direct imperfect repeats of six nucleotides separated by a three-nucleotide spacer. The VDR occupies the 3' half-site, while the RXR binds to the 5' half-site [##REF##9525333##26##]. Sequence variations have been detected in the 3' half-element, the 5' half-element, the spacer, and in the sequences flanking the VDREs and these variations appear to be important in receptor-binding efficiency [##REF##8674817##27##,##REF##10775803##28##]. The identification of VDREs has only been possible in a very limited number of vitamin D-regulated genes [##REF##9525333##26##,##REF##8674817##27##]. Our group has reported the first identification of a functional VDRE overlapped by a downstream AP-2-like site that specifically binds VDR in the human IR gene promoter [##REF##12711007##29##]. This VDRE accounted for the transcriptional induction of this gene by <italic>1,25D</italic>₃in U-937 human cells [##REF##12125099##7##,##REF##12711007##29##].</p>", "<p>With these antecedents, the aim of the present investigation was to study the effects of the treatment with <italic>1,25D</italic>₃[150 IU/Kg (3.75 μg/Kg) one a day, for 15 days] to non-diabetic and streptozotocin-induced diabetic rats. The results indicated that while treatment with <italic>1,25D</italic>₃had practically no effect on non-diabetic rats, the same treatment in streptozotocin-induced diabetic rats corrected in part the over-expression of the IR gene in liver and adipose tissue, although it did not revert the hyperglycemia, hypoinsulinemia, glycosuria or ketonemia of these diabetic animals. At the same time, it produced normalization of the IR number without alterations in the receptor affinity and with an improvement of the insulin response in terms of glucose transport in isolated adipocytes of these diabetic animals. In addition, a computer search in the rat insulin receptor gene promoter revealed the existence of two DNA sequences: -256/-219 bp and -653/-620 bp, the first overlapped by three and the second by four AP-2-like sites. These sequences represent two candidate VDREs that could respond to <italic>1,25D</italic>₃via activation of VDR, although this remains to be further investigated.</p>" ]

[ "<title>Methods</title>", "<title>Animals and sampling</title>", "<p>Four groups of male Wistar rats weighing 200–220 g at the onset were used in this study. They were kept under standard conditions of light and temperature and given free access to tap water and standard laboratory chow (Panlab, A04) containing 8.8 mg/g of calcium, 5.9 mg/g of phosphorus and 1.5 IU/g of vitamin D. The first group comprised of non-diabetic rats receiving sham-treatments during the 15 days of the experimental period (<italic>Control-rats</italic>). The second group included non-diabetic rats i.<italic>p</italic>. injected with 1,25D₃(Calcijex, Abbot) [150 IU/Kg (3.75 μg/Kg) one a day, for 15 days] (<italic>1,25D</italic>₃-<italic>rats</italic>). The third group consisted of rats rendered diabetic by a single injection of streptozotocin (Sigma) [65 mg/kg] on the 8th day of the 15-day period (<italic>STZ-rats</italic>). Finally, the fourth group comprised of rats rendered diabetic by streptozotocin on the 8th day, and i.p. injected with <italic>1,25D</italic>₃[150 IU/Kg (3.75 μg/Kg) one a day, for 15 days] (<italic>STZ</italic>+<italic>1,25D</italic>₃-<italic>rats</italic>). The National Guide for the Care and Use of Laboratory Animals was strictly followed during this study.</p>", "<p>Urine samples were collected before decapitation without anaesthesia and trunk blood samples were recovered for plasma measurements. On sacrifice, the kidney, liver and epididymal adipose tissue were immediately removed, individually packed and frozen in liquid nitrogen for nucleic acid and protein determinations. Isolated adipocytes were obtained from fresh epididymal adipose tissue for insulin binding and insulin activity assays.</p>", "<title>Analytical methods</title>", "<p>Different parameters in urine were measured using the Combur Test (Roche). Plasma insulin levels were determined by radioimmunoassay using rat insulin as standard. Plasma concentrations of glucose, 25-hydroxyvitamin D₃, calcium, phosphorus and proteins were estimated using commercially available techniques. The total DNA and RNA contents of the homogenates from individual tissues were estimated by spectrofluorometry. The protein content was determined by the Bradford method.</p>", "<title>RNA blot assays</title>", "<p>For Northern blot assays, RNA samples (40 μg) of kidney, liver and epididymal adipose tissue, were denatured, electrophoresed in 1.1% agarose-formaldehyde gels and blotted onto nylon membranes [##REF##9089644##36##]. Ethidium bromide staining of the 28S and 18S ribosomal RNAs was routinely checked before blotting as a control of sample loading and after blotting as a control of RNA transfer. RNA blots were prehybridized, hybridized with excess [³²P]-labelled probe (the 0.98-Kb rat IR specific EcoR1 fragment of the p16 clone, gift from Prof. Goldstein), washed under stringent conditions and finally autoradiographed, as described previously [##REF##9089644##36##]. The autoradiographs were scanned with a laser densitometer and the readings normalized with the respective amounts of 28S rRNA, as revealed by ethidium bromide.</p>", "<title>Insulin binding assays</title>", "<p>Binding assays were carried out as described previously [##REF##3061857##48##]. In brief, isolated adipocytes (0.25 × 10⁶cells/ml) were incubated at 30°C for 30 min in 400 μl Krebs-Hepes buffer pH 7.6 with mono-[¹²⁵I]-insulin (0.2 × 10^-9M) (NEN Life Sciences), in the presence or absence of increasing concentrations (0.2 × 10^-10to 0.5 × 10^-7M) of unlabelled insulin (Sigma). Non-specific binding was determined in the presence of 1 × 10^-7M unlabelled insulin and subtracted from each point. The adipocyte number was determined with a Neubauer-type hemocytometer and cell viability (assessed by 0.2% trypan blue exclusion) was greater than 90%. Measurements of the cell diameter were also performed. The data were analysed by the Scatchard method using the LIGAND program of Munson and Rodbard [##REF##6254391##49##].</p>", "<title>Glucose transport determinations</title>", "<p>Glucose transport was measured in isolated adipocytes by determining the uptake of D-[¹⁴C(U)]-glucose (NEN Life Sciences) at trace concentrations using our modification of the method of Kashiwagi et al. [##REF##6355180##50##]. This method is based on the premise that glucose uptake provides a measure of glucose transport when studies are carried out at very low glucose concentrations. In brief, 0.25 × 10⁶adipocytes/ml were incubated with 5 × 10^-7M D-[¹⁴C(U)]-glucose for 1 h at 37°C in the presence and absence of a concentration of insulin (10^-8M) that gave maximal glucose transport in earlier experiments [##REF##11075718##6##]. Assays were stopped by washing the cells with ice-cold PBS. The cells were solubilized in 0.5% SDS and 0.1 N NaOH and the associated radioactivity was counted.</p>", "<title>Computer analysis of DNA sequences in the rat IR gene promoter</title>", "<p>The identification of virtual VDRE sequences in the <italic>Rattus norvegicus </italic>partial IR promoter [GenBank:<ext-link ext-link-type=\"gen\" xlink:href=\"AJ006071\">AJ006071</ext-link>] was carried out by homology with two <italic>consensus </italic>VDRE sequences using the SEQFIND programme developed in our laboratory and previously employed in the computer analysis of various DNA sequences including VDREs [##REF##12711007##29##]. The first <italic>consensus </italic>(5'<bold>GGGTCA</bold>NNG<bold>GGGGCA</bold>3') had been compiled and reported by us from a series of described functional VDRE sequences in various 1,25D₃-stimulated promoters [##REF##12711007##29##]. The second <italic>consensus </italic>VDRE sequence (5'<bold>PuGGTCA</bold>NNPu<bold>PuGTTCA</bold>3') had been proposed by Colnot <italic>et al</italic>. [##REF##8674817##27##], and Haussler <italic>et al</italic>. [##REF##9525333##26##]. The presence of AP-1 and AP-2 sites flanking or overlapping the potential VDREs were identified by their homology with the AP-1 site <italic>consensus </italic>(5'TGAC/GTCA3') [##REF##3034433##31##] and the AP-2 site <italic>consensus </italic>(5'GCCN3GGG3') [##REF##11137286##32##] respectively, also using the SEQFIND programme.</p>", "<title>Statistical analysis</title>", "<p>Unless otherwise indicated, the data are expressed as the mean ± SEM. A comparison between the groups was carried out using the two-tailed, unpaired Student <italic>t</italic>-test and/or ANOVA comparison, as appropriate. Differences were considered statistically significant when p &lt; 0.05.</p>" ]

[ "<title>Results</title>", "<title>Body weights and plasma values of glucose, insulin, 25-hidroxyvitamin D₃, calcium, phosphorus and proteins</title>", "<p>Treatment with <italic>1,25D</italic>₃to non-diabetic rats increased the rat body weights but did not affect any parameter in the plasma of these animals (Table ##TAB##0##1##). The injection of streptozotocin reduced body weights and induced hyperglycemia and hypoinsulinemia (Table ##TAB##0##1##). Treatment with <italic>1,25D</italic>₃to streptozotocin-induced diabetic rats increased body weights but did not revert the hyperglycemia and the hypoinsulinemia induced by the diabetes, while calcium and phosphorus plasma levels were increased and plasma levels of 25-hydroxyvitamin D₃were maintained in the normal range [##REF##3838933##30##] (Table ##TAB##0##1##).</p>", "<title>Values of specific gravity, pH, leukocytes, nitrite, protein, glucose, ketones, urobilinogen, bilirubin and blood in urine</title>", "<p>Treatment with <italic>1,25D</italic>₃to non-diabetic rats did not affect any of the above cited urine parameters under study. Streptozotocin injection induced glycosuria (&gt; 500 mg/dl) and the appearance of leukocytes, ketones and blood in the urine of these animals. Treatment with <italic>1,25D</italic>₃to streptozotocin-induced diabetic rats did not modify the glycosuria and the ketonuria of the diabetes, but partially decreased the number of leukocytes in urine (from 10–25 to 5–10 leukocytes/μl).</p>", "<title>Total DNA, RNA and protein content of the kidney, liver and epididymal adipose tissue</title>", "<p>Treatment with <italic>1,25D</italic>₃to non-diabetic rats increased both the protein content and the indicator of cell size (protein/DNA) in the kidney, liver and adipose tissue of these animals while the DNA and RNA content remained unaltered (Table ##TAB##1##2##). Streptozotocin injection incremented the protein content in these three tissues, accompanied in the case of kidney with an increment of the protein/DNA ratio and in the case of the liver with an increment of the DNA content. Treatment with <italic>1,25D</italic>₃of streptozotocin-induced diabetic rats practically did not modify neither the high values of protein induced by the diabetes in the three tissues nor the increases in the DNA content in kidney and liver. In addition, this treatment did not alter the RNA content in any of these three tissues (Table ##TAB##1##2##).</p>", "<title>Insulin receptor mRNA levels in the kidney, liver and epididymal adipose tissue</title>", "<p>Northern blot assays of kidney (Figure ##FIG##0##1##, panels A and B), liver (Figure ##FIG##0##1##, panels C and D) and epididymal adipose tissue (Figure ##FIG##0##1##, panels E and F) from control animals revealed two major IR mRNA species of approximately 9.5 and 7.5 Kb in size. The relative amounts of the two IR mRNA species measured as 9.5 Kb/7.5 Kb ratio were: 2.2 ± 0.2 in kidney, 1.2 ± 0.2 in liver and 1.1 ± 0.03 in adipose tissue. Treatment with <italic>1,25D</italic>₃to non-diabetic rats did not affect any of the two IR mRNA species in any of the three tissues studied (Figure ##FIG##0##1##, panels A to F). IR mRNA species were expressed per unit of RNA in view of the observation (Table ##TAB##1##2##) that treatment with <italic>1,25D</italic>₃to both non-diabetic and diabetic rats did not alter RNA content per gram of tissue in any of the three tissues. The induction of diabetes by streptozotocin produced important increments in the levels of both IR mRNA species in liver and adipose tissue but not in kidney. In particular, 78% increase corresponding to the 9.5 Kb and 69% to the 7.5 Kb species in liver (Figure ##FIG##0##1##, panel D), and 58% increase corresponding to the 9.5 Kb and 92% increase to the 7.5 Kb species in the adipose tissue (Figure ##FIG##0##1##, panel F). Treatment with <italic>1,25D</italic>₃to diabetic rats partially prevented the over-expression of the IR mRNA in the liver (Figure ##FIG##0##1##, panels C and D) and adipose tissue (Figure ##FIG##0##1##, panels E and F) of these animals. The 9.5 Kb species was particularly decreased in liver and the 7.5 kb species in the adipose tissue. However, treatment with <italic>1,25D</italic>₃to diabetic rats did not affect IR expression in kidney (Figure ##FIG##0##1##, panels A and B).</p>", "<title>Insulin binding in isolated adipocytes</title>", "<p>Insulin binding studies showed that treatment with <italic>1,25D</italic>₃to non-diabetic rats did not alter IR number or IR affinity as reflected by the ED:50 value in adipocytes of these animals (Figure ##FIG##1##2## and Table ##TAB##2##3##). Streptozotocin injection produced a 64% decrease in the number of IRs without altering IR affinity. Treatment with <italic>1,25D</italic>₃to streptozotocin-induced diabetic rats inverted the decrease in the IR number induced by the diabetes to values not significantly different from those observed in control adipocytes and without altering IR affinity (Figure ##FIG##1##2## and Table ##TAB##2##3##).</p>", "<p>Given that both the weight of the adipose tissue and the diameter of the adipocytes were markedly decreased by the diabetes (Table ##TAB##2##3##), we also calculated the IR number per cell surface area (μm²). In these conditions, although the effect of the diabetes decreasing IR number was somewhat minor (44%), the tendency of <italic>1,25D</italic>₃to normalize the IR number was also evident (Table ##TAB##2##3##).</p>", "<title>Glucose transport in isolated adipocytes</title>", "<p>Treatment with <italic>1,25D</italic>₃to non-diabetic rats did not alter basal or insulin-stimulated glucose transport in adipocytes from these animals (Figure ##FIG##2##3##). The injection of streptozotocin clearly decreased both basal and insulin-stimulated glucose transport. The treatment with 1,25D₃to streptozotocin-induced diabetic rats, improved by 107% the decreased basal glucose transport and by 71% the insulin-stimulated glucose transport in adipocytes from these diabetic animals (Figure ##FIG##2##3##).</p>", "<title>Computer analysis of DNA sequences in the rat IR gene promoter</title>", "<p>We performed a computer search of virtual VDRE sequences in the rat IR gene promoter by homology with two <italic>consensus </italic>VDRE sequences. The first <italic>consensus </italic>(5'<bold>GGGTCA</bold>NNG<bold>GGGGCA</bold>3') was previously compiled and reported by us from a series of described functional VDRE sequences in various 1,25D₃-stimulated promoters [##REF##12711007##29##]. The search with this first <italic>consensus </italic>indicated no sequence identical to this VDRE sequence, and neither was any sequence found to show 1, 2, or 3 base variations. Nevertheless, we detected three virtual VDRE sequences showing a difference of four bases from this <italic>consensus</italic>: -249/-235, -637/-623 and -916/-902 (Figure ##FIG##3##4##, panel A). Comparison of the 3' half-element of each of these sequences with the 3' half-element of this first VDRE <italic>consensus</italic>, indicated complete homology only in the case of the -916/-902 sequence, and two base variations in the other two (Figure ##FIG##3##4##, panel A).</p>", "<p>The second computer search of virtual VDREs, now by homology with the second <italic>consensus </italic>VDRE sequence (5'<bold>PuGGTCA</bold>NNPu<bold>PuGTTCA</bold>3'), was also performed. Our <italic>consensus </italic>and the latter proposed by Colnot <italic>et al. </italic>[##REF##8674817##27##], and Haussler <italic>et al. </italic>[##REF##9525333##26##] have identical 5' half-elements, but ours contains two GG instead of two TT, in the 3' half-element. The spacers are identical, with a G or Pu at position 3, which appears to be important in VDR binding. The results of this computer search by homology with this second <italic>consensus </italic>indicated no sequence identical to this VDRE, and neither was any sequence found to show 1, 2, or 3 base variations. Nevertheless, we detected three virtual VDRE sequences showing a difference of four bases from this second <italic>consensus</italic>: -25/-11, -247/-233 and -881/-867 (Figure ##FIG##3##4##, panel A). Of these, the -247/-233 sequence was in part the same as that of the -249/-235 sequence detected with the first <italic>consensus </italic>(Figure ##FIG##3##4##, panel A). However, the 3' half-element of this second sequence has only one base variation instead of two that the first sequence had (Figure ##FIG##3##4##, panel A). Therefore, in principle there were six virtual VDRE sequences (Figure ##FIG##3##4##, panel A).</p>", "<p>The next step was to check for the possibility of cis-elements related to these virtual VDRE sequences in the rat IR gene promoter. Thus, a computer search of AP-1 and AP-2 sites by homology with the AP-1 site consensus (5'TGAC/GTCA3') [##REF##3034433##31##], and the AP-2 site consensus (5'GCCN3GGG3') [##REF##11137286##32##], respectively, was performed.</p>", "<p>Exploring the AP-1 sites, we detected two AP-1-like sites showing a difference of only one base from this consensus: the -374/-368 (5'TGTCTCA3') and the -963/-957 (5'TGAGCCA3'). However, these two sites are neither adjacent nor overlapped in any of our virtual VDRE sequences. Nevertheless, as far as we know, this is the first mention of these AP-1-like sites in the rat IR gene promoter. With respect to the AP-2 sites, we detected seventeen AP-2-like sites showing a difference of only one base from this <italic>consensus</italic>: -71/-63, -79/-71, -171/-163, -195/-187, -230/-222, -245/-237, -253/-245, -405/-397, -465/-457, -480/-472, -545/-537, -634/-626, -635/-627, -642/-634, -650/-642, -1087/-1079, -1088/-1080. Of these AP-2-like sites, only seven are included in panel B of the Figure ##FIG##3##4##. These were those flanking or overlapping some of the potential VDRE sequences indicated in panel A of the Figure ##FIG##3##4##. Of them, the -230/-222 (1) was located downstream, and the -245/-237 (2) and the -253/-245 (3) were overlapping the virtual VDRE sequences: -249/-235 and -247/-233, as shown in Figure ##FIG##3##4##, panel C. Moreover, the AP-2-like sites: -634/-626 (4) and -635/-627 (5) (Figure ##FIG##3##4##, panel B) were overlapping the virtual VDRE sequence: -637/-623 (Figure ##FIG##3##4##, panel C), and the AP-2-like sites: -642/-634 (6) and -650/-642 (7) (Figure ##FIG##3##4##, panel B) were located <italic>in tandem </italic>upstream of this last virtual VDRE sequence: -637/-623 (Figure ##FIG##3##4##, panel C).</p>", "<p>Therefore, we postulate as candidate VDRE sequences overlapped by AP-2-like sites those represented in Figure ##FIG##3##4##, panel C. One, located between -256 and -219 bp with three AP-2-like sites, and the other extending from -653 and -620 bp of the rat IR gene promoter with four AP-2-like sites. Separately or together, these VDRE sequences could form a locus that may respond to 1,25D3 via activation of VDR.</p>" ]

[ "<title>Discussion</title>", "<title>Effects of treatment with <italic>1,25D</italic>₃to non-diabetic rats</title>", "<p>Treatment with <italic>1,25D</italic>₃to non-diabetic rats increased body weights but it did not affect any parameter measured in the plasma or urine of these animals, including the plasma levels of calcium, phosphorus and 25-hydroxyvitamin D₃. Indeed, we did not detect significant differences in calcium and phosphorus, despite the increases of approximately 30%. In relation to 25-hydroxyvitamin D₃a number of studies [##UREF##0##33##] has shown that human serum values of 25-hydroxyvitamin D₃were maintained within a normal range across vitamin D supplies from 80 to 1000 IU daily, indicating the existence of an important homeostatic control system regulating human 25(OH)D₃serum concentration. Considering this, in the present study we hypothesize the existence of a similar counter regulatory mechanism in the rat. This mechanism could maintain rat plasma values of 25-hydroxyvitamin D₃within the observed normal range of 15–19 ng/ml, with vitamin D supplies of 150 IU daily administered by injection plus the vitamin D consumed in the diet.</p>", "<p>Treatment with <italic>1,25D</italic>₃increased the protein content and the indicator of cell size (protein/DNA) in the kidney, liver, and adipose tissue of non-diabetic rats, although the DNA and RNA content remained unaltered. These increments might indicate hypertrophy. However, although treatment with <italic>1,25D</italic>₃increased rat body weights, it did not particularly augment the weight of these tissues. Another explanation is that <italic>1,25D</italic>₃might cause hyperplasia. Regarding this, conflicting results have been reported [##REF##1966737##34##,##REF##8612483##35##].</p>", "<p>We also observed that treatment with <italic>1,25D</italic>₃to non-diabetic rats did not affect the expression of any of the two 9.5 and 7.5 Kb IR mRNA species present in kidney, liver, and adipose tissue. Our group and others have previously reported the presence of these species and their relative proportions in rat tissues [##REF##1280238##21##,##REF##9089644##36##]. Since there is only one rat IR gene, variation in transcript length may reflect alternative RNA processing events, such as polyadenylation at different 3' end sites in the final untranslated exon [##REF##2072925##37##].</p>", "<p>In addition, we detected that treatment with <italic>1,25D</italic>₃to non-diabetic rats affected neither the IR number nor the insulin response in terms of glucose transport in isolated adipocytes from these animals.</p>", "<p>The lack of effects of <italic>1,25D</italic>₃on IR expression and insulin activity in tissues of non-diabetic rats could be related to the absence of VDR in these tissues. However, both the VDR protein and VDR mRNA have been detected in the rat kidney [##UREF##1##38##,##REF##9528970##39##] and liver [##UREF##2##40##,##REF##9156521##41##], and while a VDR-like protein was identified in 3T3-L1 adipose cells [##REF##2840436##42##] VDR mRNA has been found in rat perirenal adipose tissue and 3T3-L1 adipose cells [##UREF##3##43##]. Another possibility could be the absence of regulation of VDR by its ligand in these tissues. Such lack of regulation has been described in the kidney [##REF##9528970##39##]. In the liver there are no previous data, while in adipose cells treatment with <italic>1,25D</italic>₃led to an increase in VDR mRNA levels [##UREF##3##43##] and a decrease in VDR protein [##REF##8612483##35##]. Regarding this and given our previous experience in determining both VDR protein and VDR mRNA expression and their regulation by <italic>1,25D</italic>₃in human cells [##REF##11075718##6##,##REF##12125099##7##], we determined the expression of this receptor and its regulation by 1,25D₃at the protein level in the kidney, liver and adipose tissue of non-diabetic rats. Our results showed a VDR protein only in the kidney (data not shown). More experiments are necessary to confirm this.</p>", "<title>Effects of treatment with <italic>1,25D</italic>₃to streptozotocin-induced diabetic rats</title>", "<p>Treatment with <italic>1,25D</italic>₃to streptozotocin-induced diabetic rats increased their body weight, but it did not revert the hyperglycemia and the hypoinsulinemia provoked by the streptozotocin. We also detected normal levels of 25-hydroxyvitamin D₃in streptozotocin-induced diabetic rats in accordance with other authors [##REF##3838933##30##]. These unchanged levels of 25-hydroxyvitamin D₃in streptozotocin-induced diabetic rats neither were altered by the treatment with <italic>1,25D</italic>₃, possibly due to the postulated plasma homeostatic control system that could regulate 25-hydroxyvitamin D₃plasma levels also in the absence of insulin in both groups of diabetic rats. In addition, we observed increased plasma levels of calcium and phosphorus after the treatment with <italic>1,25D</italic>₃. With regard to these last parameters, it is known that calcium <italic>per se </italic>is important for insulin secretion, as well as for correction of glucose intolerance [##REF##6283897##44##,##REF##8380563##45##]. Moreover, calcium and phosphorus have been described as regulators of VDR in renal and intestinal tissues [##REF##9528970##39##,##REF##7754807##46##].</p>", "<p>Treatment with <italic>1,25D</italic>₃to streptozotocin-diabetic rats did not modify the glycosuria and the ketonuria induced by the diabetes, but partially decreased the number of leukocytes in urine. This latter effect could be due to the broadly reported anti-inflammatory effects of <italic>1,25D</italic>₃and/or to a possible anti-proliferative effect of this hormone, as observed by our group and others in cell culture [##REF##11075718##6##,##UREF##0##33##].</p>", "<p>With respect to tissue effects of <italic>1,25D</italic>₃in streptozotocin-induced diabetic rats, we observed the permanence of high protein and DNA content values induced by the diabetes, accompanied by slight increases of the protein/DNA ratio values in the kidney and liver, but not in the adipose tissue. This suggests a certain hypertrophy produced by the action of <italic>1,25D</italic>₃in these two tissues. However, although treatment with <italic>1,25D</italic>₃to streptozotocin-diabetic rats increased rat body weights, it did not increase the weight of the kidney, liver or adipose tissue. The other explanation, hyperplasia, has conflicting results [##UREF##0##33##,##REF##1966737##34##].</p>", "<p>The induction of diabetes by streptozotocin did not affect IR mRNA species in the kidney. This finding was not in accord with previous results of increased renal IR mRNA levels, quantified by slot blot, in a similar rat model [##REF##1386819##22##]. This may reflect the failure to alter VDR number in the kidney of streptozotocin-diabetic rats [##REF##1966737##34##]. The treatment with 1,25D₃to streptozotocin-induced diabetic rats neither affected IR mRNA species in the kidney.</p>", "<p>Contrarily, streptozotocin produced an important increment in the levels of both IR mRNA species in liver. These findings agree with previous results of increased IR mRNA levels quantified by slot blot [##REF##1386819##22##], Northern blot analysis [##REF##1280238##21##,##REF##10037256##47##] and increased IR gene transcription [##REF##1280238##21##] in the liver of streptozotocin-induced diabetic rats. Streptozotocin also increased the levels of both IR mRNA species in the adipose tissue. This increment represents the first demonstration of an <italic>in vivo </italic>regulation of IR mRNA levels in the adipose tissue of streptozotocin-induced diabetic rats.</p>", "<p>In addition, we observed that the treatment with <italic>1,25D</italic>₃to streptozotocin-induced diabetic rats partially prevented the over-expression of IR mRNA induced by the diabetes in the liver and adipose tissue of these animals. The 9.5 Kb IR mRNA species was particularly decreased by <italic>1,25D</italic>₃in the liver, and the 7.5 Kb species in the adipose tissue. Therefore, these results provide the first demonstration of an <italic>in vivo </italic>tissue-specific regulation of rat IR mRNA levels by <italic>1,25D</italic>₃in streptozotocin-induced diabetic rats.</p>", "<p>A role for VDR in such genomic actions of <italic>1,25D</italic>₃in liver and adipose tissue is possible, although to our knowledge the regulation of VDR by <italic>1,25D</italic>₃has not yet been studied in these tissues of streptozotocin-diabetic rats.</p>", "<p>The correction by <italic>1,25D</italic>₃of the over-expression of IR expression at the RNA level in the liver and adipose tissue of streptozotocin-diabetic animals could represent a beneficial fact of amelioration of the diabetes. In this sense, other facts of amelioration of diabetes were the almost normalization of the number of IRs in adipocytes of streptozotocin-diabetic animals treated with <italic>1,25D</italic>₃, without alterations in receptor affinity, and the improvement of both basal and insulin-stimulated glucose transport in adipocytes of these diabetic animals treated with <italic>1,25D</italic>₃.</p>", "<p>The present <italic>in vivo </italic>results do not allow to determine whether the prevention by <italic>1,25D</italic>₃of the over-expression of IR mRNA induced by the diabetes in liver and adipose tissue of streptozotocin-diabetic animals is due to transcriptional and/or post-transcriptional regulation. Nevertheless, our previous <italic>in vitro </italic>results demonstrated that <italic>1,25D</italic>₃increased IR mRNA levels via mechanisms involving direct transcriptional activation of the human IR gene [##REF##7698218##5##, ####REF##11075718##6##, ##REF##12125099##7####12125099##7##]. Moreover, we identified an active VDRE in the human IR gene promoter that accounted for the transcriptional induction of this gene by <italic>1,25D</italic>₃in U-937 cells [##REF##12711007##29##].</p>", "<p>Therefore, in support of the possible participation of a transcriptional regulation by <italic>1,25D</italic>₃in the present in vivo processes, we have detected by computer search in the rat IR gene promoter two candidate VDREs overlapped by various AP-2-like sites. One VDRE is located between -256 and -219 bp, while the other extends from -653 and -620 bp of the rat IR gene promoter (Figure ##FIG##3##4##, panel C). Individually or in conjunction, these potential VDREs could form a locus that may respond to <italic>1,25D</italic>₃via activation of VDR. This locus may mediate the cross-talk between vitamin D and insulin signalling, although this remains to be determined.</p>" ]

[ "<title>Conclusion</title>", "<p>The present results indicate that while treatment with <italic>1,25D</italic>₃had practically no effect on non-diabetic rats, the same treatment in streptozotocin-induced diabetic rats corrected in part the over-expression of the IR gene in liver and adipose tissue, although it did not revert the hyperglycemia, hypoinsulinemia, glycosuria or ketonemia of these diabetic animals. At the same time, it produced normalization of the IR number without alterations in the receptor affinity, but with an improvement of the insulin response in terms of glucose transport in adipocytes of these diabetic animals. These genomic actions of <italic>1,25D</italic>₃could represent beneficial facts of amelioration of diabetes via mechanisms, possibly involving direct transcriptional activation of the rat IR gene. A computer search in the rat IR promoter revealed the existence of two candidate vitamin D response element (VDRE) sequences, located at -256/-219 bp and -653/-620 bp and overlapped by three and four AP-2-like sites respectively. These candidate VDREs may respond to <italic>1,25D</italic>₃via activation of the vitamin D receptor, although this remains to be investigated.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>this study set out to examine the effects of the treatment with 1,25-dihydroxyvitamin D₃(<italic>1,25D</italic>₃) [150 IU/Kg (3.75 μg/Kg) one a day, for 15 days] to non-diabetic rats and in rats rendered diabetic by a single injection of streptozotocin [65 mg/kg].</p>", "<title>Results</title>", "<p>treatment with <italic>1,25D</italic>₃to non-diabetic rats did not affect the biochemical parameters measured in the plasma and urine of these animals. Likewise, insulin receptor expression in the kidney, liver, or adipose tissue and insulin-stimulated glucose transport in adipocytes from these animals were not affected either. Treatment with <italic>1,25D</italic>₃to streptozotocin-induced diabetic rats did not correct the hyperglycemia, hypoinsulinemia, glycosuria or ketonemia induced by the diabetes, although it partially reversed the over-expression of the insulin receptor gene in the liver and adipose tissue, without altering the normal expression of this gene in the kidney. These effects were accompanied by a normalization of the number of insulin receptors without altering receptor affinity but improving the insulin response to glucose transport in adipocytes from these diabetic animals. Moreover, a computer search in the rat insulin receptor promoter revealed the existence of two candidate vitamin D response element (VDRE) sequences located at -256/-219 bp and -653/-620 bp, the first overlapped by three and the second by four AP-2-like sites.</p>", "<title>Conclusion</title>", "<p>these genomic actions of <italic>1,25D</italic>₃could represent beneficial effects associated with the amelioration of diabetes via mechanisms that possibly involve direct transcriptional activation of the rat insulin receptor gene. The candidate VDREs identified may respond to <italic>1,25D</italic>₃via activation of the vitamin D receptor, although this remains to be investigated.</p>" ]

[ "<title>List of abbreviations</title>", "<p>1,25D₃: 1,25-dihydroxyvitamin D₃; IR: insulin receptor; VDR: vitamin D receptor; RXR: retinoid × receptor; VDRE: vitamin D response element; STZ: streptozotocin.</p>", "<title>Authors' contributions</title>", "<p>CC conceived the concept and design of the study, carried out the binding and activity studies, analyzed the computer data of DNA sequences and provided drafting of the article. BM carried out the rat models and the determinations in plasma and urine of these animals and performed the RNA blot assays. She also supplied statistical expertise, collected and assembled the data. MG-A participated in the identification by computer analysis of potential VDRE sequences and provided computer support. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>We are most grateful to Prof. B. J. Goldstein, for providing the rat IR cDNA probe. This work was supported by Funds from the Comunidad de Madrid/Universidad Complutense (14134/2005) and the Ministerio de Sanidad y Consumo (P1061744/2006).</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Renal, hepatic and adipose IR mRNA levels</bold>. Insulin receptor (IR) mRNA levels in the kidney (A, B), liver (C, D) and epididymal adipose tissue (E, F) of sham-treated rats (<italic>Control</italic>), rats treated with <italic>1,25D</italic>₃[150 IU/Kg (3.75 μg/Kg) one a day, for 15 days](<italic>1,25D</italic>₃), streptozotocin-induced diabetic rats (<italic>STZ</italic>) and streptozotocin-induced diabetic rats treated with <italic>1,25D</italic>₃(<italic>STZ</italic>+<italic>1,25D</italic>₃). On the right, the autoradiograph of a representative Northern blot experiment using the kidney (B), liver (D) and adipose tissue (F) of two rats. The sizes of the two major IR mRNA species are shown in the margin. On the left, densitometric analysis of three independent Northern blot experiments on kidney (A), liver (C), and adipose tissue (E). IR mRNA species were quantified separately, normalized to the respective 28S rRNA values and expressed per unit of RNA as a percentage of the values obtained in <italic>Control</italic>-rats (mean ± SEM). ^ap &lt; 0.05 vs. <italic>Control</italic>-rats; ^bp &lt; 0.05 vs. <italic>1,25D</italic>₃-rats; and ^cp &lt; 0.05 vs. <italic>STZ</italic>-rats.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Insulin receptor number per cell</bold>. Insulin receptor number in epididymal adipocytes from sham-treated rats (<italic>Control</italic>), rats treated with <italic>1,25D</italic>₃[150 IU/Kg (3.75 μg/Kg) one a day, for 15 days] (<italic>1,25D</italic>₃), streptozotocin-induced diabetic rats (<italic>STZ</italic>) and streptozotocin-induced diabetic rats treated with <italic>1,25D</italic>₃(<italic>STZ</italic>+<italic>1,25D</italic>₃). Values are the mean ± SEM of 3–9 determinations in each group. ^ap &lt; 0.05 vs. <italic>Control</italic>-rats; ^bp &lt; 0.05 vs. <italic>1,25D</italic>₃-rats; and ^cp &lt; 0.05 vs. <italic>STZ</italic>-rats.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Basal and insulin stimulated glucose transport</bold>. Basal and insulin stimulated glucose transport in epididymal adipocytes from sham-treated rats (<italic>Control</italic>), rats treated with <italic>1,25D</italic>₃[150 IU/Kg (3.75 μg/Kg) one a day, for 15 days] (<italic>1,25D</italic>₃), streptozotocin-induced diabetic rats (<italic>STZ</italic>) and streptozotocin-induced diabetic rats treated with <italic>1,25D</italic>₃(<italic>STZ</italic>+<italic>1,25D</italic>₃). Values are the mean ± SEM of 6–7 determinations within each group. ^ap &lt; 0.05 vs. <italic>Control</italic>-rats; ^bp &lt; 0.05 vs. <italic>1,25D</italic>₃-rats; and ^cp &lt; 0.05 vs. <italic>STZ</italic>-rats.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Panel A: Computer search of vitamin D response elements (VDREs) in the rat insulin receptor (IR) gene promoter</bold>. Potential VDRE sequences in the rat IR gene promoter were identified by their homology with two <italic>consensus </italic>VDRE sequences using the SEQFIND programme developed in our laboratory. Nucleotides identical to the first VDRE <italic>consensus </italic>are labelled by filled circles (●) and nucleotides identical to the second VDRE <italic>consensus </italic>are labelled by empty circles (○). Panel B: Computer search of AP-2-like sites in the rat insulin receptor (IR) gene promoter. Potential AP-2-like sites in the rat IR gene promoter were identified by their homology with the AP-2 <italic>consensus </italic>(5'GCCN3GGG3') [##REF##11137286##32##] using the SEQFIND programme. We detected 17 AP-2-like sites showing a difference of only 1 base from this <italic>consensus</italic>. Of these, only 7 AP-2-like sites flanking or overlapping the potential VDRE sequences in panel A are shown. Nucleotides identical to the AP-2 <italic>consensus </italic>are labelled by stars (⋆). Panel C: Candidate VDRE sequences overlapped by AP-2-like sites in the rat insulin receptor gene promoter. Nucleotides identical to the first VDRE <italic>consensus </italic>are labelled by filled circles (●) and nucleotides identical to the second VDRE <italic>consensus </italic>are labelled by empty circles (○)</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Body weights and plasma values of the four groups of rats under study</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold><italic>Control</italic></bold></td><td align=\"center\"><bold><italic>1,25D</italic></bold>₃</td><td align=\"center\"><bold><italic>STZ</italic></bold></td><td align=\"center\"><bold><italic>STZ + 1,25D</italic></bold>₃</td></tr></thead><tbody><tr><td align=\"center\"><bold>Final weight (g)</bold></td><td align=\"center\">283 ± 1</td><td align=\"center\">290 ± 2^a</td><td align=\"center\">242 ± 6^ab</td><td align=\"center\">254 ± 3^abc</td></tr><tr><td align=\"center\"><bold>Glucose (mg/dl)</bold></td><td align=\"center\">125 ± 7</td><td align=\"center\">130 ± 2</td><td align=\"center\">420 ± 36^ab</td><td align=\"center\">459 ± 16^ab</td></tr><tr><td align=\"center\"><bold>Insulin (ng/ml)</bold></td><td align=\"center\">1.6 ± 0.3</td><td align=\"center\">1.5 ± 0.2</td><td align=\"center\">0.7 ± 0.3^ab</td><td align=\"center\">0.5 ± 0.1^ab</td></tr><tr><td align=\"center\"><bold>25-Hydroxy-vitamin D₃(ng/ml)</bold></td><td align=\"center\">19 ± 1</td><td align=\"center\">15 ± 1</td><td align=\"center\">15 ± 3</td><td align=\"center\">17 ± 3</td></tr><tr><td align=\"center\"><bold>Calcium (mg/dl)</bold></td><td align=\"center\">7.5 ± 1</td><td align=\"center\">10 ± 1</td><td align=\"center\">10 ± 0.4</td><td align=\"center\">11 ± 1^a</td></tr><tr><td align=\"center\"><bold>Phosphorus (mg/dl)</bold></td><td align=\"center\">6.8 ± 1</td><td align=\"center\">8.7 ± 1</td><td align=\"center\">7.2 ± 0.5</td><td align=\"center\">11 ± 0.7^ac</td></tr><tr><td align=\"center\"><bold>Proteins (mg/dl)</bold></td><td align=\"center\">65 ± 1</td><td align=\"center\">66 ± 1</td><td align=\"center\">63 ± 0.9</td><td align=\"center\">64 ± 1</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Total DNA, RNA and protein content in different tissues from the four groups of rats</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold><italic>Control</italic></bold></td><td align=\"center\"><bold><italic>1,25D</italic></bold>₃</td><td align=\"center\"><bold><italic>STZ</italic></bold></td><td align=\"center\"><bold><italic>STZ + 1,25D</italic></bold>₃</td></tr></thead><tbody><tr><td align=\"left\"><bold><underline>Kidney</underline></bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">   <bold>DNA (mg/g)</bold></td><td align=\"center\">3.0 ± 0.1</td><td align=\"center\">2.8 ± 0.2</td><td align=\"center\">3.3 ± 0.5</td><td align=\"center\">3.4 ± 0.1^ab</td></tr><tr><td align=\"left\">   <bold>Protein(mg/g)</bold></td><td align=\"center\">69 ± 2</td><td align=\"center\">120 ± 5^a</td><td align=\"center\">134 ± 33^a</td><td align=\"center\">153 ± 29^ab</td></tr><tr><td align=\"left\">   <bold>Prot/DNA (mg/mg)</bold></td><td align=\"center\">18 ± 5</td><td align=\"center\">43 ± 2^a</td><td align=\"center\">41 ± 10^a</td><td align=\"center\">45 ± 10^a</td></tr><tr><td align=\"left\">   <bold>RNA (mg/g)</bold></td><td align=\"center\">1.3 ± 0.1</td><td align=\"center\">1.6 ± 0.2</td><td align=\"center\">1.3 ± 0.3</td><td align=\"center\">1.7 ± 0.3</td></tr><tr><td align=\"left\"><bold><underline>Liver</underline></bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">   <bold>DNA (mg/g)</bold></td><td align=\"center\">1.9 ± 0.2</td><td align=\"center\">1.5 ± 0.2</td><td align=\"center\">2.9 ± 0.6^ab</td><td align=\"center\">2.4 ± 0.2^ab</td></tr><tr><td align=\"left\">   <bold>Protein(mg/g)</bold></td><td align=\"center\">105 ± 19</td><td align=\"center\">146 ± 17^a</td><td align=\"center\">203 ± 27^ab</td><td align=\"center\">182 ± 15^a</td></tr><tr><td align=\"left\">   <bold>Prot/DNA (mg/mg)</bold></td><td align=\"center\">52 ± 9</td><td align=\"center\">99 ± 11^a</td><td align=\"center\">67 ± 5</td><td align=\"center\">76 ± 3^a</td></tr><tr><td align=\"left\">   <bold>RNA (mg/g)</bold></td><td align=\"center\">3.9 ± 0.3</td><td align=\"center\">3.2 ± 0.4</td><td align=\"center\">3.5 ± 0.4</td><td align=\"center\">3.8 ± 0.2</td></tr><tr><td align=\"left\"><bold><underline>Adipose tissue</underline></bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">   <bold>DNA (mg/g)</bold></td><td align=\"center\">0.2 ± 0.06</td><td align=\"center\">0.2 ± 0.02</td><td align=\"center\">0.3 ± 0.03</td><td align=\"center\">0.3 ± 0.04</td></tr><tr><td align=\"left\">   <bold>Protein(mg/g)</bold></td><td align=\"center\">5.1 ± 1</td><td align=\"center\">8.2 ± 0.8^a</td><td align=\"center\">7.9 ± 0.5^a</td><td align=\"center\">6.1 ± 1</td></tr><tr><td align=\"left\">   <bold>Prot/DNA (mg/mg)</bold></td><td align=\"center\">24 ± 5</td><td align=\"center\">47 ± 6^a</td><td align=\"center\">25 ± 1^b</td><td align=\"center\">20 ± 3</td></tr><tr><td align=\"left\">   <bold>RNA (mg/g)</bold></td><td align=\"center\">0.1 ± 0.02</td><td align=\"center\">0.1 ± 0.01</td><td align=\"center\">0.1 ± 0.03</td><td align=\"center\">0.1 ± 0.04</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Insulin binding parameters in epididymal adipocytes from the four groups of rats</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold><italic>Control</italic></bold></td><td align=\"center\"><bold><italic>1,25D</italic></bold>₃</td><td align=\"center\"><bold><italic>STZ</italic></bold></td><td align=\"center\"><bold><italic>STZ + 1,25D</italic></bold>₃</td></tr></thead><tbody><tr><td align=\"center\"><bold>Insulin receptor number</bold></td><td align=\"center\">122000 ± 23000</td><td align=\"center\">153000 ± 39000</td><td align=\"center\">44000 ± 6000^ab</td><td align=\"center\">115000 ± 42000^c</td></tr><tr><td align=\"center\"><bold>ED:50 value</bold></td><td align=\"center\">3.0 ± 0.4</td><td align=\"center\">4.2 ± 1.4</td><td align=\"center\">3.7 ± 1.4</td><td align=\"center\">4.8 ± 0.9</td></tr><tr><td align=\"center\"><bold>Adipose tissue weight (g)</bold></td><td align=\"center\">3.2 ± 0.2</td><td align=\"center\">3.0 ± 0.3</td><td align=\"center\">1.6 ± 0.2^ab</td><td align=\"center\">2.0 ± 0.2^a</td></tr><tr><td align=\"center\"><bold>Adipocyte diameter (μm)</bold></td><td align=\"center\">50 ± 1</td><td align=\"center\">51 ± 1</td><td align=\"center\">39 ± 3^ab</td><td align=\"center\">40 ± 1^ab</td></tr><tr><td align=\"center\"><bold>Insulin receptor number/μm</bold>²</td><td align=\"center\">16 ± 3</td><td align=\"center\">19 ± 5</td><td align=\"center\">9 ± 1^ab</td><td align=\"center\">23 ± 8</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>Sham-treated rats (<italic>Control</italic>), rats treated with <italic>1,25D</italic>₃[150 IU/Kg (3.75 μg/Kg) one a day, for 15 days] (<italic>1,25D</italic>₃), streptozotocin-induced diabetic rats (<italic>STZ</italic>) and streptozotocin-induced diabetic rats treated with <italic>1,25D</italic>₃(<italic>STZ+1,25D</italic>₃).</p><p>Values are the mean ± SEM of 6–12 determinations in each group.</p><p>^ap &lt; 0.05 vs. <italic>Control</italic>-rats; ^bp &lt; 0.05 vs. <italic>1,25D</italic>₃-rats; ^cp &lt; 0.05 vs. <italic>STZ</italic>-rats.</p></table-wrap-foot>", "<table-wrap-foot><p>Sham-treated rats (<italic>Control</italic>), rats treated with <italic>1,25D</italic>₃[150 IU/Kg (3.75 μg/Kg) one a day, for 15 days] (<italic>1,25D</italic>₃), streptozotocin-induced diabetic rats (<italic>STZ</italic>) and streptozotocin-induced diabetic rats treated with <italic>1,25D</italic>₃(<italic>STZ+1,25D</italic>₃).</p><p>Values are the mean ± SEM of 4–6 determinations in each group.</p><p>^ap &lt; 0.05 vs. <italic>Control</italic>-rats; ^bp &lt; 0.05 vs. <italic>1,25D</italic>₃-rats; ^cp &lt; 0.05 vs. <italic>STZ</italic>-rats.</p></table-wrap-foot>", "<table-wrap-foot><p>Sham-treated rats (<italic>Control</italic>), rats treated with <italic>1,25D</italic>₃[150 IU/Kg (3.75 μg/Kg) one a day, for 15 days] (<italic>1,25D</italic>₃), streptozotocin-induced diabetic rats (<italic>STZ</italic>) and streptozotocin-induced diabetic rats treated with <italic>1,25D</italic>₃(<italic>STZ+1,25D</italic>₃).</p><p>Values are the mean ± SEM of 3–9 determinations in each group.</p><p>^ap &lt; 0.05 vs. <italic>Control</italic>-rats; ^bp &lt; 0.05 vs. <italic>1,25D</italic>₃-rats; ^cp &lt; 0.05 vs. <italic>STZ</italic>-rats.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2199-9-65-1\"/>", "<graphic xlink:href=\"1471-2199-9-65-2\"/>", "<graphic xlink:href=\"1471-2199-9-65-3\"/>", "<graphic xlink:href=\"1471-2199-9-65-4\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Cardiovascular disease (CVD) is the leading cause of death among women and men. About 13 million Americans have CVD and approximately 450,000 people die each year in the US from CVD [##UREF##0##1##]. Risk factors for CVD in women include post-menopausal status, age, hypertension, smoking, and diabetes [##REF##11815309##2##]. The onset of menopause coincides with elevated serum lipids such as total cholesterol, LDL-cholesterol, and triglycerides and decreased HDL-cholesterol [##REF##2488072##3##]. Elevated serum lipids in post-menopausal women are partly due to the loss of estrogen [##REF##3675308##4##]. Hormone therapy showed a reduction in LDL-cholesterol and an increase in HDL-cholesterol [##REF##7517831##5##]. It also frequently increased triglycerides by increasing the production of triglyceride-rich VLDLs [##UREF##1##6##]. Additionally, hormone therapy increased the risk for stroke, breast cancer, and cholecystitis [##REF##12186605##7##]. Hormone replacement therapy is not for all post-menopausal women [##REF##8213496##8##].</p>", "<p>Generally, relative to drug therapy, dietary intervention is less expensive and cost effective for primary intervention of heart diseases [##REF##8609317##9##]. Consumption of foods containing dietary fiber, may improve the long-term maintenance of low atherogenic LDL-cholesterol [##REF##9497178##10##]. Psyllium (<italic>Plantago ovata</italic>) seed husk fiber, a widely used soluble fiber, has been known to reduce serum total cholesterol and LDL-cholesterol [##REF##8925793##11##, ####REF##9925120##12##, ##REF##10648260##13####10648260##13##]. Additionally, psyllium fiber supplementation with 10 mg of simvastatin (hypocholesterolemic drug) was as effective as 20 mg of simvastatin alone [##REF##15911730##14##]. This suggests that psyllium fiber is an effective adjuvant hypocholesterolemic agent.</p>", "<p>There is a paucity of data on how fiber therapy modulates the risk for CVD in women. Previous studies conducted in women indicated that dietary therapy for lowering serum lipids differs according to menopausal status [##REF##11912561##15##]. In men compared to post-menopausal women, a greater decrease was observed in total cholesterol after 4 months of high fiber diet [##UREF##2##16##] suggesting that gender differences exist in hypocholesterolemic effect of dietary fiber. Limited studies were conducted comparing lipid responses to psyllium fiber intervention in pre- and post-menopausal women. Therefore, we tested the hypothesis of whether serum lipid responses to psyllium fiber intake in post-menopausal, hypercholesterolemic women differ from pre-menopausal, hypercholesterolemic women.</p>" ]

[ "<title>Methods</title>", "<title>Subjects</title>", "<p>A total of 25 hypercholesterolemic women (13 pre-menopausal and 12 post-menopausal) were recruited for the study. Hypercholesterolemia was defined as having serum cholesterol &gt;200 mg/dL. Subjects were recruited from San Francisco and surrounding communities through posting of flyers. Participants lived in their homes during the study. The exclusion criteria were smoking, diabetes, pregnancy, breast feeding, allergic to psyllium fiber, and blood cholesterol ≤200 mg/dL. Individuals with thyroid and gastrointestinal diseases, CVD, chronic alcohol and drug abuse, individuals who were taking lipid-lowering and/or anti-hypertensive medications, and individuals who were on estrogen replacement therapy were excluded from the study. Also, strict vegetarians (vegans) were excluded from the study because their habitual fiber intake is generally higher than non-vegetarians. Post-menopausal status was defined as women who were into menopause at least one year from their last menstruation at the beginning of the study. Subjects were asked to continue their habitual diet and to maintain their pre-study physical activities and body weights. Study protocol and informed consent were approved by the Human Subjects Protection Committee at San Francisco State University.</p>", "<title>Administration of psyllium fiber</title>", "<p>Subjects were asked to consume psyllium fiber enriched cookies for 6 weeks. We incorporated 15 g of psyllium fiber into cookies (≈5 g/cookie). Subjects consumed 3 cookies/d (one in each meal). Psyllium fiber enriched cookies were made in the departmental food laboratory, according to the formulation given in Table ##TAB##0##1##. Previous studies documented that 4–6 week feeding period was adequate to observe changes in serum lipids with psyllium fiber intake [##REF##1319110##17##,##UREF##3##18##]. The reason that psyllium fiber was incorporated into cookies rather than administering as supplements because fiber has shown to produce maximum cholesterol-lowering effect when it is taken mixed with foods [##REF##8172091##19##]. Numerous products (muffins and powdered drink mixes) incorporating psyllium were tested for palatability and cookies yielded the most appetizing product. Cookies were individually supplied to subject's home in a fresh or frozen form on a weekly basis. The rationale for the amount of fiber used (15 g/d) was based on the recommendation of 25–35 g/d of fiber intake for adults [##UREF##4##20##]. Since Americans consume about 15 g/d of fiber [##REF##8925793##11##], by adding 15 g/d subjects would meet the recommendations for fiber. During the study, subjects were asked to refrain from consuming fiber supplements. If a subject was a social drinker that subject was asked to maintain her usual alcohol intake level.</p>", "<title>Serum lipid analysis</title>", "<p>Twelve-hour overnight fasting blood samples were obtained from venipuncture in the arm at the beginning and at the end of the 6-week period by a certified phlebotomist. Blood samples were centrifuged and serum was separated before lipid analysis. Serum samples were analyzed for total cholesterol, LDL-cholesterol, HDL-cholesterol, triglycerides, apolipoprotein A-1, and apolipoprotein B using an automatic analyzer (Quest Diagnostic Laboratories, Dublin, CA).</p>", "<title>Data analysis</title>", "<p>Data were presented as mean ± standard deviation. Paired t-test was used to determine the significant difference between the baseline and the post-fiber treatment values in pre- and post-menopausal women (Microsoft Excel for Windows, 1999). Statistical significance was set at P &lt; 0.05.</p>" ]

[ "<title>Results</title>", "<p>Of 25 subjects, 19 (8 pre- and 11 post-menopausal) women completed the study. Reasons for discontinuation of participation included gastrointestinal discomfort, increased frequency of bowel movements, softer stools, and personal travel and time conflicts. Subjects' characteristics were summarized in Table ##TAB##1##2##. As expected, the post-menopausal women were significantly older compared to pre-menopausal women (p &lt; 0.05). Body mass index (BMI) at baseline between pre- and post-menopausal women was not significantly different. BMI for all subjects were within the desirable low health risk range. There were no significant differences in body weight between baseline to post-study measurements in both pre- and post-menopausal women (data are not shown).</p>", "<p>The baseline and post-study serum lipid concentrations for pre- and post-menopausal women are presented in Table ##TAB##2##3##. Mean serum total cholesterol concentration with psyllium fiber intake was significantly lower compared to baseline values in post-menopausal women (218.7 mg/dL vs. 230.8 mg/dL; p &lt; 0.05) but not in pre-menopausal women (240.5 mg/dL vs. 243.6 mg/dL). The decrease in post-menopausal women with respect to total serum cholesterol was ≈5.2%.</p>", "<p>Mean HDL cholesterol was significantly lower in post-menopausal women with fiber intake (65 mg/dL to 58.4 mg/dL) (p &lt; 0.05). In contrast, no change was observed in HDL-cholesterol concentrations in pre-menopausal women with psyllium. Changes in LDL-cholesterol in pre- and post-menopausal women from baseline to post-study were not significant although there was a slight decline in values with psyllium fiber intake. The decline was ≈4.6% in post-menopausal women and was ≈3.2% in pre-menopausal women.</p>", "<p>There was a ≈7.6% and ≈18.6% increase in serum triglycerides in pre- and post-menopausal women, respectively, with psyllium fiber supplementation. However, these changes were not statistically significant. Similarly, there were no significant changes in response to psyllium fiber intake in serum apolipoprotein A-1 and apolipoprotein B concentrations in both pre- and post-menopausal women.</p>" ]

[ "<title>Discussion</title>", "<p>We investigated the effect of psyllium fiber intake at 15 g/d dosage level for 6 weeks on serum lipids in free-living, pre- and post-menopausal, hypercholesterolemic women. Psyllium fiber significantly lowered serum total cholesterol in post-menopausal women. Decreased total serum cholesterol was primarily due to changes in the HDL-cholesterol. Although LDL-cholesterol was lower (≈4.6%) with fiber intake, the decrease was not statistically significant when compared to baseline values in post-menopausal women. Concentrations of serum triglycerides, apolipoprotein A-1, and apolipoprotein B were not affected by psyllium fiber intake in post-menopausal women. Also, psyllium fiber intake had no effect on serum total cholesterol, LDL-cholesterol, HDL-cholesterol, triglycerides, apolipoprotein A-1, and apolipoprotein B in pre-menopausal, hypercholesterolemic women.</p>", "<p>Previously, we documented cholesterol-lowering effect of psyllium in normocholesterolemic humans [##UREF##3##18##]. Other studies have shown that consumption of psyllium fiber reduced serum total cholesterol and LDL-cholesterol in hypercholesterolemic subjects [##REF##9925120##12##,##REF##10648260##13##]. Several other investigators reported a 3–17% reduction of serum total cholesterol and a 4–20% reduction of LDL-cholesterol with psyllium fiber [##REF##9925120##12##,##REF##8172091##19##,##REF##9311953##21##,##REF##10837282##22##]. In a meta-analysis, Anderson et al. (2000b) reported that consumption of 10.2 g psyllium/d lowered total cholesterol by 4%. Whereas, in our study, we achieved a ≈5.2% decrease in total serum cholesterol concentrations at a dosage level of 15 g psyllium/d. Most of the psyllium fiber used in previous studies was either in the form of ready-to-eat cereal or as fiber supplement. In our study, we administered psyllium fiber in cookies. Thus, the method of administration of psyllium might account for some differences in cholesterol-lowering property of psyllium fiber in various studies.</p>", "<p>Davidson et al. [##REF##9497178##10##] reported that consumption of psyllium fiber lowered total cholesterol without modifying HDL-cholesterol concentrations. In contrast, our study did not reveal that psyllium fiber reduced serum total cholesterol without lowering HDL-cholesterol. The discrepancy in results between this study and other studies may likely be due to differences in characteristics of subjects. Also, our subjects consumed self-selected diets during the study. Thus, the diet intake might have some influence on the study outcomes.</p>", "<p>In this study, we found the hypocholesterolemic effect of psyllium fiber in post-menopausal women but not in pre-menopausal women indicating that pre- and post-menopausal women responded differently to psyllium fiber. Lack of significant hypocholesterolemic effect of psyllium fiber in pre-menopausal women suggests that the elevated cholesterol concentrations in those subjects might be due to genetic causes and thus are unresponsive to dietary modification. On the other hand, elevated cholesterol in post-menopausal women is often due to deranged lipid metabolism (non-genetic causes) arising from loss of estrogen function and thus is responsive to psyllium fiber intervention. In post-menopausal women with established CVD, higher intake of fiber has been associated with less progression of coronary atherosclerosis [##REF##16084154##23##]. Also, psyllium fiber significantly reduced post-prandial concentrations of serum triglycerides, glucose, and insulin in persons with diabetes [##REF##9747644##24##] suggesting a role for psyllium fiber in the management of diabetes.</p>", "<p>Previous studies on the effect of psyllium fiber on triglycerides yielded equivocal results [##REF##10837282##22##,##REF##2724486##25##,##REF##11566640##26##]. Psyllium fiber intake significantly reduced plasma triglycerides in men, but in post-menopausal women, it resulted in a significant increase with no change in pre-menopausal women [##REF##11566640##26##]. In our study, we found no significant effect of psyllium fiber intake on serum triglycerides in both pre- and post-menopausal women, although there was a trend towards increased serum triglycerides.</p>", "<p>The lack of effect of psyllium fiber intake on serum lipids in pre-menopausal women might be due insufficient number of subjects recruited to observe statistically significant differences. Large sample size might reveal significant total cholesterol and/or LDL-cholesterol reduction. Therefore, the results of this study must be used with caution. Although, we have instructed the subjects not to alter their dietary habits during the study, it is possible that our results may have confounded by the fiber from unintended food sources. Because psyllium fiber was incorporated into cookies, consumption of psyllium enriched cookies lead to an additional intake of ≈30 g/d fat. This may have blunted the hypocholesterolemic effect of psyllium in both pre- and post-menopausal women. Additionally, all subjects in the post-menopausal group were Asians. It is not known how this has affected the cholesterol-lowering effect of psyllium.</p>", "<p>In conclusion, post- and pre-menopausal women differed with regard to lipid responses to psyllium fiber. Serum total cholesterol was responsive to psyllium fiber in post-menopausal women. Post-menopausal women would likely benefit from addition of psyllium fiber to their diets in reducing the risk of heart diseases because every 1% reduction in serum total cholesterol concentration results in 2% reduction in risk for heart diseases [##UREF##5##27##]. Differences in cholesterol-lowering response to psyllium intake may be related to the differences in hormonal status between pre- and post-menopausal women [##REF##11912561##15##,##REF##11566640##26##]. Additionally, consumption of psyllium fiber provides a low-cost adjunct to the National Cholesterol Education Program diets for hypercholesterolemia.</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Cardiovascular disease is the leading cause of death in women and men. Psyllium, a soluble fiber has been known to reduce serum lipids. In this pilot study, we evaluated whether menopausal status would affect the serum lipid responses to psyllium fiber in women.</p>", "<title>Methods</title>", "<p>Eleven post-menopausal and eight pre-menopausal women with serum total cholesterol &gt;200 mg/dL were included in the study. Subjects consumed their habitual diet and 15 g psyllium/d for 6 weeks. Psyllium was incorporated into cookies. Each cookie contained ≈5 g of psyllium fiber. Subjects ate one cookie in each meal.</p>", "<title>Results</title>", "<p>With psyllium fiber, total cholesterol concentration was significantly lower (≈5.2%, P &lt; 0.05) in post-menopausal women but not in pre-menopausal women (≈1.3%). Also, there was a significant decrease in HDL-cholesterol in post-menopausal women (≈10.2%, P &lt; 0.05). There were no significant changes observed in concentrations of LDL-cholesterol, triglycerides, apolipoprotein A1, and apolipoprotein B in both pre- and post-menopausal women with psyllium.</p>", "<title>Conclusion</title>", "<p>In this pilot study, post- and pre-menopausal, hypercholesterolemic women responded differently to psyllium fiber supplementation. Post-menopausal women would benefit from addition of psyllium to their diets in reducing the risk for heart diseases. The results of this study should be used with caution because the study was based on a small sample size.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>VG and JK directed the research. VG was responsible for getting the support for the project. Both authors are responsible for collecting the data, analysis of data, and drafting the manuscript. All authors read and approved the manuscript before submission.</p>" ]

[ "<title>Acknowledgements</title>", "<p>The study was supported by the Zellemer Grant program from the California Dietetic Association.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Composition of psyllium fiber-enriched cookies*</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Ingredients</td><td align=\"center\">Amount^#</td></tr></thead><tbody><tr><td align=\"left\">Psyllium fiber, g</td><td align=\"center\">250</td></tr><tr><td align=\"left\">All purpose flour, g</td><td align=\"center\">252</td></tr><tr><td align=\"left\">Sugar, g</td><td align=\"center\">294</td></tr><tr><td align=\"left\">Brown sugar, g</td><td align=\"center\">316.5</td></tr><tr><td align=\"left\">Shortening, g</td><td align=\"center\">484</td></tr><tr><td align=\"left\">Eggs, g</td><td align=\"center\">200</td></tr><tr><td align=\"left\">Baking soda, g</td><td align=\"center\">4.68</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\">Vanilla extract, mg</td><td align=\"center\">3.5</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Characteristics of subjects*</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Variable</td><td align=\"center\">Pre-menopausal women</td><td align=\"center\">Post-menopausal women</td></tr></thead><tbody><tr><td align=\"left\">Race-ethnicity</td><td/><td/></tr><tr><td align=\"left\"> White/Hispanic/Asian, n</td><td align=\"center\">1/1/6</td><td align=\"center\">0/0/11</td></tr><tr><td align=\"left\">Age, y</td><td align=\"center\">34.6 ± 11.5</td><td align=\"center\">52.9 ± 2.8^#</td></tr><tr><td align=\"left\">Weight, kg</td><td align=\"center\">63.2 ± 9.3</td><td align=\"center\">57.6 ± 5.6^@</td></tr><tr><td align=\"left\">Height, m</td><td align=\"center\">1.6 ± 0.1</td><td align=\"center\">1.6 ± 0.04^@</td></tr><tr><td colspan=\"3\"><hr/></td></tr><tr><td align=\"left\">BMI, kg/m²</td><td align=\"center\">24.2 ± 4.2</td><td align=\"center\">22.6 ± 2.2^@</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Serum lipid responses to psyllium fiber intake in pre- and post-menopausal women*</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Serum lipid</td><td align=\"center\">Pre-menopausal women (n = 8)</td><td align=\"center\">Post-menopausal women (n = 11)</td></tr></thead><tbody><tr><td align=\"left\">Total cholesterol</td><td/><td/></tr><tr><td align=\"left\"> Baseline, mg/dL</td><td align=\"center\">243.6 ± 40.3</td><td align=\"center\">230.8 ± 24.4</td></tr><tr><td align=\"left\"> Post-fiber, mg/dL</td><td align=\"center\">240.5 ± 41.2</td><td align=\"center\">218.7 ± 17.0^#</td></tr><tr><td align=\"left\">LDL-cholesterol</td><td/><td/></tr><tr><td align=\"left\"> Baseline, mg/dL</td><td align=\"center\">156.0 ± 46.0</td><td align=\"center\">133.4 ± 32.3</td></tr><tr><td align=\"left\"> Post-fiber, mg/dL</td><td align=\"center\">151.0 ± 42.0</td><td align=\"center\">127.3 ± 14.3</td></tr><tr><td align=\"left\">HDL-cholesterol</td><td/><td/></tr><tr><td align=\"left\"> Baseline, mg/dL</td><td align=\"center\">61.8 ± 12.2</td><td align=\"center\">65.0 ± 17.0</td></tr><tr><td align=\"left\"> Post-fiber, mg/dL</td><td align=\"center\">61.4 ± 24.1</td><td align=\"center\">58.4 ± 14.8^#</td></tr><tr><td align=\"left\">Triglycerides</td><td/><td/></tr><tr><td align=\"left\"> Baseline, mg/dL</td><td align=\"center\">129.5 ± 52.1</td><td align=\"center\">139.5 ± 72.1</td></tr><tr><td align=\"left\"> Post-fiber, mg/dL</td><td align=\"center\">139.4 ± 77.4</td><td align=\"center\">165.4 ± 85.2</td></tr><tr><td align=\"left\">Apolipoprotein A-1</td><td/><td/></tr><tr><td align=\"left\"> Baseline, mg/dL</td><td align=\"center\">152.5 ± 30.3</td><td align=\"center\">150.6 ± 21.8</td></tr><tr><td align=\"left\"> Post-fiber, mg/dL</td><td align=\"center\">154.4 ± 39.2</td><td align=\"center\">152.3 ± 23.8</td></tr><tr><td colspan=\"3\"><hr/></td></tr><tr><td align=\"left\">Apolipoprotein B</td><td/><td/></tr><tr><td align=\"left\"> Baseline, mg/dL</td><td align=\"center\">122.3 ± 34.7</td><td align=\"center\">110.6 ± 19.7</td></tr><tr><td align=\"left\"> Post-fiber, mg/dL</td><td align=\"center\">117.1 ± 31.5</td><td align=\"center\">106.3 ± 16.5</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>* = Procedures: Mixed dry and wet ingredients well. Weighed the dough and divided by 50. Baked at 330°F for 11–14 minutes until cookies became light golden brown. # = Made 50 cookies. Each cookie contained approximately 5 g of psyllium fiber.</p></table-wrap-foot>", "<table-wrap-foot><p>* = Mean ± standard deviation. ^#= Significant difference from pre-menopausal women (p &lt; 0.05).</p><p>^@= No significant difference from pre-menopausal women.</p></table-wrap-foot>", "<table-wrap-foot><p>* = Mean ± standard deviation. # = Significantly different from pre-menopausal women, paired t-test (p &lt; 0.05).</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Dissecting the specific effect of a drug on a defined biological process is often complicated by the plethora of secondary effects that arise during extended exposure. This is highlighted by the anti-cancer drug rapamycin whose principle, although not exclusive, target is the rate limiting initiation step of protein translation [##REF##18215105##1##]. Techniques designed to analyse the effect of the drug on mRNA-ribosome association are hampered by the long exposure times required to observe changes in the polysomal mRNA populations. Translation initiation is frequently regulated via the <italic>e</italic>ukaryotic <italic>i</italic>nitiation <italic>f</italic>actor 4E (eIF4E). It can be sequestered into an inactive complex by a family of 4E-binding proteins (4E-BP1/2/3). The affinity of these proteins for eIF4E is modulated by phosphorylation via the mTORC1 (mammalian target of rapamycin complex 1) kinase [##REF##10364159##2##]. Rapamycin arrests many cells in G₁[##REF##11057898##3##], and is a potent immunosuppressant [##REF##8717522##4##]. It exerts its action by binding and inactivating the mTORC1 [##REF##14536067##5##]. Previous studies have shown that extended exposure to rapamycin also alters transcription. Among the mRNAs regulated, a number impact directly on translation (e.g. eIF2α) [##REF##12101249##6##]. These changes were minor after short drug exposure times (&lt; 60 mins), but increased markedly after 2 hrs. Translational profiling studies (which examine the mRNAs associated with ribosomes) have also been reported [##REF##14576155##7##, ####REF##11943782##8##, ##REF##17311107##9##, ##REF##17562867##10####17562867##10##]. One study, performed on the Jurkat T cell clone E6-1, revealed that after an extended exposure to rapamycin (minimum 4 hrs) almost all transcripts analysed were inhibited, and 136 of those (representing ~5% of the total) were strongly inhibited (at least 10 fold). This latter group included a number of mRNAs whose products impact directly on translation (e.g. eIF5A, eIF4A1, eEF1, eEFTu) [##REF##11943782##8##]. This would predict that lengthy exposure to the drug will influence translation by modifying the levels of initiation and elongation factors. We therefore sought to develop a technique that would permit a specific analysis on how the drug alters transcript recruitment onto free ribosomes under conditions that eliminated the secondary effects associated with extended exposure. The approach has been coupled to a high-throughput microarray screen to examine how rapamycin exposure impacts on the re-seeding of the polysomal transcript populations. Results from the array have also been validated by quantitative RT-PCR.</p>" ]

[ "<title>Methods</title>", "<title>Cell culture</title>", "<p>MRC-5 cells (Coriell Cell Repository) were cultured in Minimal Essential Medium (Gibco) supplemented with 1 mM sodium pyruvate (Sigma), 0.1 mM non-essential amino acids, 10% foetal calf serum (Brunschwig), 1% penicillin/streptomycin, in a humidified atmosphere containing 5% CO₂. For polysome analysis, cells in the growing phase (60% confluence) were hypertonically shocked by shifting to medium containing 300 mM NaCl for 50 min. They were then placed in normal isotonic medium for 30 min. When rapamycin was used, 100 nM rapamycin (LC laboratories) or 0.01% DMSO (the negative control) was added during the hypertonic shock, 20 min before the transfer back to isotonic conditions. Rapamycin and DMSO were kept on the cells throughout the 30 mins recovery period (total time of exposure to rapamycin was 50 mins). These conditions were based upon previously published work [##REF##15292274##11##], although we have independently confirmed that they can be used on a range of cell lines including 293T [##REF##17591614##12##], HeLa S3 and SK-NA5 (data not shown).</p>", "<title>Polysome gradient/RNA extraction</title>", "<p>After treatment, cells were scraped into the culture medium and pelleted for 4 min at 100 g. The pellets, consisting of 5 × 10⁶cells, were lysed for 15 min on ice in 400 μL of 100 mM KCl, 50 mM Tris-Cl pH 7.4, 1.5 mM MgCl₂, 1 mM DTT, 1 mg/mL heparin, 1.5 % NP40, 100 μM cycloheximide, 1% aprotinin, 1 mM AEBSF and 100 U/mL of RNasin. Nuclei were pelleted by centrifugation in a microfuge, 10 min at 12000 rpm. The supernatant was loaded onto a 20–60% sucrose gradient (in 100 mM KCL, 5 mM MgCl₂, 20 mM HEPES pH 7.4 and 2 mM DTT). Extracts were fractionated for 3 h 30 min at 35,000 rpm at 4°C in a Beckman SW41 rotor, and the gradients were recovered in 3 fractions [monosome, light polysome (2 to 5 ribosomes) and heavy polysome (&gt; 5 ribosomes)] using a Brandel gradient fractionator equipped with an ISCO UA-6 flow cell set to 254 nm. RNA was isolated from the light and heavy polysome fractions by adding an equal volume of TriZol (Invitrogen). Samples were mixed and incubate for 15 min. on ice, then 0.3 volumes of chloroform was added. After centrifugation, the upper phase was collected and the RNA precipitated with 0.7 volumes of isopropanol. The pellet of RNA was re-suspended in water. Prior to microarray analysis the pooled RNA fractions were further purified using the Qiagen RNeasy kit. The total yield of RNA in each pooled fraction was ~2 μg. RNA quality was checked on an Agilent 2100 bioanalyser.</p>", "<title>Microarray</title>", "<p>Total RNA (100 ng from each fraction) was first amplified using the two step amplification protocol of Affymetrix. cRNA (17.5 μg) was then used to probe the Gene Chip U133 Plus 2.0 with 54,675 probe sets, covering more than 47,000 transcripts. Three biological replicates were hybridised for each condition (light and heavy polysomes +/- rapamycin). Data were analysed using the GCOS normalisation of Affymetrix. After normalisation, we filtered out probe sets that were assayed \"absent\" in all the 12 arrays. For this, the expression levels in each control experiment (DMSO) were arbitrarily fixed to one, and the fold change of the corresponding probe in the treated samples was normalised to this. The variation was then tested using the Mann and Witney U statistical test. Using this approach 24,105 of the probe sets (44%) were flagged as absent. The data was further analysed using two approaches. Firstly, probe set intensity values below 100 were removed prior to GCOS normalisation, and fold changes ≥ 1.5 relative to the DMSO control were scored (the smallest score in the three independent experiments had to show at least a 20% change relative to the DMSO control i.e. a 1.2 fold increase or a 0.8 fold decrease). In a second approach, commencing with the entire data set, points were only scored if the mean of the fold difference was ≥ 2.5 (once again the smallest score in the three independent experiments had to show at least a 20% change relative to the DMSO control). The data from both screens were plotted onto biological networks using the GO onthology (Affymetrix) and Ingenuity Pathway Analysis Software packages <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.Ingenuity.com\"/>.</p>", "<p>The microarray data are available at ArrayExpress (Accession N° E-TABM-205).</p>", "<title>Real-Time PCR</title>", "<p>One μg of total RNA from a fourth independent experiment was reverse transcribed using random hexamers (Gibco). A 1/10 dilution of the cDNA was used to perform the PCR with the SYBR Green Reagent (Roche).</p>", "<p>Primers used were:</p>", "<p><bold>QKI </bold>AGCATCACAGTCAGAGGTCAGC, GCAGTGGCATATTAAACCAAAGC;</p>", "<p><bold>RBM7 </bold>GTTGGAAATTCAAGCCCTACCT, AATCCTGATTGATCCAGAGGTG;</p>", "<p><bold>ORMDL1 </bold>GTCTGGCAGAAACAACGTCTC, CAATGTGGTTGCTGTTCTGG;</p>", "<p><bold>FAS </bold>GATGGCGAATGAGGTTCAG, CAATCCCATATCTCCCATTAAC;</p>", "<p><bold>RBM17 </bold>GTCATCTCCGGTGATCCTTAAA, CAACCAGAGAGGCACACAGAT;</p>", "<p><bold>PAPPA </bold>GCATCAGTTTCTCTAGCTGCAA, TATCAAACAAGCACTCCCTGTC;</p>", "<p><bold>Actin </bold>CTGACGGCCAGGTCATCACCATTG, GCCGGACTCGTCATACTCCTGCTTG;</p>", "<p><bold>L27 </bold>GTGACAGCTGCCATGGGCAAG, TCAAACTTGACCTTGGCCTCCCG,</p>", "<p><bold>Cyclin D1 </bold>AAGCAGGACTTTGAGGCA AG, CCTCTGAGGTCCCTACTTTCAA.</p>", "<p>Primer sets were designed to amplify regions within the 3' UTR of each transcript since this generally corresponded to the site of the probe sets used on the Affymetrix chip. The specificity of each primer set was confirmed by standard RT-PCR on total cell RNA, followed by analysis of the DNA products by agarose gel electrophoresis (data not shown).</p>", "<title>Western blot analysis</title>", "<p>Cells were lysed in CSH buffer (50 mM Tris-Cl pH 7.5, 250 mM NaCl, 1 mM EDTA, 0.1% Triton X-100) and the nuclei were pelleted by centrifugation at 12,000 rpm for 5 mins. Twenty μgs of protein was resolved on a 15% polyacrylamide-SDS gel and electrotransfered to a PVDF membrane. Antibodies used in this study were the anti-4EBP1 (Cell Signalling), the anti-phosho4EBPI (Thr37/46) (Cell Signalling), the anti-p70 S6 kinase (Cell Signalling), the anti-phospho-p70 S6 kinase (Thr389) (Cell Signalling) and mouse anti-actin (Chemicon). Blots were developed using the Super Signal Substrate (Thermo Scientific).</p>" ]

[ "<title>Results</title>", "<title>Rapamycin delays mRNA recruitment onto polysomes</title>", "<p>To directly examine the effect of rapamycin on mRNA recruitment we decided to exploit a novel approach, an approach that analyses the ability of cellular mRNAs to compete for free ribosomes in-vivo. Hypertonic shock provokes a rapid inhibition of protein synthesis, disaggregation of polysomes (Figure ##FIG##0##1A##), dephosphorylation of eIF4E, 4E-BP1, S6 and an increased association of eIF4E and 4E-BP1 [##REF##15292274##11##,##REF##12138083##13##]. Upon restoration of isotonic conditions the polysomal fraction is rapidly reconstituted (Figure ##FIG##0##1B## and [##REF##15292274##11##]). Using this methodology it was possible to examine what effect rapamycin had on the recruitment of mRNA populations onto free ribosomes following very short drug exposure times. It was in substance an in-vivo competition assay performed under two defined physiological conditions. Drug treatment appeared to delay recruitment as evidenced by the reduction in the heavy polysome peak (≥ 6 ribosomes: compare Figure ##FIG##0##1B## and ##FIG##0##1C##), and this effect was correlated with a modification of the downstream signalling targets of mTOR, including 4E-BP1 and S6 kinase. (Figure ##FIG##0##1D##, compare the second and third lanes). This confirmed that despite the relatively short time of exposure to rapamycin, the recruitment assay monitored transcript:ribosome re-association under two conditions in which eIF4E availability was altered.</p>", "<title>A profiling screen identifies changes in the light and heavy polysomal mRNA populations</title>", "<p>Equal amounts of RNA isolated from the light (2 to 5 ribosomes) and heavy (&gt; 5 ribosomes) polysomal fractions were used to probe the Affymetrix Gene Chip U133 Plus 2. Triplicate independent gradients under each experimental condition were examined. After data analysis, two subpopulations of transcripts were clearly discriminated: those dominant in the light polysomal fraction and those dominant in the heavy polysomal fraction, an important criterion since it validated the initial experimental approach (Figure ##FIG##1##2A##). Despite the fact that the polysomal peaks were smaller in the presence of rapamycin, consistent with a global repression, the two sub-populations were essentially conserved (i.e. no major movement of mRNA populations between the two fractions as a consequence of the treatment was evident). To analyse the data we used two approaches. Firstly, we filtered out all transcripts giving low probe set intensity values on the chip using the default settings of the Agilent analysis software package (values &lt; 100). We then scored for mRNAs whose polysomal occupancy was altered by greater than ×1.5 fold (listed in Additional File ##SUPPL##0##1##). This produced 437 transcripts within which was found the majority of the repressed TOP mRNAs (see below). Curiously, within this group of transcripts almost equal proportions were up and down-regulated (46% and 54%, respectively) (Figure ##FIG##1##2B##).</p>", "<p>As a second approach to analyse the data we applied a ×2.5 fold change cut-off point to the entire data set. The rationale for this alternative analysis is based upon the fact that many of the genes that are regulated at the level of translation are frequently transcribed at low levels. This includes proto-oncogenes and other factors that regulate cell growth [##REF##10216945##14##]. The majority of these transcripts are found within the lower intensity range and we therefore tested if meaningful information could be extracted from this region by applying a more stringent selection. We observed that 1160 mRNAs (3.8%) showed increased or decreased polysomal distribution in the presence of rapamycin, suggesting that in this small fraction of transcripts the affinity for the cap binding complex was changed (Figure ##FIG##1##2C## and Additional File ##SUPPL##1##2##). Over 2/3 of the mRNAs responding to the drug were down-regulated, whereas 1/3 showed increased polysomal occupancy relevant to the non-treated control. Only a few transcripts were regulated in both the heavy and the light polysomes (55 mRNAs) (Figure ##FIG##1##2C##). These results are not unlike those reported in a translational profiling study performed on the two tumoural cell lines, LAPC-4 (prostrate cancer) and U87 (glioblastoma). Applying the 2.5 fold cut-off point, ca. 6% of the 3,000 transcripts screened showed altered polysomal occupancy, and amongst these, 60% were down-regulated and 40% up-regulated [##REF##14576155##7##].</p>", "<p>Those mRNAs showing significant redistributions (both increased and decreased: as listed in Additional Files ##SUPPL##0##1## and ##SUPPL##1##2##) were then plotted onto cellular networks using the GO onthology (Affymetrix) software package. Results revealed that transcripts up- or down-regulated affected more or less the same biological processes (Figure ##FIG##1##2D## and ##FIG##1##2E##). However, using the Ingenuity pathway analysis a number of features were immediately evident in the rapamycin treated cells: (a) A group of mRNAs involved in the inflammatory response were regulated, consistent with the immunosuppressive activity of rapamycin. Interestingly, we also observed down-regulation of several transcripts linked to phagocytosis. It has recently been reported that rapamycin down-regulates phagocytosis in a murine macrophage cell line (Table ##TAB##0##1##) [##REF##15485887##15##]. (b) We observed a number of mRNAs involved in cell growth and proliferation. Among them, a number of anti-apoptotic mRNAs were up-regulated (e.g. relA and mdm2), whereas pro-apoptotic ones were down-regulated (e.g. fas, faslg and faf1) (Table ##TAB##0##1##). This is consistent with the observation that rapamycin did not induce apoptosis in MRC-5 cells even after extended exposure (unpublished observations), and corroborated recent studies showing the anti-apoptotic properties of rapamycin [##REF##16497721##16##,##REF##14871980##17##].</p>", "<p>Rapamycin is known to have a marked effect on the expression of a subset of transcripts referred to as terminal oligopyrimidine (TOP) mRNAs [##REF##16246169##18##]. This includes ribosomal protein mRNAs (estimated to be ~15% of the total cellular mRNA) and translation elongation factors. In previous translational profiling studies, a number of TOP mRNAs were clearly repressed. We also observed repression of ribosomal transcripts, although the effects were less extensive than in the earlier reports. This may reflect both the shorter drug exposure times and the experimental approach that was employed (i.e. mRNA re-recruitment onto free ribosomes). With the cut-off threshold at ×2.5 fold we observed down regulation of rpl14 and rpl21 in the light polysomes and rplp0 in the heavy polysomes (<italic>r</italic>ibosomal <italic>p</italic>rotein <italic>l</italic>arge). However, if the threshold was reduced to ×1.5 fold the number of hits significantly increased. Most significantly, ribosomal transcripts were only ever observed down-regulated (Table ##TAB##1##2## and Additional File ##SUPPL##0##1##).</p>", "<title>Independent confirmation of the array results by RT-PCR</title>", "<p>With the aim of validating the microarray we performed real-time RT-PCR on selected transcripts across the spectrum of probe set intensities (Figure ##FIG##2##3A##). We were particularly interested in those transcripts that gave low probe set values. For the RT-PCR, RNA was extracted from a fourth independent experiment. Nine mRNAs were initially selected (Transcripts up-regulated = qki, rbm7, ormdl1. Transcripts down-regulated = fas, rbm17, pappa, Transcripts not regulated = β-actin, L27, cyclin D1: see Additional Files ##SUPPL##0##1## and ##SUPPL##1##2##). Transcripts from the lower values of the data set (see Additional File ##SUPPL##1##2## and Figure ##FIG##2##3A##) were selected because of the large fold difference between the rapamycin and DMSO control (&gt; 2.5 fold). The results, with the exception of those obtained with the pappa transcript, largely confirmed the micro-array data (Figure ##FIG##2##3B##). This was particularly encouraging for transcripts such as those coding for QKI and FAS, indicating that application of the ×2.5 fold cut-off approach permitted the extraction of useful hits even in the lower end of the probe intensity set (Figure ##FIG##2##3A##). Note that the absence of a light RT-PCR value for RBM17 simply reflected its low levels in this fraction (Ct &gt; 35 cycles). The microarray study also indicated that this transcript was regulated only within the heavy polysomal fraction, a result confirmed by the RT-PCR. In addition, these studies also demonstrated that no significant changes in total mRNA levels had occurred within the selected transcripts as a consequence of drug exposure (Figure ##FIG##2##3B##).</p>" ]

[ "<title>Discussion</title>", "<p>In this technical report, we have outlined a novel approach to translational profiling that follows the impact of a drug on the de-novo re-association of mRNAs and ribosomes in living cells in culture. An intracellular pool of free ribosomes was generated by a short hypertonic shock. Although this undoubtedly induced a stress response, previous work has demonstrated that the translation initiation machinery recovers very rapidly after the cells are transferred back to isotonic conditions as monitored by the reconstitution of the polysomal fraction [##REF##15292274##11##,##REF##12138083##13##]. The rapidity at which the polysomes are reformed has permitted us to examine the effect of a drug on this process using a short time window of exposure. This has the effect of limiting undesirable secondary effects that arise upon extended exposure, an effect highlighted by the drug selected for this study, namely rapamycin (see Introduction). The short exposure time warranted a drug concentration higher than that generally used in animal studies [##REF##8900311##19##], however, some studies in cell culture systems have employed rapamycin concentrations as high as 15–20 μM before observing an impact on cell growth [##REF##18026138##20##]. The selection was also dictated too by the fact that rapamycins' effect on translation initiation has been extensively studied. We have demonstrated that this approach can be coupled to a high-throughput analysis of the polysomal transcript populations.</p>", "<p>Treatment with rapamycin limits the availability of eIF4E via its sequestration into an inactive complex with the hypophosphorylated 4E-BPs. Such a scenario represses global translation rates but with an effect more marked on those mRNAs containing structured 5' UTRs and those containing TOP elements. Indeed, when TOP containing transcripts were identified in our array they were always down-regulated (Table ##TAB##1##2##). Additionally, although the polysome profiles demonstrated an overall translational repression, the position of the vast majority of mRNAs on the gradient (i.e. the transcript populations in both the heavy and light polysomes) was largely unperturbed by the drug (i.e. there was little movement of transcripts from heavy to light polysomes, a somewhat unanticipated response), indicating that the affinity of re-recruitment of ribosomes onto these mRNAs (as reflected by the number of ribosomes per transcript) was largely unchanged.</p>", "<p>A number of translational profiling studies examining the effect of rapamycin in mammalian cells have already been reported. In one of these, the effect of rapamycin on the polysomal distribution of mRNAs was demonstrated to be coupled to the activity of AKT [##REF##14576155##7##], a result that demonstrates the extent to which a drugs effect can be modulated by the physiological status of the cell. This interpretation has become even more convoluted following the observation that prolonged rapamycin treatment may inhibit AKT signalling by interfering with the assembly of the second mTOR complex, mTORC2 [##REF##15718470##21##]. However, with regards to the two transcripts characterised in this work as strongly up-regulated, namely cyclin D1 and c-myc, the former was down-regulated in our screen (×1.5 fold), and the latter gave values that were not considered statistically significant. Other studies have also failed to observe changes in the polysomal occupancy of the cyclin D1 mRNA in the presence of rapamycin [##REF##17311107##9##,##REF##17562867##10##,##REF##16936314##22##]. These differences may reflect the cell lines and/or the experimental procedures employed. Both cyclin D1 and c-myc were proposed to carry IRESes within the 5' UTR, and IRES activity has been reported to show cell type specificity linked to the availability of ITAFs (<italic>I</italic>RES <italic>T</italic>rans-<italic>A</italic>cting <italic>F</italic>actors) [##REF##10950867##23##]. Furthermore, the responsiveness of these IRESes to rapamycin was shown to be tightly coupled to the cellular activity of the AKT and RAF/MEK/ERK signalling cascades, features that may also show cell-type variation [##REF##15634685##24##]. However, the slight reduction that we observed in the polysomal levels of the cyclin D1 mRNA would be consistent with other reports indicating that its expression was sensitive to the levels of eIF4E [##REF##7673150##25##,##REF##8577715##26##]. Finally, the earlier profiling study followed changes in the steady-state polysomal populations after extended exposure to the drug, whilst in the current work we have followed a competitive re-association. These processes may have altered initiation factor requirements, which could impact on the mRNA populations that respond. Indeed, it has been proposed that eIF4GII but not eIF4GI is required for re-initiation subsequent to a hypertonic shock [##REF##16959778##27##]. Nonetheless, a listing of transcripts detected in both studies demonstrated that the majority behaved similarly (see Additional File ##SUPPL##2##3##).</p>" ]

[ "<title>Conclusion</title>", "<p>In summary, a major effort is underway to use high density microarray profiles to study how different drug regimes impact on the polysomal mRNA populations. These studies provide insights into how cellular gene expression is regulated at the level of translation initiation, the rate limiting step in protein expression. Changes in this read-out are a very rapid cellular response to physiological perturbations. The method that we have outlined permits a specific analysis of how a drug impacts on transcript-ribosome association. This early response almost certainly conditions subsequent cell behaviour during extended exposure. The choice of rapamycin for this \"proof-of-principle\" work was not arbitrary since the impact of this drug on translation initiation has been extensively studied. The technique offers the possibility of establishing molecular fingerprints for different tumour derived cell types and drug regimes [##REF##17057710##28##]. In addition, it provides a very powerful technique to analyse the early events in translational control at the level of mRNA:ribosome association.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Recent work, using both cell culture model systems and tumour derived cell lines, suggests that the differential recruitment into polysomes of mRNA populations may be sufficient to initiate and maintain tumour formation. Consequently, a major effort is underway to use high density microarray profiles to establish molecular fingerprints for cells exposed to defined drug regimes. The aim of these pharmacogenomic approaches is to provide new information on how drugs can impact on the translational read-out within a defined cellular background.</p>", "<title>Methods</title>", "<p>We describe an approach that permits the analysis of de-novo mRNA-ribosome association in-vivo during short drug exposures. It combines hypertonic shock, polysome fractionation and high-throughput analysis to provide a molecular phenotype of translationally responsive transcripts. Compared to previous translational profiling studies, the procedure offers increased specificity due to the elimination of the drugs secondary effects (e.g. on the transcriptional read-out). For this pilot \"proof-of-principle\" assay we selected the drug rapamycin because of its extensively studied impact on translation initiation.</p>", "<title>Results</title>", "<p>High throughput analysis on both the light and heavy polysomal fractions has identified mRNAs whose re-recruitment onto free ribosomes responded to short exposure to the drug rapamycin. The results of the microarray have been confirmed using real-time RT-PCR. The selective down-regulation of TOP transcripts is also consistent with previous translational profiling studies using this drug.</p>", "<title>Conclusion</title>", "<p>The technical advance outlined in this manuscript offers the possibility of new insights into mRNA features that impact on translation initiation and provides a molecular fingerprint for transcript-ribosome association in any cell type and in the presence of a range of drugs of interest. Such molecular phenotypes defined pre-clinically may ultimately impact on the evaluation of a particular drug in a living cell.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>RG and PJ-G prepared the polysomal RNA for the microarray analysis. RG and TA were involved in the analysis of the microarray data. RG and LM performed the RT-PCR control. JC prepared the manuscript. All authors have read and approved the manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1755-8794/1/33/prepub\"/></p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>The work was supported by grants from the \"La Ligue Genevoise Contre le Cancer\" and the \"Fonds National Suisse\". We thank Dr. Jean-Dominique Vassalli and Dr. Beatrice Conne for use of the gradient apparatus, and the input of Dr. Bernadino Conrad. Finally, we thank Dr Patrick Descombe, Dr Olivier Schaad and Didier Chollet from the Genomics Platform of the NCCR program \"Frontiers in Genetics\" <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.frontiers-in-genetics.org/genomics.htm\"/> for their invaluable help in performing and analyzing the microarray experiments.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Ribosomal re-recruitment in the presence of rapamycin</bold>. (A). High salt provokes a rapid disaggregation of polysomes. (B). Upon restoration of isotonic conditions the polysomal fraction is reconstituted (C). Pre-treatment with rapamycin delays the re-recruitment of ribosomes. The position of the ribonucleoprotein (RNP), the monosomal (Mono) and polysomal (Poly) fractions are indicated. (D) Western blot analysis of phospho-4EBP-1, 4EBP-1, phospho-S6K, S6K, and actin was performed on extracts isolated under the different conditions depicted in panels (A); (B) and (C).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Microarray analysis</bold>. (A). Hierarchical clustering of relative expression. Each column represents the different conditions. On the right the vertical bar indicates probe set intensity values (indicated as Expression) in arbitrary units. The zero value indicates absence, with blue indicating a low level and red a high level of expression (the maximum value being fixed as 5). The vertical brackets on the right indicate that transcripts have been grouped into those over-represented in the light polysomes (lower) or in the heavy polysomes (upper). (B). After removal of probe set intensity values &lt; 100, and the application of a ×1.5 fold cut-off, the regulated genes were classified into four groups. The values indicate the number of transcripts in each group. (C). In a second approach, a ×2.5 fold cut-off was applied. (D) and (E). Functional classification using Gene Ontology of the genes either up-regulated or down-regulated (as depicted in panel b and in Additional File ##SUPPL##0##1##, and panel c and in Additional File ##SUPPL##1##2##, respectively). The unknown fraction represents genes not annotated in the Gene Ontology database.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Real-Time RT-PCR analysis</bold>. (A). A schematic representation of the intensity values of transcripts selected for the RT-PCR validation. In the upper panel, the horizontal bars represented the value range for each mRNA on the array. The lower panel plots the distribution of regulated genes (×2.5 fold cut-off selection) relative to the probe set values. (B). RT-PCR values were normalised to those obtained from two housekeeping genes and the fold change indicates the difference in the DMSO and rapamycin values after normalisation. The DMSO value was arbitrarily set at 1. The results are compared with those obtained from the microarray. The variation in the total mRNA extracted is also represented. The 2.5 fold difference used for the screening is represented by the dotted lines (2.5 and 0.4). Each bar is representative of 2 independent RT-PCR assays performed in triplicate. Bars indicate the SEM.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>List of transcripts regulated by rapamycin (Ingenuity classification)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\" colspan=\"7\"><bold>Inflammation</bold></td></tr></thead><tbody><tr><td align=\"center\" colspan=\"2\"><bold>Upregulated</bold></td><td align=\"center\" colspan=\"5\"><bold>Downregulated</bold></td></tr><tr><td colspan=\"2\"><hr/></td><td colspan=\"5\"><hr/></td></tr><tr><td align=\"center\">apobec3f</td><td align=\"center\">lst1</td><td align=\"center\">adra1a</td><td align=\"center\">epha4</td><td align=\"center\">il1rn</td><td align=\"center\">ptpn22</td><td align=\"center\">tlr7</td></tr><tr><td align=\"center\">blnk</td><td align=\"center\">mdm2</td><td align=\"center\">adra2c</td><td align=\"center\">f9</td><td align=\"center\">irf2</td><td align=\"center\">ptprc</td><td align=\"center\">zeb1</td></tr><tr><td align=\"center\">cald1</td><td align=\"center\">oas1</td><td align=\"center\">alpp</td><td align=\"center\">fas</td><td align=\"center\">itgb3</td><td align=\"center\">ptprz1</td><td/></tr><tr><td align=\"center\">cmklr1</td><td align=\"center\">parvg</td><td align=\"center\">bcl11a</td><td align=\"center\">faslg</td><td align=\"center\">klf2</td><td align=\"center\">rbl2</td><td/></tr><tr><td align=\"center\">ctla4</td><td align=\"center\">pdgfc</td><td align=\"center\">bcl2l1</td><td align=\"center\">fcgr2a</td><td align=\"center\">lama3</td><td align=\"center\">rbm15</td><td/></tr><tr><td align=\"center\">cxcr4</td><td align=\"center\">rag2</td><td align=\"center\">cast</td><td align=\"center\">folr1</td><td align=\"center\">lilra2</td><td align=\"center\">rbpj</td><td/></tr><tr><td align=\"center\">cyp3a4</td><td align=\"center\">rela</td><td align=\"center\">cd28</td><td align=\"center\">fyb</td><td align=\"center\">mll</td><td align=\"center\">rel</td><td/></tr><tr><td align=\"center\">fyn</td><td align=\"center\">sat1</td><td align=\"center\">cd36</td><td align=\"center\">gal</td><td align=\"center\">mpo</td><td align=\"center\">satb1</td><td/></tr><tr><td align=\"center\">hla-g</td><td align=\"center\">tcf12</td><td align=\"center\">clec1b</td><td align=\"center\">gap43</td><td align=\"center\">pcgf2</td><td align=\"center\">siglec8</td><td/></tr><tr><td align=\"center\">ifih1</td><td align=\"center\">tfap4</td><td align=\"center\">cul4a</td><td align=\"center\">gnrh1</td><td align=\"center\">pla2g6</td><td align=\"center\">smpd1</td><td/></tr><tr><td align=\"center\">ifna2</td><td align=\"center\">thbs1</td><td align=\"center\">cxcl11</td><td align=\"center\">gnrhr</td><td align=\"center\">plcg1</td><td align=\"center\">spn</td><td/></tr><tr><td align=\"center\">il16</td><td align=\"center\">tlr7</td><td align=\"center\">cyp3a4</td><td align=\"center\">hal-dqb1</td><td align=\"center\">prdm16</td><td align=\"center\">syk</td><td/></tr><tr><td align=\"center\">il28b</td><td align=\"center\">unc119</td><td align=\"center\">ddl1</td><td align=\"center\">hck</td><td align=\"center\">pscdbp</td><td align=\"center\">thra</td><td/></tr><tr><td align=\"center\">itga4</td><td align=\"center\">vtcn1</td><td align=\"center\">dok2</td><td align=\"center\">ifne1</td><td align=\"center\">ptgs2</td><td align=\"center\">thrb</td><td/></tr><tr><td colspan=\"7\"/></tr><tr><td align=\"left\" colspan=\"7\"><bold>Cell growth and proliferation</bold></td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"center\" colspan=\"2\"><bold>Upregulated</bold></td><td align=\"center\" colspan=\"5\"><bold>Downregulated</bold></td></tr><tr><td colspan=\"2\"><hr/></td><td colspan=\"5\"><hr/></td></tr><tr><td align=\"center\">adam12</td><td align=\"center\">mdm2</td><td align=\"center\">adra1a</td><td align=\"center\">dok2</td><td align=\"center\">gnrh1</td><td align=\"center\">olig2</td><td align=\"center\">rffl</td></tr><tr><td align=\"center\">blnk</td><td align=\"center\">pappa</td><td align=\"center\">bcl2l1</td><td align=\"center\">f12</td><td align=\"center\">hck</td><td align=\"center\">p53aip1</td><td align=\"center\">sec14l2</td></tr><tr><td align=\"center\">cdc2l5</td><td align=\"center\">rag2</td><td align=\"center\">ccl27</td><td align=\"center\">fas</td><td align=\"center\">hmga2</td><td align=\"center\">pcgf2</td><td align=\"center\">syk</td></tr><tr><td align=\"center\">ctla4</td><td align=\"center\">rela</td><td align=\"center\">cd28</td><td align=\"center\">faslg</td><td align=\"center\">irf2</td><td align=\"center\">pdgfa</td><td align=\"center\">tfr2</td></tr><tr><td align=\"center\">cxcr4</td><td align=\"center\">s100b</td><td align=\"center\">cdca7</td><td align=\"center\">fbxo2</td><td align=\"center\">itgb3</td><td align=\"center\">piwil1</td><td align=\"center\">tgif1</td></tr><tr><td align=\"center\">erbb3</td><td align=\"center\">ss18</td><td align=\"center\">clca2</td><td align=\"center\">fcer2</td><td align=\"center\">klf2</td><td align=\"center\">pla2g6</td><td align=\"center\">thra</td></tr><tr><td align=\"center\">fyn</td><td align=\"center\">tcf12</td><td align=\"center\">cltc</td><td align=\"center\">fgf18</td><td align=\"center\">lzts1</td><td align=\"center\">ptgs2</td><td align=\"center\">thrb</td></tr><tr><td align=\"center\">hla-g</td><td align=\"center\">thbs1</td><td align=\"center\">csh2</td><td align=\"center\">folr1</td><td align=\"center\">mdm4</td><td align=\"center\">ptk2</td><td align=\"center\">tnfsf15</td></tr><tr><td align=\"center\">igf1r</td><td align=\"center\">unc119</td><td align=\"center\">cyp2c9</td><td align=\"center\">foxo1</td><td align=\"center\">mll</td><td align=\"center\">ptpn22</td><td align=\"center\">zfn10</td></tr><tr><td align=\"center\">il28b</td><td align=\"center\">vtcn1</td><td align=\"center\">dcc</td><td align=\"center\">gal</td><td align=\"center\">nab2</td><td align=\"center\">ptpra</td><td/></tr><tr><td align=\"center\">itga4</td><td/><td align=\"center\">ddx17</td><td align=\"center\">gap43</td><td align=\"center\">nos1</td><td align=\"center\">ptprc</td><td/></tr><tr><td align=\"center\">lst1</td><td/><td align=\"center\">dll1</td><td align=\"center\">glmn</td><td align=\"center\">nov</td><td align=\"center\">rbl2</td><td/></tr><tr><td colspan=\"7\"/></tr><tr><td align=\"left\" colspan=\"7\"><bold>Cell death</bold></td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"center\" colspan=\"2\"><bold>Upregulated</bold></td><td align=\"center\" colspan=\"5\"><bold>Downregulated</bold></td></tr><tr><td colspan=\"2\"><hr/></td><td colspan=\"5\"><hr/></td></tr><tr><td align=\"center\">blnk</td><td align=\"center\">itga2</td><td align=\"center\">abcd2</td><td align=\"center\">dll1</td><td align=\"center\">hmga2</td><td align=\"center\">p53aip1</td><td align=\"center\">rbm17</td></tr><tr><td align=\"center\">cdk6</td><td align=\"center\">mdm2</td><td align=\"center\">acvr1b</td><td align=\"center\">eraf</td><td align=\"center\">ifne1</td><td align=\"center\">pigt</td><td align=\"center\">rel</td></tr><tr><td align=\"center\">ctla4</td><td align=\"center\">rag2</td><td align=\"center\">adora2a</td><td align=\"center\">faf1</td><td align=\"center\">il1rn</td><td align=\"center\">piwil1</td><td align=\"center\">satb1</td></tr><tr><td align=\"center\">cul3</td><td align=\"center\">rela</td><td align=\"center\">atrx</td><td align=\"center\">fas</td><td align=\"center\">irf2</td><td align=\"center\">pla2g6</td><td align=\"center\">serpinb4</td></tr><tr><td align=\"center\">cxcr4</td><td align=\"center\">rnase1</td><td align=\"center\">atxn3</td><td align=\"center\">faslg</td><td align=\"center\">itgb3</td><td align=\"center\">pou4f1</td><td align=\"center\">siglec8</td></tr><tr><td align=\"center\">cyp2e1</td><td align=\"center\">sgpp1</td><td align=\"center\">bcl2l1</td><td align=\"center\">fcer2</td><td align=\"center\">klf2</td><td align=\"center\">ppp1r9b</td><td align=\"center\">smpd1</td></tr><tr><td align=\"center\">cyp3a4</td><td align=\"center\">tfap4</td><td align=\"center\">bircabp</td><td align=\"center\">foxo1</td><td align=\"center\">klra1</td><td align=\"center\">ppp2r1b</td><td align=\"center\">spn</td></tr><tr><td align=\"center\">erbb3</td><td align=\"center\">thbs1</td><td align=\"center\">cast</td><td align=\"center\">gal</td><td align=\"center\">lrp5</td><td align=\"center\">prdm2</td><td align=\"center\">stk4</td></tr><tr><td align=\"center\">erg</td><td align=\"center\">traf4</td><td align=\"center\">ccl27</td><td align=\"center\">gimap5</td><td align=\"center\">mdm4</td><td align=\"center\">ptgs2</td><td align=\"center\">syk</td></tr><tr><td align=\"center\">fyn</td><td align=\"center\">zmym2</td><td align=\"center\">cd28</td><td align=\"center\">gng2</td><td align=\"center\">mll</td><td align=\"center\">ptk2</td><td align=\"center\">thra</td></tr><tr><td align=\"center\">hla-g</td><td/><td align=\"center\">cdk6</td><td align=\"center\">gria2</td><td align=\"center\">mpo</td><td align=\"center\">ptpn22</td><td align=\"center\">tnfsf15</td></tr><tr><td align=\"center\">ifih1</td><td/><td align=\"center\">cyp2e1</td><td align=\"center\">grik2</td><td align=\"center\">nol3</td><td align=\"center\">ptprc</td><td align=\"center\">traf5</td></tr><tr><td align=\"center\">ifna2</td><td/><td align=\"center\">cyp3a4</td><td align=\"center\">grm1</td><td align=\"center\">nos1</td><td align=\"center\">ptprz1</td><td align=\"center\">trps1</td></tr><tr><td align=\"center\">igf1r</td><td/><td align=\"center\">dcc</td><td align=\"center\">hint1</td><td align=\"center\">nrtn</td><td align=\"center\">rbl2</td><td align=\"center\">znf10</td></tr><tr><td colspan=\"7\"/></tr><tr><td align=\"left\" colspan=\"7\"><bold>Phagocytosis</bold></td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"center\" colspan=\"2\"><bold>Upregulated</bold></td><td align=\"center\" colspan=\"2\"><bold>Downregulated</bold></td><td/><td/><td/></tr><tr><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td><td colspan=\"3\"/></tr><tr><td/><td/><td align=\"center\">cd36</td><td align=\"center\">klf2</td><td/><td/><td/></tr><tr><td/><td/><td align=\"center\">fas</td><td align=\"center\">mpo</td><td/><td/><td/></tr><tr><td/><td/><td align=\"center\">fcgr2a</td><td align=\"center\">pla2g6</td><td/><td/><td/></tr><tr><td/><td/><td align=\"center\">hck</td><td align=\"center\">syk</td><td/><td/><td/></tr><tr><td/><td/><td align=\"center\">itgb3</td><td/><td/><td/><td/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>List of TOP mRNAs detected in the array.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\"><bold>LIGHT POLYSOME UP</bold></td><td align=\"center\"><bold>LIGHT POLYSOME DOWN</bold></td><td align=\"center\"><bold>HEAVY POLYSOME DOWN</bold></td></tr></thead><tbody><tr><td align=\"center\">(NONE)</td><td align=\"center\">rpl5 (-1.5)</td><td align=\"center\">rplP0 (-2.7)</td></tr><tr><td/><td align=\"center\">rpl14 (-2.5)</td><td align=\"center\">rpl36 (-1.5)</td></tr><tr><td/><td align=\"center\">rpl21 (-5.1)</td><td align=\"center\">eef2 (-1.5)</td></tr><tr><td colspan=\"1\"><hr/></td><td colspan=\"2\"/></tr><tr><td align=\"center\"><bold>HEAVY POLYSOME UP</bold></td><td align=\"center\">rpl38 (-1.5)</td><td align=\"center\">(translation elongation factor)</td></tr><tr><td colspan=\"1\"><hr/></td><td colspan=\"2\"/></tr><tr><td align=\"center\">(NONE)</td><td align=\"center\">rps11 (-2.1)</td><td/></tr><tr><td/><td align=\"center\">rps19 (-1.5)</td><td/></tr><tr><td/><td align=\"center\">rps21 (-2.1)</td><td/></tr><tr><td/><td align=\"center\">rps28 (-1.6)</td><td/></tr><tr><td/><td align=\"center\">eftuD1 (-5.5)</td><td/></tr><tr><td/><td align=\"center\">(translation elongation factor)</td><td/></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional File 1</title><p>List of transcripts from filtered data set. Transcripts with probe set values &lt; 100 were removed from the data set. After GCOS normalisation a ×1.5 fold selection cut-off was applied to those that remained.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional File 2</title><p>Complete list of polysomal transcripts regulated by rapamycin. After GCOS normalisation a ×2.5 fold selection cut-off was applied to all the regulated transcripts independent of probe set intensity values.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S3\"><caption><title>Additional File 3</title><p>Comparison with the array of Gera and co-workers (7). Transcripts detected in both screens are listed with the fold change. Values are indicated in red when they differ between the two studies.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>The 1160 mRNAs identified in the microarray screen (as listed in Additional File ##SUPPL##1##2##) were plotted onto biological networks using the Ingenuity Pathway Analysis Software package. Those whose protein products are involved in inflammation, phagocytosis, cell growth/proliferation and cell death are listed.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1755-8794-1-33-1\"/>", "<graphic xlink:href=\"1755-8794-1-33-2\"/>", "<graphic xlink:href=\"1755-8794-1-33-3\"/>" ]

[ "<media xlink:href=\"1755-8794-1-33-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1755-8794-1-33-S2.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1755-8794-1-33-S3.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Many observational studies have demonstrated an inverse association between birth weight and blood pressure (BP), raising the intriguing possibility that lifelong hypertension might be programmed <italic>in utero </italic>[##REF##10930178##1##]. The mechanisms underlying this association remain unclear.</p>", "<p>One plausible hypothesis, based on animal data, proposes that a low protein diet during pregnancy leads to increased fetal glucocorticoid exposure, permanently programming both lower birth weight and elevated BP in offspring [##REF##8730887##2##]. Studies in humans have not found a consistent association between maternal protein intake and offspring blood pressure [##REF##15576466##3##, ####REF##11204301##4##, ##REF##11524398##5##, ##REF##11751704##6####11751704##6##], but few studies have examined whether fetal glucocorticoid exposure might program offspring BP.</p>", "<p>Experiments in rats have shown that maternal treatment during pregnancy with the synthetic glucocorticoid dexamethasone, which freely crosses the placenta, is associated with lower birth weight and raised BP in the offspring [##REF##11738795##7##]. Protection from exposure to maternal physiological glucocorticoids is mediated by a placental enzyme, 11-beta-hydroxysteroid dehydrogenase type 2 (11β-HSD2), which converts active glucocorticoid (corticosterone in rats, cortisol in humans) into inert 11-keto forms [##REF##11738795##7##]. The 'placental barrier' function of 11β-HSD2 is important because circulating maternal cortisol freely crosses the placenta.</p>", "<p>A relative deficiency of placental 11β-HSD2 might allow increased access of maternal glucocorticoids to the fetus, retarding growth and programming responses leading to later hypertension, as suggested by studies in rats and mice [##REF##11738795##7##, ####REF##8641724##8##, ##REF##10074485##9####10074485##9##]. Humans who are homozygous for the gene encoding 11β-HSD2, resulting in little or no enzyme activity, show much lower birth weight than their unaffected (mainly heterozygous) siblings [##REF##9661590##10##]. Limited data suggest that heterozygous carriers may have a milder phenotype indistinguishable from essential hypertension, though dissecting placental from offspring renal effects is problematic [##REF##12860834##11##].</p>", "<p>We hypothesized that a higher ratio of cortisol to cortisone in umbilical venous cord blood, reflecting decreased 11β-HSD2 activity, would be associated with lower birth weight and raised blood pressure in childhood. We examined this hypothesis within Project Viva, a longitudinal cohort of pregnant women and their offspring.</p>" ]

[ "<title>Methods</title>", "<title>Subjects</title>", "<p>We recruited women with a singleton pregnancy from April 1999 to July 2002 at eight obstetric offices in the Boston, Massachusetts, USA area [##REF##14760269##12##]. Women were study-eligible if they entered prenatal care before 22 weeks of gestation, planned ongoing obstetric care at an enrollment site, and were able to answer questionnaires in English. We collected sociodemographic and medical data through in-person interviews, self-administered questionnaires and hospital and ambulatory medical records. Human Subjects Committees of Harvard Pilgrim Health Care, Brigham and Women's Hospital, and Beth Israel Deaconess Medical Center approved study protocols. All participants gave informed consent [##REF##14760269##12##].</p>", "<p>We previously described study population, enrollment and follow up procedures [##REF##14760269##12##]. Of 657 infants with cord blood stored at 4°C for less than 30 hours, 362 infants were eligible for this analysis (mothers consented to enroll their infants into the study and completed a research visit including blood pressure at age 6 months). Of the 362 eligible infants, 286 (79%) returned for a 3-year blood pressure measurement. Therefore, we based our analyses on 286 children with 3-year blood pressure.</p>", "<title>Umbilical cord blood glucocorticoids</title>", "<p>At the time of delivery, using a needle and syringe technique, a midwife or obstetrician collected cord venous blood. Because research technicians were unable to be present at every delivery, delivery room staff refrigerated the specimen immediately after collection for a period of up to 30 hours (median: 13 hours). A research technician then separated the specimen into aliquots of serum, red blood cells, and white blood cells for storage at -70°C until analysis. We performed a small validation study to ensure that refrigeration of samples for less than 30 hours did not lead to degradation of glucocorticoid levels. We collected blood samples from 11 pregnant women and used three different methods of storage and processing. In method A (the gold standard), blood was drawn into room temperature heparinized glass tubes, spun at room temperature immediately, and the plasma was frozen at -20°C for 24 hours before being transported to a storage site and stored at -80°C for 1–6 months. In method B, blood was treated similarly except that after being drawn it was refrigerated at 4°C for 4 hours, and in method C for 24 hours. We found that within-person cortisol levels did not significantly differ by cortisol processing protocol. Comparing methods A and B, the mean within-person difference in cortisol was 22.1 nmol/liter (95%CI -129.7 to 173.8, <italic>p </italic>= 0.76). Comparing methods A and C, the mean within-person difference in cortisol was 60.7 nmol/liter (95% CI -80.0 to 204.2, <italic>p </italic>= 0.36).</p>", "<p>Cortisol and cortisone were quantified in duplicate in venous cord plasma by validated direct radioimmunoassay techniques [##UREF##0##13##,##REF##2620465##14##]. Concentrations of cortisol and cortisone were determined using a radioimmunoassay kit for cortisol (MP Biomedicals, UK) and cortisone (Immunovation Ltd, Southampton, UK) [##REF##15943831##15##]. The intra-assay coefficients of variation were 5.6% and 5.2% for cortisol and cortisone respectively.</p>", "<p>We used the mean of duplicate cortisol and cortisone assays to calculate our primary exposure, the ratio of cortisol/cortisone in umbilical venous blood, which correlates well (<italic>r </italic>= 0.5) with direct measures of 11β-HSD2 activity using placental tissue homogenates [##REF##9619527##16##].</p>", "<title>Blood pressure and anthropometric measurements</title>", "<p>At the 3-year visit, trained research assistants measured each child's blood pressure (BP) up to five times, at 1-minute intervals, using a Dinamap Pro 100 or Pro 200 (Critikon Inc., Tampa, FL, USA) automated blood pressure monitor. We recorded the conditions of measurement as previously described [##REF##15576466##3##], including room temperature, activity state of the child (crying, quiet awake, active awake), cuff size (infant, child, small adult), appendage used for BP measurement (left arm, right arm, calf), and child position (semi-reclining, seated, standing). The majority of children, 255 out of 286 (89%), had five blood pressure measurements (mean 4.8), and 87% of measurements were taken when the child was in a quiet, awake state. We defined our primary endpoint to be child systolic BP, because of its validity of measurement with an automated device [##REF##7641339##17##]. We examined diastolic BP as a secondary outcome.</p>", "<p>We measured child weight with a digital scale, and height using a stadiometer (Shorr Productions, Olney, MD, USA). We calculated a continuous measure of sex-specific birth weight for gestational age <italic>z</italic>-score (fetal growth) based on published US reference standards [##REF##12848901##18##]. We computed gestational age at birth as the number of days between the first day of last menstrual period and the delivery date and confirmed by ultrasound fetal measurements at 16–20 weeks. For date discrepancies of more than 10 days, we used the ultrasound-based gestational age.</p>", "<title>Statistical methods</title>", "<p>To assess multivariable associations between cord glucocorticoids and offspring BP, we used mixed effect regression models, incorporating each of the up to five BP measurements per child as repeated outcome measures [##REF##7168798##19##]. In contrast to standard ordinary least squares analysis, mixed effect models weight subjects based on the number of measurements and their variability, and thus result in appropriate standard errors. In our baseline crude model, we included only BP measurement conditions to reduce measurement error in child BP. Crude models stratified by gender showed similar associations among males and females; therefore we combined males and females in our models, adjusting for gender. In our multivariable model, we added gestational age at birth, child age, gender, attained length and weight, and maternal race and income [##REF##12848901##18##]. Potential confounders that did not change our effect estimates included maternal age, pre-pregnancy body mass index, maternal height, gravidity, pregnancy weight gain, third-trimester BP, induced delivery, Caesarean section, Apgar score, smoking status prior to pregnancy, education level, and marital status. Therefore, we excluded these variables from our final models. We conducted all data analyses using SAS version 9.1 (SAS Institute Inc., Cary, NC, USA).</p>" ]

[ "<title>Results</title>", "<p>Participant characteristics are shown in Table ##TAB##0##1##. Most mothers had a relatively high level of income and education. Mean venous cord cortisol was 343.9 nmol/liter (74.5–1042.9, 1st to 99th percentile). Mean venous cord cortisone was 242.7 nmol/liter (106.3–484.0, 1st to 99th percentile). Mean <italic>F/E </italic>ratio was 1.4 (0.5–3.8, 1st to 99th percentile). The Pearson correlation of the <italic>F/E </italic>ratio with cortisol was 0.8, and with cortisone was -0.1. The correlation of <italic>F/E </italic>ratio with birth weight for gestational age <italic>z</italic>-score was 0.02, and with gestational age at birth was 0.1. Mean systolic blood pressure was 93.3 mm Hg at age 3 years. Mean diastolic blood pressure was 58.9 mm Hg.</p>", "<p>Compared with the 657 mother-offspring pairs with cord blood, mothers included in our analyses had a slightly higher level of education (69% vs 62% completed a college or graduate degree) and had a higher household income (60% vs 54% reported income of more than $70,000 per year). Mean birth weight was similar among infants included in our analyses compared with all eligible infants (3545 g vs 3529 g). Child systolic and diastolic blood pressure, length and weight were similar among included and eligible children.</p>", "<p>Bivariate analyses showing systolic BP as a function of <italic>F/E </italic>ratio are shown in Figure ##FIG##0##1##. In multivariable analyses, each one unit increment in <italic>F/E </italic>ratio was associated with a 1.6 mm Hg (95% CI 0.0 to 3.1, <italic>p </italic>= 0.05) increment in systolic BP at age 3 years (Table ##TAB##1##2##). The direction of this effect estimate at age 3 years was consistent with our hypothesis that higher <italic>F/E </italic>ratio, reflecting lower 11β-HSD2 activity, would be associated with higher blood pressure. The effect estimate for diastolic BP at 3 years was 0.8 mm Hg (95% CI -0.4 to 2.0), adjusted for other covariates.</p>", "<p>In unadjusted analyses, the <italic>F/E </italic>ratio was not associated with birth weight; the effect estimate for birth weight was 47.1 g per one unit increment in the <italic>F/E </italic>ratio estimating placental 11β-HSD2 activity (95% CI -42.9 to 137.1). After adjustment for gestational age and sex, there was no association between <italic>F/E </italic>ratio and birth weight (-0.2 g for each one unit increment in <italic>F/E </italic>ratio, 95% CI -76.1 to 75.7). After additional adjustment for maternal age, BMI, race, income, and smoking status, the effect estimate was -12.1 g per unit increment in <italic>F/E </italic>ratio (95% CI -87.0 to 62.8). In unadjusted analyses, the effect estimate for the association between <italic>F/E </italic>ratio and gestational age at birth was 0.2 weeks per one unit increment in the <italic>F/E </italic>ratio (95% CI -0.0 to 0.5, <italic>p </italic>= 0.05). Additional adjustment for maternal age, BMI, race, income, and smoking status did not materially change the effect estimate (0.2 weeks, 95% CI -0.05 to 0.5, <italic>p </italic>= 0.11).</p>" ]

[ "<title>Discussion</title>", "<p>Our study is the first in humans to examine the relationship between umbilical cord concentrations of glucocorticoids and offspring blood pressure. We found that a one unit increment in <italic>F/E </italic>ratio, a proxy for reduced placental 11β-HSD2 activity, was associated with a 1.6 mm Hg higher systolic BP at 3 years of age. Although this effect estimate may appear to be small, a shift of this magnitude in the mean population blood pressure could result in a clinically meaningful difference in the prevalence of hypertension. A meta-analysis of 1 million adults estimated that a 2 mm Hg reduction in systolic blood pressure could result in a 10% decrease in mortality from stroke, and a 7% decrease in mortality from ischemic heart disease [##REF##12493255##20##]. We did not find an association between <italic>F/E </italic>ratio and fetal growth, suggesting that lower activity levels of 11β-HSD2 might program higher offspring blood pressure without restricting overall fetal growth.</p>", "<p>Our finding is consistent with data in animals showing that decreased placental 11β-HSD2 activity programs higher offspring blood pressure. Rats and sheep treated during pregnancy with the glucocorticoid dexamethasone, a relatively poor substrate for 11β-HSD2, have offspring with reduced birth weight and elevated blood pressure in offspring [##REF##11738795##7##]. Pregnant rats given an 11β-HSD2 inhibitor (carbenoxolone) have offspring with lower birth weight and raised blood pressure [##REF##8641724##8##]. In addition, pregnant rats fed a low protein diet have been shown to have both decreased placental 11β-HSD2 activity and higher offspring BP [##REF##8730887##2##]. Within Project Viva, we did not find a relationship either between protein intake and infant BP [##REF##15576466##3##] or between protein intake and <italic>F/E </italic>ratio (data not shown).</p>", "<p>No human studies have directly examined the relationship of placental 11β-HSD2 activity or cord glucocorticoid levels with later blood pressure. A few studies in humans have examined the effect of prenatal exogenous steroid administration on later blood pressure. In one observational study, preterm infants exposed to antenatal betamethasone had higher systolic BP and diastolic BP in adolescence than unexposed infants [##REF##10657267##21##]. However, follow-up data from two randomized controlled trials using two doses of prenatal betamethasone found that prenatal betamethasone use either did not affect offspring blood pressure [##REF##15342900##22##,##REF##15924982##23##], or was associated with a lower systolic blood pressure in offspring [##REF##10835090##24##]. It is possible that the effects of 11β-HSD2 activity could differ among preterm and term infants. Several [##REF##9619527##16##,##REF##11454515##25##,##REF##9363218##26##], but not all [##REF##7132352##27##, ####REF##3465674##28##, ##REF##12852864##29####12852864##29##], studies suggest that human placental 11β-HSD2 expression or activity rises with increasing gestational age until term; our data suggested a direct correlation between <italic>F/E </italic>ratio and gestational age at birth, although the effect estimates did not reach statistical significance. Preterm infants may also be predisposed to higher blood pressures in later life for other reasons [##REF##11149114##30##]. Our study excluded infants with a gestational age of less than 33 weeks and may therefore be more generalizable than prior studies focusing on preterm infants.</p>", "<p>In animals, lower placental 11β-HSD2 activity is associated with lower birth weight [##REF##8641724##8##]. In humans, similar associations have been shown in preterm infants [##REF##9619527##16##,##REF##12519895##31##]. Among term infants, two studies did not find an association between 11β-HSD2 activity and birth weight [##REF##9140081##32##,##REF##11234613##33##]. One study found that among asthmatic pregnant women with a female fetus, lower birth weight was associated with reduced placental 11β-HSD2 activity [##REF##14500261##34##]; this association was not present among women pregnant with a male fetus [##REF##14500261##34##], suggesting that there may be sex-specific differences in regulation of 11β-HSD2 activity. Our data (mean gestational age 39.7 weeks) showed little association between the <italic>F/E </italic>ratio and fetal growth, suggesting that a relative deficiency in placental 11β-HSD2 activity might program higher offspring blood pressure without affecting birth weight. Stratification by child gender made no difference to our analyses. Because birth weight has many determinants, it is not surprising that physiologic programming of blood pressure might occur without affecting birth weight [##REF##11980782##35##].</p>", "<p>Our results must be interpreted with caution, because we used <italic>F/E </italic>ratio in venous cord blood as a proxy for placental 11β-HSD2 activity rather than directly measuring enzyme activity. Glucocorticoids in the human fetal circulation may originate from the fetal adrenal gland, or from maternal cortisol, which freely crosses the placenta. Data show that 75–100% of cord blood cortisone originates from placental metabolism of maternal cortisol [##REF##4704743##36##]. Cortisol levels in venous umbilical cord blood may be affected by increased production of cortisol by fetal adrenal glands at term, as well as maternal or fetal stress and other factors [##REF##6345146##37##]. Adjustment for mode of delivery and induction of labor made no difference to our estimates; other variables relating to labor and delivery, such as duration of labor, or the use of instrumentation were not available for these analyses. Prior studies have not found an effect of labor on 11β-HSD2 mRNA or activity levels [##REF##12852864##29##,##REF##6951829##38##].</p>", "<p>Among healthy term infants, the ratio of cortisol/cortisone in umbilical venous cord blood appears to correlate well with direct measures of placental 11β-HSD2 activity using placental tissue homogenates [##REF##9619527##16##]. A study that measured placental cortisol metabolism using a similar venous cord ratio, <italic>E</italic>/(<italic>E</italic>+<italic>F</italic>), was not improved by creation of an 11β-HSD2 activity index incorporating arterial cord glucocorticoid measurements [##REF##7621576##39##], which presumably reflect the fetal adrenal contribution (Pearson correlation between <italic>E</italic>/(<italic>E</italic>+<italic>F</italic>) and <italic>F/E </italic>was 0.9 using our data). The cord <italic>F/E </italic>ratio could also be affected by the activity of either placental 11β-hydroxysteroid dehydrogenase type 1 enzyme (11β-HSD1) or fetal 11β-HSD2. 11β-HSD1, an enzyme isoform that can convert cortisone to cortisol, has been detected in human perfused placenta [##REF##9950140##40##], and placental 11β-HSD1 activity levels may increase during gestation [##REF##14557491##41##]. However, studies suggest that type 2 enzymatic activity predominates during pregnancy [##REF##7132352##27##,##REF##9408088##42##], with one study reporting no detectable 11β-HSD1 activity [##REF##9135697##43##]. Several fetal tissues express 11β-HSD2 until mid-gestation [##REF##8200959##44##]. Dy <italic>et al</italic>. recently reported that among term infants born with intrauterine growth restriction, the ratio of cortisone to cortisol in the umbilical artery was lower than in the umbilical vein, suggesting either attenuated fetal 11β-HSD2 activity, or reduced fetal glucocorticoid clearance [##REF##18061258##45##]. Term infants with a birth weight appropriate for gestational age showed no difference in cortisone to cortisol ratio between the umbilical artery and vein [##REF##18061258##45##]. Since most 11β-HSD2 expression is lost from the fetus by term (at least in rodents and probably in humans) we think this is unlikely to have contributed to levels of cortisol and cortisone in cord blood, emphasizing the placental contribution.</p>", "<p>Among infants born preterm or small for gestational age, the <italic>F/E </italic>ratio may not be a valid proxy for placental 11β-HSD2 activity. A study of preterm infants born at less than 32 weeks gestational age was unable to detect a correlation between venous cord blood glucocorticoid concentrations and placental 11β-HSD2 activity [##REF##12519895##31##]. A recent study reported that compared with term infants with a birth weight appropriate for gestational age, infants with intrauterine growth restriction had reduced placental homogenate 11β-HSD2 activity, but similar umbilical cord venous cortisone to cortisol ratio [##REF##18061258##45##]. The reasons underlying the lack of correlation between the venous cord <italic>F/E </italic>ratio and placental 11β-HSD2 activity among preterm and small for gestational age infants are unclear, but might be explained by a greater contribution of placental 11β-HSD1 activity to circulating glucocorticoid levels, or decreased glucocorticoid clearance by the placenta or fetal tissues [##REF##18061258##45##]. Our study included only infants born later than 33 weeks gestation (mean 39.7 weeks), and the majority of our participants were healthy term infants with a birth weight appropriate for gestational age (mean birth weight 3545 g), supporting the use of <italic>F/E </italic>as a valid proxy for placental 11β-HSD2 activity in our study.</p>", "<p>Strengths of our study include our adjustment for multiple relevant confounders and careful blood pressure measurements. Our study has several limitations. One limitation is that we measured glucocorticoids only at a single time point (delivery), and thus were unable to measure the potential impact of presumed fetal glucocorticoid exposure at earlier time windows during pregnancy. By definition, use of a proxy for enzymatic activity results in some misclassification of the exposure. A non-differential misclassification should have biased our findings toward the null; therefore, our data may actually underestimate the association between fetal glucocorticoid exposure and child BP. Our study did have some loss to follow-up, and therefore may be subject to selection bias. However, child BP among participants excluded from our analyses was similar to those included in our analyses. It seems unlikely that placental 11β-HSD2 activity would systematically differ in participants excluded from our analyses, although we cannot exclude this possibility.</p>" ]

[ "<title>Conclusion</title>", "<p>In conclusion, our data show for the first time that a higher <italic>F/E </italic>ratio in venous cord blood is associated with higher offspring systolic blood pressure in humans. The higher <italic>F/E </italic>ratio may reflect reduced placental 11β-HSD2 activity, resulting in increased fetal glucocorticoid exposure and higher offspring systolic blood pressure. Our study goes beyond previous data focusing on the birth weight-blood pressure relationship to explore a potential prenatal mechanism for the programming of later blood pressure, but our findings are limited by the use of an indirect measure of enzyme activity. Additional studies using direct measures of enzymatic activity are needed to definitively determine whether placental 11β-HSD2 activity does program blood pressure in childhood.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Animal data show that decreased activity of placental 11-beta-hydroxysteroid dehydrogenase type 2 (11β-HSD2), which potently inactivates glucocorticoids (e.g. cortisol) to inert forms (cortisone), allows increased access of maternal glucocorticoids to the fetus and 'programs' hypertension. Data in humans are limited. We examined in humans the association between venous umbilical cord blood glucocorticoids, a potential marker for placental 11β-HSD2 enzyme activity, and blood pressure at age 3 years.</p>", "<title>Methods</title>", "<p>Among 286 newborns in Project Viva, a prospective pre-birth cohort study based in eastern Massachusetts, we measured cortisol (<italic>F</italic>) and cortisone (<italic>E</italic>) in venous cord blood and used the ratio of <italic>F/E </italic>as a marker for placental 11β-HSD2 activity. We measured blood pressure (BP) when the offspring reached age 3 years. Using mixed effects regression models to control for BP measurement conditions, maternal and child characteristics, we examined the association between the <italic>F/E </italic>ratio and child BP.</p>", "<title>Results</title>", "<p>At age 3 years, each unit increase in the <italic>F/E </italic>ratio was associated with a 1.6 mm Hg increase in systolic BP (95% CI 0.0 to 3.1). The <italic>F/E </italic>ratio was not associated with diastolic blood pressure or birth weight for gestational age <italic>z</italic>-score.</p>", "<title>Conclusion</title>", "<p>A higher <italic>F/E </italic>ratio in umbilical venous cord blood, likely reflecting reduced placental 11β-HSD2 activity, was associated with higher systolic blood pressure at age 3 years. Our data suggest that increased fetal exposure to active maternal glucocorticoids may program later systolic blood pressure.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>SYH designed research question, led data analysis, and drafted the manuscript. RA directed glucocorticoid assays, assisted in study design and data analysis, and contributed to the manuscript. KPK designed the statistical analysis and contributed to the manuscript. JWR–E obtained funding, helped direct study operations, and contributed to the manuscript. JRS participated in the study design and contributed to the manuscript. MWG participated in the study design, obtained funding, directed study operations, and contributed to the manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1741-7015/6/25/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We thank the participants and staff of Project Viva, and Scott Denham for excellent laboratory support. This work was supported by grants from the US National Institutes of Health (HD34568, HL64925, HL68041), and by Harvard Medical School, the Harvard Pilgrim Health Care Foundation, and a program grant from the Wellcome Trust. Susanna Huh was an American Academy of Pediatrics, American Pediatric Society and Hospital for Sick Children Fellow of the Pediatric Scientist Development Program (NICHD Grant Award K12-HD00850-17) when she completed this work. This work was presented in part at the Third World Congress on Developmental Origins of Health and Disease, Toronto, Canada, 2005.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Systolic blood pressure (BP) at age 3 years as a function of cortisol/cortisone (<italic>F/E</italic>) ratio</bold>. Bivariable regression line (solid line), unadjusted for covariates, and its 95% confidence limits (dotted lines) are shown.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Cord blood glucocorticoids and other characteristics of 286 participants with 3-year blood pressure.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><italic>n</italic></td><td align=\"center\">Mean (SD) or %</td></tr></thead><tbody><tr><td align=\"left\"><bold>Maternal characteristics at enrollment</bold></td><td/><td/></tr><tr><td align=\"left\"> Age (years)</td><td align=\"center\">286</td><td align=\"center\">32.2 (4.9)</td></tr><tr><td align=\"left\"> Pre-pregnancy BMI (kg/m²)</td><td align=\"center\">284</td><td align=\"center\">24.6 (5.1)</td></tr><tr><td align=\"left\"> Race/ethnicity (%)</td><td/><td/></tr><tr><td align=\"left\">  White</td><td align=\"center\">204</td><td align=\"center\">71%</td></tr><tr><td align=\"left\">  Black</td><td align=\"center\">31</td><td align=\"center\">11%</td></tr><tr><td align=\"left\">  Hispanic</td><td align=\"center\">20</td><td align=\"center\">7%</td></tr><tr><td align=\"left\">  Asian</td><td align=\"center\">13</td><td align=\"center\">5%</td></tr><tr><td align=\"left\">  Other</td><td align=\"center\">18</td><td align=\"center\">6%</td></tr><tr><td align=\"left\"> Marital status (%)</td><td/><td/></tr><tr><td align=\"left\">  Married</td><td align=\"center\">238</td><td align=\"center\">83%</td></tr><tr><td align=\"left\">  Partner</td><td align=\"center\">26</td><td align=\"center\">9%</td></tr><tr><td align=\"left\">  Other</td><td align=\"center\">22</td><td align=\"center\">8%</td></tr><tr><td align=\"left\"> Education (%)</td><td/><td/></tr><tr><td align=\"left\">  High school or less</td><td align=\"center\">17</td><td align=\"center\">6%</td></tr><tr><td align=\"left\">  Less than 4 years of college</td><td align=\"center\">73</td><td align=\"center\">26%</td></tr><tr><td align=\"left\">  4 years of college</td><td align=\"center\">100</td><td align=\"center\">35%</td></tr><tr><td align=\"left\">  Graduate degree</td><td align=\"center\">96</td><td align=\"center\">34%</td></tr><tr><td align=\"left\"> Annual household income (%)</td><td/><td/></tr><tr><td align=\"left\">  &lt;$20 000</td><td align=\"center\">3</td><td align=\"center\">1%</td></tr><tr><td align=\"left\">  $20 000–$40 000</td><td align=\"center\">30</td><td align=\"center\">10%</td></tr><tr><td align=\"left\">  $40 000–$70 000</td><td align=\"center\">65</td><td align=\"center\">23%</td></tr><tr><td align=\"left\">  &gt;$70 000</td><td align=\"center\">173</td><td align=\"center\">61%</td></tr><tr><td align=\"left\">  Missing</td><td align=\"center\">15</td><td align=\"center\">5%</td></tr><tr><td align=\"left\"> Smoked 3 months prior to pregnancy (%)</td><td align=\"center\">23</td><td align=\"center\">8%</td></tr><tr><td align=\"left\"> Pregnancy weight gain (kg)</td><td align=\"center\">283</td><td align=\"center\">15.6 (5.5)</td></tr><tr><td align=\"left\"> 3^rdtrimester systolic BP (mm Hg)</td><td align=\"center\">285</td><td align=\"center\">110.9 (8.3)</td></tr><tr><td align=\"left\"> Spontaneous delivery (vs induced)</td><td align=\"center\">186</td><td align=\"center\">65%</td></tr><tr><td align=\"left\"> Caesarean section (vs vaginal delivery)</td><td align=\"center\">47</td><td align=\"center\">16%</td></tr><tr><td align=\"left\"><bold>Child characteristics</bold></td><td/><td/></tr><tr><td align=\"left\"> Male (%)</td><td align=\"center\">146</td><td align=\"center\">51%</td></tr><tr><td align=\"left\"> Birth weight (g)</td><td align=\"center\">286</td><td align=\"center\">3545 (501)</td></tr><tr><td align=\"left\"> Birth weight for gestational age <italic>z</italic>-value (units)</td><td align=\"center\">286</td><td align=\"center\">0.2 (0.9)</td></tr><tr><td align=\"left\"> Gestational age (weeks)</td><td align=\"center\">286</td><td align=\"center\">39.7 (1.3)</td></tr><tr><td align=\"left\"> Cortisol (nmol/liter)</td><td align=\"center\">286</td><td align=\"center\">343.9 (209.3)</td></tr><tr><td align=\"left\"> Cortisone (nmol/liter)</td><td align=\"center\">286</td><td align=\"center\">242.7 (81.3)</td></tr><tr><td align=\"left\"> Cortisol/cortisone ratio (units)</td><td align=\"center\">286</td><td align=\"center\">1.4 (0.6)</td></tr><tr><td align=\"left\"> Systolic BP at age 3 years (mm Hg)</td><td align=\"center\">286</td><td align=\"center\">93.3 (10.2)</td></tr><tr><td align=\"left\"> Diastolic BP at age 3 years (mm Hg)</td><td align=\"center\">286</td><td align=\"center\">58.9 (8.0)</td></tr><tr><td align=\"left\"> Weight at age 3 years (kg)</td><td align=\"center\">286</td><td align=\"center\">15.8 (2.6)</td></tr><tr><td align=\"left\"> Height at age 3 years (cm)</td><td align=\"center\">286</td><td align=\"center\">97.6 (4.6)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Multivariable mixed effect models showing change in 3-year child BP per unit increment in venous cord <italic>F/E </italic>ratio.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"left\" colspan=\"2\">Change in child BP (mm Hg) per one unit increment in <italic>F/E </italic>ratio (95% CI)</td></tr><tr><td/><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\">Models*</td><td align=\"center\">Systolic BP</td><td align=\"center\">Diastolic BP</td></tr></thead><tbody><tr><td align=\"left\">Model 1: <italic>F/E </italic>ratio (unit increment)</td><td align=\"center\">1.2 (-0.4 to 2.7)<break/><italic>p </italic>= 0.14</td><td align=\"center\">0.7 (-0.5, 1.8)<break/><italic>p </italic>= 0.27</td></tr><tr><td align=\"left\">Model 2: Model 1 + infant sex, gestational age at birth, age, 3 year weight, length, maternal race/ethnicity, income</td><td align=\"center\">1.6 (0.0 to 3.1)<break/><italic>p </italic>= 0.05</td><td align=\"center\">0.8 (-0.4, 2.0)<break/><italic>p </italic>= 0.20</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>BP, blood pressure.</p></table-wrap-foot>", "<table-wrap-foot><p>Data from 286 mother-offspring pairs within Project Viva.</p><p>*All models were adjusted for blood pressure measurement order, cuff size, appendage, position, state, and machine model. BP, blood pressure; <italic>F</italic>/<italic>E</italic>, cortisol/cortisone ratio in venous cord blood.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1741-7015-6-25-1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Linkage analysis of family data have been extensively used in the past in the search for genetic determinants. Nowadays, investigators favor large epidemiological studies of unrelated individuals, however several family datasets are currently being re-analyzed and/or pooled (e.g. [##REF##16691188##1##]). The persistance of interest for linkage is partly triggered by the advent of single-nucleotide-polymorphisms (SNP) array genotyping technology in the field, indeed SNP arrays hold the promise of more reliable linkage maps [##REF##15311375##2##,##REF##15514889##3##]. Although less prone to genotyping error than microsatellites when viewed as singlepoint markers, SNP arrays heavily rely on multipoint algorithms for accurate determination of the identical by descent (IBD) status of alleles. The gain in singlepoint reliability might therefore be annihilated by the propagation of errors across the many SNPs required to infer IBD status.</p>", "<p>In the search for genetic determinants of complex traits by linkage, the use of selective designs appears to be an efficient way to gain adequate power for detection of typically small gene effects. A few authors have shown by simulation that the impact of genotyping error on evidence for linkage could be particularly severe in affected sib-pair (ASP) designs [##REF##10739757##4##, ####REF##11313746##5##, ##REF##16059746##6####16059746##6##], virtually masking most of the evidence for linkage. The impact of error on quantitative traits appears to be less dramatic in random samples, however it is unclear whether the same dramatic power losses hold in selected samples.</p>", "<p>A method of choice is now emerging for the analysis of quantitative traits arising from selected sib pairs. This method is essentially a regression through the origin of excess identical by descent (IBD) sharing on a function of the trait value, whose slope is an estimate of the linkage parameter. It was first proposed by Sham et al. [##REF##11353401##7##] and turns out to be equivalent to a score test [##REF##12933546##8##]. In a numerical comparison of methods for selected samples, Skatkiewicz et al. [##REF##12970846##9##] and Cuenco et al. [##REF##12970847##10##] showed that this method had good properties in finite samples for extreme proband ascertained sib-pair and discordant sib-pair designs. By use of simple genotyping error models (<italic>population frequency error model </italic>and <italic>false homozygosity model</italic>), we show analytically what effects such error generating processes (occurring at rate ϵ per sib pair) induce for an idealized fully informative marker. It is shown that it results in a reduction of the slope estimate (i.e. of the estimated linkage parameter) by a factor 1 - whether sib pairs are selected or not. Since the genotyping error rate ϵ is typically small, the previous effect on the linkage test is minimal. In addition to this slope effect, the regression's intercept is modified and this may have a much more sizable effect on the test for linkage depending on the sampling scheme used to select sib pairs. Surprisingly, this simple result allows us to predict that in extremely concordant (EC) sib pairs designs and in ASP designs, the effect of genotyping error will be milder as the selection becomes more extreme. In extreme discordant (ED) designs, the effect can in theory be either increased type I error or decreased power depending on the definition of discordance, the genotyping error rate and the true linkage effect; in practice however, for small quantitative trait locus (QTL) effects, the result will be an increased type I error. We argue that the basic error generating mechanisms assumed provide reasonable approximations of real-life situations. In the next section, we first describe some common error-generating processes and quantify their effect on IBD sharing in an idealized situation where marker information is complete. We then briefly sketch the inverse regression approach to linkage, we show analytically what the effect of genotyping error is on this regression and quantify the subsequent bias, power and type I error in common selective designs. We argue that under certain assumptions regarding the error model, one can easily implement a linkage test that incorporates a genomic control for genotyping error. Finally, we discuss some assumptions made in our study and the practical relevance of our findings. In particular, we argue that our results generalize to situations where marker information is incomplete and that the smaller error rates observed in SNP chip array compared to microsatellites offer no protection against bias in analysis.</p>" ]

[]

[ "<title>Results</title>", "<title>Genotyping error models</title>", "<p>We consider two mechanisms for the generation of errors in marker data, namely the <italic>population frequency error model </italic>and the <italic>false homozygosity model</italic>. In those two models, we consider a single marker with <italic>m </italic>alleles and further assume that a maximum of one allelic error per sib pair can be made and that this happens with probability ϵ. This restriction to 'one error per sib pair' is just a first order approximation, for small ϵ, of a process where all four alleles would be allowed to be independently erroneous and does not restrict the generalizability of our results.</p>", "<p>The <italic>population frequency error model </italic>re-assigns the erroneous allele (chosen at random among the four forming the sib-pair genotype) to one of the possible <italic>m </italic>alleles with probability equal to population allele frequency. One mathematical advantage of this model is that the marginal distribution of alleles and genotypes is unaltered. The <italic>false homozygosity model </italic>keeps homozygotes unchanged but re-assigns heterozygotes to homozygotes with alleles equal to one of the two original alleles chosen according to probabilities proportional to population allele frequencies.</p>", "<p>To our knowledge, <italic>false homozygosity </italic>is a common type of error: fairly rare alleles go un-reported in samples. The <italic>population frequency error model </italic>provides an approximation to a process whereby alleles are misread. Errors at the two alleles of a marker's genotype might be correlated, we do not consider this type of process in details here although the effect on linkage will be qualitatively the same as in the two other models. We refer the reader to Sobel et al. [##REF##11791215##11##] for a detailed exposé on genotyping error mechanisms. Note that the two models that we have chosen have been used in the past in order to identify potential genotyping errors [##REF##10739757##4##,##REF##11791215##11##].</p>", "<title>Impact on IBD sharing</title>", "<p>Let's denote by <italic>π </italic>the proportion of alleles shared identical by descent (IBD) at a certain locus by two siblings. Tests for linkage are based on the IBD sharing distribution and although errors as described earlier are made at the genotype level (<italic>G </italic>is read as <italic>G</italic>^ϵ), the effect of errors on linkage will be entirely mediated via the distortion of the IBD distribution (the true IBD status <italic>π </italic>of two siblings may be incorrectly inferred as <italic>π</italic>^ϵ). We are therefore interested in deriving the probability distribution <bold>P</bold>(<italic>π</italic>^ϵ|<italic>π</italic>), this is done by conditioning on both the true and observed genotypes as follows:</p>", "<p></p>", "<p>Let us consider the case of complete information. This can be conceptualized by means of an idealized marker whose number of alleles is infinite, in particular identity by state (IBS) status is equivalent to IBD status. The unordered genotypes of a sib pair can be partitioned into seven exclusive classes denoted <italic>ii</italic>/<italic>ii</italic>, <italic>ii</italic>/<italic>ij</italic>, <italic>ii</italic>/<italic>jj</italic>, <italic>ii</italic>/<italic>jk</italic>, <italic>ij</italic>/<italic>ij</italic>, <italic>ij</italic>/<italic>ik </italic>and <italic>ij</italic>/<italic>kl </italic>depending on the number of homozygous sibs in the pair and the number of distinct alleles in the sib-pair genotype. Sharing 0 alleles IBD corresponds to a sib-pair genotype of the <italic>ij</italic>/<italic>kl </italic>class, should an error occur according to the <italic>population frequency error model </italic>then one of the four alleles would be transformed into yet another type (since the number of alleles is infinite, the probability that the new allele is read as one of <italic>i</italic>, <italic>j</italic>, <italic>k </italic>or <italic>l </italic>tends to 0), therefore the sib pair genotype will remain in the <italic>ij</italic>/<italic>kl </italic>class and the observed IBD status <italic>π</italic>^ϵwill still be 0. For the same starting genotype, an error according to the <italic>false homozygosity model </italic>produces an <italic>ii</italic>/<italic>jk </italic>class and <italic>π</italic>^ϵalso equals 0 therefore <bold>P</bold>(<italic>π</italic>^ϵ= 0|<italic>π </italic>= 0) = 1 whatever the genotyping error mechanism considered previously. The same line of reasoning leads to <bold>P</bold>(<italic>π</italic>^ϵ= 0.5|<italic>π </italic>= 0.5) = 1 - , <bold>P</bold>(<italic>π</italic>^ϵ= 0|<italic>π </italic>= 0.5) = , <bold>P</bold>(<italic>π</italic>^ϵ= 1.0|<italic>π </italic>= 1.0) = 1 - ϵ, <bold>P</bold>(<italic>π</italic>^ϵ= 0.5|<italic>π </italic>= 1.0) = ϵ. Those results can be summarized by the transition matrix below, where the (<italic>i</italic>, <italic>j</italic>) element is equal to <bold>P</bold>(<italic>π</italic>^ϵ= (<italic>j </italic>- 1)/2|<italic>π </italic>= (<italic>i </italic>- 1)/2)</p>", "<p></p>", "<p>The overall effect of genotyping error is thus to reduce the observed IBD sharing, indeed <bold>E</bold>(<italic>π</italic>^ϵ|<italic>π</italic>) = (1 - ϵ/2)<italic>π </italic>and <bold>E</bold>(<italic>π</italic>^ϵ) = - ϵ/4 while the variance is practically unchanged since . In selected samples of extremely concordant sib pairs (EC) where linkage is evidenced by an excess in IBD sharing, it therefore seems logical to expect a decrease in power. Conversely, in selected samples of extremely discordant sib pairs (ED) where linkage is evidenced by a reduction in IBD sharing, the test might lead to increased type I error. In the next subsection, we formally quantify this bias in selective samples schemes for quantitative traits under the usual assumption of a normal variance components model.</p>", "<title>Impact on linkage testing</title>", "<title>Regression-based linkage testing</title>", "<p>We assume that the sib pair phenotypic data <bold>x </bold>= (<italic>x</italic>₁, <italic>x</italic>₂)<italic>' </italic>have been adjusted for any relevant covariates (e.g. sex, age, country, ...) and have been standardized so that the (known) population mean, variance and sib-sib correlation are 0, 1 and <italic>ρ </italic>respectively. Under the additive variance components model, <bold>x </bold>given IBD information <italic>p </italic>follows a bivariate normal distribution with zero mean and variance-covariance matrix given by</p>", "<p></p>", "<p>where <italic>γ </italic>≥ 0 denotes the proportion of total variance explained by the putative locus. Under this model, an optimal testing strategy first advocated in [##REF##11353401##7##] (and sometimes referred to as the optimal Haseman-Elston regression) is to regress (through the origin) excess IBD sharing <italic>π </italic>- on the following <italic>C </italic>function of the trait values:</p>", "<p></p>", "<p>This test turns out to be a score test for the linkage parameter <italic>γ </italic>[##REF##12933546##8##] and is based upon the following approximate relation which is valid for small locus effects [##REF##14624721##12##]:</p>", "<p></p>", "<p>where = var₀(<italic>π</italic>). In a set of sibships indexed by <italic>i</italic>, an efficient estimate of the linkage parameter <italic>γ </italic>is . It is approximately unbiased <bold>E</bold>() = <italic>γ </italic>and has variance var₀() = 1/ where is the corresponding Fisher's information. The test statistic is given by , it is one-sided, only positive values being regarded as evidence for linkage. For small QTL effects, power of this test can be computed as Φ (Φ^-1(<italic>α</italic>) + <italic>γ</italic>^1/2). Fisher's information , which depends on sample size and study design, therefore controls power. In the design phase of a study, should be used as a criterion to differentiate between alternative designs rather than sample size only [##REF##14624721##12##,##REF##15305326##13##].</p>", "<title>Impact of genotyping error on regression</title>", "<p>By conditioning on the true IBD sharing values, we can compute <bold>P</bold>(<italic>π</italic>^ϵ|<bold>x</bold>, <italic>γ</italic>, ϵ) = ∑_{<italic>π</italic>}<bold>P</bold>(<italic>π</italic>^ϵ|<italic>π</italic>) <bold>P</bold>(<italic>π</italic>|<bold>x</bold>, <italic>γ</italic>), using the transition probabilities <bold>P</bold>(<italic>π</italic>^ϵ|<italic>π</italic>) derived earlier, while the <bold>P</bold>(<italic>π</italic>|<bold>x</bold>, <italic>γ</italic>)'s are given in [##REF##14624721##12##]. This permits computation of the new regression line in presence of genotyping error as</p>", "<p></p>", "<p>As mentioned earlier, the corresponding variance under the null hypothesis is only slightly altered. The effect of genotyping error is thus to shrink the regression line by a factor 1 - and to shift the intercept by -. If we ignore genotyping error i.e. we estimate <italic>γ </italic>using , this results in a biased estimator with . The resulting testing statistic would then have power equal to</p>", "<p></p>", "<p>Note that taking <italic>γ </italic>= 0 in this formula gives the type I error rate. Since increases with sample size, the impact of genotyping error on both power and type I error will be larger as the sample size increases. In terms of Y versus X regression, the intuition is that the regression through the origin is not affected by a general shift in the Y-variable (IBD sharing) if the X-variable (<italic>C </italic>variable) has average 0, or takes values far away from 0. The further away the X-variable <italic>C </italic>is from 0, the smaller <italic>A</italic>, hence the smaller the bias.</p>", "<title>Bias and impact on power and type I error</title>", "<p>Since and <italic>γ </italic>is typically small, the distortion of the usual linkage test in presence of genotyping error heavily depends on the design-specific quantity . Unfortunately, there is little intuition about the distribution of <italic>C </italic>(hence about the distribution of <italic>A</italic>) in the whole population or in a selected sample. Nevertheless, Monte Carlo simulations can be used to determine the characteristics of the <italic>C </italic>and <italic>A </italic>distributions in the whole population or for a specific ascertainment scheme. In random samples and under the variance components model, <italic>C </italic>is a score function hence <bold>E</bold>(<italic>C</italic>) = 0 therefore its sample estimate will be close to 0; one can also check that its distribution is negatively skewed (unless <italic>ρ </italic>= 0). The result is that the bias will be small for random samples. The same finding would hold for any ascertainment scheme where = 0. An optimal selection scheme [##REF##14624721##12##] that would select sib pairs based on Fisher's information (i.e. such that |<italic>C</italic>| ≥ <italic>C</italic>₀) does not warrant that = 0 because of the skewness of <italic>C</italic>. In EC designs (both siblings have trait values either larger than a positive threshold or smaller than a negative threshold), tends to be positive while it tends to be negative in ED designs (one sibling's trait value is larger than a positive threshold while the other sibling's trait value is smaller than a negative threshold), the linkage test will therefore have reduced power in EC designs and increased type I error in ED designs.</p>", "<p>In the left-hand side of Table ##TAB##0##1##, we have computed the values of <italic>A </italic>and for the three selective schemes considered. The designs are indexed by the sib-sib correlation <italic>ρ </italic>and the degree of selection. One obvious way to correct for the shift in the intercept induced by genotyping error would be to leave the regression unconstrained, this would correct for most of the bias. Unfortunately, in selected designs where the variance of <italic>C </italic>is reduced, this results in a very inefficient estimator of the linkage parameter <italic>γ</italic>. The right-hand side of Table ##TAB##0##1## displays the variance of the linkage parameter estimates in constrained () and unconstrained () regressions. Efficiency losses of unconstrained versus constrained regressions in EC and ED designs are unacceptably large even for moderately extreme selection schemes.</p>", "<p>In Table ##TAB##1##2##, we report the power and type I error for realistic genotyping error rates [##REF##10924406##14##] equal to 0.005 and 0.01 for the same designs as in Table ##TAB##1##2##. The equivalent sample size used corresponds to samples with Fisher's information equal to 2500 which provides 90% power to detect a QTL explaining 10% of the total variance in absence of genotyping error (pointwise nominal error rate = 10^-4). The most visible impact is on type I error rates in ED design which is up to 7 times its nominal value. The design that combines EC and ED sib pairs appears to be fairly immune to genotyping error while EC designs do not incur power losses greater than 20%. Finally, those computations confirm the intuition expressed earlier that the effect of genotyping error is less severe in more extreme selection schemes.</p>", "<title>Genomic control for genotyping error</title>", "<p>As we have seen in previous sections, the main effect of genotyping error is to modify the intercept in the regression used to test for linkage. Although an unconstrained regression would correct most of the bias due to genotyping error, the inefficiency of this strategy makes it impractical. In order to obtain an efficient and robust inference, it therefore seems natural to try and constrain the regression through its correct origin <italic>a</italic>. In this section, we propose a completely data-driven strategy for doing this.</p>", "<p>At any position, the sample mean IBD sharing has variance 1/8<italic>n </italic>where <italic>n </italic>is the number of sib pairs available. If we knew that the position is unlinked or if the sample of sib pairs was random then the deviation of this mean from would provide an estimate of the intercept <italic>a </italic>in the linkage regression.</p>", "<p>Unfortunately, detection of a position-specific intercept corresponding to typical error rates would require a sample size of order 10⁴, a number that is almost never reached in linkage studies. In order to obtain an intercept estimate with sufficient precision, it is therefore essential to combine information across positions. The value of IBD sharing at positions outside of the neighborhood of influencing loci (those positions are subsequently referred to as unlinked) across the genome may serve as control in the test for linkage, this concept of genomic control has been used to make the analysis of association studies more robust [##REF##11315092##15##].</p>", "<p>Let's assume that the proportions of alleles shared IBD <italic>π </italic>is computed at a series of approximately regular positions indexed by <italic>t </italic>across the whole genome. Let <italic>y</italic>_{<italic>t </italic>}be the sample mean (among families) excess IBD at position <italic>t </italic>i.e. . Under the variance components model and for small QTL effect <italic>γ</italic>, equation (3) implies that</p>", "<p></p>", "<p>In random samples or in any sample where ≃ 0, taking the average of <italic>y</italic>_{<italic>t </italic>}across positions provides an estimate of <italic>a</italic>. In selected samples, we can use a trimmed version of the mean of <italic>y</italic>, for example a 20%-trimmed mean of the (<italic>y</italic>_{<italic>t</italic>})_{<italic>t </italic>}series (i.e. the mean of the <italic>y</italic>_{<italic>t </italic>}values after removing the 20% lowest and and 20% highest values) will provide a robust genomic estimate of <italic>a</italic>. Because <italic>a </italic>≤ 0 and is positive and negative in EC designs and ED designs respectively, could be refined by trimming off only the 20% highest and lowest <italic>y</italic>_{<italic>t </italic>}values respectively before taking the mean. Of course, how much we trim is arbitrary but 20% can safely be taken as a conservative value for oligogenic traits (Indeed, a 3500 cM genome contains approximately 70 quasi-independent loci, so a 20% trimming of <italic>y</italic>_{<italic>t </italic>}values discards 14 positions (including all active gene positions if less than 14 genes) from the sample used to estimate intercept <italic>a</italic>.). An ad-hoc implementation of the concept of genomic control is then to plug in the estimate of the intercept into the linkage regression (3). Since most of the bias in the inference is due to the intercept mis-specification, the precise estimate obtained by pooling across the genome will eliminate it. The implicit assumption that we make in this genomic control approach is that the regression intercept is the same at all positions, this will be challenged in the next section.</p>" ]

[ "<title>Discussion</title>", "<p>Under two basic error models, we were able to predict quantitatively the consequences of genotyping error on inference in linkage analysis. In the idealized situation of complete IBD information, both error models have the same impact on linkage analysis. As we have seen, the effect is due to a decrease in IBD sharing. A contrario, an error process which would increase IBD sharing would produce opposite results. The true error processes involved in practice are complicated mixtures of the models alluded to here. In our experience however, it seems that processes which lower IBD sharing are predominant. Because genotyping error tends to decrease the estimated number of alleles shared IBD, the effect on evidence for linkage is opposite in EC (reduced power) and ED (increased type I error) designs, it can be dramatic in typical designs and paradoxically less severe for more extreme ascertainment schemes. By analogy, for a dichotomous trait, this means that the effect of genotyping error is less severe in ASP designs for rare diseases than for common diseases. Remarkably, in designs combining both ED and EC pairs like the (or EDAC designs), the competing effects of genotyping error tend to cancel each other out. We have considered here only three types of basic selection schemes however the approach can be straightforwardly applied to any arbitrary selection scheme. Under the widely accepted variance components model, the important quantity which determines bias, type I error and power is and it can be easily estimated by Monte Carlo simulations. Note that the bias is proportional to the error rate so that Equation (4) can easily be adapted to different error rates than those considered in Table ##TAB##1##2##.</p>", "<p>Our study used an idealized model where IBD information is assumed to be complete. In practice, IBD is uncertain and it is inferred using marker data and multipoint algorithms as implemented in publicly available software [##REF##8651312##16##,##REF##11731797##17##], the general effect is to shrink the IBD estimate towards 0.5. The linkage regression (2) is changed into where can be either estimated from the data or by simulations. The effect of genotyping error is again mediated via the shift of the intercept in this regression but no general formula can be obtained because it depends in a very complex manner on the whole marker map configuration. Nevertheless, we can quantify this shift under realistic scenarios and compare it to its theoretical value when IBD information is complete. We simulated two different marker maps in 1 million sib pairs without parents and quantified by how much IBD sharing was reduced on average under the <italic>population frequency error model </italic>(error rate = 0.01). The microsatellites map (MS) had 13 equi-frequent ten-allele markers (heterozygozity = 90%) located 10 cM apart (spanning the 0–120 cM chromosomal region) and the SNP map had 41 equi-frequent SNPs (heterozygozity = 50%) spanning the 50–70 cM chromosomal region (this smaller region was chosen to keep simulation time acceptable). The resulting average reduction in IBD sharing for an error rate of 0.01 was measured every 2 cM in the 50–70 cM region, it ranged from 0.4974 to 0.4976 in the MS map and from 0.4945 to 0.4955 in the SNP map. For these two maps which mimic the two most widespread genotyping paradigms nowadays, those simulations confirm results derived under the complete marker information assumption with a reduction in IBD sharing from 0.5 to 0.5 – 0.01/4 = 0.4975. Our results therefore appear to be applicable to real-life situations where IBD information is incomplete.</p>", "<p>The genomic-control strategy that we have proposed, although triggered by the specific issue of genotyping error, potentially offers a general robust method for carrying out linkage analysis. It is nonetheless important to recognize its limitations. Firstly, if the trait is highly polygenic with contributing genes scattered across the genome, the high correlation between linkage positions will make it impossible to estimate the IBD sharing at null positions. The genomic control strategy should therefore only be considered with oligogenic traits. Secondly, the concept of genomic control relies on the assumption that the genotyping error rates are similar across markers. For markers with a similar degree of polymorphism (number of alleles and frequencies), this assumption might be acceptable. In a multipoint setting, an additional assumption required to ensure the validity of a genomic control strategy is that inter-marker distances be approximately equal. With microsatellite markers, both these assumptions might fail resulting in differences in the IBD sharing reduction across markers. The 'regression-based linkage testing' view allows one to qualitatively assess how deviation from these assumptions will impact linkage testing. For example, in ASP or EC designs, wrongly assuming that IBD is uniformly reduced across markers will result in inflated type I error at marker positions with low genotyping error rate compared to other markers. The advent of SNP chips in linkage mapping holds the promise of regular marker maps with less variable information content than in classical microsatellites maps [##REF##15311375##2##,##REF##15514889##3##]. The many SNPs used are likely to be subject to similar genotyping error processes, this makes the critical assumption of the genomic control strategy all the more plausible. Alternatives to this genomic-control strategy are possible and they also consist in constraining the linkage regression through a new origin as in the ad-hoc method, the estimation procedure can be adapted to suit particular circumstances. Firstly, in random samples, the assumption regarding exchangeability of positions might be relaxed. Indeed, the reduction in IBD sharing at each position may be used as estimates of the position-specific intercepts (a study sufficiently powered to detect linkage in random samples should have a huge sample size which would ensure sufficient precision of the position-specific intercepts). However, it must be stressed that the advantage of using a genomic control in random samples is limited because the impact of genotyping error is small in such designs. Secondly, one could use previous lab data to estimate by how much IBD sharing deviates from its expected value, this could also be done at each position separately provided sufficient data are available. In practice, such data might not be available or they might not trustfully reflect current error mechanisms.</p>", "<p>Elston et al. [##REF##15540158##18##] have pointed out that the implicit assumption made in ASP designs, that randomly sampled sib pairs share half of their alleles IBD, might not hold in practice and have argued for including discordant pairs in such studies. The genomic control approach suggested here may be an alternative solution to this issue. Finally we note that, although we have only considered designs involving sib pairs, the approach naturally extends to other types of relative pairs.</p>" ]

[ "<title>Conclusion</title>", "<p>Under realistic genotyping error scenarios, power losses observed in extremely concordant designs are modest but the effect on type I error in extremely discordant designs can be dramatic. Our analytic approach provides some understanding of the differences in influence of genotyping errors across study designs. The advent of SNP arrays does not eliminate the impact of genotyping errors but it makes genomic control a feasible option with the potential to deliver more robust inference in linkage analysis data subject to genotyping errors or other mechanisms distorting the IBD signal.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Despite the current trend towards large epidemiological studies of unrelated individuals, linkage studies in families are still thoroughly being utilized as tools for disease gene mapping. The use of the single-nucleotide-polymorphisms (SNP) array technology in genotyping of family data has the potential to provide more informative linkage data. Nevertheless, SNP array data are not immune to genotyping error which, as has been suggested in the past, could dramatically affect the evidence for linkage especially in selective designs such as affected sib pair (ASP) designs. The influence of genotyping error on selective designs for continuous traits has not been assessed yet.</p>", "<title>Results</title>", "<p>We use the identity-by-descent (IBD) regression-based paradigm for linkage testing to analytically quantify the effect of simple genotyping error models under specific selection schemes for sibling pairs. We show, for example, that in extremely concordant (EC) designs, genotyping error leads to decreased power whereas it leads to increased type I error in extremely discordant (ED) designs. Perhaps surprisingly, the effect of genotyping error on inference is most severe in designs where selection is least extreme. We suggest a genomic control for genotyping errors via a simple modification of the intercept in the regression for linkage.</p>", "<title>Conclusion</title>", "<p>This study extends earlier findings: genotyping error can substantially affect type I error and power in selective designs for continuous traits. Designs involving both EC and ED sib pairs are fairly immune to genotyping error. When those designs are not feasible the simple genomic control strategy that we suggest offers the potential to deliver more robust inference, especially if genotyping is carried out by SNP array technology.</p>" ]

[ "<title>Abbreviations</title>", "<p>ASP: affected sib pair; EC: extremely concordant; ED: extremely discordant; EDAC: extremely concordant and extremely discordant; IBD: identical-by-descent; QTL: quantitative trait locus; SNP: single-nucleotide-polymorphism.</p>", "<title>Authors' contributions</title>", "<p>JJPL participated in the method development, carried out the simulations summarized in Table ##TAB##0##1##, drafted and finalized the manuscript. HP participated in method development and in drafting the manuscript. JJH-D and HCvH both participated in method development. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>This paper originates from the GENOMEUTWIN project which is supported by the European Union Contract No. QLG2-CT-2002-01254. We are grateful to Dr. Bas Heijmans from the section Molecular Epidemiology, Dept. of Medical Statistics and Bioinformatics, Leiden University Medical Center for discussions on genotyping error mechanisms.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Bias in selective designs</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td align=\"center\" colspan=\"3\"><italic>A</italic></td><td align=\"center\" colspan=\"3\"></td><td align=\"center\" colspan=\"3\">var_con</td><td align=\"center\" colspan=\"3\">var_uncon</td></tr><tr><td/><td/><td colspan=\"3\"><hr/></td><td colspan=\"3\"><hr/></td><td colspan=\"3\"><hr/></td><td colspan=\"3\"><hr/></td></tr><tr><td align=\"center\">Selection</td><td align=\"center\"><italic>ρ</italic></td><td align=\"center\">EC</td><td align=\"center\">ED</td><td align=\"center\"></td><td align=\"center\">EC</td><td align=\"center\">ED</td><td align=\"center\"></td><td align=\"center\">EC</td><td align=\"center\">ED</td><td align=\"center\"></td><td align=\"center\">EC</td><td align=\"center\">ED</td><td align=\"center\"></td></tr></thead><tbody><tr><td align=\"center\">1%</td><td align=\"center\">0.1</td><td align=\"center\">0.27</td><td align=\"center\">-0.23</td><td align=\"center\">-0.07</td><td align=\"center\">3.45</td><td align=\"center\">-3.93</td><td align=\"center\">-1.61</td><td align=\"center\">0.08</td><td align=\"center\">0.06</td><td align=\"center\">0.04</td><td align=\"center\">1.13</td><td align=\"center\">0.68</td><td align=\"center\">0.05</td></tr><tr><td/><td align=\"center\">0.2</td><td align=\"center\">0.29</td><td align=\"center\">-0.21</td><td align=\"center\">-0.13</td><td align=\"center\">3.28</td><td align=\"center\">-4.25</td><td align=\"center\">-3.19</td><td align=\"center\">0.09</td><td align=\"center\">0.05</td><td align=\"center\">0.04</td><td align=\"center\">1.46</td><td align=\"center\">0.52</td><td align=\"center\">0.07</td></tr><tr><td/><td align=\"center\">0.3</td><td align=\"center\">0.30</td><td align=\"center\">-0.19</td><td align=\"center\">-0.15</td><td align=\"center\">3.15</td><td align=\"center\">-4.72</td><td align=\"center\">-4.63</td><td align=\"center\">0.10</td><td align=\"center\">0.04</td><td align=\"center\">0.03</td><td align=\"center\">1.82</td><td align=\"center\">0.38</td><td align=\"center\">0.10</td></tr><tr><td/><td align=\"center\">0.4</td><td align=\"center\">0.31</td><td align=\"center\">-0.17</td><td align=\"center\">-0.14</td><td align=\"center\">3.06</td><td align=\"center\">-5.29</td><td align=\"center\">-6.00</td><td align=\"center\">0.10</td><td align=\"center\">0.03</td><td align=\"center\">0.02</td><td align=\"center\">2.27</td><td align=\"center\">0.27</td><td align=\"center\">0.17</td></tr><tr><td/><td align=\"center\">0.5</td><td align=\"center\">0.32</td><td align=\"center\">-0.14</td><td align=\"center\">-0.12</td><td align=\"center\">3.01</td><td align=\"center\">-6.10</td><td align=\"center\">-7.44</td><td align=\"center\">0.11</td><td align=\"center\">0.02</td><td align=\"center\">0.02</td><td align=\"center\">2.38</td><td align=\"center\">0.18</td><td align=\"center\">0.23</td></tr><tr><td/><td align=\"center\">0.6</td><td align=\"center\">0.31</td><td align=\"center\">-0.12</td><td align=\"center\">-0.10</td><td align=\"center\">3.02</td><td align=\"center\">-7.33</td><td align=\"center\">-9.33</td><td align=\"center\">0.10</td><td align=\"center\">0.02</td><td align=\"center\">0.01</td><td align=\"center\">1.92</td><td align=\"center\">0.12</td><td align=\"center\">0.19</td></tr><tr><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"center\">10%</td><td align=\"center\">0.1</td><td align=\"center\">0.47</td><td align=\"center\">-0.40</td><td align=\"center\">-0.06</td><td align=\"center\">1.71</td><td align=\"center\">-1.87</td><td align=\"center\">-0.40</td><td align=\"center\">0.28</td><td align=\"center\">0.22</td><td align=\"center\">0.14</td><td align=\"center\">1.48</td><td align=\"center\">0.88</td><td align=\"center\">0.14</td></tr><tr><td/><td align=\"center\">0.2</td><td align=\"center\">0.50</td><td align=\"center\">-0.36</td><td align=\"center\">-0.11</td><td align=\"center\">1.66</td><td align=\"center\">-1.99</td><td align=\"center\">-0.81</td><td align=\"center\">0.30</td><td align=\"center\">0.18</td><td align=\"center\">0.13</td><td align=\"center\">1.84</td><td align=\"center\">0.66</td><td align=\"center\">0.14</td></tr><tr><td/><td align=\"center\">0.3</td><td align=\"center\">0.52</td><td align=\"center\">-0.32</td><td align=\"center\">-0.14</td><td align=\"center\">1.64</td><td align=\"center\">-2.14</td><td align=\"center\">-1.30</td><td align=\"center\">0.32</td><td align=\"center\">0.15</td><td align=\"center\">0.11</td><td align=\"center\">2.20</td><td align=\"center\">0.48</td><td align=\"center\">0.13</td></tr><tr><td/><td align=\"center\">0.4</td><td align=\"center\">0.53</td><td align=\"center\">-0.28</td><td align=\"center\">-0.16</td><td align=\"center\">1.63</td><td align=\"center\">-2.35</td><td align=\"center\">-1.86</td><td align=\"center\">0.32</td><td align=\"center\">0.12</td><td align=\"center\">0.09</td><td align=\"center\">2.37</td><td align=\"center\">0.35</td><td align=\"center\">0.12</td></tr><tr><td/><td align=\"center\">0.5</td><td align=\"center\">0.52</td><td align=\"center\">-0.24</td><td align=\"center\">-0.17</td><td align=\"center\">1.64</td><td align=\"center\">-2.61</td><td align=\"center\">-2.61</td><td align=\"center\">0.31</td><td align=\"center\">0.09</td><td align=\"center\">0.06</td><td align=\"center\">2.05</td><td align=\"center\">0.23</td><td align=\"center\">0.11</td></tr><tr><td/><td align=\"center\">0.6</td><td align=\"center\">0.47</td><td align=\"center\">-0.19</td><td align=\"center\">-0.15</td><td align=\"center\">1.68</td><td align=\"center\">-3.01</td><td align=\"center\">-3.64</td><td align=\"center\">0.28</td><td align=\"center\">0.06</td><td align=\"center\">0.04</td><td align=\"center\">1.33</td><td align=\"center\">0.15</td><td align=\"center\">0.10</td></tr><tr><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"center\">30%</td><td align=\"center\">0.1</td><td align=\"center\">0.65</td><td align=\"center\">-0.53</td><td align=\"center\">-0.04</td><td align=\"center\">0.96</td><td align=\"center\">-1.01</td><td align=\"center\">-0.11</td><td align=\"center\">0.68</td><td align=\"center\">0.52</td><td align=\"center\">0.31</td><td align=\"center\">1.80</td><td align=\"center\">1.12</td><td align=\"center\">0.32</td></tr><tr><td/><td align=\"center\">0.2</td><td align=\"center\">0.69</td><td align=\"center\">-0.46</td><td align=\"center\">-0.07</td><td align=\"center\">0.95</td><td align=\"center\">-1.03</td><td align=\"center\">-0.23</td><td align=\"center\">0.73</td><td align=\"center\">0.44</td><td align=\"center\">0.29</td><td align=\"center\">2.15</td><td align=\"center\">0.85</td><td align=\"center\">0.29</td></tr><tr><td/><td align=\"center\">0.3</td><td align=\"center\">0.71</td><td align=\"center\">-0.39</td><td align=\"center\">-0.09</td><td align=\"center\">0.96</td><td align=\"center\">-1.06</td><td align=\"center\">-0.36</td><td align=\"center\">0.74</td><td align=\"center\">0.37</td><td align=\"center\">0.25</td><td align=\"center\">2.33</td><td align=\"center\">0.63</td><td align=\"center\">0.26</td></tr><tr><td/><td align=\"center\">0.4</td><td align=\"center\">0.69</td><td align=\"center\">-0.32</td><td align=\"center\">-0.11</td><td align=\"center\">0.97</td><td align=\"center\">-1.13</td><td align=\"center\">-0.52</td><td align=\"center\">0.71</td><td align=\"center\">0.28</td><td align=\"center\">0.20</td><td align=\"center\">2.17</td><td align=\"center\">0.45</td><td align=\"center\">0.21</td></tr><tr><td/><td align=\"center\">0.5</td><td align=\"center\">0.62</td><td align=\"center\">-0.25</td><td align=\"center\">-0.11</td><td align=\"center\">1.00</td><td align=\"center\">-1.22</td><td align=\"center\">-0.73</td><td align=\"center\">0.62</td><td align=\"center\">0.21</td><td align=\"center\">0.15</td><td align=\"center\">1.64</td><td align=\"center\">0.30</td><td align=\"center\">0.16</td></tr><tr><td/><td align=\"center\">0.6</td><td align=\"center\">0.50</td><td align=\"center\">-0.19</td><td align=\"center\">-0.10</td><td align=\"center\">1.05</td><td align=\"center\">-1.35</td><td align=\"center\">-1.01</td><td align=\"center\">0.47</td><td align=\"center\">0.14</td><td align=\"center\">0.10</td><td align=\"center\">0.98</td><td align=\"center\">0.19</td><td align=\"center\">0.11</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Impact of genotyping error (rate = ϵ) on type I error and power</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\">Error rate ϵ</td><td align=\"center\">Selection</td><td align=\"center\"><italic>ρ</italic></td><td align=\"left\" colspan=\"2\">EC</td><td align=\"left\" colspan=\"2\">ED</td><td align=\"left\" colspan=\"2\"></td></tr><tr><td/><td/><td/><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td></tr><tr><td/><td/><td/><td align=\"center\">Power</td><td align=\"center\">Type I Error × 10^-4</td><td align=\"center\">Power</td><td align=\"center\">Type I Error × 10^-4</td><td align=\"center\">Power</td><td align=\"center\">Type I Error × 10^-4</td></tr></thead><tbody><tr><td align=\"center\">0.005</td><td align=\"center\">1%</td><td align=\"center\">0.1</td><td align=\"center\">0.87</td><td align=\"center\">0.6</td><td align=\"center\">0.92</td><td align=\"center\">1.6</td><td align=\"center\">0.90</td><td align=\"center\">1.1</td></tr><tr><td/><td/><td align=\"center\">0.2</td><td align=\"center\">0.87</td><td align=\"center\">0.6</td><td align=\"center\">0.92</td><td align=\"center\">1.5</td><td align=\"center\">0.91</td><td align=\"center\">1.3</td></tr><tr><td/><td/><td align=\"center\">0.3</td><td align=\"center\">0.87</td><td align=\"center\">0.5</td><td align=\"center\">0.91</td><td align=\"center\">1.5</td><td align=\"center\">0.91</td><td align=\"center\">1.3</td></tr><tr><td/><td/><td align=\"center\">0.4</td><td align=\"center\">0.87</td><td align=\"center\">0.5</td><td align=\"center\">0.91</td><td align=\"center\">1.4</td><td align=\"center\">0.91</td><td align=\"center\">1.3</td></tr><tr><td/><td/><td align=\"center\">0.5</td><td align=\"center\">0.87</td><td align=\"center\">0.5</td><td align=\"center\">0.91</td><td align=\"center\">1.4</td><td align=\"center\">0.91</td><td align=\"center\">1.3</td></tr><tr><td/><td/><td align=\"center\">0.6</td><td align=\"center\">0.87</td><td align=\"center\">0.5</td><td align=\"center\">0.91</td><td align=\"center\">1.3</td><td align=\"center\">0.91</td><td align=\"center\">1.2</td></tr><tr><td/><td align=\"center\">10%</td><td align=\"center\">0.1</td><td align=\"center\">0.85</td><td align=\"center\">0.4</td><td align=\"center\">0.93</td><td align=\"center\">2.2</td><td align=\"center\">0.90</td><td align=\"center\">1.1</td></tr><tr><td/><td/><td align=\"center\">0.2</td><td align=\"center\">0.85</td><td align=\"center\">0.4</td><td align=\"center\">0.93</td><td align=\"center\">2.0</td><td align=\"center\">0.91</td><td align=\"center\">1.2</td></tr><tr><td/><td/><td align=\"center\">0.3</td><td align=\"center\">0.84</td><td align=\"center\">0.3</td><td align=\"center\">0.92</td><td align=\"center\">1.9</td><td align=\"center\">0.91</td><td align=\"center\">1.3</td></tr><tr><td/><td/><td align=\"center\">0.4</td><td align=\"center\">0.84</td><td align=\"center\">0.3</td><td align=\"center\">0.92</td><td align=\"center\">1.7</td><td align=\"center\">0.91</td><td align=\"center\">1.4</td></tr><tr><td/><td/><td align=\"center\">0.5</td><td align=\"center\">0.84</td><td align=\"center\">0.3</td><td align=\"center\">0.92</td><td align=\"center\">1.6</td><td align=\"center\">0.91</td><td align=\"center\">1.4</td></tr><tr><td/><td/><td align=\"center\">0.6</td><td align=\"center\">0.85</td><td align=\"center\">0.4</td><td align=\"center\">0.91</td><td align=\"center\">1.5</td><td align=\"center\">0.91</td><td align=\"center\">1.3</td></tr><tr><td/><td align=\"center\">30%</td><td align=\"center\">0.1</td><td align=\"center\">0.83</td><td align=\"center\">0.3</td><td align=\"center\">0.94</td><td align=\"center\">2.8</td><td align=\"center\">0.90</td><td align=\"center\">1.1</td></tr><tr><td/><td/><td align=\"center\">0.2</td><td align=\"center\">0.82</td><td align=\"center\">0.2</td><td align=\"center\">0.93</td><td align=\"center\">2.4</td><td align=\"center\">0.90</td><td align=\"center\">1.1</td></tr><tr><td/><td/><td align=\"center\">0.3</td><td align=\"center\">0.82</td><td align=\"center\">0.2</td><td align=\"center\">0.93</td><td align=\"center\">2.1</td><td align=\"center\">0.91</td><td align=\"center\">1.2</td></tr><tr><td/><td/><td align=\"center\">0.4</td><td align=\"center\">0.82</td><td align=\"center\">0.2</td><td align=\"center\">0.92</td><td align=\"center\">1.9</td><td align=\"center\">0.91</td><td align=\"center\">1.2</td></tr><tr><td/><td/><td align=\"center\">0.5</td><td align=\"center\">0.83</td><td align=\"center\">0.3</td><td align=\"center\">0.92</td><td align=\"center\">1.6</td><td align=\"center\">0.91</td><td align=\"center\">1.2</td></tr><tr><td/><td/><td align=\"center\">0.6</td><td align=\"center\">0.85</td><td align=\"center\">0.4</td><td align=\"center\">0.91</td><td align=\"center\">1.5</td><td align=\"center\">0.91</td><td align=\"center\">1.2</td></tr><tr><td/><td/><td/><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"center\">0.01</td><td align=\"center\">1%</td><td align=\"center\">0.1</td><td align=\"center\">0.84</td><td align=\"center\">0.3</td><td align=\"center\">0.93</td><td align=\"center\">2.4</td><td align=\"center\">0.91</td><td align=\"center\">1.3</td></tr><tr><td/><td/><td align=\"center\">0.2</td><td align=\"center\">0.83</td><td align=\"center\">0.3</td><td align=\"center\">0.93</td><td align=\"center\">2.2</td><td align=\"center\">0.92</td><td align=\"center\">1.7</td></tr><tr><td/><td/><td align=\"center\">0.3</td><td align=\"center\">0.83</td><td align=\"center\">0.3</td><td align=\"center\">0.93</td><td align=\"center\">2.1</td><td align=\"center\">0.92</td><td align=\"center\">1.8</td></tr><tr><td/><td/><td align=\"center\">0.4</td><td align=\"center\">0.83</td><td align=\"center\">0.3</td><td align=\"center\">0.92</td><td align=\"center\">1.9</td><td align=\"center\">0.92</td><td align=\"center\">1.7</td></tr><tr><td/><td/><td align=\"center\">0.5</td><td align=\"center\">0.83</td><td align=\"center\">0.3</td><td align=\"center\">0.92</td><td align=\"center\">1.7</td><td align=\"center\">0.92</td><td align=\"center\">1.6</td></tr><tr><td/><td/><td align=\"center\">0.6</td><td align=\"center\">0.83</td><td align=\"center\">0.3</td><td align=\"center\">0.92</td><td align=\"center\">1.6</td><td align=\"center\">0.91</td><td align=\"center\">1.5</td></tr><tr><td/><td align=\"center\">10%</td><td align=\"center\">0.1</td><td align=\"center\">0.78</td><td align=\"center\">0.1</td><td align=\"center\">0.95</td><td align=\"center\">4.5</td><td align=\"center\">0.91</td><td align=\"center\">1.3</td></tr><tr><td/><td/><td align=\"center\">0.2</td><td align=\"center\">0.78</td><td align=\"center\">0.1</td><td align=\"center\">0.95</td><td align=\"center\">3.9</td><td align=\"center\">0.91</td><td align=\"center\">1.5</td></tr><tr><td/><td/><td align=\"center\">0.3</td><td align=\"center\">0.77</td><td align=\"center\">0.1</td><td align=\"center\">0.94</td><td align=\"center\">3.4</td><td align=\"center\">0.92</td><td align=\"center\">1.7</td></tr><tr><td/><td/><td align=\"center\">0.4</td><td align=\"center\">0.77</td><td align=\"center\">0.1</td><td align=\"center\">0.94</td><td align=\"center\">2.9</td><td align=\"center\">0.92</td><td align=\"center\">1.9</td></tr><tr><td/><td/><td align=\"center\">0.5</td><td align=\"center\">0.77</td><td align=\"center\">0.1</td><td align=\"center\">0.93</td><td align=\"center\">2.5</td><td align=\"center\">0.92</td><td align=\"center\">1.9</td></tr><tr><td/><td/><td align=\"center\">0.6</td><td align=\"center\">0.78</td><td align=\"center\">0.1</td><td align=\"center\">0.93</td><td align=\"center\">2.1</td><td align=\"center\">0.92</td><td align=\"center\">1.8</td></tr><tr><td/><td align=\"center\">30%</td><td align=\"center\">0.1</td><td align=\"center\">0.73</td><td align=\"center\">0.1</td><td align=\"center\">0.96</td><td align=\"center\">7.1</td><td align=\"center\">0.90</td><td align=\"center\">1.2</td></tr><tr><td/><td/><td align=\"center\">0.2</td><td align=\"center\">0.71</td><td align=\"center\">0.1</td><td align=\"center\">0.96</td><td align=\"center\">5.6</td><td align=\"center\">0.91</td><td align=\"center\">1.3</td></tr><tr><td/><td/><td align=\"center\">0.3</td><td align=\"center\">0.71</td><td align=\"center\">0.0</td><td align=\"center\">0.95</td><td align=\"center\">4.4</td><td align=\"center\">0.91</td><td align=\"center\">1.4</td></tr><tr><td/><td/><td align=\"center\">0.4</td><td align=\"center\">0.71</td><td align=\"center\">0.1</td><td align=\"center\">0.94</td><td align=\"center\">3.4</td><td align=\"center\">0.91</td><td align=\"center\">1.5</td></tr><tr><td/><td/><td align=\"center\">0.5</td><td align=\"center\">0.74</td><td align=\"center\">0.1</td><td align=\"center\">0.93</td><td align=\"center\">2.6</td><td align=\"center\">0.91</td><td align=\"center\">1.5</td></tr><tr><td/><td/><td align=\"center\">0.6</td><td align=\"center\">0.78</td><td align=\"center\">0.1</td><td align=\"center\">0.93</td><td align=\"center\">2.1</td><td align=\"center\">0.91</td><td align=\"center\">1.5</td></tr></tbody></table></table-wrap>" ]

[ "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1471-2156-9-57-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mi>ϵ</mml:mi><mml:mn>2</mml:mn></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M2\" name=\"1471-2156-9-57-i2\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>P</mml:mi></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:msup><mml:mi>π</mml:mi><mml:mi>ϵ</mml:mi></mml:msup><mml:mo>|</mml:mo><mml:mi>π</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munder><mml:mo>∑</mml:mo><mml:mrow><mml:msup><mml:mi>G</mml:mi><mml:mi>ϵ</mml:mi></mml:msup></mml:mrow></mml:munder><mml:mrow><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>P</mml:mi></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:msup><mml:mi>π</mml:mi><mml:mi>ϵ</mml:mi></mml:msup><mml:mo>|</mml:mo><mml:msup><mml:mi>G</mml:mi><mml:mi>ϵ</mml:mi></mml:msup><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mstyle><mml:mstyle displaystyle=\"true\"><mml:munder><mml:mo>∑</mml:mo><mml:mi>G</mml:mi></mml:munder><mml:mrow><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>P</mml:mi></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:msup><mml:mi>G</mml:mi><mml:mi>ϵ</mml:mi></mml:msup><mml:mo>|</mml:mo><mml:mi>G</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mtext> </mml:mtext><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>P</mml:mi></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>G</mml:mi><mml:mo>|</mml:mo><mml:mi>π</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mstyle><mml:mo>.</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M3\" name=\"1471-2156-9-57-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mi>ϵ</mml:mi><mml:mn>2</mml:mn></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M4\" name=\"1471-2156-9-57-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mi>ϵ</mml:mi><mml:mn>2</mml:mn></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M5\" name=\"1471-2156-9-57-i3\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>P</mml:mi></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:msup><mml:mi>π</mml:mi><mml:mi>ϵ</mml:mi></mml:msup><mml:mo>|</mml:mo><mml:mi>π</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mn>1</mml:mn></mml:mtd><mml:mtd><mml:mn>0</mml:mn></mml:mtd><mml:mtd><mml:mn>0</mml:mn></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mfrac><mml:mi>ϵ</mml:mi><mml:mn>2</mml:mn></mml:mfrac></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mfrac><mml:mi>ϵ</mml:mi><mml:mn>2</mml:mn></mml:mfrac></mml:mrow></mml:mtd><mml:mtd><mml:mn>0</mml:mn></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mn>0</mml:mn></mml:mtd><mml:mtd><mml:mi>ϵ</mml:mi></mml:mtd><mml:mtd><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>ϵ</mml:mi></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mo>.</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M6\" name=\"1471-2156-9-57-i4\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M7\" name=\"1471-2156-9-57-i5\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>var</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:msup><mml:mi>π</mml:mi><mml:mi>ϵ</mml:mi></mml:msup><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mn>8</mml:mn></mml:mfrac><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mrow><mml:mn>16</mml:mn></mml:mrow></mml:mfrac><mml:msup><mml:mi>ϵ</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M8\" name=\"1471-2156-9-57-i6\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mn>1</mml:mn></mml:mtd><mml:mtd><mml:mrow><mml:mi>γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>π</mml:mi><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>+</mml:mo><mml:mi>ρ</mml:mi></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mi>γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>π</mml:mi><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>+</mml:mo><mml:mi>ρ</mml:mi></mml:mrow></mml:mtd><mml:mtd><mml:mn>1</mml:mn></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mo>,</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M9\" name=\"1471-2156-9-57-i4\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM1\"><label>(1)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M10\" name=\"1471-2156-9-57-i7\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>C</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mn>1</mml:mn></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mn>2</mml:mn></mml:msub><mml:mo>,</mml:mo><mml:mi>ρ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>+</mml:mo><mml:msup><mml:mi>ρ</mml:mi><mml:mn>2</mml:mn></mml:msup><mml:mo stretchy=\"false\">)</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mn>1</mml:mn></mml:msub><mml:msub><mml:mi>x</mml:mi><mml:mn>2</mml:mn></mml:msub><mml:mo>−</mml:mo><mml:mi>ρ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:msubsup><mml:mi>x</mml:mi><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:msubsup><mml:mo>+</mml:mo><mml:msubsup><mml:mi>x</mml:mi><mml:mn>2</mml:mn><mml:mn>2</mml:mn></mml:msubsup><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>+</mml:mo><mml:mi>ρ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:msup><mml:mi>ρ</mml:mi><mml:mn>2</mml:mn></mml:msup><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:msup><mml:mi>ρ</mml:mi><mml:mn>2</mml:mn></mml:msup><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfrac><mml:mo>.</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM2\"><label>(2)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M11\" name=\"1471-2156-9-57-i8\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>E</mml:mi></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>π</mml:mi><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac><mml:mo>|</mml:mo><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>x</mml:mi></mml:mstyle><mml:mo>,</mml:mo><mml:mi>γ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mi>γ</mml:mi><mml:mn>8</mml:mn></mml:mfrac><mml:mi>C</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>x</mml:mi></mml:mstyle><mml:mo>,</mml:mo><mml:mi>ρ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>,</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M12\" name=\"1471-2156-9-57-i9\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mn>1</mml:mn><mml:mn>8</mml:mn></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M13\" name=\"1471-2156-9-57-i10\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mn>8</mml:mn><mml:mfrac><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac><mml:mo stretchy=\"false\">)</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle></mml:mrow><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:msubsup><mml:mi>C</mml:mi><mml:mi>i</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:mstyle></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M14\" name=\"1471-2156-9-57-i11\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M15\" name=\"1471-2156-9-57-i11\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M16\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M17\" name=\"1471-2156-9-57-i13\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi mathvariant=\"script\">I</mml:mi><mml:mo>=</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mn>8</mml:mn></mml:mfrac><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:msubsup><mml:mi>C</mml:mi><mml:mi>i</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M18\" name=\"1471-2156-9-57-i14\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:msqrt><mml:mi mathvariant=\"script\">I</mml:mi></mml:msqrt></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M19\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M20\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M21\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM3\"><label>(3)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M22\" name=\"1471-2156-9-57-i15\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>E</mml:mi></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:msup><mml:mi>π</mml:mi><mml:mi>ϵ</mml:mi></mml:msup><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac><mml:mo>|</mml:mo><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>x</mml:mi></mml:mstyle><mml:mo>,</mml:mo><mml:mi>γ</mml:mi><mml:mo>,</mml:mo><mml:mi>ϵ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mo>−</mml:mo><mml:mfrac><mml:mi>ϵ</mml:mi><mml:mn>4</mml:mn></mml:mfrac><mml:mo>+</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mfrac><mml:mi>ϵ</mml:mi><mml:mn>2</mml:mn></mml:mfrac><mml:mo stretchy=\"false\">)</mml:mo><mml:mfrac><mml:mi>γ</mml:mi><mml:mn>8</mml:mn></mml:mfrac><mml:mi>C</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>x</mml:mi></mml:mstyle><mml:mo>,</mml:mo><mml:mi>ρ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>.</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M23\" name=\"1471-2156-9-57-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mi>ϵ</mml:mi><mml:mn>2</mml:mn></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M24\" name=\"1471-2156-9-57-i16\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mi>ϵ</mml:mi><mml:mn>4</mml:mn></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M25\" name=\"1471-2156-9-57-i17\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msup><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mi>ϵ</mml:mi></mml:msup><mml:mo>=</mml:mo><mml:mn>8</mml:mn><mml:mfrac><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac><mml:mo stretchy=\"false\">)</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle></mml:mrow><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:msubsup><mml:mi>C</mml:mi><mml:mi>i</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:mstyle></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M26\" name=\"1471-2156-9-57-i18\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtext>bias</mml:mtext><mml:mo stretchy=\"false\">(</mml:mo><mml:msup><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mi>ϵ</mml:mi></mml:msup><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>E</mml:mi></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:msup><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mi>ϵ</mml:mi></mml:msup><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>−</mml:mo><mml:mi>γ</mml:mi><mml:mo>=</mml:mo><mml:mo>−</mml:mo><mml:mi>ϵ</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mfrac><mml:mi>γ</mml:mi><mml:mn>2</mml:mn></mml:mfrac><mml:mo>+</mml:mo><mml:mn>2</mml:mn><mml:mi>A</mml:mi></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M27\" name=\"1471-2156-9-57-i19\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>A</mml:mi><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:msub><mml:mi>C</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle></mml:mrow><mml:mrow><mml:msubsup><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:mo>∑</mml:mo><mml:mi>C</mml:mi></mml:mstyle></mml:mrow><mml:mi>i</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover><mml:mrow><mml:mover accent=\"true\"><mml:mrow><mml:msup><mml:mi>C</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M28\" name=\"1471-2156-9-57-i20\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msup><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mi>ϵ</mml:mi></mml:msup><mml:msup><mml:mi mathvariant=\"script\">I</mml:mi><mml:mrow><mml:mn>1</mml:mn><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM4\"><label>(4)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M29\" name=\"1471-2156-9-57-i21\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>Φ</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msup><mml:mi>Φ</mml:mi><mml:mrow><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:msup><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>+</mml:mo><mml:mi>γ</mml:mi><mml:msup><mml:mi mathvariant=\"script\">I</mml:mi><mml:mrow><mml:mn>1</mml:mn><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:msup><mml:mo>+</mml:mo><mml:mtext>bias</mml:mtext><mml:mo stretchy=\"false\">(</mml:mo><mml:msup><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mi>ϵ</mml:mi></mml:msup><mml:mo stretchy=\"false\">)</mml:mo><mml:msup><mml:mi mathvariant=\"script\">I</mml:mi><mml:mrow><mml:mn>1</mml:mn><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:msup></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mo>.</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M30\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M31\" name=\"1471-2156-9-57-i22\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtext>bias</mml:mtext><mml:mo stretchy=\"false\">(</mml:mo><mml:msup><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mi>ϵ</mml:mi></mml:msup><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mo>−</mml:mo><mml:mi>ϵ</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mfrac><mml:mi>γ</mml:mi><mml:mn>2</mml:mn></mml:mfrac><mml:mo>+</mml:mo><mml:mn>2</mml:mn><mml:mi>A</mml:mi></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M32\" name=\"1471-2156-9-57-i23\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>A</mml:mi><mml:mo>=</mml:mo><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover><mml:mo>/</mml:mo><mml:mover accent=\"true\"><mml:mrow><mml:msup><mml:mi>C</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M33\" name=\"1471-2156-9-57-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M34\" name=\"1471-2156-9-57-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M35\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M36\" name=\"1471-2156-9-57-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M37\" name=\"1471-2156-9-57-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M38\" name=\"1471-2156-9-57-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M39\" name=\"1471-2156-9-57-i25\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mrow><mml:mi>var</mml:mi></mml:mrow><mml:mrow><mml:mtext>con</mml:mtext></mml:mrow></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mn>1</mml:mn><mml:mo>/</mml:mo><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:msubsup><mml:mi>C</mml:mi><mml:mi>i</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M40\" name=\"1471-2156-9-57-i26\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mrow><mml:mi>var</mml:mi></mml:mrow><mml:mrow><mml:mtext>uncon</mml:mtext></mml:mrow></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mn>1</mml:mn><mml:mo>/</mml:mo><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M41\" name=\"1471-2156-9-57-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M42\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M43\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M44\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M45\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M46\" name=\"1471-2156-9-57-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M47\" name=\"1471-2156-9-57-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M48\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M49\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M50\" name=\"1471-2156-9-57-i4\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M51\" name=\"1471-2156-9-57-i27\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>a</mml:mi><mml:mo>ˆ</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M52\" name=\"1471-2156-9-57-i28\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>y</mml:mi><mml:mi>t</mml:mi></mml:msub><mml:mo>≡</mml:mo><mml:mover accent=\"true\"><mml:mrow><mml:msubsup><mml:mi>π</mml:mi><mml:mi>t</mml:mi><mml:mi>ϵ</mml:mi></mml:msubsup></mml:mrow><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover><mml:mo>−</mml:mo><mml:mn>0.5</mml:mn></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M53\" name=\"1471-2156-9-57-i29\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>E</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>y</mml:mi><mml:mi>t</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>≃</mml:mo><mml:mrow><mml:mo>{</mml:mo><mml:mrow><mml:mtable columnalign=\"left\"><mml:mtr columnalign=\"left\"><mml:mtd columnalign=\"left\"><mml:mrow><mml:mi>a</mml:mi><mml:mo>,</mml:mo></mml:mrow></mml:mtd><mml:mtd columnalign=\"left\"><mml:mrow><mml:mtext>if position </mml:mtext><mml:mi>t</mml:mi><mml:mtext> is unlinked</mml:mtext><mml:mo>,</mml:mo></mml:mrow></mml:mtd></mml:mtr><mml:mtr columnalign=\"left\"><mml:mtd columnalign=\"left\"><mml:mrow><mml:mi>a</mml:mi><mml:mo>+</mml:mo><mml:mfrac><mml:mi>b</mml:mi><mml:mn>8</mml:mn></mml:mfrac><mml:mi>γ</mml:mi><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover><mml:mo>,</mml:mo></mml:mrow></mml:mtd><mml:mtd columnalign=\"left\"><mml:mrow><mml:mtext>if position </mml:mtext><mml:mi>t</mml:mi><mml:mtext> is linked</mml:mtext><mml:mo>.</mml:mo></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow></mml:mrow></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M54\" name=\"1471-2156-9-57-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M55\" name=\"1471-2156-9-57-i27\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>a</mml:mi><mml:mo>ˆ</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M56\" name=\"1471-2156-9-57-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M57\" name=\"1471-2156-9-57-i27\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>a</mml:mi><mml:mo>ˆ</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M58\" name=\"1471-2156-9-57-i27\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>a</mml:mi><mml:mo>ˆ</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M59\" name=\"1471-2156-9-57-i12\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">I</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M60\" name=\"1471-2156-9-57-i23\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>A</mml:mi><mml:mo>=</mml:mo><mml:mover accent=\"true\"><mml:mi>C</mml:mi><mml:mo>¯</mml:mo></mml:mover><mml:mo>/</mml:mo><mml:mover accent=\"true\"><mml:mrow><mml:msup><mml:mi>C</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M61\" name=\"1471-2156-9-57-i30\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>π</mml:mi><mml:mo>ˆ</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M62\" name=\"1471-2156-9-57-i31\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>E</mml:mi></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:mover accent=\"true\"><mml:mi>π</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac><mml:mo>|</mml:mo><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>x</mml:mi></mml:mstyle><mml:mo>,</mml:mo><mml:mi>γ</mml:mi><mml:mo>,</mml:mo><mml:mi>ϵ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>≃</mml:mo><mml:msub><mml:mrow><mml:mi>var</mml:mi><mml:mo>⁡</mml:mo></mml:mrow><mml:mn>0</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mover accent=\"true\"><mml:mi>π</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>γ</mml:mi><mml:mtext> </mml:mtext><mml:mi>C</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mstyle mathsize=\"normal\" mathvariant=\"bold\"><mml:mi>x</mml:mi></mml:mstyle><mml:mo>,</mml:mo><mml:mi>ρ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M63\" name=\"1471-2156-9-57-i32\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mrow><mml:mi>var</mml:mi></mml:mrow><mml:mn>0</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mover accent=\"true\"><mml:mi>π</mml:mi><mml:mo>ˆ</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>&lt;</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mn>8</mml:mn></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>Left-hand side: Values of <italic>A</italic>, quantities influencing the effect of genotyping error for a variety of selective designs indexed by degree of Selection and sib-sib trait correlation <italic>ρ</italic>) – Right-hand side: Comparison of efficiency in constrained and unconstrained regressions – See text for definitions of <italic>A</italic>, , var_conand var_uncon</p></table-wrap-foot>", "<table-wrap-foot><p>Impact of genotyping error (rate = ϵ) on power and type I error of linkage test in selective designs (indexed by degree of Selection and sib-sib trait correlation <italic>ρ</italic>) – Nominal error rate = 10^-4, QTL effect <italic>γ </italic>= 0.1 and sample size equivalent to a Fisher's information = 2500 in each design (provides 90% power in absence of genotyping error)</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Two crises dominate the health sectors of sub-Saharan countries: those of human resources and of HIV. In principle, both these crises magnify each other. HIV places a significant additional load on the health workforce and contributes to attrition from it through illness, caring for family members who have developed AIDS and death. And the impact of the HIV crisis is accentuated because health workers are unavailable to implement anti-HIV interventions.</p>", "<p>A particular source of recent concern has been the impact on workforce distribution of increased levels of support for HIV/AIDS programmes and especially treatment. This paper seeks to explore this interaction in more detail. It reviews the continent-wide distribution of the two phenomena and initial evidence of the impact of expanded treatment programmes, before looking in depth at the case of Malawi, a country with one of the lowest densities of human resources for health and one of the highest prevalence rates of HIV.</p>" ]

[ "<title>Methods</title>", "<p>This paper is based on data derived from published literature; the global atlas of the health workforce, a database compiled by the World Health Organisation (WHO); and grey literature, particularly concerning Malawi. In addition, one of the authors (DM) was part of a nine member external team established by the Government of Malawi and the UK Department for International Development to evaluate the country's antiretroviral therapy (ART) programme in September 2006. During the evaluation a number of health facilities were visited and informal interviews and discussions with service providers, managers and policy makers were conducted.</p>" ]

[]

[]

[ "<title>Conclusion</title>", "<p>Malawi is one of a few countries in extreme crisis in relation to both human resources and HIV. Despite this, it is making remarkable progress both in tackling the causes of human resource problems and in providing antiretroviral therapy. The case highlights how both synergies and conflicts between the two strategies have been realised in this context out of the many possible configurations that could be theoretically predicted.</p>", "<p>Among the evident synergies is the contribution of the EHRP to increasing the availability of staff for the expanding ART programme both in total and in underserved areas. This is consistent with the growing acknowledgement that the basic human infrastructure of health systems in SSA must be strengthened if global ART coverage targets are to be met.</p>", "<p>However, while Malawi has been fortunate enough to receive donor support for its EHRP, it is clear that more needs to be done. Donors and the government still need to secure additional resources to increase staff recruitment and training, and to improve retention if WHO's minimum HRH standards are to be met. In addition greater attention will need to be paid to the dynamics of the domestic labour market and the disparities between private and public sector remuneration if Malawi is to ensure a more rational and needs-based deployment of health workers across the country.</p>", "<p>The impact of external project funding on workforce balance has been suggested to lead to the need for health workforce impact assessments as part of project appraisal: 'It could be envisaged that at country level, public and private health services, NGOs and international agencies that would like to start up a new programme or activity would have to demonstrate the impact of their plan on the current health workforce to the Ministry of Health. Similarly, organizations applying for funding at international donor agencies would be asked the same' [##UREF##32##38##].</p>", "<p>The adoption of a low-resource model of ART provision, underscored by task shifting and the use of clinical officers and nurses to diagnose and treat AIDS patients is also consistent with protecting staff availability to also meet other demands on the health system. Moreover, the effective delivery of ART should reduce demands on the health care system for the treatment of opportunistic infections and end-of-life care, although this may only be a short term impact. As the ART programme matures, the number of treatment failures to first-line regimens will grow and unless there is the capacity to fund and supervise the use of second-line treatment, the health care system could see a 'rebound effect' of returning AIDS patients.</p>", "<p>Another notable achievement of the Malawi ART programme has been its ability to incorporate private sector providers into the national ART programme. This has been achieved through a quid pro quo arrangement whereby private sector providers agree to adopt national treatment and monitoring policies in exchange for medicines that are mainly paid for by the public purse.</p>", "<p>The potential for synergy in the strengthening of drug supply systems has not yet been realised. The procurement and supply system for ARVs described earlier operates in isolation from the procurement and supply system for other medicines and commodities, which includes a number of other stand-alone, vertical systems.</p>", "<p>Among the evident conflicts are the demands placed on service delivery by additional in-service training activities and the conflicts between dedicating resources to ensuring the excellence of the antiretroviral therapy programme, and ensuring excellence in prevention – especially of mother-to-child transmission which could reduce the demands for a paediatric antiretroviral therapy programme. There is perhaps a more important tension between the patient-centred focus required for ART, and the community centred approach required for prevention programmes that target the broader determinants of HIV infection and other health problems. This may prove more difficult to resolve.</p>", "<p>Planning for the rapid scale-up of antiretroviral therapy whilst simultaneously strengthening health systems and delivering the broader EHP will require prioritizing amongst the different health care needs and careful consideration of coverage targets and timeframes set by the antiretroviral therapy programme because it takes time (not just resources) to develop positive synergies between antiretroviral therapy scale-up and health systems strengthening. It will also require the right balance between the use of 'dedicated' antiretroviral therapy cadres of health workers and an antiretroviral therapy programme operated through 'generalist' health workers capable of providing comprehensive care. The former may result in faster and more effective antiretroviral therapy coverage, but the latter may be more sustainable and be less harmful to other health care services.</p>", "<p>A greater emphasis on sustainability and health systems strengthening may compromise the speed of up-scaling in the short-term but could derive greater benefits in the long-term by, for example, ensuring a high level of treatment adherence, keeping patients on first line treatment for longer and deferring the use of more expensive second-line treatments.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<p>Two crises dominate the health sectors of sub-Saharan African countries: those of human resources and of HIV. Nevertheless, there is considerable variation in the extent to which these two phenomena affect sub-Saharan countries, with a few facing extreme levels of both: Lesotho, Zimbabwe, Zambia, Mozambique, the Central African Republic and Malawi.</p>", "<p>This paper reviews the continent-wide situation with respect to this double burden before considering the case of Malawi in more detail. In Malawi, there has been significant concurrent investment in both an Emergency Human Resource Programme and an antiretroviral therapy programme which was treating 60,000 people by the end of 2006. Both areas of synergy and conflict have arisen, as the two programmes have been implemented. These highlight important issues for programme planners and managers to address and emphasize that planning for the scale-up of antiretroviral therapy while simultaneously strengthening health systems and the human resource situation requires prioritization among compelling cases for support, and time (not just resources).</p>" ]

[ "<title>Findings</title>", "<title>The twin human resource and HIV burden</title>", "<p>The 2006 World Health Report (WHR) defined health workers as 'the people whose job it is to protect and improve the health of their communities' [##REF##17178522##1##]. While recognising the important role of unpaid carers such as mothers and voluntary health workers, its analysis is restricted to people engaged in paid activities. Among those, two categories are identified: 'health service providers' who deliver services; and 'health management and support workers' who are not engaged in any direct provision of services. Table ##TAB##0##1##, reproduced from the WHR, summarises data on the availability of health workers by region and by the categories mentioned above. It suggests that regions with more health workers have proportionately more managerial and support workers. However, better data are required before any conclusions can be made about the number and relative availability of 'health management and support workers'.</p>", "<p>The WHR also identified those countries with a 'critical shortage' of health workers (see Figure ##FIG##0##1##). Critical shortage was defined as having less than 2.28 doctors, nurses and midwives per 1000 population, a threshold derived from an analysis of workforce density associated with key public health outcomes by the Joint Learning Initiative [##UREF##0##2##] (see Table ##TAB##0##1##).</p>", "<p>The WHR suggest that there are critical shortages of health workers in many countries. The absolute shortage is greatest in Asia, where there is a shortfall of 1.16 million doctors, nurses and midwives and perhaps 2.1 million of all types of health workers, dominated by the shortages in Bangladesh, India and Indonesia. The relative shortage is greatest in sub-Saharan Africa where an increase of 139% is required [##REF##17178522##1##]. The countries with the lowest ratios of health workers per 1000 population are mainly in sub-Saharan Africa but some others such as Indonesia and Papua New Guinea also have densities below one half of the proposed critical shortage threshold.</p>", "<p>The focus on doctors, nurses and midwives reflects the greater reliability of estimated numbers of these cadres. In practice, the contribution of other cadres such as pharmacists, laboratory technicians and 'non-physician clinicians', is just as critical. Indeed, 'non-physician clinicians' (often known as 'clinical officers' or 'medical assistants') have been trained in some countries to compensate for the lack of doctors and are active in 25 of 47 sub-Saharan African countries included in a recent study [##UREF##1##3##]. In nine countries there are more non-physician clinicians than physicians and they are reported to play prominent roles in primary health care and HIV/AIDS treatment in five of the worst affected sub-Saharan countries. However, in no country do they add more than 0.2 to the health worker per thousand population ratio, so they do not significantly alter the relative position of different countries from WHO's analyses.</p>", "<p>Further analysis of data from WHO's global atlas of the health workforce identifies the countries in Table ##TAB##1##2## as having ratios of doctors, nurses and midwives lower than 0.5 per 1000 population. All of these are in sub-Saharan Africa. Malawi has a slightly higher overall health worker density than these countries at 0.61 per thousand population, but a physician density level as low as the least well served of these countries (Niger) at 0.02 per thousand population [##UREF##1##3##] (see Table ##TAB##1##2##).</p>", "<p>The focus on nurses, doctors and midwives also runs the risk of neglecting the importance of 'health management and support workers'. Clinical workers require management and administrative systems to work if they are to be effective. And ART programmes require, in particular, effective drug procurement and supply systems, laboratory support and information management.</p>", "<p>When countries with low HR levels are assessed in terms of HIV prevalence, all the non-African countries with a critical human resource shortage are found to have relatively low adult HIV prevalence rates. According to UNAIDS [##UREF##2##4##], adult HIV prevalence ranges from less than 0.1 to 1.6% in these countries except Haiti, where prevalence is 3.8%. The twin burden of HRH crisis and HIV/AIDS crisis is therefore an African phenomenon. Figure ##FIG##1##2## plots total numbers of doctors, nurses and midwives against adult HIV prevalence across all African countries for which both statistics are available. It identifies 6 countries with an HRH crisis as defined by WHO and with adult HIV prevalence rates greater than 10%. These are Lesotho, Zimbabwe, Zambia, Mozambique, the Central African Republic and Malawi (see Figure ##FIG##1##2##).</p>", "<p>Hirschhorn et al. [##REF##16438710##5##] estimated that the additional health workforce required to deliver ART to 1000 patients amounted to 1–2 physicians, 2–7 nurses, &lt;1 to 3 pharmacy staff and an unquantified number of counsellors and treatment supporters. On this basis, Mozambique, which needs to provide ART to about 200,000 patients, would require 200–400 doctors from its' total stock of 514, and 400–1400 nurses from its total stock of 3947.</p>", "<p>Other estimates of the workforce required to scale up ART suggest even more stark results. Smith [##UREF##3##6##] calculated that seven out of fourteen countries included in his study would be unable to meet needs even if they used 100% of their current workforce. Figure ##FIG##2##3## shows Smith's estimates of human resource requirements for full coverage of population with antiretroviral therapy. Only two of the 'twin burden' countries are considered in Smith's analysis – Mozambique and Zambia (Malawi was not included). Both are among the three countries whose current medical personnel situation appears least adequate for antiretroviral therapy expansion according to Smith. The third is Rwanda, one of the most human resource constrained countries (see Table ##TAB##1##2##), but with a relatively low estimated HIV prevalence rate. This estimate has recently fallen from a reported rate of 8.9% to 3.1% following the expansion of sentinel HIV surveillance to rural sites [##UREF##4##7##]; it is possible, but not clear, if Smith used the higher rate in his calculation (see Figure ##FIG##2##3##).</p>", "<p>In part, these stark estimates reflect the clinical complexity and chronic nature of treating patients with AIDS. Even in the absence of antiretroviral therapy, HIV increases the needs for skilled human intervention in the health system, particularly due to the incidence of opportunistic infections. For example, one study in Rwanda estimated that 60% of hospital beds were occupied by AIDS patients being treated for opportunistic infections [##UREF##5##8##].</p>", "<p>A comprehensive HIV/AIDS programme also includes a range of interventions unrelated to the treatment of people with AIDS such as HIV prevention strategies, including the comprehensive management of patients with other sexually transmitted infections, voluntary counselling and testing (VCT) services and the prevention of vertical transmission. All these interventions also require skilled health workers.</p>", "<p>The HR requirements of ART programmes therefore have to be met within a severely limited pool of human resources. It is therefore unsurprising that the volume of additional funding and energy directed at HIV/AIDS programmes should threaten less well supported activities. Furthermore, the delivery of HIV/AIDS interventions through non-government organisations (NGOs) and private providers that are able to offer better pay and working conditions to health workers can lead to attrition from the public sector and other areas of health care [##UREF##6##9##,##UREF##7##10##].</p>", "<title>A case study of Malawi</title>", "<title>Background</title>", "<p>With an estimated GDP per head of US$646 in 2004, Malawi is one of the poorest countries in Africa [##UREF##8##11##]. Over half the 12 million population is food insecure and 65.3% were unable to meet their daily consumption needs in 1998 [##UREF##9##12##]. Life expectancy at birth is 39.8 years. HIV prevalence in Malawi was 14.1% (CI: 6.9 – 21.4) in 2005 [##UREF##10##13##]. The country is heavily dependent on aid which contributed 31.2% of Gross National Income in 2003, a higher proportion than most other countries in sub-Saharan Africa [##UREF##11##14##].</p>", "<p>Malawi's health care indicators are poor [##UREF##12##15##, ####UREF##13##16##, ##UREF##14##17####14##17##]:</p>", "<p>• Only 10% of health facilities were able in 2002 to deliver a basic minimum standard of care, with many being in poor condition, lacking an operational water source, electricity or a working telecommunications system.</p>", "<p>• Full immunisation coverage has fallen from a rate of 81.8% in 1990 to 64.4% in 2004.</p>", "<p>• The maternal mortality ratio is one of the highest in the world, standing at 984 per 100 000 live births in 2004.</p>", "<p>• Only 46% of the population live within 5 km of a formal health facility and only 20% live within 25 km of a hospital</p>", "<p>• 73% of households lacked an insecticide-treated bednet in 2004. According to one survey, more than 50% of malaria cases do not get treatment at health facilities.</p>", "<p>There are however some notable achievements. Neonatal tetanus and polio have been eliminated through immunisation programmes and TB cure rates are over 70% [##UREF##15##18##]. And, as discussed later, there has been a great increase in the number of people living with AIDS receiving anti-retroviral therapy.</p>", "<p>Malawi's health system is severely under-financed. In 2001, total health expenditure was US$ 12.4 per person [##UREF##16##19##]. At that time, the cost of delivering an 'essential health package' (EHP) of eleven cost-effective health services was estimated at $17.53 per capita, nearly 50% more than existing total health spending [##UREF##17##20##]. Furthermore, the cost estimate of this EHP was based on only 67% coverage for some services and did not include the costs of central level management and supervision, central hospital activity, or the provision of antiretroviral therapy.</p>", "<p>According to WHO's National Health Accounts database, per capita total health expenditure in 2005 had risen to US$ 23. The government accounted for 24.3% of total health spending; donors/external funding for 51.5%; and private expenditure for 24.2% [##UREF##18##21##]. The organization of health care finances in Malawi has improved since 2005 as a consequence of a Sector Wide Approach (SWAp) which several donors, particularly DFID (UK), have agreed to support. Under the SWAp, a six-year programme of work was established, with the delivery of the EHP being at the core. However, not all external funding is channelled through the SWAp. USAID and PEPFAR are notable bilateral donors operating outside the SWAp framework.</p>", "<p>In line with the focused international attention on HIV/AIDS, Malawi established a separate National AIDS Commission (NAC) to manage the significant amount of dedicated HIV/AIDS funding (including grants from the Global Fund) and to provide oversight over the country's HIV/AIDS plan. When first established, tension existed between the NAC and the Ministry of Health, partly because the NAC employed staff at higher salaries than the Ministry and because of the Ministry's loss of direct control over HIV/AIDS funding. According to a draft copy of Malawi's 2004/05 National Health Accounts, the Ministry of Health's share of public finance has decreased between 2002/03 and 2004/05 while that of the NAC increased (see Table ##TAB##2##3##).</p>", "<p>Health services are provided by a multiplicity of providers. Of 'formal' health facilities, 60% are government-run; and 26% are mission facilities (mainly found in the rural areas). There is a small private-for-profit health sector (including three private hospitals) limited mainly to urban areas, as well as services provided by private companies for their employees. There is also a substantial traditional health sector. Nearly a quarter of deliveries are attended by a traditional birth attendant.</p>", "<p>Mission facilities tend to operate independently of each other but within a loose association called the Christian Health Association of Malawi (CHAM). A formal agreement exists with the Ministry of Health whereby most of the CHAM workforce is paid from the government payroll. Other providers include islamic health facilities, NGOs, grocery stores, pharmacies and community-based distribution agents for contraception. The share of total health care expenditure in 1998/9 amongst different providers is shown in Figure ##FIG##3##4##. Since then, NGO health care provision has expanded, particularly NGOs providing HIV/AIDS services. There are also a number of clinical research projects, particularly related to HIV/AIDS in the health care system – these provide services to research subjects but also consume a significant number of the country's scarce skilled health workforce (see Figure ##FIG##3##4##).</p>", "<p>In theory, health care providers in Malawi are organized according to a system of five 'zones' and 28 'health districts'. Each district is supposed to have an integrated health plan that incorporates the public sector, CHAM facilities and NGO providers. In practice, this does not always happen. Zonal offices which are supposed to provide support and supervision to district level services are relatively new and do not yet have the capacity to effectively support health districts. And in many districts, public, CHAM, NGO and private providers operate independently of each other.</p>", "<p>Presently, local government assemblies provide a small amount of health services. However, there are plans to devolve primary health care provision to local assemblies, including the transfer of budgets and human resource employment responsibilities.</p>", "<title>Human resources</title>", "<p>As noted in the introduction, the total density of doctors and nurses (including midwives) in Malawi is 0.61 per thousand population, a shade higher than the threshold of 0.5 that defines the 10 worst served sub-Saharan African countries. As in other countries, the national average masks extreme inequities of provision within the country. In November 2004, 15 out of Malawi's 26 districts had less than 1.5 nurses per facility, and five had less than one. Only 13% of all health facilities had 24-hour midwifery coverage. Of 28 600 health worker posts in Ministry of Health (MoH) and CHAM facilities in 2005, about 38% were vacant [##UREF##19##22##]. Half of Malawi's doctors work at one of four central hospitals (although this partly due to deployment of newly qualified doctors to the central hospitals for the period of internship).</p>", "<p>The HRH situation was described in April 2004 by the MoH as \"dangerously close to collapse\" and as a \"major, persistent and deepening crisis\" [##UREF##20##23##]. An independent review of a safe motherhood project concluded that in spite of \"extensive staff training and support\" to midwives, problems with staff retention would remain an important obstacle to increasing coverage of births by skilled attendants [##UREF##21##24##].</p>", "<p>Three notable features of the health workforce in Malawi are the extensive use of clinical officers, medical assistants and about 4500 community-based health surveillance assistants (HSAs). Clinical officers receive four years of training and provide a range of medical services, including diagnosis and treatment, surgery and anaesthesia, and mending fractures. They form the cornerstone of hospital care in many rural areas. Medical assistants receive two years of training and mainly provide medical care in health centres and the outpatient departments of district hospitals. HSAs receive 10 weeks of training and are responsible for a variety of different tasks ranging from health promotion activities to TB defaulter tracing.</p>", "<p>There are several reasons for Malawi's health worker crisis. One is its low resource base which has made it difficult for the government to adequately fund the training, employment and retention of health staff. Even after establishing a medical school in 1991, Malawi produced only 20 doctors per year until 2005. Although it produced about 40–60 registered nurses and 300–350 enrolled nurses annually in the early 2000s [##UREF##22##25##], this is small compared to an establishment of 8,963 public sector nurses (including CHAM) [##UREF##20##23##].</p>", "<p>Another reason is HIV/AIDS. A 2002 study showed annual death rates of 2% among hospital health care workers [##REF##11925986##26##]. Fear of exposure to HIV, particularly as shortages of gloves and other supplies hampers adherence to universal precautions, is also said to have contributed to staff leaving the sector [##UREF##23##27##]. Staff time is also lost to funeral attendance, care of sick family members and prolonged periods of illness. The increased workload caused by HIV/AIDS has also contributed to further demotivation, although according to local informants, the ability to treat patients with antiretroviral therapy is said to have improved staff morale.</p>", "<p>Staffing problems are more acute in the public sector. Whereas 20 years ago, public sector health worker salaries were considered attractive, wages for civil servants in Malawi have not kept up with rising consumer prices [##UREF##24##28##]. Job opportunities in the better-paid private sector (including NGOs and research institutions) and abroad, particularly in the UK, have been another potent 'pull factor'.</p>", "<p>Other reasons staff leave the public sector include poor working conditions; infrequent supervision and support; the lack of essential drugs, supplies and equipment; limited career progression opportunities; unequal access to training; an unclear deployment policy; and poor housing [##UREF##25##29##]. A study conducted in early 2006 identified up to 740 'inactive' professional health workers, including 469 nurses and 164 Clinical Officers, who had either resigned or retired from the health sector [##UREF##26##30##].</p>", "<title>The Emergency Human Resources Programme</title>", "<p>Malawi has implemented a variety of initiatives to solve its health worker shortages over the years. However, it was only after Peter Piot, Executive Director of UNAIDS, and Suma Chakrabarti, Permanent Secretary of the UK Department for International Development (DfID), visited Malawi in 2004 and witnessed first hand the hopeless staffing situation of many facilities that a substantial human resources plan was pulled together. The result was a shift from piecemeal donor support to a comprehensive six-year \"Emergency Human Resources Programme\" (EHRP).</p>", "<p>Costed at US$272 million, with major funding from DFID and some from the Global Fund, the EHRP aims to raise Malawi's staffing levels (see Table ##TAB##3##4##) to a point where it could deliver the EHP (the planned targets do not therefore cater for the additional staff needed to provide antiretroviral therapy services). Although the EHRP would significantly boost staffing levels, the targets still fall short of the WHO-recommended minimum (on a rough estimate the EHRP would increase the total doctor and nurse density to 1.51 compared to the 2.28 threshold used by the 2006 WHR to define a 'critical shortage') (Table ##TAB##3##4##).</p>", "<p>The EHRP takes a five-pronged approach:</p>", "<p>• Improving incentives for recruitment and retention of public sector and CHAM staff through a 52% salary top-up for 11 professional and technical cadres, coupled with a major initiative to recruit and re-engage qualified Malawian staff.</p>", "<p>• Expanding domestic training capacity, including doubling the number of nurses and tripling the number of doctors in training.</p>", "<p>• Using international volunteer doctors and nurse tutors as a short-term measure to fill critical posts while Malawians are being trained.</p>", "<p>• Providing technical assistance to bolster Ministry of Health (MoH) capacity in human resources planning, management and development.</p>", "<p>• Establishing robust human resources monitoring and evaluation capacity.</p>", "<p>In addition, the programme explicitly recognises the importance of improving policies on postings and promotions; training and career development; and incentives for deploying staff to underserved areas (which includes a major effort to improve staff housing). Technical assistance to the MoH in the form of human resources experts was therefore arranged. The government has also introduced a period of compulsory public health service for enrolled nurses trained at public expense.</p>", "<p>Contrary to initial fears, other public servants did not protest at the improved pay for health workers partly because of the careful way in which the government and others had made the case for higher pay for health workers [##REF##16713877##31##]. However, any further improvements to the pay and working conditions of health workers are likely to be resisted without improvements for other civil servants.</p>", "<p>Since its implementation, anecdotal reports indicate that the salary rise had helped stem the flow of staff, particularly nurses, out of the public sector [##REF##16713877##31##]. In addition, by the last quarter of 2005, 591 'inactive' staff had been recruited and more than 1,100 staff had been promoted (mostly nurses whose promotions had been blocked by civil service rules following a change to the nursing curriculum).</p>", "<p>The number of health professionals trained annually increased from 400/year in 2004 to over 1000/year in 2006. The College of Medicine increased its first-year Medical Doctor intake for 2005 to 60 students [##UREF##19##22##]. By mid-2006, health-training institutions were running at full capacity, albeit with a need to improve tutor: student ratios. To further increase the output of nurse training institutions, proposals exist to reduce the length of time required for basic nurse training from four to three years (longer than in most other African countries). In 2006, 51 expatriate doctors and 15 nurse tutors were scheduled to be in post [##REF##16713877##31##].</p>", "<p>However, a recent evaluation of the EHRP in 2006 concluded that there were still difficulties in attracting tutors, doctors and nurses and that the EHP would not succeed if \"radical action is not taken to dramatically improve retention rates\", particularly in rural areas [##UREF##27##32##]. Another illustration of on-going problems was the observation that although expatriate doctors had been recruited successfully through the UN Volunteer Programme, in 2004/05 and 2005/06 less than eight medical graduates had joined the MoH whilst several other junior doctors had resigned [##UREF##28##33##].</p>", "<p>According to Medecins Sans Frontieres (MSF), in Chiradzulu district, there were 50 nurses working at the district hospital in 2006; that number had dropped to 28 by 2007 [##UREF##29##34##]. MSF also noted the experience of retired nurses who had been attracted to return to the workforce having trouble getting contracts and payment due to administrative delays.</p>", "<p>One problem was that the promised 52% salary top-up was not translated into a 52% increase in take-home pay because of changes to the tax and allowance structure of public sector health workers. Furthermore, in spite of the salary top-ups, non-government employers still offer much better rates of pay, particularly for scarce health worker cadres such as doctors, laboratory technicians and pharmacists.</p>", "<p>The fragmented and competitive provider market, coupled with the pressure on funders and policy makers to achieve ambitious coverage targets, has caused the labour market to become extremely uneven. Scarce skills appear to be concentrated in urban areas and in NGO/research projects that are able to offer higher remuneration. According to MSF, external financing is also associated with workshops and training programmes which public health workers are paid with per diems and stipends to attend. A five-day training workshop can increase a nurse's basic monthly salary by 25–40% [##UREF##29##34##]. Although training workshops are necessary, the competition for stipends can disrupt service delivery and increase absence from facilities.</p>", "<title>HIV/AIDS and the provision of antiretroviral therapy</title>", "<p>In spite of its significant health systems constraints, Malawi has made exceptional progress in expanding access to ART. At the end of 2006, there were about 60 000 people on treatment in the country, with plans to expand coverage to 245 000 people by 2010.</p>", "<p>This progress is argued to have been achieved because of several factors [##UREF##30##35##]:</p>", "<p> A strong rights-based international advocacy movement</p>", "<p> Earmarked funding for antiretroviral therapy services from a range of donors.</p>", "<p> Support from international NGOs and research organisations to deliver ART services</p>", "<p> Strong technical leadership and management within the Ministry of Health</p>", "<p> A vertical management and delivery system which has included:</p>", "<p>◦ dedicated ART training programmes for various cadres of health workers (see below)</p>", "<p>◦ A stand-alone system for financing, procuring and distributing antiretroviral therapy drugs. This involves drugs procured by UNICEF from India being flown to Copenhagen where they are individually packed for each ART clinic, and then flown to Malawi where they are couriered to each ART clinic.</p>", "<p>◦ A stand-alone information system to enable high-quality monitoring and evaluation</p>", "<p>◦ Quarterly supervision and support visits to all ART clinics</p>", "<p> A 'low-resource approach' which includes using a single first-line and second-line regimen for all patients and providers; using clinical staging to determine eligibility for treatment (not CD4 counts); using fixed-dose combination tablets; and using clinical signs only to monitor treatment response.</p>", "<p>Under the direction of the HIV/AIDS unit within the MoH, an agreement has been reached that ART providers will be supplied with government-procured drugs, whether in the public or private sector, provided they attend a 5-day training course and formal assessment. In the private sector, in addition to paying private consultation fees, patients pay a fee of MK 500 (at time of writing, US$ 1 = MK 140) per month for the medicines, of which MK 200 is retained by the private provider and MK 300 is paid into a revolving fund managed by the Malawi Business Coalition Against HIV/AIDS which is then remitted to the National AIDS Council. The cost of ART on the government procurement scheme is approximately MK 1820 per month, although this excludes the costs of supply and distribution logistics [##UREF##31##36##].</p>", "<p>Before new ART sites are established, those responsible for establishing the sites and providing care must also spend two weeks attached to one of the specialist HIV centres within Malawi after completing the 5-day training course. Through the provision of subsidised medicines and using this model of structured training, the government has been able to harness the private sector to support the national ART programme.</p>", "<p>Human resource plans to further expand ART coverage involve four main strategies [##UREF##30##35##]:</p>", "<p>1) minimizing the health worker: patient ratio by changing the requirement for all patients to be seen by a clinician when they come for repeat prescriptions.</p>", "<p>2) 'task-shifting' to enable nurses to diagnose and prescribe ART and 'lower' cadres of health workers (in particular HSAs) to dispense ART. Plans exist to overcome legal and professional restrictions on prescribing and dispensing, and to train and equip HSAs with the competencies to provide ART drugs; keep accurate patient records; and question carers and patients so that an appropriate treatment regime and referral pattern can be established. [However, it should be noted that this is being resisted by the Pharmacy, Medicines and Poisons Board]</p>", "<p>3) increasing the number of health workers involved in the ART programme by including volunteers/unpaid workers.</p>", "<p>4) decentralizing management and supervision to zonal and district health management structures.</p>", "<p>In Thyolo district, MSF has been working with the government to provide more than 10 000 people with ART by the end of 2007. One of its strategies has been to support 600 volunteer community home-based caregivers to assist community nurses with the management of common HIV-related conditions, support people on ART, and trace defaulters [##REF##16962622##37##]. Nurses are also being used to manage 'stable patients' (defined as non-pregnant adults who have been on first-line treatment for at least one year with no complications or adherence problems).</p>", "<p>However, as Malawi contemplates the further expansion of antiretroviral therapy (whilst sustaining its current gains) within the context of limited resources and many shortfalls in the provision of other essential health services, there are concerns that these other services could be harmed. In addition, the intention to decentralize new responsibilities to zonal and district offices might compromise this management capacity which is already struggling to oversee and support other health programmes.</p>", "<p>The external evaluation of Malawi's ART programme in 2006 noted concerns that antiretroviral services resembled 'islands of excellence in a sea of problems' [##UREF##30##35##]. While the ART programme's achievements were impressive, other services (including the prevention of vertical transmission) showed signs of stagnation. One contrast was the excellent supply of ART drugs compared with the abysmal supply of other essential health commodities; another was the plans to up-scale paediatric ART when it was clear that the country's programme to reduce vertical transmission had stalled. It was also noted that the ART programme's focus on individual treatment had under-emphasised the potential for treatment services to act as an engine for HIV prevention.</p>", "<p>However, the ART programme could also impact positively on the health system by, for example, helping keep HIV-positive health workers healthy and preventing facilities from being overwhelmed by the needs of people dying from AIDS. In addition, the political and civic energy and additional resources directed at the scale up of ART provides an opportunity to strengthen health systems. For example, the impetus to reduce vertical HIV transmission can be harnessed to improve the quality of ante-natal and obstetric care as a whole.</p>", "<p>The recent introduction of eight paediatricians from the United States to help increase coverage of paediatric ART is another example of how the current international focus on AIDS treatment could be harnessed to strengthen the health system as a whole. As well as increasing paediatric ART coverage, these physicians could be deployed to support the improvement and expansion of other child health services.</p>", "<title>Abbreviations</title>", "<p>ART: antiretroviral therapy; CHAM: Christian Hospital Association of Malawi; EHP: essential health package; EHRP: Emergency Human Resources Programme; GDP: gross domestic product; HSA: health surveillance assistants; MSF: Medecins Sans Frontieres; NAC: National AIDS Commission; NGO: non-government organisations; VCT: voluntary counselling and testing; WHO: World Health Organisation; WHR: World Health Report; SWAp: Sector Wide Approach.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>DM and BM drafted the paper on the basis of two separate papers they had individually authored. VM made significant contributions to the original draft with particular reference to the Malawi case study.</p>" ]

[ "<title>Acknowledgements</title>", "<p>This paper is derived from two background papers prepared for meeting of the UNAIDS/World Bank Economics Reference Group, 3 &amp; 4 May 2007. The case study of Malawi is partly derived from the participation of one author (DM) in an external evaluation of the country's ART programme which was funded and supported by the Government of Malawi, the UK Department for International Development, USAID, the World Health Organisation and UNICEF.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Countries with a critical shortage of health service providers (doctors, nurses and midwives)</bold>. (Source: World Health Report (2006), Working Together for Health, Geneva: WHO [##REF##17178522##1##]).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Total numbers of doctors, nurses and midwives against adult HIV prevalence across African countries for which both statistics are available</bold>. (Source: Authors' analysis based on HRH global atlas and UNAIDS data).</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Percentage of existing doctor workforce required for full coverage in 10 years</bold>. (Source: Smith 2005 [##UREF##3##6##]).</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Health expenditure in Malawi by provider sector, 1998/9 FY</bold>. (Source: Government of Malawi, Ministry of Health and Population: Malawi National Health Accounts: a broader perspective of the Malawian Health Sector, 2001).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Global health workforce, by density</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>WHO Region</bold></td><td align=\"center\" colspan=\"2\"><bold>Total health workforce</bold></td><td align=\"center\" colspan=\"2\"><bold>Health service providers</bold></td><td align=\"center\" colspan=\"2\"><bold>Health management and support workers</bold></td></tr><tr><td/><td colspan=\"6\"><hr/></td></tr><tr><td/><td align=\"center\"><bold>Number</bold><break/></td><td align=\"center\"><bold>Density </bold><break/><bold>(per 1000 population)</bold></td><td align=\"center\"><bold>Number</bold><break/></td><td align=\"center\"><bold>% of total health </bold><break/><bold>workforce</bold></td><td align=\"center\"><bold>Number</bold><break/></td><td align=\"center\"><bold>% of total health </bold><break/><bold>workforce</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Africa</bold></td><td align=\"center\">1 640 000</td><td align=\"center\">2.3</td><td align=\"center\">1 360 000</td><td align=\"center\">83</td><td align=\"center\">280 000</td><td align=\"center\">17</td></tr><tr><td align=\"left\"><bold>Eastern Mediterranean</bold></td><td align=\"center\">2 100 000</td><td align=\"center\">4.0</td><td align=\"center\">1 580 000</td><td align=\"center\">75</td><td align=\"center\">520 000</td><td align=\"center\">25</td></tr><tr><td align=\"left\"><bold>South-East Asia</bold></td><td align=\"center\">7 040 000</td><td align=\"center\">4.3</td><td align=\"center\">4 730 000</td><td align=\"center\">67</td><td align=\"center\">2 300 000</td><td align=\"center\">33</td></tr><tr><td align=\"left\"><bold>Western Pacific</bold></td><td align=\"center\">10 070 000</td><td align=\"center\">5.8</td><td align=\"center\">7 810 000</td><td align=\"center\">78</td><td align=\"center\">2 260 000</td><td align=\"center\">23</td></tr><tr><td align=\"left\"><bold>Europe</bold></td><td align=\"center\">16 630 000</td><td align=\"center\">18.9</td><td align=\"center\">11 540 000</td><td align=\"center\">69</td><td align=\"center\">5 090 000</td><td align=\"center\">31</td></tr><tr><td align=\"left\"><bold>Americas</bold></td><td align=\"center\">21 740 000</td><td align=\"center\">24.8</td><td align=\"center\">12 460 000</td><td align=\"center\">57</td><td align=\"center\">9 280 000</td><td align=\"center\">43</td></tr><tr><td align=\"left\"><bold>World</bold></td><td align=\"center\">59 220 000</td><td align=\"center\">9.3</td><td align=\"center\">39 470 000</td><td align=\"center\">67</td><td align=\"center\">19 750 000</td><td align=\"center\">33</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Countries in the deepest human resource crisis according to their numbers of doctors, nurses and midwives: ratios per thousand population.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Physicians</bold></td><td align=\"center\"><bold>Nurses</bold></td><td align=\"center\"><bold>Midwives</bold></td><td align=\"center\"><bold>TOTAL</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Burundi</bold></td><td align=\"center\">0.03</td><td align=\"center\">0.19</td><td align=\"center\">0</td><td align=\"right\">0.22</td></tr><tr><td align=\"left\"><bold>Ethiopia</bold></td><td align=\"center\">0.03</td><td align=\"center\">0.2</td><td align=\"center\">0.02</td><td align=\"right\">0.25</td></tr><tr><td align=\"left\"><bold>Niger</bold></td><td align=\"center\">0.02</td><td align=\"center\">0.2</td><td align=\"center\">0.03</td><td align=\"right\">0.25</td></tr><tr><td align=\"left\"><bold>Chad</bold></td><td align=\"center\">0.04</td><td align=\"center\">0.24</td><td align=\"center\">0.04</td><td align=\"right\">0.32</td></tr><tr><td align=\"left\"><bold>Liberia</bold></td><td align=\"center\">0.03</td><td align=\"center\">0.17</td><td align=\"center\">0.13</td><td align=\"right\">0.33</td></tr><tr><td align=\"left\"><bold>Mozambique</bold></td><td align=\"center\">0.03</td><td align=\"center\">0.21</td><td align=\"center\">0.12</td><td align=\"right\">0.36</td></tr><tr><td align=\"left\"><bold>Senegal</bold></td><td align=\"center\">0.06</td><td align=\"center\">0.25</td><td align=\"center\">0.07</td><td align=\"right\">0.38</td></tr><tr><td align=\"left\"><bold>United Republic of Tanzania</bold></td><td align=\"center\">0.02</td><td align=\"center\">0.3</td><td align=\"center\">0.07</td><td align=\"right\">0.39</td></tr><tr><td align=\"left\"><bold>Togo</bold></td><td align=\"center\">0.04</td><td align=\"center\">0.33</td><td align=\"center\">0.05</td><td align=\"right\">0.42</td></tr><tr><td align=\"left\"><bold>Rwanda</bold></td><td align=\"center\">0.05</td><td align=\"center\">0.42</td><td align=\"center\">0.01</td><td align=\"right\">0.48</td></tr><tr><td align=\"left\"><bold>Central African Republic</bold></td><td align=\"center\">0.08</td><td align=\"center\">0.23</td><td align=\"center\">0.18</td><td align=\"right\">0.49</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Share of public finance managed by different segments of the health system</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\"><bold>Budget management</bold></td><td align=\"center\" colspan=\"3\"><bold>Year</bold></td></tr><tr><td/><td colspan=\"3\"><hr/></td></tr><tr><td/><td align=\"center\"><bold>2002/03</bold></td><td align=\"center\"><bold>2003/04</bold></td><td align=\"center\"><bold>2004/05</bold></td></tr></thead><tbody><tr><td/><td align=\"center\"><bold>%</bold></td><td align=\"center\"><bold>%</bold></td><td align=\"center\"><bold>%</bold></td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\">Ministry of Health</td><td align=\"center\">60.2</td><td align=\"center\">49.5</td><td align=\"center\">51.6</td></tr><tr><td align=\"left\">National AIDS Commission</td><td align=\"center\">1.8</td><td align=\"center\">3.5</td><td align=\"center\">11.9</td></tr><tr><td align=\"left\">CHAM</td><td align=\"center\">4.2</td><td align=\"center\">2.9</td><td align=\"center\">4.2</td></tr><tr><td align=\"left\">Other NGOs</td><td align=\"center\">4.3</td><td align=\"center\">7.9</td><td align=\"center\">6.4</td></tr><tr><td align=\"left\">Donors</td><td align=\"center\">10.6</td><td align=\"center\">20</td><td align=\"center\">10.9</td></tr><tr><td align=\"left\">Other</td><td align=\"center\">18.9</td><td align=\"center\">16.2</td><td align=\"center\">15</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Selected EHRP staffing targets (F/Y 2005–2006 Stock Indicator)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Category</bold></td><td align=\"center\" colspan=\"3\"><bold>Combined Ministry of Health &amp; CHAM</bold></td></tr><tr><td/><td colspan=\"3\"><hr/></td></tr><tr><td/><td align=\"center\"><bold>EHRP Target</bold></td><td align=\"center\"><bold>Filled</bold></td><td align=\"center\"><bold>Vacancy Rate (%)</bold></td></tr></thead><tbody><tr><td align=\"left\">Physician/Specialist</td><td align=\"center\">433</td><td align=\"center\">162</td><td align=\"center\">63%</td></tr><tr><td align=\"left\">Nurse (all categories)</td><td align=\"center\">8,440</td><td align=\"center\">3,416</td><td align=\"center\">60%</td></tr><tr><td align=\"left\">Clinical Officer</td><td align=\"center\">1,405</td><td align=\"center\">1,033</td><td align=\"center\">26%</td></tr><tr><td align=\"left\">Medical Assistant</td><td align=\"center\">1,500</td><td align=\"center\">491</td><td align=\"center\">67%</td></tr><tr><td align=\"left\">Radiography/Technician</td><td align=\"center\">270</td><td align=\"center\">58</td><td align=\"center\">79%</td></tr><tr><td align=\"left\">Pharmacy/Technician</td><td align=\"center\">269</td><td align=\"center\">134</td><td align=\"center\">50%</td></tr><tr><td align=\"left\">Medical Laboratory Technician</td><td align=\"center\">507</td><td align=\"center\">182</td><td align=\"center\">64%</td></tr><tr><td align=\"left\">Environmental Health Officer</td><td align=\"center\">1,662</td><td align=\"center\">223</td><td align=\"center\">87%</td></tr><tr><td align=\"left\">Dental Technician/Therapist</td><td align=\"center\">470</td><td align=\"center\">138</td><td align=\"center\">71%</td></tr><tr><td align=\"left\">Physiotherapy</td><td align=\"center\">168</td><td align=\"center\">22</td><td align=\"center\">87%</td></tr><tr><td align=\"left\">Medical Engineering</td><td align=\"center\">60</td><td align=\"center\">24</td><td align=\"center\">60%</td></tr><tr><td align=\"left\">Health Surveillance Assistant</td><td align=\"center\">11,000</td><td align=\"center\">4,664</td><td align=\"center\">58%</td></tr></tbody></table></table-wrap>" ]

[ "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>Note: All data for latest available year. For countries where data on the number of health management and support workers were not available, estimates have been made based on regional averages for countries with complete data.</p><p>Data sources: World Health Organization. <italic>Global Atlas of the Health Workforce </italic><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.who.int/globalatlas/default.asp\"/></p></table-wrap-foot>", "<table-wrap-foot><p>Source: Author's analysis of HRH global atlas, latest year available <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.who.int/globalatlas/default.asp\"/></p></table-wrap-foot>", "<table-wrap-foot><p>(Source: Malawi 2004/05 National Health Accounts – draft copy (September 2006). Lilongwe: Malawi Ministry of Health)</p></table-wrap-foot>", "<table-wrap-foot><p>(Source: Government of Malawi, Ministry of Health (July 2006). Strategic human resources for health framework for the health sector)</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1478-4491-6-16-1\"/>", "<graphic xlink:href=\"1478-4491-6-16-2\"/>", "<graphic xlink:href=\"1478-4491-6-16-3\"/>", "<graphic xlink:href=\"1478-4491-6-16-4\"/>", "<inline-graphic xlink:href=\"1478-4491-6-16-i1.gif\"/>", "<inline-graphic xlink:href=\"1478-4491-6-16-i1.gif\"/>", "<inline-graphic xlink:href=\"1478-4491-6-16-i1.gif\"/>", "<inline-graphic xlink:href=\"1478-4491-6-16-i1.gif\"/>", "<inline-graphic xlink:href=\"1478-4491-6-16-i1.gif\"/>", "<inline-graphic xlink:href=\"1478-4491-6-16-i1.gif\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>1 Introduction</title>", "<p>Human T-lymphotropic viruses (HTLV-1 to -4) belong to the <italic>Deltaretrovirus </italic>genera of the Orthoretrovirinae subfamily. HTLV-1 was the first discovered human retrovirus in the early eighties [##REF##6261256##1##]. HTLV-2 was described two years later [##REF##2992932##2##] whereas HTLV-3 and -4 subtypes were isolated only recently [##REF##15882466##3##,##REF##15911757##4##]. HTLV-1 is the etiological agent of an aggressive leukemia called adult T-cell leukemia/lymphoma (ATL) and a neurodegenerative disease, tropical spastic paraparesis/HTLV associated myelopathy (TSP/HAM). Isolated from a case of hairy-cell leukemia, HTLV-2 is by far less pathogenic although its involvement in the development of TSP has been reported [##REF##16155606##5##,##REF##15236397##6##]. HTLV-3 and -4 have not yet been associated to any pathology, likely due to their recent identification and to the low number of isolates. Three HTLV subtypes have closely related simian viruses (named STLV-1, -2 and -3) while a STLV-5 strain is presently still devoid of a human counterpart [##REF##17976676##7##]. Another related deltaretrovirus, bovine leukemia virus (BLV) is the etiological agent of enzootic bovine leukemia. BLV infection of sheep has been used as an animal model for HTLV [##REF##17362524##8##].</p>", "<p>The genome of the HTLV viruses contain typical structural and enzymatic genes (<italic>gag</italic>, <italic>prt</italic>, <italic>pol </italic>and <italic>env</italic>) flanked by two long terminal repeats (LTRs) but also harbors an additional region called pX located between the <italic>env </italic>gene and the 3'-LTR. This region contains at least four partially overlapping reading frames (ORFs) encoding accessory proteins (p12^I, p13/p30^II), the Rex post-transcriptional regulator (ORF III) and the Tax protein (ORF IV). The complementary strand of the HTLV-1 proviral genome is also transcribed, yielding spliced isoforms of the Hbz factor [##REF##12438606##9##, ####REF##16512901##10##, ##REF##17384582##11####17384582##11##]. Hbz interacts with factors JunB, JunD, CREB and CBP/p300 to modulate gene transcription [##REF##17306025##12##, ####REF##18078517##13##, ##REF##18599479##14####18599479##14##]. There is an inverse relantionship between high Hbz and low Tax expresssion in primary ATL [##REF##18426605##15##].</p>", "<p>Among proteins encoded by HTLV-1, Tax1 exerts an essential role in viral transcription as well as in cell transformation [##REF##17384582##11##,##REF##16155604##16##, ####REF##2888190##17##, ##REF##2403646##18####2403646##18##]. These pleiotropic functions are directed by a very wide spectrum of interactions with cellular proteins. In this review, we summarize the current knowledge pertaining to the Tax1 interactome and focus more particularly on its impact on transcription, viral persistence and transformation.</p>" ]

[]

[]

[]

[ "<title>8 Conclusion</title>", "<p>The most intriguing point relating to the Tax1 interactome is the very high number of cellular proteins to which this viral oncogene is able to interact. Today, about 100 Tax1-binding proteins are identified (Table ##TAB##0##1##) and this number is permanently growing (see <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.htlv-tax.com/\"/> for regular updates). Is it possible that a single protein modulates such a wide variety of functions? Are these interactions all relevant for the viral life cycle or pathogenesis? As schematized on Figure ##FIG##4##5##, the vast majority of these interactions contributes to viral or cellular gene expression and promotes infected cell proliferation or survival, required for maintaining viral load <italic>in vivo </italic>[##REF##17483473##231##,##REF##18042431##305##]. On the other hand, checkpoint abrogation allows proliferation of cells with DNA lesions and progressive accumulation of chromosomal abnormalities as frequently observed in ATL [##REF##16155605##220##]. Even if one might entertain doubts about the biological relevance of some Tax1 partners, the Tax1 interactome as a whole likely contributes to the viral life cycle as well as to development of pathogenesis.</p>", "<p>Other viral oncogenes such as Kaposi's sarcoma-associated herpesvirus-encoded LANA and adenovirus E1A also interact with numbers of cellular proteins (e.g. more than 40 for E1A and 100 for LANA) [##REF##17434559##306##,##REF##16299528##307##]. Interestingly, some of these proteins are targeted both by Tax1 and E1A (such as ATF, CBP, p300 or Smad), indicating that similar signaling pathways are involved in distinct viral systems to achieve cell transformation. In particular, Tax1 and E1A share common properties that include regulation of transcriptional activation, chromatin remodeling, interference with p53 activity, regulation of proteasome function and cooperation with Ras in cell transformation [##REF##11753665##308##].</p>", "<p>How are these different activities controlled temporarily and spatially? Additional studies are definitely required to address this point. Currently, Tax1 is known to shuttle between cytoplasm and nucleus, to form intranuclear speckles along with a series of cellular proteins (e.g. NF-κB factors [##REF##9094620##309##], sc35 [##REF##8709263##141##] and chk2 [##REF##12842897##239##]) and to target specialized structures such as the centrosome [##REF##16365316##249##,##REF##16767081##250##]. Moreover, Tax1 localisation and protein interactions are altered under stress conditions [##REF##16775353##291##,##REF##18081936##292##].</p>", "<p>Despite numbers of attempts, Tax1 3-D crystallographic structure is intriguingly still unsolved suggesting that Tax1 adopts a rather undefined conformation. In this context, the concept of intrinsically disordered proteins (IDP) has recently emerged [##REF##17158572##310##]. IDPs contrast to \"ordered\" proteins that fold into a unique and structured state, which represents a kinetically accessible and energetically favorable conformation. IDP proteins contain one or multiple disordered regions that exist as dynamic ensembles in which atom positions and backbone Ramachandran angles vary significantly with no specific equilibrium values [##REF##17158572##310##]. The presence of short (&lt; 30 residues) and long (&gt; 30 residues) ID regions confer conformationnal flexibility thereby facilitating post-translational modifications and enabling a protein to functionally interact with many cellular partners [##REF##17158572##310##,##REF##16218947##311##]. Consistently, IDPs are frequently highly connected 'hubs' in the protein-protein networks [##REF##16218947##311##, ####REF##18364713##312##, ##REF##15190252##313####15190252##313##]. In fact, Tax1 contains many proline (n = 40), serine (n = 25) and glycine (n = 25) residues that are known to promote disorder [##REF##17158572##310##]. According to the VSL1 prediction programme (PONDR^®, <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.pondr.com\"/>), Tax1 contains multiple ID regions (n = 6) (Figure ##FIG##5##6##). In particular, Tax1 contains a long disordered region (spanning amino-acids 76 to 121), in contrast to the well structured capsid (p24), transmembrane (gp21) and surface (gp46) proteins (data not shown). Interestingly, other viral oncogenes such as HPV E6 and E7 are also predicted to contain significant intrinsic disorder [##REF##16889404##314##].</p>", "<p>On the other hand, Tax1 is modified by phosphorylation, ubiquitination and sumoylation that potentially modulate its functions, localisation and interactions [##REF##16287853##76##,##REF##16923801##315##]. Tax1 also contains 8 cysteines that may form disulfite bonds or coordinate zinc ions and 48 leucines that are considered as order-promoting residues [##REF##17158572##310##]. Tax1 thus appears as a flexible structure formed by a series of small modular domains that are relatively independent of surrounding sequences and that permits wide conformational changes depending upon its subcellular environment.</p>", "<p>We propose that, similarly to the hubs, the ID-based structure of Tax1 allows a wide variety of conformational changes enabling binding diversity and recognition of differently shaped protein partners. Flexible accommodation at various binding interfaces would then allow interaction of more structured domains such as the Tax1 zinc finger and leucine containing helices. This hypothetical model provides a rationale to the very broad range of Tax1 interacting proteins identified so far.</p>", "<p>In conclusion, the Tax1 interactome network with the associated biochemical studies reported here provides a molecular basis for understanding viral persistence and pathogenesis, paving the way for the design of compounds to antagonize its ability to mediate cell transformation.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<p>The Tax1 oncoprotein encoded by Human T-lymphotropic virus type I is a major determinant of viral persistence and pathogenesis. Tax1 affects a wide variety of cellular signalling pathways leading to transcriptional activation, proliferation and ultimately transformation. To carry out these functions, Tax1 interacts with and modulates activity of a number of cellular proteins. In this review, we summarize the present knowledge of the Tax1 interactome and propose a rationale for the broad range of cellular proteins identified so far.</p>" ]

[ "<title>2 Interaction of Tax1 with transcription factors and post-transcriptional regulators</title>", "<p>In eukaryotes, initiation and elongation of gene transcription requires decondensation of the locus, nucleosome remodeling, histone modifications, binding of transcriptional activators and coactivators to enhancers and promoters and recruitment of the basal transcription machinery to the core promoter [##REF##11092823##19##,##REF##12651739##20##]. Tax1 is a pleiotropic transcription factor that interferes with several of these mechanisms and modulates transcription of a wide range of cellular genes. In fact, Tax1 deregulates expression of more than one hundred genes [##REF##11494144##21##] through interactions with transcriptional activators, basal transcription factors and proteins involved in chromatin remodeling. Moreover, Tax1 associates with proteins involved post-transcriptionnal control of mRNAs and further modulates gene expression.</p>", "<title>2.1 Transcriptional activators and repressors</title>", "<title>2.1.1 CREB/ATF factors</title>", "<p>Tax1 was initially described as an activator of LTR-directed transcription [##REF##2992082##22##]. Three imperfectly conserved 21-base-pair (bp) repeat sequences called (TxRE) located in the U₃region of the LTR are required and sufficient to confer Tax1 responsiveness [##REF##3035218##23##]. The TxRE element contains an octamer motif TGACG(T/A)(C/G)(T/A) that is flanked by a G stretch and a C stretch at the 5' and 3' sides, respectively [##REF##3263510##24##]. Interestingly this octamer shares homology with the consensus cAMP-responsive element (CRE) 5'-TGACGTCA-3' [##REF##3263510##24##]. Nevertheless, Tax1 exhibits poor affinity for DNA and does not bind directly to the TxRE element [##REF##2768259##25##] but interacts with CRE-binding/activating transcription factors (CREB/ATF). In fact, Tax1 interacts <italic>in vitro </italic>with a number of proteins of the CREB/ATF family of transcription factors: CREB, CREM, ATF1, ATF2, ATF3, ATF4 (CREB2) and XBP1 (X-box-binding protein 1) [##REF##1386673##26##, ####REF##8007991##27##, ##REF##8628284##28##, ##REF##9190894##29##, ##REF##8407959##30##, ##REF##18287238##31####18287238##31##]. These proteins share a common cluster of basic residues allowing DNA binding and a leucine zipper (b-Zip) domain involved in homo- and heterodimerization. Dimer formation modulates their DNA binding specificity and transcriptional activity [##REF##11483355##32##]. Biochemical studies revealed that Tax1 promotes formation of a Tax1-CREB/ATF-TxRE ternary complex <italic>in vitro </italic>by interacting with the b-Zip domain of CREB/ATF factors. Mechanistically, Tax1 enhances the dimerization of CREB/ATF factors, increases their affinity for the viral CRE [##REF##7637811##33##, ####REF##8211160##34##, ##REF##8065935##35##, ##REF##8950264##36####8950264##36##] and further stabilizes the ternary complex through direct contact of the GC-rich flanking sequences [##REF##9593719##37##,##REF##9668114##38##]. Tax1 then recruits co-activators (CBP/p300 and P/CAF) to facilitate transcriptional initiation (see 2.3.1). The ability of Tax1 to dimerize is required for efficient ternary complex formation and for optimal transactivation [##REF##9016568##39##,##REF##8970957##40##]. Interaction of Tax1-CREB/ATF with the LTR promoter DNA was further explored by chromatin immunoprecipitation (ChIP) [##REF##12386157##41##]. In HTLV-1 infected human T-cells (SLB-1), Tax1 and a plethora of CREB/ATF factors as well as other b-Zip proteins bind to the LTR promoter, further confirming interaction <italic>in vivo</italic>. The fact that Tax1 interacts with ATFx adds another level of complexity since this factor represses Tax1-mediated LTR activation [##REF##15890932##42##]. Tax1 is thus able to interact with positive as well as with negative CREB/ATF factors to modulate LTR promoter-directed activity.</p>", "<p>Tax1 also binds to CREB co-activator proteins called transducers of regulated CREB activity (TORCs). In fact, Tax1 interacts with the three members of this family (TORC1, TORC2 and TORC3) [##REF##15466468##43##,##REF##16809310##44##] and TORCs cooperate with Tax1 to activate the LTR in a CREB and p300-dependent manner. Thus, TORCs are thought to associate with the Tax1 ternary complex and participate to transcriptional activation.</p>", "<p>CREB/ATF members play a role in cell growth, survival and apoptosis by regulating CRE-directed gene transcription in response to environmental signals such as growth factors or stress [##REF##11483355##32##,##REF##12766482##45##]. Furthermore, CREB/ATF proteins also have significant impact on cancer development [##REF##12766482##45##]. Depending on the cell type, Tax1 mutants deficient for CREB activation are incompetent for transformation or induction of aneuploidy [##REF##9160887##46##, ####REF##10233947##47##, ##REF##9506967##48##, ##REF##9151835##49##, ##REF##16822311##50####16822311##50##]. Tax1 activates a variety of cellular genes through its interactions with CREB/ATF proteins, for example those encoding interleukin 17 or c-fos [##REF##15218177##51##,##REF##1827666##52##]. Conversely, Tax1 also represses expression of genes like cyclin A, p53 and c-myb by targeting CREB/ATF factors [##REF##11160720##53##, ####REF##9765430##54##, ##REF##11080802##55####11080802##55##]. Transcriptomic profiling of cells expressing either a wild-type or a CREB-deficient Tax1 protein revealed several cellular genes controlled by CRE elements activated by Tax1 [##REF##16822311##50##]. Among these, Sgt1 (suppressor of G2 allele of SKP1) and p97(Vcp) (valosin containing protein) have functions in spindle formation and disassembly, respectively.</p>", "<p>Together, these reports thus demonstrate that Tax1 interacts with a series of CREB/ATF factors and modulates expression of viral and cellular genes through CRE elements. The specific contribution of each CREB/ATF member in Tax1-mediated gene transcription remains unclear.</p>", "<title>2.1.2 Serum responsive factor and members of the ternary complex factor</title>", "<p>HTLV-1 infected T-cell lines expressing Tax1 display increased expression of AP1 (activator protein 1), a homo- or heterodimeric complex of Fos (c-Fos, FosB, Fra1 and Fra2) and Jun (c-Jun, JunB and JunD) [##REF##1906155##56##,##REF##11080798##57##]. Fos and Jun are under the transcriptional control of the serum responsive factor (SRF) in response to various stimuli such as cytokines, growth factors, stress signals and oncogenes. SRF binds to the SRF responsive element (SRE) located in the Fos/Jun promoters which contains two binding sites: a CarG box (CC(A/T)₆GG) and an upstream Ets box (GGA(A/T)). Once SRF occupies the CArG box, the ternary complex factor (TCF) establishes protein interaction with SRF and subsequently binds the upstream Ets site. This complex then recruits the co-activators P/CAF and CBP/p300 to activate transcription.</p>", "<p>In reporter assays, Tax1 activates transcription of promoters under the control of SRE motifs [##REF##1827666##52##,##REF##1906155##56##,##REF##1833716##58##] without direct binding to the DNA but through interactions with transcription factors associated with the SRF pathway. Tax1 has been shown to bind directly to SRF [##REF##1427072##59##, ####REF##7624133##60##, ##REF##8361755##61####8361755##61##] and to various members of the TCF complex such as Sap1 (SRF accessory protein 1), Elk1, Spi1 (spleen focus forming virus (SFFV) proviral integration oncogene 1) and Ets1 [##REF##9151835##49##,##REF##9030555##62##,##REF##11070040##63##]. Tax1 interaction with SRF results in increased binding of SRF to the SRE and altered site selection [##REF##17376895##64##]. Once the complexes are stabilized, Tax1 recruits the co-activators CBP/p300 and P/CAF (see 2.3.1) and mediates transactivation [##REF##11070040##63##].</p>", "<p>It thus appears that Tax1 activates transcription from CREB- and SRF-responsive sites through a similar mechanism which involves its interaction with transcription factors resulting in enhanced DNA binding, altered site selection and coactivator recruitment [##REF##16155604##16##].</p>", "<title>2.1.3 Nuclear factors κB (NF-κB)</title>", "<p>HTLV-1 infected cells display increased expression of various cytokines and cytokine receptors such as interleukin 2 (IL2) and the α-subunit of its high-affinity receptor complex (IL2Rα) [##REF##3024966##65##, ####REF##3030566##66##, ##REF##3026643##67##, ##REF##3037548##68####3037548##68##]. Induction of IL2 and IL2Rα expression is mediated by Tax1 activation of the NF-κB/Rel family of transcription factors [##REF##2843985##69##,##REF##8455941##70##]. By modulating expression of a wide range of genes involved in apoptosis, proliferation, immune response and inflammation, NF-κB is thought playing a central role in Tax1-mediated cell transformation [##REF##16155604##16##].</p>", "<p>In mammals, the NF-κB family of transcription factors is composed of five structurally related members, RelA, RelB (p65), c-Rel, NF-κB1 (p50/p105) and NF-κB2 (p52/p100) which form various dimeric complexes that transactivate or repress target genes bearing a κB enhancer [##REF##7888182##71##,##REF##17183360##72##]. p105 and p100 are precursor proteins that are processed proteolytically to the mature p50 and p52 forms, respectively. These factors share a common Rel-homology domain (RHD) mediating their dimerization, DNA binding and nuclear localization. In non-activated cells, NF-κB dimers are trapped in the cytoplasm by inhibitory proteins called IκBs such as p105, p100, IκBα, IκBβ and IκBγ (C-terminal region of p105), that mask the nuclear localization signal of NF-κB factors through physical interaction [##REF##7888182##71##,##REF##17183360##72##]. NF-κB activation involves phosphorylation of IκB inhibitors by the IκB kinase (IKK), which triggers their ubiquitination and subsequent proteasomal degradation, resulting in nuclear translocation of NF-κB dimers [##REF##17183360##72##,##REF##10837071##73##].</p>", "<p>Tax1 associates with RelA, c-Rel, p50 and p52 after their translocation in the nucleus [##REF##8361755##61##,##REF##7856081##74##,##REF##7936632##75##] but also directly recruits RelA from the cytoplasm [##REF##16287853##76##,##REF##15806143##77##]. After interaction with these NF-κB factors, Tax1 increases their dimerization which is essential for binding to target promoters [##REF##8361755##61##,##REF##7936632##75##,##REF##8918533##78##]. When the complex is bound to the promoter, Tax1 recruits the CBP/p300 and PCAF co-activators [##REF##9528808##79##,##REF##9774518##80##], leading to transcriptional activation</p>", "<title>2.1.4 Other transcription factors</title>", "<p>Tax1 has been shown to associate with CCAAT binding proteins such as NF-YB (nuclear factor YB subunit) and C/EBPβ (CCAAT/enhancer-binding protein β) [##REF##9376596##81##, ####REF##9032250##82##, ##REF##14972524##83####14972524##83##]. Through its binding to NF-YB, Tax1 activates the major histocompatibility complex class II promoter [##REF##9032250##82##]. Besides, C/EBPβ acts as a transcriptional repressor by preventing Tax1 binding to the LTR [##REF##14972524##83##]. On the other hand, Tax1 increases binding of C/EBPβ to and activates the IL-1β promoter [##REF##9376596##81##]. It is noteworthy that C/EBP factors have been implicated in regulation of cellular proliferation and differentiation but also in tumor formation and leukemia development [##REF##17658261##84##].</p>", "<p>Tax1 forms ternary complexes <italic>in vitro </italic>with Sp1 (specificity protein 1)/Egr1 (early growth response 1) [##REF##9341193##85##] and Sp1/Ets1 [##REF##9030555##62##], thereby participating directly in transcriptional activation of the c-sis/PDGF-B (platelet-derived growth factor B) proto-oncogene and PTHrP (parathyroid hormone-related protein) P2 promoters, respectively. Of note, PTHrP is up-regulated during immortalization of T-lymphocytes by HTLV-1 and plays a primary role in the development of humoral hypercalcemia of malignancy that occurs in the majority of patients with ATL [##REF##11395400##86##,##REF##18541021##87##].</p>", "<p>Tax1 further associates with nuclear respiratory factor 1 (NRF1) and activates the CXCR4 chemokine receptor promoter [##REF##10381170##88##].</p>", "<p>Finally, the transcriptional repressor MSX2 (msh homeobox homolog 2) impairs Tax1 mediated transactivation through direct binding [##REF##15970589##89##].</p>", "<title>2.2 Basal transcription factors</title>", "<p>Tax1 interacts with TF_IIA (transcription factor II A) and with two subunits of TF_IID: TBP (TATAA-binding protein) and TAF_II28 (TBP-associated factor II 28) [##REF##8223437##90##, ####REF##9108034##91##, ##REF##8756622##92####8756622##92##]. These basal transcription factors compose the preinitiation transcription complex responsible for the recruitment of RNA polymerase II. Owing to this interaction, Tax1 increases the binding of TBP to the TATAA site and further stimulates transcription initiation from the LTR [##REF##15285791##93##].</p>", "<title>2.3 Chromatin modifying enzymes</title>", "<p>Structural variations of chromatin range from condensed heterochromatin to more open euchromatin, a process that depends on antagonistic effects between multiple protein complexes. Among the complexes affecting chromatin structure, there are those who are capable of altering the histones themselves, the histone deacetylases (HDAC), acetyltransferases (HAT), demethylases (HDM) and methyltransferases (HMT), and those that use the energy of ATP to change the structure of the nucleosome as the SWI/SNF complex [##REF##11943761##94##, ####REF##11937040##95##, ##REF##18037899##96####18037899##96##]. Tax1 expression and HTLV-1 infection both reduce histone levels in T cells [##REF##18237376##97##]. Moreover, Tax1 interacts directly and recruits several proteins involved in chromatin remodeling to modulate gene transcription. The involvement of Tax1-binding proteins in transcriptional activation has been primarily described in the context of the viral LTR. Nevertheless, similar mechanisms are also likely to participate in the activation of cellular promoters.</p>", "<title>2.3.1 HATs</title>", "<p>Acetylation of lysine residues located in the N-terminal tails of histone proteins by HATs is a crucial step for activation of gene transcription. Tax1 interacts with several HATs: p300, its homologous CREB binding protein (CBP) and p300/CBP associated factor (P/CAF) [##REF##9271393##98##, ####REF##9447968##99##, ##REF##8602268##100##, ##REF##10567539##101##, ##REF##10766811##102####10766811##102##]. Tax1 recruits the CBP/p300 and P/CAF once the Tax1-CREB-TxRE complex is stabilized (see 2.1.1), each of which being able to enhance Tax1-mediated transactivation of a transiently transfected LTR reporter. CBP/p300 and P/CAF bind independently on different regions of Tax1 and interaction of Tax1 with these two cofactors is required for optimal transcriptional activity from transiently transfected but also stably integrated LTR reporters [##REF##10567539##101##, ####REF##10766811##102##, ##REF##12438582##103####12438582##103##]. Surprisingly, P/CAF but not CBP/p300 is able to enhance transcription from the LTR independently of its HAT activity [##REF##10567539##101##,##REF##12438582##103##]. Tax1 mediates recruitment of CBP/p300 on reconstituted chromatin templates and facilitates transactivation in a HAT-activity dependent manner [##REF##11739728##104##,##REF##12052856##105##]. CBP/p300 presence at the LTR template correlates with histone H3 and H4 acetylation as well as increased binding of basal transcription factors and RNA polymerase II. ChIP analyses with HTLV-1 infected T cell lines indicate that Tax1, CBP/p300 and acetylated histone H3 and H4 are indeed associated with the LTR promoter [##REF##12386157##41##,##REF##12052856##105##].</p>", "<p>There is a long lasting debate about how Tax1 recruits CBP/p300 at the Tax1-CREB/ATF-TxRE complex. Phosphorylation of CREB at serine 133 by protein kinases A or C is required for CBP/p300 recruitment via physical interaction with the KIX domain [##REF##8552098##106##, ####REF##9413984##107##, ##REF##15337521##108####15337521##108##]. It has long been suggested that Tax1 bypasses the requirement for CREB phosphorylation to recruit coactivators [##REF##9271393##98##,##REF##8602268##100##]. Nevertheless, recent reports indicate that Tax1 rather cooperates with phosphorylated CREB (pCREB) to induce transactivation [##REF##15302573##109##,##REF##17449469##110##]. High levels of pCREB are detected in Tax1 expressing cells and in HTLV-1-infected human T-lymphocytes [##REF##17449469##110##]. Tax1 and pCREB interact simultaneously at two distinct binding sites on the KIX domain forming a very stable complex with the viral CRE [##REF##17449469##110##,##REF##17707401##111##]. Both CREB phosphorylation and Tax1 binding are needed for efficient interaction of full-length CBP to pCREB and subsequent transcriptional activation [##REF##18070920##112##].</p>", "<p>Finally, Tax1 is able to repress the activity of some transcription factors by competitive usage of CBP, p300 and P/CAF. As mentionned above, stable complex formation between Tax1, a transcription factor (e.g. CREB or SRF) and CBP/p300 contributes to transcriptional activation. On the contrary, when Tax1 has poor affinity for a transcription factor (e.g. p53, MyoD or STAT2), it interferes with co-activator recruitment and prevents their activition [##REF##11782467##113##, ####REF##10744749##114##, ##REF##11244044##115##, ##UREF##0##116####0##116##]. Although controversial, this mechanism termed trans-repression could partipate to p53 inactivation in Tax1 expressing cells and HTLV-1 infected lymphocytes (for a review see [##REF##16308315##117##]).</p>", "<title>2.3.2 HDACs</title>", "<p>Among three HDACs (-1, -2 and -3) interacting with the viral LTR in HTLV infected cell lines [##REF##15226416##118##], Tax1 binds directly to HDAC1. HDAC1 overexpression represses Tax1-mediated transactivation owing to its HDAC activity [##REF##12370815##119##]. Nevertheless, the presence of Tax1 and HDAC1 on the viral promoter is mutually exclusive [##REF##15226416##118##,##REF##15194748##120##]. HDAC1 binds to the non-activated LTR and is released from the promoter through physical interaction with Tax1 allowing recruitment of co-activators and transcription initiation. Tax1 is also able to tether HDAC1 to the tyrosine phosphatase SHP1 promoter and selectively down-regulate gene expression [##REF##17540846##121##].</p>", "<p>HDACs form multiprotein complexes together with DNA-histone binding proteins such as SMRT (silencing mediator for retinoid and thyroid receptor) and MBD2 (methyl-CpG-binding domain 2) that both interact with Tax1 and are involved in Tax1 transcriptional activities [##REF##12642864##122##,##REF##15674330##123##]. It thus seems that Tax1, through direct association with HDACs and HDAC-containing complexes is able to selectively activate or repress viral and cellular genes expression.</p>", "<title>2.3.3 HMTs and HDMs</title>", "<p>Mono-, di- and tri-methylation of histone H3 at lysine 9 (H3K9) play a crucial role in structural modification of chromatin. Tax1 associates with two enzymes involved in regulation of H3K9 methylation: SUV39H1 (suppressor of variegation 3–9 homologue 1), a HMT and JMJD2A (Jumonji containing domain 2A), a HDM [##REF##16409643##124##,##REF##15927959##125##]. Methylated H3K9 is a hallmark of transcriptionally inactive chromatin whereas demethylation rather promotes transcriptional activation [##REF##17320507##126##]. SUV39H1 interacts with Tax1 and represses Tax1-mediated transactivation of the LTR [##REF##16409643##124##]. JMJD2A is highly expressed in HTLV-1 infected cell lines but its role on Tax1-mediated transcription is currently unknown [##REF##15927959##125##].</p>", "<p>Methylation of histone H3 at arginine residues is another important regulatory mechanism of transcriptionnal regulation. Tax1 associates with coactivator-associated arginine methyltransferase (CARM1), which preferentially induces methylation at residues R2, R17 and R26 of histone H3 [##REF##17005681##127##]. CARM1 is recruited by Tax1 to the LTR and increases Tax1-mediated transactivation of the LTR. Consistently, silencing of CARM1 impairs Tax1 transcriptional activation, R2-, R17- and R26-methylated histone H3 proteins being present on the LTR promoter in HTLV-1 infected cells.</p>", "<p>Tax1 thus interacts with different histone methyltranferases and demethylases to modulate histone methylation and regulate gene expression.</p>", "<title>2.3.4 The SWI/SNF complex</title>", "<p>The SWI/SNF (Switch/Sucrose Non Fermentable) complex utilizes the energy of ATP hydrolysis to remodel chromatin structures, thereby allowing transcription factors to gain access to DNA during initiation and elongation steps of transcription [##REF##17320508##128##,##REF##12596907##129##]. Tax1 interacts with different components of SWI/SNF: BRG1, BAFs 53, 57 and 155 [##REF##14530271##130##]. Overexpression and silencing of BRG1 increments and impedes Tax1 transactivation of the LTR, respectively [##REF##14530271##130##]. It was first suggested that Tax1 targets BRG1/BRM downstream of RNA polymerase II in order to prevent stalling of transcription. This model was apparently contradicted by the capacity of Tax1 to efficiently activate transcription from chromosomally integrated LTR and NF-κB promoter in a BRG1/BRM deficient cell line [##REF##16840326##131##]. Nevertheless, this observation does not exclude that factors of the SWI/SNF complex cooperate with Tax1 to promote gene transcription. Consistent with this idea, Tax1 cooperates with SWI/SNF complex and RNA polymerase II to promote nucleosome eviction during transactivation [##REF##16547351##132##]. Histone eviction increases accessibility of DNA to transcription factors and requires activity of SWI/SNF and RNA polymerase II [##REF##17320508##128##,##REF##16882978##133##]. Of note, Tax1 may also impact indirectly on SWI/SNF function [##REF##11172715##134##] by interaction with DNA topoisomerase I [##REF##10792988##135##].</p>", "<p>Tax1 is thus able to target SWI/SNF complex components to promote nucleosome displacement and participate to transcriptional activation.</p>", "<title>2.4 Positive transcription elongation factor b (P-TEFb) and sc35</title>", "<p>The switch from initiation of transcription to elongation requires promoter clearance and phosphorylation of the RNA polymerase II carboxyl-terminal domain (CTD) [##REF##11092823##19##]. Phosphorylaton of CTD on serine 5 (S5) and 2 (S2) requires the kinase activities of the basal transcription factor TF_IIH and CDK9, respectively. In the cell, CDK9 together with regulatory subunits cyclin T1, -T2, or -K compose the positive transcription elongation factor b (P-TEFb) that ensures the elongation phase of transcription by RNA polymerase II [##REF##16885020##136##,##REF##10733565##137##]. Tax1 recruits P-TEFb to the viral promoter by interacting with cyclin T1 and CDK9 silencing or depletion inhibits Tax1-mediated transactivation [##REF##17686863##138##,##REF##16641271##139##]. In fact, recruitment of P-TEFb activity to the LTR promoter increases CTD phosphorylation at serine S2 (but not S5) and allows transcriptional activation [##REF##17686863##138##].</p>", "<p>Recent data suggest that the splicing factor sc35 has a critical role in P-TEFb recruitment and positively impacts on transcription [##REF##18641664##140##]. Tax1 binds and colocalizes with sc35 and P-TEFb in nuclear transcriptional hot spots termed speckled structures [##REF##8709263##141##].</p>", "<title>2.5 Nuclear receptors</title>", "<p>Nuclear receptors (NR) belong to a large family of ligand-activated transcription factors that regulate gene expression in response to steroids, retinoids, and other signaling molecules [##REF##15520817##142##]. Tax1 functions as a general transcriptional repressor of nuclear receptors such as glucocorticoid receptors (GR) [##REF##10077562##143##]. A Tax1-binding protein referred to as Tax1BP1 and identified in a yeast two hybrid screen acts as a transcriptional co-activator for NR. Tax1 represses GR signaling by dissociating Tax1BP1 from the receptor-protein containing complex. Consistently, Tax1BP1 overexpression restores GR signaling in Tax1-expressing cells [##REF##17283140##144##].</p>", "<title>2.6 Post-transcriptional and translational regulators</title>", "<p>Tax1-directed gene expression is further regulated at the post-transcriptional and translational levels through protein-protein interactions. Among these, Tax1 associates with TTP, Int6 and TRBP.</p>", "<title>2.6.1 Tristetraprolin (TTP)</title>", "<p>TTP belongs to a family of adenine/uridine-rich element (ARE)-binding proteins that contain tandem CCCH zinc finger RNA-binding domains [##REF##12440952##145##]. TTP is therefore an important player in posttranscriptional regulation of mRNA containing ARE elements. Indeed, TTP delivers ARE-containing mRNAs in discrete cytoplasmic regions, called RNA granules, involved in regulation of translation or decay of these transcripts [##REF##16520386##146##]. The repertoire of ARE-containing genes includes Tumor Necrosis Factor α (TNFα) and Granulocyte Macrophage-Colony-Stimulating Factor (GM-CSF) [##REF##12440952##145##] involved in cell signaling, metabolism, cell proliferation, immune response, death, differentiation and morphogenesis [##REF##16381826##147##].</p>", "<p>Tax1 interacts with TTP and redirects TTP from the cytoplasm to the nuclear compartment as well as in a region surrounding the nucleus [##REF##14679154##148##]. Through its interactions with TTP, Tax1 stabilizes TNFα mRNA and indirectly increases TNFα protein expression. This observation is of importance for the cell transformation process induced by HTLV-1, because TNFα overexpression plays a central role in pathogenesis.</p>", "<title>2.6.2 Int6 and TRBP</title>", "<p>Tax also binds Int6 (Integration site 6) and TRBP (TAR binding protein) that regulate translation and RNA interference, respectively. In fact, Int6 is a subunit of translation initiation factor eIF3, which regulates mRNA binding to the ribosome [##REF##8688078##149##] while TRBP (TAR binding protein) is a componant of RISC (RNA-induced silencing complex) that mediates RNA interference [##REF##9060615##150##]. Currently, the role of these interactions remains unclear.</p>", "<title>2.7 A global model of Tax1 transactivation</title>", "<p>Most of the data summarized in the former paragraphs relate to transcriptional activation of the LTR by Tax1 although it is likely that similar mechanisms also pertain to cellular promoters. Figure ##FIG##0##1## recapitulates the mechanisms of transactivation: Tax1 relieves transcriptional repression through direct interaction with HDAC (i.e. HDAC1) and/or HMT (panel A). Tax1 interacts with CREB/ATF factors (CA) and enhances their binding to the LTR (panel B). When complexes are stabilized on the promoter, Tax1 recruits histone modifying enzymes and chromatin remodelers. This step affects chromatin structure and allows binding of basal transcription factors on the TATA box that is further stabilized by Tax1 interaction with TF_IIA, TF_IID and TBP (panel C). Once the initiation complex is formed, Tax1 recruits the P-TEFb factor, leading to CTD phosphorylation and processive elongation (panel D). Finally, interaction of Tax1 with SWI/SNF prevents stalling of transcription elongation.</p>", "<title>3 Tax1 interaction with proteins involved in cell signaling</title>", "<title>3.1 NF-κB signaling</title>", "<p>NF-κB can be activated by a series of stimuli such as antigens or cytokines that trigger two alternative pathways (so-called canonical and non-canonical). The canonical pathway is engaged in response to inflammatory stimuli (such as TNF-α and interleukin 1 IL-1), T-cell receptor activation or exposure to lipopolysaccharide (LPS). This pathway begins with the phosphorylation of IκB inhibitors by the IκB kinase (IKK), a complex of IKKα, IKKβ and IKKγ/NEMO (NF-κB Essential Modulator). IKK is activated by a mitogen-activated protein kinase kinase kinase (MAP3K) that phosphorylates the IKKα and IKKβ subunits. Phosphorylation of IκB inhibitors triggers their ubiquitination and subsequent degradation by the 26S proteasome, resulting in nuclear translocation of NF-κB dimers (e.g. p50/relA) [##REF##17183360##72##,##REF##10837071##73##]. The non-canonical pathway, which can be induced by stimuli such as CD40 ligand, involves IKKα activation upon phosphorylation by NF-κB inducing kinase (NIK). IKKα then phosphorylates the C-terminal region of p100 leading to subsequent processing of the p100/RelB complex into p52/RelB and its translocation into the nucleus [##REF##16970925##151##]. Interestingly, p52/RelB and p50/RelA dimers target distinct κβ enhancers thereby activating different gene subsets.</p>", "<p>Tax1 stimulates both canonical and non-canonical pathways and constitutively activates NF-κB in HTLV-1 infected cells [##REF##17715223##152##, ####REF##16155602##153##, ##REF##11726516##154####11726516##154##]. The above mentionned interactions of Tax1 with NF-κB transcription factors (see 2.1.3) only explains part of Tax1-mediated NF-κB activation since this completion of this process also requires cytoplamic events. In the canonical pathway, Tax1 associates with the IKKγ/NEMO subunit [##REF##10438454##155##,##REF##10364167##156##] as well as with activating upstream kinases such as MAPK/ERK kinase kinase 1 (MEKK1) and TGF-β activating kinase 1 (TAK1) [##REF##9630230##157##,##REF##17363973##158##] (see 3.2). Tax1 thus connects activated kinases to the IKK complex and forces the phosphorylation of IKKα and IKKβ leading to degradation of IκBα and IκBβ [##REF##10438454##155##,##REF##10364167##156##]. In addition, Tax1 binds directly to the IKKα and IKKβ subunits and activates their kinase activity independently of the upstream kinases [##REF##9632633##159##]. Consistently, silencing of MEKK1 and TAK1 does not impair Tax1-induced NF-κB activation [##REF##17418100##160##]. A third level of Tax1 interference with the canonical pathway is its direct binding to IκBs and their degradation independently of IKK phosphorylation [##REF##8170951##161##,##REF##7700645##162##]. Tax1 further interacts two subunits of the 20S proteasome (HsN3 and HC9), favors anchorage of p105 and accelerates its proteolysis [##REF##8692272##163##]. Tax1 thus leads to IκB degradation at multiple levels, thereby allowing nuclear translocation of NF-κB independently of external stimuli. Besides, activation of the non-canonical pathway by Tax1 requires its interaction with IKKγ and p100 [##REF##17715223##152##,##REF##11726516##154##]. Through these interactions, Tax1 targets IKKα to p100, induces p100 processing and nuclear translocation of the p52/RelB dimer. It thus appears that IKKγ is an important Tax1 docking site for activation of both pathways.</p>", "<p>Post-translationnal modifications of IKKγ such as phosphorylation and K63 ubiquitination fine-tune NF-κB signaling [##REF##16056267##164##,##REF##17072324##165##] and are modulated by Tax1 through complex formation. In fact, PP2A activates the IKK complex by promoting dephosphorylation of IKKγ serine 68 [##REF##16126728##166##,##REF##17977820##167##]. Tax1 complexes with PP2A and IKKγ, maintaining the IKK complex in an active state that is required for activating NF-κB [##REF##12419799##168##,##REF##17314097##169##]. Ubiquitination is targeted by Tax1 through interaction with Ubc13 and Tax1BP1 [##REF##17942533##170##,##REF##18246070##171##]. Ubc13, an E2 ubiquitin-conjugating enzyme, is required for Tax1 interaction with IKKγ and subsequent NF-κB activation. Tax1BP1 participates to the formation of an ubiquitin-editing complex together with the deubiquitin enzyme (DUB) A20 and plays a pivotal role in termination of NF-κB and JNK signaling by regulating the activity of A20 [##REF##18246070##171##, ####REF##17703191##172##, ##REF##18239685##173####18239685##173##]. A20 inhibits IKK activation by cleaving K63 linked polyubiquitin chains on tumor necrosis factor receptor (TNFR) signaling-associated factor 6 (TRAF6), receptor interacting protein 1 (RIP1) and IKKγ [##REF##15653317##174##]. By disruption of complex A20-Tax1BP1, Tax1 inactivates DUB function of A20 and prevents downregulation of IKKγ ubiquitination. Consistent with this model, IKKγ is ubiquitinated in Tax1-expressing cells and in a series of HTLV-1 infected cell lines [##REF##17418100##160##,##REF##18246070##171##] providing a rationale for the constitutive activation of NF-κB pathway.</p>", "<title>3.2 Mitogen-activated kinases (MAPKs)</title>", "<p>MAPKs are serine/threonine-specific protein kinases that respond to external mitogen stimuli such as growth factors, cytokines or physical stress. MAPK signaling relies on a sequential phosphorylation cascade that goes through MAP kinase kinase kinase (MAP3K) to MAP kinase kinase (MAP2K) and finally to MAPK. The MAPK family includes the extracellular signal-regulated kinase protein homologues 1 and 2 (ERK1/2), ERK5, the c-Jun N-terminal Kinase 1, 2 and 3 (JNK1/2/3) also known as stress-activated protein kinase-1 (SAPK-1), the p38 isoforms (p38α/β/δ), ERK6, ERK3/4 and ERK7/8 [##REF##17957138##175##]. Tax1 interacts with two MAP3Ks: MEKK1 and TAK1 [##REF##9630230##157##,##REF##17363973##158##].</p>", "<title>3.2.1 MEKK1</title>", "<p>MEKK1 primarily regulates JNK and ERK1/2 but also contributes to the NF-κB pathway [##REF##9836645##176##,##REF##10852963##177##]. Tax1 binds to the amino terminus of MEKK1 and stimulates MEKK1 kinase activity [##REF##9630230##157##]. As a result, Tax1 expression increases IKKβ activity, leading to phosphorylation and degradation of IκBα. Dominant negative mutants of both IKKβ and MEKK1 prevent Tax1 activation of the NFκB pathway but, intriguingly, silencing of MEKK1 does not affect Tax1-induced NF-κB activation [##REF##17418100##160##].</p>", "<title>3.2.2 TAK1</title>", "<p>TAK1 is involved in JNK, TGF-β and NF-κB dependent signaling pathways [##REF##17218788##178##]. TAK1 acts in concert with TAK1 binding proteins (TABs) which link TAK1 to the upstream activating TNF receptor associated factor (TRAFs) proteins. TAK1 phosphorylates IKKβ and MKK6, thereby activating NF-κB and JNK [##REF##17496917##179##].</p>", "<p>TAK1 is constitutively activated in Tax1-expressing cells and in HTLV-1 infected lymphocytes [##REF##17363973##158##,##REF##17418100##160##,##REF##17626013##180##]. Tax1 activates TAK1 through complexation with TAK1 and TAB2 and connects TAK1 onto the IKK complex thereby stimulating IKK activity [##REF##17626013##180##,##REF##17986383##181##]. Consistently, overexpression of TAK1 or TAB2 increases Tax1 transactivation of a NF-κB reporter [##REF##17626013##180##,##REF##17986383##181##]. However, RNA interference of TAK1 suppresses activation of JNK and p38 but not NF-κB. Constitutive activation of TAK1 is thus not absolutely required for NF-κB activation [##REF##17418100##160##,##REF##17626013##180##]. TAK1 rather participates to JNK signaling, which is constitutively activated in Tax1-expressing cells, in Tax1-transformed murine fibroblasts and in human lymphocytes transformed with HTLV-1 [##REF##8700539##182##, ####REF##9325311##183##, ##REF##12393612##184##, ##REF##15169569##185####15169569##185##].</p>", "<title>3.3 GPS-2</title>", "<p>By linking the nuclear co-receptor (NCoR)-HDAC3 complex to intracellular JNK signaling, G protein pathway suppressor 2 (GPS2) suppresses Ras/MAPK signaling and JNK1 activation [##REF##9325311##183##,##REF##8943324##186##,##REF##11931768##187##]. Indeed, the NCoR-HDAC3 deacetylase activity represses transcription of genes involved in JNK signaling [##REF##11931768##187##]. Through interaction with GPS2, Tax1 potently inhibits GPS2-mediated inactivation of JNK signaling [##REF##9325311##183##]. Tax1 thus targets multiple proteins (i.e. TAK1 and GPS2) to constitutively activate JNK signaling.</p>", "<title>3.4 GTP-binding proteins</title>", "<p>The guanine nucleotide-binding proteins GTPases are molecular switches that cycle between active (GTP-bound) and inactive (GDP-bound) states. The G protein family includes Ras-related GTPases (or small GTPases) and heterotrimeric G proteins (α, β and γ subunits) that are activated by G protein-coupled receptors.</p>", "<title>3.4.1 Rho GTPases and the cytoskeleton proteins</title>", "<p>Tax1 complexes with several members of the small GTPase Rho family such as RhoA, Rac, Gap1m and Cdc42 [##REF##14530271##130##]. Rho GTPases are activated in response to external stimuli (e.g. growth factor, stress, cytokines) and exert a wide range of biochemical functions like cytoskeleton organization, regulation of enzymatic activities as well as gene expression [##REF##16212495##188##]. Notably, Tax1 binds to proteins involved in cytoskeleton structure and dynamics: α-internexin, cytokeratin, actin, gelsolin, annexin and γ-tubulin [##REF##14530271##130##,##REF##9435256##189##, ####REF##11725133##190##, ##REF##17891179##191####17891179##191##]. Through these interactions, Tax1 might thus connect Rho GTPases to their targets and affect cytoskeleton organization. Consistent with this idea, Tax1 localizes around the microtubule organization center (MTOC) and in the cell-cell contact region [##REF##15975923##192##]. Thereby, Tax1 provides an intracellular signal that synergizes with ICAM1 engagement to cause the T-cell microtubule polarization and formation of the virological synapse. Through the formation of complexes with both Rho GTPases and their targets, Tax1 could thus favor HTLV-1 cell-to-cell transmission.</p>", "<p>Since Rho GTPases modulate a wide range of signaling networks (SRF, JNK, p38 and NF-kB) [##REF##16212495##188##], complex formation with Tax1 is also likely to modulate transcription.</p>", "<title>3.4.2 Heterotrimeric Gβ subunit</title>", "<p>Heterotrimeric G proteins are the molecular switches that turn on intracellular signaling cascades in response to activation of G protein coupled receptor (GPCR). After binding of an agonist, the activated GPCR induces an exchange of GDP to GTP on the Gα subunit and facilitates the dissociation of GTP-bound Gβγ and Gα subunits [##REF##18043707##193##]. Through its interaction with Gβ, Tax1 affects SDF-1 dependent activation of the CXCR4 GPCR chemokine receptor. Tax1 enhances response to SDF-1 resulting in MAPK pathway over-activation and increased cell chemotaxis. The HTLV-1 associated pathologies (ATL, HAM/TSP and dermatitis) are characterized by invasion and accumulation of infected T-cells in organs such as lymph nodes, central nervous system or dermis [##REF##17388788##194##]. These results thus provide a rationale for the mechanisms of cell migration observed in HTLV-1 associated pathologies.</p>", "<title>3.5 Phosphatidylinositol 3-kinase and AP-1</title>", "<p>Phosphatidylinositol 3-kinase (PI3K) and its downstream effector Akt play a pivotal role in regulation of nutrient metabolism, cell survival, motility, proliferation and apoptosis. The PI3K family comprises eight members divided into three classes according to their sequence homology and substrate preference [##REF##9759495##195##,##REF##10579926##196##]. PI3K activation results in phosphorylation of Akt at Ser⁴⁷³which in turn triggers a broad range of regulatory proteins and transcription factors like AP1 [##REF##17979625##197##].</p>", "<p>PI3K-Akt is activated in Tax1-transformed murine fibroblasts and is required for cell transformation [##REF##11420661##198##]. Tax1 complexes with the p85α regulatory subunit of PI3K [##REF##16436385##199##] and inhibits activity of the p110α catalytic protein. p85α/p110α belong to the class IA PI3Ks and are activated by receptor tyrosine kinases, by Ras and Rho family GTPases and by Gβγ subunits from heterotrimeric G-proteins [##REF##12613196##200##]. Since monomeric p110 is unstable and is rapidly degraded, activation of p85α/p110α does not involve the complex dissociation but would rather depend on conformational changes [##REF##9804776##201##,##REF##9488453##202##]. Tax1 targets p85α and disrupts the p85α/p110α complex leading to increased PI3K activity [##REF##11784871##203##], Akt Ser⁴⁷³phosphorylation, AP1 activation and ultimately cell proliferation. Consistent with this model, ATL cells display constitutive activation of AP1 [##REF##16436385##199##,##REF##10845928##204##,##REF##11145887##205##].</p>", "<title>3.6 Smad proteins</title>", "<p>Transforming growth factor β (TGFβ) inhibits T cell growth in mid-G1 but can also promote tumorigenesis [##REF##17906455##206##]. TGFβ binds to a heterodimeric complex composed of type I (TβRI) and type II (TβRII) serine/threonine kinase receptors [##REF##18000526##207##]. Upon binding of a TGFβ ligand, TβRII recruits and activates TβRI, which, in turn, phosphorylates downstream targets such as Smad proteins (Smad1-2-3-5-8, receptor activated R-Smad). Common mediator Co-Smad (Smad4) containing complexes then translocate to the nucleus and activate transcription of genes under the control of a Smad-binding element. Signal termination is further mediated by inhibitory Smad (I-Smad) Smad6 and Smad7 [##REF##18000526##207##].</p>", "<p>Due to constitutive AP1 activation, ATL cells produce high levels of TGFβ in the sera of HTLV-1 infected patients [##REF##2358774##208##]. TGFβ does not inhibit growth of HTLV-1 infected CD4+ cells but affects CD8-dependent response a mechanism that may impact on immune surveillance [##REF##7586743##209##]. Furthermore, TGFβ stimulates cell surface expression of proteins involved in HTLV binding and fusion (Glut1), leading to enhanced virus transmission and production [##REF##12116037##210##,##REF##15778389##211##].</p>", "<p>Tax1 inhibits Smad-dependent signaling, thereby promoting resistance of HTLV-1 infected cells to TGFβ [##REF##12393612##184##,##REF##12097320##212##,##REF##11264182##213##]. This inhibition is mediated by Tax1 interaction with the aminoterminus of Smad2, Smad3, and Smad4 [##REF##12097320##212##]. Through these interactions, Tax1 inhibits complexation and DNA binding of Smad3-Smad4 [##REF##12393612##184##,##REF##12097320##212##]. Furthermore, Tax1 may compete with Smads for the recruitment of CBP/p300 [##REF##11264182##213##].</p>", "<title>3.7 Cas-L and p130Cas</title>", "<p>Proteins belonging to Crk-associated substrate (Cas) family are multiadaptator and scaffold molecules that spatially and temporally control signal transduction downstream of integrins, receptors protein tyrosine kinase, estrogen receptors and GPCRs. Upon binding of a ligand to these receptors, Cas proteins are tyrosine phosphorylated and recruit adaptors and effectors (such as small GTPase) to activate downstream targets such as JNK and ERK. As a result, Cas proteins regulate cell survival, apoptosis and migration. Furthermore, deregulation of Cas functions has been linked to cell transformation, invasion and cancer [##REF##16581250##214##].</p>", "<p>Among Cas proteins, Tax1 associates with p130Cas and CasL (lymphocyte type) [##REF##15592516##215##]. CasL, which is preferentially expressed in lymphocytes [##REF##8879209##216##], is phosphorylated and over-expressed in Tax1-expressing cells, in Tax1-transgenic mice as well as in primary lymphocytes isolated from ATL patients [##REF##15592516##215##,##REF##12847683##217##]. The Tax1 and CasL interplay results in enhanced motility of Tax1-expressing lymphocytes in response to fibronectin and CD3 [##REF##15592516##215##]. Since CasL also participates in RhoGTPase activation, Tax1 could interconnect cytoskeleton proteins, stimulate cytoskeleton rearrangement and enhance the motility of leukemic cells.</p>", "<title>3.8 Global effects of Tax1-mediated deregulation of cell signaling pathways</title>", "<p>As schematized on figure ##FIG##1##2##, Tax1 interactions with a series of components of several signaling pathways (MAPK, JNK, NF-κB, G proteins, AP1 and TGFβ) affect multiple cellular processes among which cellular activation, proliferation, cytoskeleton rearrangement, cell migration and formation of the virological synapse.</p>", "<title>4 Interaction of Tax1 with cell cycle associated proteins</title>", "<title>4.1 Cyclin D-CDK4/6 complexes, Rb and CDK inhibitors</title>", "<p>Cell cycle progression is a tightly regulated process controlled by cyclins associated with cyclin-dependent kinases (CDK). Cyclins D and E cooperate with CDK4/6 and CDK2 to mediate passage through G1 phase and G1/S transition, respectively [##REF##15576929##218##]. Cyclin D-CDK4/6 and Cyclin E-CDK2 complexes target the Rb retinoblastoma protein (Figure ##FIG##2##3##). In its hypophosphorylated form, Rb is bound to the transcription factor E2F1, and upon phosphorylation, Rb frees E2F1, which activates transcription of genes required for transition from G1 to S. G1/S progression can be inhibited by CDK inhibitors (CDKI) such as p15^INK4b, p16^INK4a, p18^INK4cand p19^INK4dby preventing cyclin D/CDK4/6 complex formation. Tax1 reprograms cell cycle progression, particularly at G1/S transition, through different mechanisms pertaining to transcriptional activation or repression, post-translational modifications and protein-protein interactions [##REF##15090550##219##,##REF##16155605##220##].</p>", "<p>Tax1 is able to interact with cyclins-D1, -D2 and -D3 as well as with CDK4 and CDK6 but not with CDK1 or CDK2 [##REF##11971966##221##, ####REF##11080810##222##, ##REF##16164752##223##, ##REF##9584203##224####9584203##224##]. Through these interactions, Tax1 stabilizes the cyclin D2/CDK4 complex and enhances its kinase activity, leading to hyperphosphorylation of retinoblastoma protein (Rb). Tax1 also associates with p15^INK4band p16^INK4aand counteracts their inhibitory activity of CDK4 [##REF##8612584##225##, ####REF##9209282##226##, ##REF##9032327##227##, ##REF##10388662##228####10388662##228##]. Finally, Tax1 binds to and targets Rb for proteosomal degradation [##REF##15580311##229##]. Consistently, HTLV-1 infected cell lines and freshly isolated ATL cells display decreased levels of Rb protein.</p>", "<p>Figure ##FIG##2##3A## illustrates Tax1 interactions with components of the cyclin D/CDK complexes and provides a mechanistic model for increased G1-S phase transition efficiency as well as the accelerated cell proliferation measured <italic>in vivo </italic>[##REF##11435441##230##,##REF##17483473##231##].</p>", "<title>4.2 DNA repair pathway associated proteins</title>", "<p>DNA insults and replication stress activate the DNA damage response (DDR) pathway in S and G2/M phases of the cell cycle. Activation of the DDR pathway leads to cell cycle delay or even apoptosis of severely damaged cells, and activates the DNA repair pathway. ATM (Ataxiatelangiectasia mutated) and ATR (ATM-Rad3) proteins and their respective downstream targets Chk2 (checkpoint kinase 2) and Chk1 (checkpoint kinase 1) proteins play a central role in the DDR pathway [##REF##11100718##232##]. In mammals, Chk1 and Chk2 regulate Cdc25, Wee1 and p53 that ultimately inactivate CDKs which inhibit cell-cycle progression. Double-strand breaks usually activate the ATM/Chk2-dependent pathway whereas ATR/Chk1 responds to a wide variety of lesions and replication blocks [##REF##14508077##233##,##REF##16314342##234##].</p>", "<p>Tax1-expressing and ATL cells display DNA damages suggesting that Tax1 abrogates cellular checkpoint and DNA repair [##REF##12145525##235##, ####REF##10969065##236##, ##REF##16014171##237####16014171##237##]. Tax1 binds Rad51 [##REF##14530271##130##] and DNA Topoisomerase 1 [##REF##10792988##135##] that are both directly involved in DNA repair processes [##REF##11100718##232##,##REF##18285803##238##]. Moreover, Tax1 associates and colocalizes with Chk1 and Chk2 proteins [##REF##12842897##239##, ####REF##15107832##240##, ##REF##16158050##241##, ##REF##17698850##242####17698850##242##]. <italic>De novo </italic>Tax1 expression causes phosphorylation of Chk2 resulting in accumulation of cells in S-G2/M [##REF##12842897##239##,##REF##11907241##243##]. However, upon gamma irradiation, Tax1 inhibits Chk1 and Chk2 kinase activities and attenuates G2/M arrest and apoptosis, respectively [##REF##15107832##240##,##REF##16158050##241##]. Tax1 thus activates and represses checkpoint controls depending on the experimental conditions (figure ##FIG##2##3B##). In fact, Tax1 sequesters phosphorylated Chk2 within chromatin after gamma irradiation-induced DNA-damage [##REF##17698850##242##]. Tax1 thereby impedes phosphorylated Chk2 chromatin egress, a mechanism required for further signal amplification and transmission [##REF##16150728##244##]. Tax1 thus targets multiple components of DNA damage repair pathway and promotes DNA abnormalities.</p>", "<title>4.3 Centrosome associated proteins and spindle assembly checkpoint</title>", "<p>One of the hallmarks of Tax1-expressing cells particularly in ATL is chromosomal instability and severe aneuploidy [##REF##12145525##235##], suggesting that mechanisms monitoring chromosomal segregation during mitosis are subverted by Tax1. Tax1 interacts with 4 proteins involved in centrosome amplification or in mitotic spindle assembly checkpoint (SAC) (Figure ##FIG##2##3C##).</p>", "<title>4.3.1 RanBP1 and Tax1BP2</title>", "<p>The presence of two centrosomes at mitosis is crucial for formation of bipolar mitotic spindles and correct chromosome segregation. Multipolar mitosis which happens when more than two centrosomes emerge in one cell is a possible cause of aneuploidy in solid tumors and leukemias [##REF##12415252##245##]. Supernumerary centrosomes are observed in approximately 30% of ATL cells [##REF##16805820##246##, ####REF##11986936##247##, ##REF##18004399##248####18004399##248##]. Tax1 colocalizes with the centrosome during mitosis and causes centrosome amplification through physical interaction with Ran/Ran Binding protein 1 (RanBP1) and Tax1BP2 [##REF##16365316##249##,##REF##16767081##250##]. RanBP1 is involved in the Ran GTP cycle that controls microtubule nucleation and/or stabilization and centrosome cohesion during mitosis [##REF##10965850##251##,##REF##12840069##252##]. Centrosome fragmentation requires direct Tax1/RanBP1 interaction and Tax1's ability to transactivate NF-κB. Tax1BP2 is thought to act as an intrinsic block to centrosome overreplication [##REF##12766773##253##]. Consistently, overexpression of Tax1BP2 abolishes Tax1-induced centrosome amplification. On the other hand, a Tax1 mutant unable to bind to Tax1BP2 is impaired in centrosome overreplication [##REF##16767081##250##].</p>", "<title>4.3.2 Mad1 and cdc20</title>", "<p>In eukaryotes, the mitotic spindle assembly checkpoint (SAC) monitors the fidelity of chromosome segregation [##REF##16631173##254##]. SAC functioning requires complex formation between Mad1-2-3 and Bub1-2-3 proteins that arrest mitosis in response to microtubule damage [##REF##14699129##255##]. At the molecular level, SAC activation involves formation of inhibitory complexes between Mad2 and/or Mad3/BubR1 and Cdc20, preventing Cdc20 from activating the anaphase promoting complex/cyclosome (APC/C). APC/C is active during mitosis where it mediates ubiquitination and degradation of an inhibitory chaperone of separase called securin. Once liberated from its inhibitor, separase triggers anaphase by hydrolysing cohesin leading to subsequent separation of sister chromatin. Furthermore, APC/C regulates the degradation of mitotic cyclin, activates CDK1 and, ultimately, promotes mitotic exit [##REF##17612486##256##].</p>", "<p>Through physical interactions with Mad1 and Cdc20, Tax1 subverts activation of SAC and APC/C. Tax1 inhibits Mad1 homodimerization, a process that is required for formation of a inhibitory complex between Mad2 and Cdc20 [##REF##11729202##257##, ####REF##9546394##258##, ##REF##16636141##259####16636141##259##]. Consistently, ATL cells exhibit a defect in the mitotic spindle assembly checkpoint [##REF##11729202##257##]. On the other hand, Tax1 associates with and activates Cdc20-associated APC/C leading to unscheduled degradation of securin and cyclin B1, a delay or failure in mitotic entry and progression, and faulty chromosome transmission [##REF##15623561##260##,##REF##12861013##261##]. Tax1-induced premature activation of APC/C provokes permanent G1 arrest and senescence [##REF##16601696##262##,##REF##17535428##263##]. Finally, coexpression of Tax1 and securin enhances chromosomal instability and favors cell transformation <italic>in vitro </italic>and <italic>in vivo </italic>[##REF##17507465##264##].</p>", "<title>5 Interaction of Tax1 with PDZ-containing proteins</title>", "<p>The PSD-95/<italic>Drosophila </italic>Discs Large/Zona Occludens-I (PDZ) domain containing proteins form signaling complexes at the inner surface of the cell membrane and are involved in a very broad range of functions like cell signaling, adhesion, tight-junction integrity, molecular scaffolding for protein complexes and tumor suppression [##REF##9604925##265##, ####REF##10449334##266##, ##REF##10079096##267####10079096##267##]. Numerous PDZ proteins have been shown to form a complex with Tax1 owing to its PDZ binding motif (PBM) located in the C-terminus (ETEV) [##REF##9482110##268##]: Pro-IL16 (precursor of interleukin 16) [##REF##12620798##269##], hDLG (<italic>Drosophila </italic>Discs Large) [##REF##9192623##270##,##REF##10557085##271##], PSD-95, beta-syntrophin, lin-7 [##REF##9482110##268##], Tip1 (Tax1 Interaction protein 1) [##REF##10940294##272##], MAGI3 (Membrane Associated Guanylate kinase with inverted orientation 3) [##REF##15003862##273##], hTid1 [##REF##11719219##274##] and hScrib [##REF##17855372##275##]. Interaction of Tax1 with these PDZ proteins frequently leads to their delocalization [##REF##15003862##273##,##REF##17855372##275##,##REF##14972558##276##]. Functionally, PDZ proteins such as hTid1 and hScrib participate to Tax1-mediated activation of NF-κB and NFAT pathways, respectively [##REF##11719219##274##,##REF##17855372##275##].</p>", "<p>A Tax1-binding protein, hDLG, has been particularly studied owing to its ability to act as a tumor suppressor. hDLG acts downstream of the Wnt/frizzled pathway and binds to the adenomatous polyposis complex (APC) which mediates cell cycle progression [##REF##10656683##277##,##REF##16158975##278##]. APC-hDLG complex formation negatively regulates G1 to S transition and plays an important role in transducing the APC cell cycle blocking signal [##REF##10656683##277##]. Besides, hDLG is also involved in maintenance and modulation of T cell polarity [##REF##17053799##279##]. Through PBM/PDZ domain interaction, Tax1 induces hyperphosphorylation of hDLG, affects its localization [##REF##14972558##276##] and prevents its binding to APC [##REF##10557085##271##]. Interestingly, hDLG inactivation increases the ability of Tax1 to transform a mouse T-cell line [##REF##17042961##280##].</p>", "<p>The Tax1 PBM is critically involved in transformation of rat fibroblasts and IL2 independent growth of mouse lymphocytes [##REF##14972558##276##,##REF##16042787##281##] and to promote virus-mediated T-cell proliferation <italic>in vitro </italic>and persistence <italic>in vivo </italic>[##REF##16263794##282##]. In contrast, HTLV-2 Tax2 protein which does not harbor a PBM has a lower transforming activity than Tax1 [##REF##11861831##283##].</p>", "<title>6 Tax1 interaction with nuclear pore and secretory pathway proteins</title>", "<p>Tax1 shuttles between the cytoplasm and the nucleus by virtue of a nuclear localization sequence (NLS) and a nuclear export signal (NES) [##REF##1736534##284##, ####REF##10666266##285##, ##REF##12670929##286####12670929##286##]. In the nucleus, Tax1 is primarily located in interchromatin granules or spliceosomal speckles [##REF##8709263##141##]. In the cytoplasm, Tax1 localizes to organelles associated with the cellular secretory process including the endoplasmic reticulum and Golgi complex [##REF##15975923##192##,##REF##15659397##287##]. Tax1 is also secreted in the supernatant of HTLV-1 infected cells isolated from HAM-TSP patients [##REF##15659397##287##, ####REF##1751450##288##, ##REF##17897946##289####17897946##289##] and may behave as an extracellular cytokine. Tax1 shuttling is mediated through interaction with proteins involved in nuclear import, cytoplasmic export and secretory pathways [##REF##17897946##289##, ####REF##17344183##290##, ##REF##16775353##291##, ##REF##18081936##292##, ##REF##17395420##293####17395420##293##].</p>", "<title>6.1 Nucleoporins</title>", "<p>Nucleoporins of the nuclear pore complex (NPC) form a channel spanning the double lipid bilayer of the nuclear envelope. Nuclear pore complexes allow passive diffusion of ions and small proteins but translocation of cargoes larger than 40 kDa generally requires specific transport proteins [##REF##14504656##294##]. Import of cargo proteins containing a classical NLS is mediated by the importin α/β dimer and requires metabolic energy which is provided by Ran GTP [##REF##18048681##295##]. In contrast, carrier-independent translocation of proteins into the nucleus is energy independent and requires direct interactions with nucleoporins [##REF##18048681##295##].</p>", "<p>Nuclear import and export of Tax1 are both carrier and energy independent but relies on the interaction between Tax1 and the p62 nucleoporin [##REF##17344183##290##]. This interaction is mediated by the aminoterminal zinc-finger motif of Tax1. Consistently, mutations within this motif abolishes Tax1 interaction with p62 and nuclear import [##REF##17344183##290##].</p>", "<title>6.2 Proteins involved in Tax1 nuclear export and secretion</title>", "<p>Proteins containing a NES domain like Tax1 are expected to interact with the chromosome region maintenance 1 protein (CRM1), a member of the importin β family [##REF##17317185##296##]. Under stress conditions (i.e. UV irradiation), Tax1 interacts with CRM1 and is exported outside of the nucleus, a mechanism that is inhibited by leptomycin B [##REF##16775353##291##,##REF##18081936##292##]. In the absence of stress however, leptomycin B does not alter subcellular distribution of Tax1 [##REF##12670929##286##], suggesting that Tax1 is not exclusively exported through the CRM1 pathway.</p>", "<p>Tax1 nucleo-cytoplasmic shuttling and secretion is directed by associations with proteins involved in nuclear export (calreticulin, RanBP2, p97), in ER to Golgi transport (the coat proteins (COP) βCOP and COPII) and in movement from Golgi to plasma membrane (secretory carrier membrane protein 23 (SNAP23), secretory carrier membrane protein 1 and 2 (SCAMP1, SCAMP2)) [##REF##17897946##289##,##REF##17395420##293##,##REF##16228291##297##]. Calreticulin, which is overexpressed in HTLV-1 infected cells, functions similarly to CRM1 by transporting proteins via NES interactions [##REF##17395420##293##,##REF##11149926##298##]. Tax1 secretion involves a secretory signal located in the C-terminal domain and requires interaction with SNAP23, SCAMP1 and COPII [##REF##17897946##289##].</p>", "<p>Tax1 thus targets different cellular factors involved in protein transport to shuttle between nucleus, cytoplasm and extracellular environment.</p>", "<title>7 Binding domains in Tax1</title>", "<p>To interact with such a broad range of cellular targets, Tax1 contains multiple protein-binding domains (Figure ##FIG##3##4##). Among these, the N-terminal zinc finger motif associates with transcription factors (CREB/ATF [##REF##7637025##299##], TBP [##REF##8223437##90##], Ets1 [##REF##9030555##62##], NF-YB [##REF##9032250##82##], Egr1 [##REF##9341193##85##]), cyclins [##REF##11971966##221##], nucleoproteins (p62) [##REF##17344183##290##], proteasome subunits [##REF##8692272##163##] and phosphatase PP2A [##REF##12419799##168##]. Mutations within this zinc finger affects Tax1-mediated CREB transactivation as well as subcellular localization due to the presence of a NLS [##REF##1736534##284##]. A domain encompassing residues 55 to 95 regulate interaction of Tax1 with CBP/p300, Chk2 and Gβ2 [##REF##10766811##102##,##REF##16158050##241##,##REF##16990599##300##]. The middle of Tax1 harbors a region required for dimerization, two leucine zipper-like motifs (aa 116–145 and 213–248) [##REF##9016568##39##,##REF##10906125##301##,##REF##14645559##302##] and a NES sequence [##REF##16775353##291##]. Substitutions within the first leucine zipper (such as T130A and L131S in mutant M22) affect Tax1 interactions with NF-κB [##REF##9630230##157##,##REF##10906125##301##], proteasome subunits [##REF##8692272##163##] and PP2A [##REF##12419799##168##]. Another mutation (S132A) abolishes Tax1 binding to coil-coiled domain containing proteins [##REF##11080811##303##] (i.e. Mad1, Tax1BP1, Tax1BP2 and GPS2). A region located between the two leucine zippers is required for interaction with CARM-1, Chk2 and Gβ2 [##REF##17005681##127##,##REF##16158050##241##,##REF##16990599##300##]. Amino acids 233–246, located within the second leucine zipper regulates Tax1 association with p15^INK4b[##REF##10388662##228##], p16^INK4a[##REF##9209282##226##], DNA topoisomerase [##REF##10792988##135##] and IκBγ [##REF##8170951##161##]. Consistently, the central region of Tax1 is indeed involved in NF-κB activation. Finally, the carboxyterminal region of Tax1 contains an activation domain (residues 289–332) [##REF##7853523##304##] as well as motifs required for Tax1 localization within the Golgi (residues 312–315) and secretion (residues 330–332) [##REF##16228291##297##]. The carboxyterminal domain is involved in Tax1 binding to Rb [##REF##15580311##229##], PI3K [##REF##16436385##199##], P/CAF [##REF##10766811##102##], P-TEFb [##REF##17686863##138##] and PDZ containing-proteins [##REF##9482110##268##]. In particular, Tax1 mutant M47 (L319R, L320S) is impaired for interaction with P/CAF[##REF##10766811##102##].</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>MB and LW collected data from the literature and wrote the paper, JT, SL and RK suggested comments, FD provided technical help. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>This work was supported by the Sixth Research Framework Programme of the European Union (project INCA LSHC-CT-2005-018704), the Belgian Foundation against Cancer, the Bekales Foundation, and the \"Fonds National de la Recherche Scientifique\" (FRS-FNRS and Télévie). MB (\"FRIA\" fellow), SL (\"FRIA\" fellow), RK (Research Director), LW (Research Director), JCT (Post-doctoral fellow) and JFD (Technician) are members of the FNRS. We thank Françoise Bex for comments and suggestions.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Global model of Tax1 mediated transactivation</bold>. Tax1 relieves transcriptional repression of the LTR through direct interaction with HDAC (i.e. HDAC1) and/or HMT (panel A). Tax1 recruits CREB/ATF transcription factors (CA in panel B), histone modifying enzymes and chromatin remodelers (SWI/SNF, P/CAF and CBP/p300). Tax1 then allows binding of basal transcription factors on the TATA box (panel C). Once the initiation complex is formed, Tax1 recruits the P-TEFb factor, leading to CTD phosphorylation and processive elongation (panel D). Finally, interaction of Tax1 with SWI/SNF prevents stalling of transcription elongation. Adapted from [##REF##15194748##120##,##REF##16547351##132##,##REF##17686863##138##,##REF##16155601##316##].</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Overview of cell signaling proteins targeted by Tax1</bold>. Tax1 interacts with components of several signaling pathways (MAPK, JNK, NF-κB, AP-1 and TGF-β) and promotes cellular activation, proliferation, cytoskeleton rearrangement, cell migration and formation of the virological synapse.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Effect of Tax1 on cell cycle progression</bold>. Through a series of interactions with cell-cycle associated proteins, Tax1 accelerates G1/S transition (<bold>A</bold>), attenuates Chk1/2 activity (<bold>B</bold>), induces supernumerary centrosomes and impedes SAC (spindle assembly checkpoint) activity (<bold>C</bold>).</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Functional regions of Tax1 and interaction domains</bold>. NLS (nuclear localization sequence), NES (nuclear export sequence), G (Golgi localization motif), S (secretion motif), LZR (leucine-zipper-like region), P (PDZ binding domain). Adapted from [##REF##14530271##130##].</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p>Overview of the Tax1 interactome.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p><bold>Identification of disordered regions in Tax1 according to the VSL1 algorithm</bold>. (PONDR^®, <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.pondr.com\"/>). A domain encompassing residues 76 to 121 (black bar) corresponds to a long disordered region within Tax1.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Cellular proteins interacting with Tax1</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\"><bold>Transcription and translation</bold></td><td align=\"left\"><bold>Cell signaling</bold></td></tr><tr><td colspan=\"1\"><hr/></td><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\">ATF1 [##REF##8628284##28##]</td><td align=\"left\">CasL [##REF##15592516##215##]</td></tr><tr><td align=\"left\">ATF2 [##REF##8407959##30##]</td><td align=\"left\">Cdc42 [##REF##14530271##130##]</td></tr><tr><td align=\"left\">ATF3 [##REF##8007991##27##]</td><td align=\"left\">Gap1m [##REF##14530271##130##]</td></tr><tr><td align=\"left\">ATF4 [##REF##9190894##29##]</td><td align=\"left\">GPS2 [##REF##9325311##183##]</td></tr><tr><td align=\"left\">ATFx [##REF##15890932##42##]</td><td align=\"left\">Gβ2 [##REF##16990599##300##]</td></tr><tr><td align=\"left\">BAF155 [##REF##14530271##130##]</td><td align=\"left\">IKKα [##REF##9632633##159##]</td></tr><tr><td align=\"left\">BAF53 [##REF##14530271##130##]</td><td align=\"left\">IKKβ [##REF##9632633##159##]</td></tr><tr><td align=\"left\">BAF57 [##REF##14530271##130##]</td><td align=\"left\">IKKγ [##REF##10364167##156##]</td></tr><tr><td align=\"left\">BRG1 [##REF##14530271##130##]</td><td align=\"left\">IκBα [##REF##7700645##162##]</td></tr><tr><td align=\"left\">C/EBPβ [##REF##9376596##81##]</td><td align=\"left\">IκBγ [##REF##8170951##161##]</td></tr><tr><td align=\"left\">CARM1 [##REF##17005681##127##]</td><td align=\"left\">MEKK1 [##REF##9630230##157##]</td></tr><tr><td align=\"left\">CREB [##REF##1386673##26##]</td><td align=\"left\">p100 [##REF##8289813##317##]</td></tr><tr><td align=\"left\">c-Rel [##REF##7936632##75##]</td><td align=\"left\">p105 [##REF##8692272##163##]</td></tr><tr><td align=\"left\">CREM [##REF##8628284##28##]</td><td align=\"left\">p130Cas [##REF##15592516##215##]</td></tr><tr><td align=\"left\">Cyclin T1 [##REF##17686863##138##]</td><td align=\"left\">p85α [##REF##16436385##199##]</td></tr><tr><td align=\"left\">DNA topoisomerase I [##REF##10792988##135##]</td><td align=\"left\">PP2A [##REF##12419799##168##]</td></tr><tr><td align=\"left\">Egr1 [##REF##9341193##85##]</td><td align=\"left\">Rac [##REF##14530271##130##]</td></tr><tr><td align=\"left\">Elk1 [##REF##11070040##63##]</td><td align=\"left\">RhoA [##REF##14530271##130##]</td></tr><tr><td align=\"left\">Ets1 [##REF##9030555##62##]</td><td align=\"left\">Smad2 [##REF##11264182##213##]</td></tr><tr><td align=\"left\">HDAC1 [##REF##12370815##119##]</td><td align=\"left\">Smad3 [##REF##11264182##213##]</td></tr><tr><td align=\"left\">Int6 [##REF##8688078##149##]</td><td align=\"left\">Smad4 [##REF##11264182##213##]</td></tr><tr><td align=\"left\">JMJD2A [##REF##15927959##125##]</td><td align=\"left\">TAB2 [##REF##17986383##181##]</td></tr><tr><td align=\"left\">MBD2 [##REF##15674330##123##]</td><td align=\"left\">TAK1 [##REF##17363973##158##]</td></tr><tr><td align=\"left\">MSX2 [##REF##15970589##89##]</td><td align=\"left\">Tax1BP1 [##REF##18246070##171##]</td></tr><tr><td align=\"left\">NF-YB [##REF##9032250##82##]</td><td align=\"left\">Ubc13 [##REF##17942533##170##]</td></tr><tr><td align=\"left\">NRF1 [##REF##10381170##88##]</td><td/></tr><tr><td align=\"left\">p/CAF [##REF##10567539##101##]</td><td/></tr><tr><td align=\"left\">p300/CBP [##REF##8602268##100##]</td><td/></tr><tr><td/><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\">p50 [##REF##8361755##61##]</td><td align=\"left\"><bold>PDZ proteins</bold></td></tr><tr><td/><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\">p52 [##REF##7856081##74##]</td><td align=\"left\">beta-syntrophin [##REF##9482110##268##]</td></tr><tr><td align=\"left\">RelA [##REF##7936632##75##]</td><td align=\"left\">hDLG [##REF##9192623##270##]</td></tr><tr><td align=\"left\">RPL6 [##REF##12007002##318##]</td><td align=\"left\">hScrib [##REF##17855372##275##]</td></tr><tr><td align=\"left\">Sap1 [##REF##11070040##63##]</td><td align=\"left\">hTid1 [##REF##11719219##274##]</td></tr><tr><td align=\"left\">Sc35 [##REF##8709263##141##]</td><td align=\"left\">lin7 [##REF##9482110##268##]</td></tr><tr><td align=\"left\">SMRT [##REF##12642864##122##]</td><td align=\"left\">MAGI3 [##REF##15003862##273##]</td></tr><tr><td align=\"left\">Sp1 [##REF##9030555##62##]</td><td align=\"left\">Pro-IL16 [##REF##12620798##269##]</td></tr><tr><td align=\"left\">Spi-1 [##REF##9376596##81##]</td><td align=\"left\">PSD95 [##REF##9482110##268##]</td></tr><tr><td align=\"left\">SRF [##REF##1427072##59##]</td><td align=\"left\">Tip1 [##REF##10940294##272##]</td></tr><tr><td align=\"left\">SUV39H1 [##REF##16409643##124##]</td><td/></tr><tr><td align=\"left\">TAF_II28 [##REF##9108034##91##]</td><td/></tr><tr><td align=\"left\">Tax1BP1 [##REF##17283140##144##]</td><td/></tr><tr><td/><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\">TBP [##REF##8223437##90##]</td><td align=\"left\"><bold>Transport</bold></td></tr><tr><td/><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\">TF_IIA [##REF##8756622##92##]</td><td align=\"left\">Calreticulin [##REF##17897946##289##]</td></tr><tr><td align=\"left\">TORC1 [##REF##16809310##44##]</td><td align=\"left\">COPII [##REF##17897946##289##]</td></tr><tr><td align=\"left\">TORC2 [##REF##16809310##44##]</td><td align=\"left\">CRM1 [##REF##16775353##291##]</td></tr><tr><td align=\"left\">TORC3 [##REF##15466468##43##]</td><td align=\"left\">p62 [##REF##17344183##290##]</td></tr><tr><td align=\"left\">TRBP [##REF##9060615##150##]</td><td align=\"left\">p97 [##REF##17897946##289##]</td></tr><tr><td align=\"left\">TTP [##REF##14679154##148##]</td><td align=\"left\">RanBP2 [##REF##17897946##289##]</td></tr><tr><td align=\"left\">XBP1 [##REF##18287238##31##]</td><td align=\"left\">SCAMP1 [##REF##17897946##289##]</td></tr><tr><td/><td align=\"left\">SCAMP2 [##REF##17897946##289##]</td></tr><tr><td/><td align=\"left\">SNAP23 [##REF##17897946##289##]</td></tr><tr><td colspan=\"1\"><hr/></td><td/></tr><tr><td align=\"left\"><bold>Cell cycle and DNA repair</bold></td><td align=\"left\">βCOP [##REF##17897946##289##]</td></tr><tr><td colspan=\"1\"><hr/></td><td/></tr><tr><td align=\"left\">CDC20 [##REF##15623561##260##]</td><td/></tr><tr><td align=\"left\">CDK4[##REF##11971966##221##]</td><td/></tr><tr><td align=\"left\">CDK6 [##REF##11971966##221##]</td><td/></tr><tr><td/><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\">Chk1 [##REF##15107832##240##]</td><td align=\"left\"><bold>Cytoskeleton</bold></td></tr><tr><td/><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\">Chk2 [##REF##12842897##239##]</td><td align=\"left\">Actin [##REF##14530271##130##]</td></tr><tr><td align=\"left\">Cyclin D1 [##REF##9584203##224##]</td><td align=\"left\">Annexin [##REF##14530271##130##]</td></tr><tr><td align=\"left\">Cyclin D2 [##REF##11971966##221##]</td><td align=\"left\">Cytokeratin [##REF##11725133##190##]</td></tr><tr><td align=\"left\">Cyclin D3 [##REF##9584203##224##]</td><td align=\"left\">Gelsolin [##REF##14530271##130##]</td></tr><tr><td align=\"left\">Mad1 [##REF##9546394##258##]</td><td align=\"left\">α-internexin [##REF##9190894##29##]</td></tr><tr><td align=\"left\">p15^INK4b[##REF##10388662##228##]</td><td align=\"left\">γ-tubulin [##REF##17891179##191##]</td></tr><tr><td align=\"left\">p16^INK4a[##REF##8612584##225##]</td><td/></tr><tr><td align=\"left\">Rad51 [##REF##14530271##130##]</td><td/></tr><tr><td align=\"left\">RanBP1 [##REF##16365316##249##]</td><td/></tr><tr><td/><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\">Rb [##REF##15580311##229##]</td><td align=\"left\"><bold>Proteasome</bold></td></tr><tr><td/><td colspan=\"1\"><hr/></td></tr><tr><td align=\"left\">Tax1BP2 [##REF##16767081##250##]</td><td align=\"left\">HC9 [##REF##8692272##163##]</td></tr><tr><td align=\"left\">Topoisomerase 1 [##REF##10792988##135##]</td><td align=\"left\">HSN3 [##REF##8692272##163##]</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1742-4690-5-76-1\"/>", "<graphic xlink:href=\"1742-4690-5-76-2\"/>", "<graphic xlink:href=\"1742-4690-5-76-3\"/>", "<graphic xlink:href=\"1742-4690-5-76-4\"/>", "<graphic xlink:href=\"1742-4690-5-76-5\"/>", "<graphic xlink:href=\"1742-4690-5-76-6\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Bovine Leukemia Virus (BLV) and Human T-Cell Leukemia Virus Type-1 (HTLV-1) are closely related deltaretroviruses that cause aggressive lymphoproliferative disorders in a small percentage of infected individuals [##UREF##0##1##, ####UREF##1##2##, ##REF##17362524##3####17362524##3##]. In order to efficiently enter cells, both viruses are dependent on a fusion event between viral and cell membranes. As with other retroviruses, fusion is catalyzed by the virally encoded Env complex, which is synthesized as a polyprotein precursor and is subsequently cleaved to yield the surface glycoprotein (SU) and transmembrane glycoprotein (TM) subunits. On the surface of the virus or infected cell, Env is displayed as a trimer, with three SU subunits linked by disulphide bonds to a spike of three TM subunits.</p>", "<p>The amino-acid sequences of the HTLV-1 and BLV envelope glycoproteins are strikingly similar [##REF##12368329##4##] and, in common with other oncoretroviruses, share a characteristic modular structure [##REF##12368329##4##, ####REF##7742034##5##, ##REF##10851227##6##, ##REF##12134000##7##, ##REF##14972546##8####14972546##8##]. A receptor-binding domain is located at the amino-terminal end of SU and is connected to a C-terminal domain by a proline-rich linker [##REF##12368329##4##,##REF##10851227##6##,##REF##15575958##9##]. The C-terminal domain includes a conserved CXCC sequence and is required for interactions with TM [##REF##10684314##10##, ####REF##14685283##11##, ##REF##15650193##12####15650193##12##]. The modular nature of envelope extends into TM, and it is here that the homology between retroviruses and phylogenetically diverse viral isolates is most apparent. The functional regions of TM include a hydrophobic fusion peptide linked to an isoleucine/leucine heptad repeat, a membrane spanning segment and a cytoplasmic tail of variable length. These conserved modules identify retroviral TM proteins as members of a diverse family of virally expressed class 1 membrane fusion proteins.</p>", "<p>Accumulating evidence advocates a conserved mechanism of retroviral envelope-mediated membrane fusion [##REF##10332732##13##, ####REF##11395423##14##, ##REF##14595397##15####14595397##15##]. SU binds to the cellular receptor, which is accompanied by isomerisation of the disulphide linkages between SU and TM [##REF##14685283##11##,##REF##15650193##12##], and triggers a conformational change in TM. The N-terminal hydrophobic fusion peptide of TM is then inserted into the target cell membrane, while the C-terminus remains anchored in the viral or host cell membrane. This transient rod-like conformation, referred to as a \"pre-hairpin\" intermediate, is stabilized by the assembly of a trimeric coiled coil composed of one alpha helix from each of the three adjacent TM monomers. A more C-terminal region of the TM ecto-domain, which in HTLV-1 includes an extended non-helical leash and short α-helix [##REF##10200260##16##], then folds onto the coiled coil to generate a six-helix bundle or trimer-of-hairpins [##REF##10200260##16##, ####REF##9108481##17##, ##REF##9689046##18##, ##REF##11447278##19####11447278##19##]. These dramatic conformational changes draw the opposing membranes together, destabilise the lipid bilayers, promote lipid mixing and culminate in membrane fusion [##REF##10332732##13##,##REF##11395423##14##].</p>", "<p>Despite the sequence homology and conserved modular structure, there are notable differences in primary sequence, size, and function of the HTLV-1 and BLV envelope proteins. It is likely that these differences contribute in a substantial way to the species-specificity, and the distinctive patterns of tissue tropism and pathogenesis that are observed for these viruses [##UREF##1##2##,##REF##17362524##3##]. Consequently, comparative analysis of the envelope glycoproteins will provide significant insight into the determinants of species- and tissue-specific tropism, the strategies for immune modulation, and the mechanisms of membrane fusion that are adopted by these viruses. Information derived from such studies will aid the development of effective vaccines and small-molecule inhibitors of viral entry and cell-to-cell viral transfer.</p>", "<p>Significantly, our laboratory [##REF##12584351##20##, ####REF##17251585##21##, ##REF##18305034##22####18305034##22##], and others [##REF##8627675##23##], have demonstrated that synthetic peptides that mimic the C-terminal non-helical <underline>l</underline>eash and α-<underline>h</underline>elical <underline>r</underline>egion (LHR) of HTLV-1 TM are inhibitory to envelope-mediated membrane fusion. Prototypic α-helical TM-mimetic inhibitory peptides have also been characterized for a number of highly divergent enveloped viruses, including HIV and paramyxoviruses [##REF##1438243##24##, ####REF##8312047##25##, ##REF##8700906##26##, ##REF##8521809##27####8521809##27##]. The HTLV-derived peptide binds to the coiled coil of TM and, in a <italic>trans</italic>-dominant negative manner, blocks resolution of the pre-hairpin intermediate to the trimer-of-hairpins, thus impairing the fusogenic activity of TM. The potency of these inhibitors makes them attractive leads for antiviral therapeutics.</p>", "<p>Although the HTLV-1 peptide inhibitor also blocks viral entry of the divergent HTLV-2 it is inactive against a variety of heterologous viral envelope proteins [##REF##12584351##20##,##REF##8627675##23##]. However, the molecular features that determine the target specificity, activity, and potency of these peptide inhibitors is only beginning to be understood [##REF##12584351##20##, ####REF##17251585##21##, ##REF##18305034##22####18305034##22##]. In this study, we examine the target specificity and activity of peptide inhibitors derived from the conserved C-terminal leash and α-helical region (LHR) of the HTLV-1 and BLV trans-membrane glycoproteins. We demonstrate that a synthetic peptide that mimics the BLV LHR is a potent antagonist of BLV envelope-mediated membrane fusion. Surprisingly, despite the high level of identity between the HTLV-1 and BLV derived peptides, the inhibitory activity of the peptides is limited exclusively to the virus from which they were derived. While the peptides display remarkable target specificity, the activity of each peptide is nevertheless dependent upon the interaction of conserved amino acid side chains with their respective targets. An amino acid substitution analysis reveals that several conserved residues within the BLV LHR play a critical role in determining peptide potency and identifies a single amino acid substitution within the BLV peptide that yields a more potent inhibitor. Finally, based on homology with HTLV-1 TM, the inhibition data and amino acid substitution analysis support a model for the BLV trimer-of-hairpins.</p>" ]

[ "<title>Materials and methods</title>", "<title>Cells</title>", "<p>HeLa and BLV-FLK (a kind gift of Dr Arsène Burny and Dr Luc Willems; Universitaire des Sciences Agronomiques de Gembloux, Belgium) cells were maintained in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum (FBS).</p>", "<title>Plasmids</title>", "<p>The Plasmid HTE-1 [##REF##2677310##28##] and pRSV-Rev [##REF##2120472##29##] have been described. The plasmid pCMV-BLV<italic>env</italic>-RRE was constructed by replacing a fragment of the HIV-1 envelope open reading frame in pCMVgp160ΔSA [##REF##8709194##30##] with a genomic fragment spanning the entire BLV envelope. In brief, pCMVgp160ΔSA was digested with EcoR I, which cuts the recipient vector after the CMV early promoter but prior to the initiating ATG of the HIV-1 <italic>env </italic>sequences. The vector was subsequently digested with BglII, which removes the HIV-1 SU region but retains the HIV RRE. A fragment encompassing the entire BLV envelope open reading frame between a 5' Xho I site and a 3' BamH I site (nucleotides 4347–6997 of NC_001414) was ligated into the vector backbone using an EcoR I-Xho I linker. The resulting plasmid encodes BLV <italic>env </italic>including the natural BLV <italic>env </italic>stop codon placed upstream of the HIV RRE; the transcription unit is terminated by the SV40 poly A site and is expressed from the CMV early promoter.</p>", "<title>Peptides</title>", "<p>Peptides (Table ##TAB##0##1##) were synthesized using standard solid-phase Fmoc chemistry and unless stated otherwise have acetylated N-termini and amidated C-termini. The peptides were purified by reverse-phase high-pressure liquid chromatography and verified for purity by MALDI-TOF mass spectrometry. All peptides were dissolved in dimethyl sulfoxide (DMSO), the concentration of peptide stock solutions was confirmed where possible by absorbance at 280 nm in 6 M guanidine hydrochloride and peptides were used at the final concentrations indicated. For the peptide P^BLV-ΔN, peptide concentration was estimated by Bradford assay at 5 two-fold serial dilutions from a stock solution using the P^BLV-ΔC peptide in concentrations verified by absorbance at 280 nm in 6 M guanidine hydrochloride to plot a standard curve. The HTLV-1-derived peptides are based on the sequence of HTLV-1 strain CR and conform closely to the consensus sequence for HTLV-1 and HTLV-2 strains, the BLV peptides conform to the consensus sequence for most BLV isolates.</p>", "<title>Peptide biotinylation</title>", "<p>Peptides to be biotinylated were reduced using immobilized Tris [2-carboxyethyl] phosphine (TCEP) reducing agent (Pierce), and subsequent biotinylation was carried out with EZ-Link^®Iodoacetyl-PEO₂-Biotin (Pierce), in both cases according to the manufacturer's protocols. The biotinylation reaction was quenched with cysteine. The biotinylated peptide was incubated for 30 mins at room temperature with either streptavidin-agarose (Gibco-BRL) or amylose-agarose (New England Biolabs) in a spin-column. Unbound peptide was recovered by centrifugation, the flow-through was re-applied to the column, and the incubation and centrifugation was repeated. The flow-through from the second centrifugation was used in syncytium interference assays; the peptide concentration of the amylose-agarose flow through was established by UV spectrometry as described above, and added to tissue culture medium to produce the final assay concentrations as indicated. In the case of the flow-through from the streptavidin-agarose column, volumes equivalent to those used with the amylose-agarose flow-through were added to the wells.</p>", "<title>Determination of relative peptide solubility</title>", "<p>A two-fold serial dilution of peptide in DMSO was performed, and added in duplicate to 96-well microplates. Filtered PBS was added to give a total volume of 200 μl and a final DMSO concentration of 1.5 % in all wells. The plates were incubated at room temperature for 1 hr and the relative solubility of peptides was established by measuring forward scattered light using a NEPHELOstar laser-based microplate nephelometer (BMG LABTECH). Wells containing PBS and 1.5 % DMSO only were used as blanks. Data analysis was carried out using ActivityBase, and peptides giving readings up to and including 3-fold higher than the average reading for the DMSO control were considered to be in solution at the concentrations specified.</p>", "<title>Syncytium Interference Assays</title>", "<p>Syncytium interference assays were performed by standard methods [##REF##12584351##20##,##REF##11520577##31##]. Briefly, HeLa cells for use as effector cells were transfected with the envelope expression vector pHTE-1 or with equal amounts of pCMV-BLV<italic>env</italic>-RRE and pRSV-Rev using the Genejuice™ transfection reagent (Novagen) in accordance with the manufacturer's instructions. 24 h later, 3 × 10⁵effector cells were added to 7 × 10⁵untransfected HeLa target cells in six-well dishes (Nunc). Where appropriate, the co-culture was incubated in the presence of peptides at the concentrations specified. To assess the ability of the peptides to inhibit fusion induced by virally expressed BLV envelope, 2 × 10⁵BLV-infected FLK cells were used as effectors and added to 8 × 10⁵uninfected HeLa cells. After incubation at 37°C for 16 h, cells were washed twice with PBS and fixed in PBS + 3% paraformaldehyde. Assays were performed in triplicate and the number of syncytia (defined as multinucleated cells with 4 or more nuclei) from 10 low-power fields (LPF) per replicate was scored by light microscopy; some assays were stained using Giemsa. A syncytium formation value of 100% is defined as the number of syncytia formed in the absence of peptide but in the presence of 1.5% DMSO. The peptide concentration required to give 50% inhibition (IC₅₀) of syncytium formation was calculated using GraphPad Prism 4.</p>" ]

[ "<title>Results</title>", "<title>Amino acid residues conserved between the HTLV-1 and BLV TM ectodomains map to the interacting surfaces of the LHR and coiled-coil</title>", "<p>Although there are considerable differences in the amino acid sequence of class-1 fusion proteins from diverse viral groups there is exceptional conservation of secondary and tertiary structure. To compare the class-1 fusion proteins from the related retroviruses BLV and HTLV-1, the predicted coiled-coil regions of the BLV TM were identified using the program LearnCoil-VMF [##REF##10438601##32##] and the BLV and HTLV-1 amino acid sequences were aligned using Clustal-W [##REF##7984417##33##] (Figure ##FIG##0##1A##). The alignment revealed that for the TM 33% of the residues are identical and a further 10% are conservative substitutions. The homology is particularly evident in the predicted coiled-coil region incorporating the heptad repeat and in the LHR of the TM ectodomain (Figure ##FIG##0##1A##), the LHR lies distal to a CX₆CC motif common to oncoretroviral fusion proteins. The crystal structure of the HTLV-1 six-helix-bundle has been solved and the structure spans these regions of homology [##REF##10200260##16##].</p>", "<p>Using the crystal structure of the HTLV-1 TM as a template, we mapped on the coiled coil and LHR the location of amino acid residues that are conserved between the ectodomain of HTLV-1 and BLV TM (Figure ##FIG##0##1B##). Using this approach, we observed that for the core coiled-coil the majority of conserved residues map along the grooves formed by the interface of each pair of interacting N helices. Importantly, these grooves act as docking sites for the LHR as TM folds from the pre-hairpin intermediate to the trimer-of-hairpins. Moreover, many of the conserved amino acids of the LHR are located on the face of the LHR that interacts with the grooves on the coiled coil. By examining the location of substituted residues on the HTLV-1 TM it becomes clear that where there are amino acid substitutions on the BLV LHR there are complimentary or accommodating amino acid changes within the hydrophobic grooves of the core coiled coil (Figure ##FIG##0##1C##). For example, leucines 413 and 419 in the HTLV-1 LHR are conserved in BLV, and these leucines interact with eight coiled coil residues of which seven are identical in BLV and one is a conservative substitution (Figure ##FIG##0##1C##). In contrast, HTLV-1 LHR residues H409 and R416 interact with the side chains of six residues of the coiled coil, but H409 and R416 are not conserved in BLV and of the six interacting coiled coil residues four have diverged and only one residue is semi-conserved (Figure ##FIG##0##1C##). Overall, the analysis indicates that the majority of the conserved residues occupy positions that form the interacting surfaces of the trimer-of-hairpins. In agreement with these observations, those residues that do not involve the interacting surfaces of the TM are invariably solvent exposed on the trimer-of-hairpins and are subject to the highest degree of variation between the two viruses.</p>", "<p>A synthetic peptide, P^cr-400, which mimics the LHR of the HTLV-1 TM is a potent inhibitor of envelope-catalysed membrane fusion [##REF##12584351##20##]. This peptide interacts directly and specifically with a recombinant coiled coil derived from HTLV-1 TM and substitution of critical amino acid residues within the peptide disrupts coiled coil binding and impairs the biological activity of the peptide [##REF##12584351##20##, ####REF##17251585##21##, ##REF##18305034##22####18305034##22##]. These findings are consistent with the view that the peptide blocks membrane fusion by binding to the coiled coil of fusion-active envelope. As illustrated above, there are remarkable similarities in the interacting surfaces of the coiled coil and LHR between HTLV-1 and BLV (Figure ##FIG##0##1##). Considering the noted differences, it was not clear if the HTLV-1-derived synthetic peptide could inhibit membrane fusion mediated by BLV envelope. The HTLV-1 peptide inhibits viral entry by the divergent HTLV-2 but does not inhibit membrane fusion catalysed by a number of heterologous viral envelopes including HIV-1, feline immunodeficiency virus and vesicular stomatitis virus G protein [##REF##12584351##20##,##REF##8627675##23##] (our unpublished results). Moreover, the HTLV-1 inhibitory peptide is unusual among C helix-based fusion inhibitors in that it includes both α-helical and extended non-helical peptide segments. It was therefore uncertain if peptides based on the LHR of BLV would, like the HTLV-mimetic peptide, display anti-fusogenic activity. We therefore compared the fusogenic activity of HTLV-1 and BLV envelope and examined the sensitivity of BLV envelope to inhibition by peptide inhibitors.</p>", "<title>A robust BLV Env-mediated membrane fusion assay</title>", "<p>Preliminary experiments with a variety of BLV envelope expression constructs produced only low levels of BLV envelope expression and little fusogenic activity in syncytium formation assays (data not shown); this may, in part, be due to the nuclear retention of the envelope transcripts as observed for HIV-1 and HTLV-1. Therefore, we developed an envelope expression vector whereby BLV <italic>env </italic>was inserted downstream of the strong cytomegalovirus (CMV) early promoter, and immediately upstream of the human immunodeficiency virus Rev-response element (RRE). The RRE forms a region of extensive secondary structure in the mRNA that is recognized by Rev and the resulting ribonucleoprotein complex is subsequently exported out of the nucleus. The BLV envelope expression construct was examined for envelope-induced membrane fusion in syncytium formation assays. Briefly, HeLa cells were either transfected with pCMV-BLV<italic>env</italic>-RRE or pRSV-Rev individually, or cotransfected with equal amounts of both vectors. These cells were then used as effector cells to induce syncytia when co-cultured with non-transfected cells. Neither vector induced syncytium formation when transfected alone, but cotransfection of effector cells with pCMV-BLV<italic>env</italic>-RRE and pRSV-Rev resulted in the widespread formation of large syncytia (Figure ##FIG##1##2##). Furthermore, BLV envelope expressed in this system produced levels of syncytia that were comparable to that of HTLV-1 envelope expressed from pHTE-1 and consequently this approach was used to express BLV envelope for these studies.</p>", "<title>Inhibition of envelope-mediated membrane fusion by LHR-mimetic peptides is limited to the parental virus</title>", "<p>To compare the inhibitory properties and specificity of LHR-based synthetic peptides from HTLV-1 and BLV a peptide based on the LHR of BLV was generated. The synthetic peptide designated P^BLV-391 includes residues Cys391 to Gln419 of BLV Env and spans a region that is equivalent to the HTLV-1 LHR-derived peptide P^cr-400 (Table ##TAB##0##1##). To aid comparison with TM, we refer to the residues of each peptide using the co-ordinates for the full-length envelope precursor (thus for the BLV-derived peptide residue 1 is referred to as Cys391). The BLV and HTLV-1 peptides share 45 % identity (Figure ##FIG##0##1A, B##), but it should be noted that only a fragment of the HTLV-1 LHR that is mimicked by P^cr-400 is resolved in the available HTLV-1 TM crystal structure (Table ##TAB##0##1##, Figure ##FIG##0##1##) [##REF##12584351##20##].</p>", "<p>Both HTLV-1 and BLV envelope induced widespread syncytium formation in cultures incubated in the absence of peptide inhibitors or in the presence of inactive control peptides (Figure ##FIG##2##3A, B##). However, in keeping with previous studies [##REF##12584351##20##, ####REF##17251585##21##, ##REF##18305034##22####18305034##22##], HTLV envelope-mediated syncytium formation was robustly blocked in a dose-dependent manner by P^cr-400 with an IC₅₀of 0.28 ± 0.01 μM (Figure ##FIG##2##3A##). However, despite the marked conservation of amino acid sequence between the LHRs and coiled coils of HTLV-1 and BLV, P^cr-400 failed to inhibit membrane fusion induced by BLV envelope even at concentrations up to 15 μM (Figure ##FIG##2##3B##) and above (data not shown). Also, like the inactive control peptides, the BLV LHR-mimetic peptide at concentrations up to 20 μM (Figure ##FIG##2##3A##) and above (data not shown) failed to inhibit membrane fusion induced by HTLV-1 envelope. By contrast, the peptide P^BLV-391 specifically antagonized BLV envelope-mediated membrane fusion (Figure ##FIG##2##3B##) with a calculated IC₅₀of 3.49 ± 0.03 μM; control peptides including C34 and P^cr-400 L/A did not interfere with BLV Env-induced membrane fusion (Figure ##FIG##2##3B##). In addition, P^BLV-391 robustly antagonized membrane fusion induced by virally expressed envelope as shown by the inhibition of syncytium formation between chronically BLV infected FLK cells and target cells (Figure ##FIG##2##3C##); whereas, the HTLV-1 peptide inhibitor did not block BLV-induced membrane fusion. Thus, it appears that the inhibitory properties of the LHR-mimetic peptides are highly specific to the virus from which they were derived.</p>", "<title>The C- and N-terminal regions of P^BLV-391 are necessary but not individually sufficient to block membrane fusion</title>", "<p>Our group recently demonstrated that truncations at the N- or C-terminal end of P^cr-400 abolished fusion-inhibitory function [##REF##2120472##29##]. To test whether or not the N- and C-terminal leash regions are required for the activity of P^BLV-391, we synthesized two peptides, P^BLV-ΔN and P^BLV-ΔC, which lack nine amino acid residues at the N-terminus or C-terminus respectively (Table ##TAB##0##1##). The peptides retain an eleven-residue overlap, and have solubility profiles comparable to the parental peptide P^BLV-391 (Table ##TAB##0##1##). Unlike the parental peptide, the peptide derivatives P^BLV-ΔN and P^BLV-ΔC lacked detectable inhibitory activity in syncytium interference assays (Figure ##FIG##3##4A##). These data illustrate that amino acid residues contained within the regions Cys391 to Asp399, and Ser411 to Gln419, are critical to the activity of the mimetic peptide, and that both the amino-terminal and C-terminal regions are necessary but not sufficient for antagonism of membrane fusion. Importantly, the data also demonstrate that the central 11-residue region of the BLV peptide, equivalent to Ser400-Leu410 and homologous to the short C-terminal α-helix of the HTLV-1 trimer-of-hairpins is not sufficient for inhibition of syncytium formation.</p>", "<p>Moreover, the BLV peptide was remarkably intolerant of even relatively small deletions. For example, a peptide, P^BLV-ΔCCF, in which only 3 amino acids were deleted from the N-terminus exhibited dramatically reduced ability to inhibit membrane fusion (Figure ##FIG##3##4B##). The P^BLV-ΔCCF peptide blocked syncytium formation by only 30% at 20 μM (Figure ##FIG##3##4B##), compared to &gt; 95% for the parental peptide, and even at a concentration of 30 μM peptide P^BLV-ΔCCF achieved only 40% inhibition (data not shown). These results can be explained only in part by the decrease in peptide solubility at concentrations above 11 μM that is associated with the loss of the three N-terminal amino acid residues (Table ##TAB##0##1##). At peptide concentrations below 11 μM, P^BLV-ΔCCF is soluble under the conditions used in the syncytium interference assays and yet fails to inhibit membrane fusion (Figure ##FIG##3##4B##). It should be noted that disulphide formation between the peptide and envelope is not required for inhibitory activity, as reduction of P^BLV-391 and subsequent modification of the cysteine residues with the sulfhydryl reactive agent Iodoacetyl-PEO₂-Biotin failed to disrupt the inhibitory properties of the peptide (Figure ##FIG##3##4C##). Moreover, the activity of the biotinylated peptide was indistinguishable from that of the unmodified P^BLV-391, indicating that potential dimerization of the peptide through inter-molecular disulphide bonding does not influence peptide potency (Figure ##FIG##3##4C##). The first 3 amino acids of the BLV peptide, which includes the two cysteine residues and an adjacent phenylalanine, are conserved between HTLV-1 and BLV. Given the data obtained for the BLV peptide it is surprising to note that substitution of the cysteines with alanine did not affect the activity of the HTLV-1 inhibitor P^cr-400 [##REF##18305034##22##]. Thus it seems that, at least for the BLV peptide, the first 3 amino acids aid peptide solubility and contribute in an important but, as yet, ill-defined way to the binding or orientation of the peptide within the target-binding site on TM.</p>", "<title>Two conserved leucines are essential for the inhibitory activity of P^BLV-391</title>", "<p>Leucine residues in P^cr-400 play a key functional role in peptide activity [##REF##12584351##20##]. The crystal structure of the HTLV-1 TM [##REF##10200260##16##] reveals that within the LHR several leucine and isoleucine residues reach down into deep pockets within the groove of the coiled coil. It appears that the LHR-derived peptide P^cr-400 makes similar contacts with the coiled coil and that these contacts are necessary for stable binding of the peptide to the coiled coil and thus are critical to the inhibitory activity of the peptide [##REF##18305034##22##]. Intriguingly, some but not all of these leucine and isoleucine residues are conserved between the LHRs of HTLV-1 and BLV. We therefore sought to determine the importance of these conserved residues to the inhibitory properties of the BLV LHR-mimetic peptide. Two peptides were synthesized, P^BLV-L/A in which all leucines were substituted with alanine, and P^BLV-L404/410A in which the Leu404 and Leu410 of BLV envelope were replaced by alanine (Table ##TAB##0##1##) these particular leucines are equivalent to the well-conserved Leu413 and Leu419 of HTLV-1 isolates. Syncytium interference assays revealed that compared to the parental peptide (P^BLV-391) the alanine-substituted peptides were severely compromised in their ability to inhibit membrane fusion (Figure ##FIG##3##4D##); in particular, P^BLV-L/A did not exhibit any discernible inhibition up to 20 μM (Figure ##FIG##3##4D##) or above (data not shown). Hence, the leucine residues are important to peptide function. Moreover, although P^BLV-L404/410A was just as soluble as the parental peptide (Table ##TAB##0##1##), P^BLV-L404/410A also failed to display any fusion-blocking activity up to 20 μM (Figure ##FIG##3##4D##); indicating that the leucines equivalent to BLV envelope residues 404 and 410 are particularly important to the inhibitory properties of the LHR-mimetic peptide.</p>", "<title>A model for the BLV trimer-of-hairpins</title>", "<p>Our analysis reveals that for the ectodomain of the TM the majority of the amino acid residues that are conserved between HTLV-1 and BLV map to the interacting surfaces of the trimer-of-hairpins. Moreover, a BLV homologue of the HTLV-1 LHR-derived peptide inhibitor also exhibits robust but highly specific inhibitory activity against BLV-induced membrane fusion. Significantly, conserved leucine residues are critical to the inhibitory activity of both peptides. Encouraged by these results and to gain greater insight into the mechanism of fusion and the likely contacts made by P^BLV-391 with the coiled coil, we constructed a homology model of the BLV trimer-of-hairpins that is based on the crystal structure of the HTLV-1 TM (Figure ##FIG##0##1B##) [##REF##10200260##16##].</p>", "<p>Having identified the predicted BLV coiled-coil (Figure ##FIG##0##1A##), the Clustal-W alignment of the TM ectodomain sequences of BLV and HTLV-1 (Figure ##FIG##0##1A##) permitted the substitution of the BLV residues onto the HTLV-1-derived scaffold, consisting of the complete trimer of N-helices and a single LHR. The geometry of the crude model was improved by simulated annealing and energy minimisation in explicit solvent with the GROMACS (Groningen Machine for Chemical Simulations) package using the GROMOS96 43a1 force field [##UREF##2##34##]. It should be noted that, compared to the HTLV-1 trimer of hairpins, there are two additional residues in the predicted BLV chain-reversal region at positions 380 and 381 of BLV envelope. Since these residues are within a flexible loop there is insufficient information to model these residues with any degree of accuracy therefore these residues are omitted in the current model. Nonetheless, the restraint provided by the disulphide bond between Cys384 and Cys391 coupled with a high level of sequence conservation within the heptad repeat region and within the LHR suggests that the model is likely to be a reasonably accurate representation of the interaction between the LHR and the coiled coil. The model for the BLV coiled coil and LHR is presented in Figure ##FIG##4##5A##.</p>", "<p>Consistent with the sequence alignment and the structure of the HTLV-1 TM ectodomain (Figure ##FIG##0##1##), the BLV TM model indicates that Leu394 and Ile396 likely project into a hydrophobic pocket at the membrane-distal end of the core coiled-coil (Figure ##FIG##4##5B##). It also implies that Ile401, Leu404 and Leu407, which all lie on the same side of the putative α-helix of the LHR, are oriented such that they project into the groove of the coiled coil. Notably, Leu410 is predicted to make a significant contact with a deep pocket situated towards the membrane-proximal end of the core coiled-coil. Therefore, the BLV coiled coil and LHR model is highly consistent with the experimental data and provides a molecular explanation for the loss of activity associated with substitutions in the BLV LHR-derived peptide.</p>", "<title>Substituting an arginine residue for an alanine in P^BLV-391 results in a more potent peptide inhibitor</title>", "<p>The accumulated experimental data correlate well with the structural model, implying that predications based on the BLV trimer-of-hairpins model are likely to be informative. The homology model of the BLV TM ectodomain (Figure ##FIG##5##6##) suggests that Arg403, a residue within the predicted α-helix of the LHR and mimicked by P^BLV-391 peptide, may be electrostatically unfavourable for efficient binding of the C-terminal LHR into the groove of the core coiled-coil. We predicted that removing this unfavourable charge interaction would improve the binding of the peptide to the BLV coiled coil and thereby improve the inhibitory activity of the peptide. We therefore synthesized a peptide, P^BLV-R403A, which incorporated an alanine residue in place of the arginine equivalent to Arg403 of Env (Table ##TAB##0##1##). As anticipated, substitution of the arginine residue resulted in a modest but highly consistent and significant (<italic>p </italic>&lt; 0.0001, Student's <italic>t</italic>-test) improvement in peptide potency when compared to P^BLV-391. The peptide P^BLV-R403A is more than twice as potent as P^BLV-391 in syncytium interference assays, with a calculated IC₅₀of 1.56 ± 0.05 μM compared to 3.49 μM ± 0.03 μM for P^BLV-391 (Figure ##FIG##5##6##). The data show that a single amino-acid substitution in the predicted short α-helix of the LHR-mimetic peptide increases the ability of the peptide to block membrane fusion and provides further support for the utility of the model of the BLV TM core.</p>" ]

[ "<title>Discussion</title>", "<p>Experimental evidence points towards a remarkably conserved mechanism by which virally encoded envelope glycoproteins catalyse membrane fusion and facilitate delivery of the viral core into the target cell [##REF##10332732##13##,##REF##11395423##14##]. The structures of several class 1 fusion proteins reveal a characteristic \"trimer-of-hairpins\" motif believed to represent a late or post-fusion conformation [##REF##10200260##16##, ####REF##9108481##17##, ##REF##9689046##18##, ##REF##11447278##19####11447278##19##,##REF##9844633##35##, ####REF##10077567##36##, ##REF##11038187##37####11038187##37##]. Investigating the way in which envelope proteins fold from a rod-like, pre-hairpin intermediate into the trimer-of-hairpins to pull the viral and cellular membranes together is important not only for our understanding of viral entry but also for the development of therapeutically relevant inhibitors of this process.</p>", "<p>The protein sequences of the TM ectodomains of BLV and HTLV-1 display a striking level of conservation. By scrutinizing the position of conserved residues in the context of the HTLV-1 six-helix-bundle structure, we have found that the majority of the conserved residues map to the interacting surfaces of the LHR and core coiled-coil. It is interesting to note that there are several non-conserved residues within the LHR of each virus; significantly, these modifications are mirrored by compensating substitutions within the specific area of the core coiled-coil with which the variant residue interacts (Figure ##FIG##0##1C##) and consequently, the association with the coiled coil is maintained. It appears that in order to support variation and speciation but to maintain biological function complementary regions of the fusion proteins have evolved in parallel. The greatest functional constraint and therefore most highly conserved regions map along the interacting surfaces of the trimer-of-hairpins. Conversely, regions of the TM that are likely exposed to the aqueous environment both during and after fusion exhibit considerable divergence and display relatively few amino acids in common. Such changes may reflect strong selective pressures exerted on the virus, perhaps due to the need for particular regions of the TM to interact functionally with the relatively divergent surface glycoproteins of the respective viruses. Alternatively, the selective pressure may be due to the differing immunological environments of the respective hosts. It is worth noting, that the TM and the trimer-of-hairpins of HTLV-1 are immunogenic [##REF##9223472##38##,##REF##17376912##39##], that antibodies targeting TM often recognise non-neutralizing conformational epitopes [##REF##17376912##39##,##REF##17940280##40##], and that trimer-of-hairpin structures are frequently displayed on the surface of infected cells [##REF##17940280##40##]. Whether or not these features of the TM contribute to the pathogenesis or immune evasion of leukaemia viruses remains to be determined.</p>", "<p>The HTLV-1-derived LHR-based peptide is able to inhibit membrane fusion mediated by the divergent envelope of HTLV-2 and, given the level of conservation between the HTLV-1 and BLV TM ectodomain, we anticipated that the HTLV-1-derived peptide P^cr-400 would also inhibit the fusogenic activity of BLV envelope. Surprisingly, although P^cr-400 is an extremely effective inhibitor of HTLV-1-mediated fusion, the peptide had no detectable activity in BLV syncytium interference assays. Moreover, the BLV LHR-based peptide P^BLV-391 does not inhibit HTLV-1 envelope-catalysed syncytium formation. Sequence alignment and homology modelling (Figures ##FIG##0##1## and ##FIG##4##5##) indicate that within the first eight residues only two residues differ between the HTLV-1 and BLV peptides and these residues are likely to be solvent exposed and unable to contribute to the interaction with the core coiled-coil. The residues that determine the specificity of inhibition are therefore located within or overlapping the short α-helix or C-terminal leash segments of the peptide. In terms of peptide function, it is clear that the putative α-helix within the central region of these peptides is important for inhibitory activity. Nonetheless, both the N- and to the C-terminal leash residues contribute to the inhibitory properties of the peptide as deletion of these regions severely attenuates inhibitory activity. The structure of residues C-terminal of Asn421 in the HTLV-1 TM (equivalent to Gln412 of BLV Env) has not been resolved [##REF##10200260##16##]. Consequently, it is not yet possible to account in molecular terms for the conserved interactions beyond this point. However, our data highlight a number of features that play a key role in the biological activity of the BLV-derived peptide. The first three N-terminal amino acid residues appear to be critical to activity. Given the orientation of the phenylalanine residue in the BLV TM model and the equivalent Phe402 in the crystal structure of HTLV-1 TM, it is unlikely that this side chain directly contributes to the interaction with the coiled coil. Consistent with this view, Maerz et al. [##REF##10864675##41##] have demonstrated that Phe402 likely plays a structural role in pre-fusogenic envelope and is required for envelope processing, but likely becomes solvent exposed during assembly of the fusion-associated trimer-of-hairpins structure [##REF##10864675##41##]. Furthermore, although disulphide bonding regulates TM function [##REF##14685283##11##,##REF##15650193##12##] and association with the SU subunit [##REF##10684314##10##], the adjacent cysteines at the N-terminus of P^cr-400 are not required for disulphide formation, for binding to the coiled-coil, or for inhibitory activity [##REF##18305034##22##]. Similarly, modification of the adjacent cysteine residues in the BLV-derived peptide reveals that disulphide formation is not required for coiled coil binding or inhibition of membrane fusion. The apparent requirement for the cysteine residues for functional activity of the BLV-derived peptide may reflect an intrinsic difference between BLV and HTLV-1 peptide target interactions. Currently, our preferred view is that the N-terminus of the BLV peptide aids alignment of the adjacent peptide sequences relative to the target-binding site on the coiled coil.</p>", "<p>A recurring theme in the interaction of the C-terminal helix of the trimer-of-hairpins with the coiled coil of viral fusion proteins is the interaction of non-polar side chains with deep pockets on the coiled coil [##REF##10200260##16##, ####REF##9108481##17##, ##REF##9689046##18####9689046##18##,##REF##9844633##35##,##REF##10077567##36##,##REF##10430879##42##]. The model for the BLV trimer-of-hairpins suggests that this is also the case for BLV and this interpretation is supported by the peptide inhibition data. The model suggests that a series of leucine residues, which include L404, L407 and L410, make contact with the coiled coil. Moreover, the inhibitory activity of P^BLV-391 is completely abrogated following substitution of all the leucine residues with alanine. Similar results have been observed for the P^cr-400 inhibitor of HTLV-1 [##REF##12584351##20##]. In particular, two leucines, Leu413 and Leu419, are important for the inhibitory activity of P^cr-400 [##REF##18305034##22##]. Leucine 413 is situated within the short α-helix, whereas Leu419 is situated within the C-terminal leash-like domain. Significantly, both of these leucine residues are conserved in BLV, at positions 404 and 410 respectively, and the model for the BLV trimer-of-hairpins suggests that they are located in areas of similar structure. Importantly, substitution of these residues in P^BLV-391 results in a non-functional peptide. This is a significantly more dramatic outcome than is observed for specific substitutions at each of these residues in P^cr-400 [##REF##18305034##22##] and suggests that disruption of both of the potential contacts made with the coiled coil has a profound cumulative effect on loss of peptide activity. Given that these leucines are critical to the inhibitory properties of the LHR-mimetic we suspect, and are currently testing the view, that within envelope such substitutions would severely impair envelope-mediated membrane fusion. Our data also reveal that P^BLV-391 is significantly less potent against BLV than the comparable peptide (P^cr-400) against HTLV-1. The structure and model of the HTLV-1 and BLV TM suggests a plausible explanation for this observation in that, relative to the HTLV-derived peptide, the BLV peptide displays a smaller surface area available for interaction with the core coiled-coil. In addition, non-conserved residues within the HTLV-1 peptide may contribute disproportionately to the stability of the interaction between the HTLV-1 peptide and the core coiled-coil. The model and accumulated data also underscore the importance of a deep pocket that is situated towards the membrane-proximal end of the trimer-of-hairpins and is conserved between leukaemia viruses. The peptide inhibitors engage this pocket and this interaction appears to contribute substantially to the stability of peptide association with the coiled coil and is required for optimal inhibitory activity. The data provides further validation of the BLV coiled coil and LHR model and reveals that conserved hydrophobic amino acid side-chains within the helical and non-helical regions mediate interaction of the peptide inhibitors with their target.</p>", "<p>An intriguing finding of this study is that, directed by analysis of the model structure, an improved inhibitor of BLV envelope-mediated membrane fusion can be generated by the substitution of a single amino acid residue, Arg403, with alanine. A similar observation has been made for the Ile412 residue of the HTLV-1 fusion inhibitors [##REF##18305034##22##]. Interestingly, the relative location of these beneficial substitutions is conserved: the BLV residue Arg403 and the HTLV-1 residue Ile412 are immediately N-terminal of an important coiled-coil contact mediated by a conserved leucine residue. It is likely that the substitutions relieve a steric and/or electrostatic clash between the peptides and the relevant viral core coiled-coil, and thereby allow the adjacent leucine residue to dock more effectively with the coiled coil. For BLV, the clash between Arg403 and the coiled coil is highlighted in the model of the trimer-of-hairpins (Figure ##FIG##4##5B##), and this structure is validated by the collected experimental data. Surprisingly, the data derived from the peptide inhibitors identifies a conserved position at which a residue impedes assembly of the trimer-of-hairpins. It appears that during evolution two related but diverging viruses have maintained non-optimal residues within the LHR and that the LHR has not been selected for the best possible fit with the coiled coil. It seems strange not only that such clashes occur, but that they occur in ostensibly the same place. Perhaps, these non-optimal residues act to modulate the fusogenic activity of the TM. It is worth noting that highly fusogenic or readily activated fusion proteins have been described for a number of viruses and these proteins display an array of mutations or deletions, implying that fusogenic activity is modulated by multiple regions of envelope [##REF##12502812##43##, ####REF##9499049##44##, ##REF##11711617##45##, ##REF##15914218##46####15914218##46##]. Of course, it is also possible that the non-optimal residues for LHR association with the coiled coil modulate envelope activity at an earlier pre-fusogenic stage of envelope assembly. Studies are currently underway to test these ideas. Importantly, the ability to remove residues that hinder LHR:coiled coil interaction provides an opportunity to design peptides with \"super-binding\" characteristics and thereby pave the way towards more drug-like HTLV-1 entry inhibitors.</p>", "<p>BLV is prevalent among cattle throughout many regions of the world [##REF##17362524##3##]. The combined effect of decreased milk production, mortality due to lymphoma, reduced productive lifespan and increased susceptibility of infected cattle to opportunistic pathogens has significant economic ramifications [##REF##17362524##3##]. Our data indicate that the core coiled-coil of gp30 is exposed at least transiently during the fusion process and is accessible to a small inhibitory peptide and that inhibitory peptides will be of significant utillity in the analysis of BLV entry into cells. Moreover, it will be interesting to determine if the BLV coiled coil is also accessible to neutralising antibodies and whether coiled-coil-based immunogens could be of value as components of a subunit vaccine to prevent BLV transmission between animals. Although retroviral TM displays significant resistance to neutralisation by coiled-coil-specific antibodies [##REF##14760422##47##,##REF##17940280##40##] recent efforts indicate that such hurdles can be successfully overcome [##REF##16203977##48##]. Moreover, attenuated BLV strains provide long-term protection against experimental BLV infection of cattle [##REF##10725421##49##]; and an HTLV-1 envelope-derived subunit vaccine candidate provides significant protection against virus challenge in primate models [##REF##9010298##50##]. The accumulating evidence therefore suggests that a subunit vaccine based on viral envelope may be an achievable objective for prophylactic treatment against leukaemia virus infections.</p>", "<p>Our data further define a membrane-proximal region of TM that is conserved between BLV and HTLV-1, which has potential as an anti-HTLV-1 drug target. This study demonstrates that comparative analysis of BLV and HTLV-1 induced membrane fusion will provide significant insight into envelope function and ultimately will be of value in the quest for compounds that block HTLV-1 entry into cells.</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Human T-cell leukaemia virus (HTLV-1) and bovine leukaemia virus (BLV) entry into cells is mediated by envelope glycoprotein catalyzed membrane fusion and is achieved by folding of the transmembrane glycoprotein (TM) from a rod-like pre-hairpin intermediate to a trimer-of-hairpins. For HTLV-1 and for several virus groups this process is sensitive to inhibition by peptides that mimic the C-terminal α-helical region of the trimer-of-hairpins.</p>", "<title>Results</title>", "<p>We now show that amino acids that are conserved between BLV and HTLV-1 TM tend to map to the hydrophobic groove of the central triple-stranded coiled coil and to the leash and C-terminal α-helical region (LHR) of the trimer-of-hairpins. Remarkably, despite this conservation, BLV envelope was profoundly resistant to inhibition by HTLV-1-derived LHR-mimetics. Conversely, a BLV LHR-mimetic peptide antagonized BLV envelope-mediated membrane fusion but failed to inhibit HTLV-1-induced fusion. Notably, conserved leucine residues are critical to the inhibitory activity of the BLV LHR-based peptides. Homology modeling indicated that hydrophobic residues in the BLV LHR likely make direct contact with a pocket at the membrane-proximal end of the core coiled-coil and disruption of these interactions severely impaired the activity of the BLV inhibitor. Finally, the structural predictions assisted the design of a more potent antagonist of BLV membrane fusion.</p>", "<title>Conclusion</title>", "<p>A conserved region of the HTLV-1 and BLV coiled coil is a target for peptide inhibitors of envelope-mediated membrane fusion and HTLV-1 entry. Nevertheless, the LHR-based inhibitors are highly specific to the virus from which the peptide was derived. We provide a model structure for the BLV LHR and coiled coil, which will facilitate comparative analysis of leukaemia virus TM function and may provide information of value in the development of improved, therapeutically relevant, antagonists of HTLV-1 entry into cells.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>DL performed the experiments and helped to draft the manuscript, AS provided technical expertise in molecular modeling, DvA provided assistance and technical expertise in structural analysis, DWB designed the experiments and wrote the manuscript. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Dr Arsène Burny and Dr Luc Willems for kindly supplying reagents. The Leukaemia Research Fund generously supported this work through a project grant (LRF-354) to D.W.B. D.L. is the recipient of a Medical Research Council studentship. D.v.A. is supported by a Wellcome Trust Senior Research Fellowship. We thank Clare Connolly, Dr Daniella Zheleva and Cyclacel Pharmaceuticals, Inc. for assistance with laser nephelometry.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Analysis of the conserved regions of BLV and HTLV-1 TM</bold>. (A) Alignment of the BLV and HTLV-1 TM sequences, the predicted coiled coil of BLV TM is indicated between the arrow heads; the LHR is in bold; the helical regions of the HTLV-1 TM are boxed; the limits of the HTLV-1 crystal structure are marked by asters; and the membrane spanning region is underlined. (B) The HTLV-1 core coiled-coil and, on the right, the leash and α-helical region that is mimicked by the HTLV-1 inhibitory peptide (from PDB <ext-link ext-link-type=\"pdb\" xlink:href=\"1MG1\">1MG1</ext-link>). The face of the peptide that interacts with the coiled coil is shown. For the sequence alignment and structural renderings, residues identical between BLV and HTLV-1 are shown in red, conservative substitutions are blue, and non-conserved are rendered white. Amino acid coordinates refer to the full-length envelope precursor. (C) Detail of the predicted interaction of the HTLV-1 LHR-mimetic peptide (ribbon structure) with the surface of the coiled coil (space filling form) based on the structure of Kobe <italic>et al</italic>. [##REF##10200260##16##]; shading as above.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>BLV Env-induced syncytia</bold>. Mock transfected HeLa cells (Mock) or HeLa cells transfected with pRSV-Rev alone (<italic>rev</italic>), pCMV-BLVenv-RRE alone (<italic>env</italic>), or both pRSV-Rev and pCMV-BLVenv-RRE (<italic>rev </italic>+ <italic>env</italic>) were co-cultured with target untransfected HeLa cells. Cells were stained with Giemsa and typical syncytia profiles are shown.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>The specificity of peptide inhibitors of Envelope-mediated membrane fusion is limited to the parental virus</bold>. HeLa cells expressing HTLV-1 (A) or BLV (B) envelope were used as effector cells and co-cultured with untransfected HeLa cells. Cells were incubated in the presence of the peptides P^cr-400, P^BLV-391, P^cr-400 L/A a non-functional derivative of P^cr-400 [##REF##12584351##20##], or the control HIV C helix mimetic peptide C34 [##REF##9861018##51##]. (C) Syncytia formation between BLV infected FLK cells and non-infected HeLa cells. Syncytia were counted in 10 low-power light microscope fields. Data points show the mean ± SD of triplicate assays.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Deletions or substitutions of specific amino acids in P^BLV-391 have a detrimental effect on inhibitory activity</bold>. Syncytium interference assays using BLV envelope-expressing HeLa cells as effectors. (A) The inhibitory properties of P^BLV-391, P^BLV-ΔN, P^BLV-ΔC and the P^cr-400 control were examined. (B) The activity of P^BLV-391, the derivative P^BLV-ΔCCF, and the control peptide P^cr-400 were compared. (C) The activity of P^BLV-391 was compared to Bio-P^BLV-391^Ara biotinylated peptide recovered from the flow-through of an amylose column (see methods), Bio-P^BLV-391^Sdthe same peptide depleted over a streptavidin column (volumes of column buffer equal to those required to give the specified concentrations of Bio-P^BLV-391^Arwere used), and the control peptide C34. (D) The inhibitory properties of P^BLV-391, P^BLV-L/A, P^BLV-L404/410A and the control P^cr-400 were compared. Syncytia were counted in 10 low-power light microscope fields. Data points show the mean ± SD of triplicate assays.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>Homology model of the BLV core coiled-coil and the interacting LHR</bold>. The protein sequence of BLV TM was modelled onto the HTLV-1 TM ectodomain structure (PDB ID <ext-link ext-link-type=\"pdb\" xlink:href=\"1MG1\">1MG1</ext-link>). (A) The predicted BLV core coiled-coil is shown as a space-filling model in grey with the LHR in green. (B) Detail of the coiled coil in blue, grey and red, with the C-terminal section mimicked by P^BLV-391 shown as a green ribbon, the predicted position of relevant side chains are shown as sticks. The membrane proximal region is uppermost. The arrowhead marks the position of Leu404.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p><bold>Substitution of a single arginine residue with alanine yields an improved inhibitor</bold>. The syncytium inhibition activity of the peptides P^cr-400, P^BLV-391 and the derivative peptide P^BLV-R403A was examined. The percentage syncytium inhibition following co-incubation of cells with the peptides is shown. Syncytia were counted in 10 low-power light microscope fields. Data points show the mean ± SD of triplicate assays. The asters show the data points for which the <italic>p </italic>values were calculated (see main text).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Peptides used in this study.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\">Peptide</td><td align=\"center\">Amino Acid Position</td><td align=\"center\">Sequence</td><td align=\"left\">MW</td><td align=\"center\">Maximum Solubility (μM)*</td></tr></thead><tbody><tr><td align=\"left\">P^cr-400</td><td align=\"center\">gp21 400–429</td><td align=\"left\"><monospace>CCFLNITNSHVSILQERPPLENRVLTGWGL       </monospace></td><td align=\"left\">3,411</td><td align=\"center\">&gt; 90.00</td></tr><tr><td align=\"left\">P^cr-400 L/A</td><td align=\"center\">gp21 400–429</td><td align=\"left\"><monospace>---A---------A-----A----A----A       </monospace></td><td align=\"left\">3,200</td><td align=\"center\">45.00</td></tr><tr><td align=\"left\">P^BLV-391</td><td align=\"center\">gp30 391–419</td><td align=\"left\"><monospace>CCFLRIQNDSIIRLGDLQPLSQRVSTDWQ        </monospace></td><td align=\"left\">3,447</td><td align=\"center\">&gt; 90.00</td></tr><tr><td align=\"left\">P^BLV-ΔN</td><td align=\"center\">gp30 400–419</td><td align=\"left\"><monospace>         S-------------------        </monospace></td><td align=\"left\">2,312</td><td align=\"center\">&gt; 90.00</td></tr><tr><td align=\"left\">P^BLV-ΔC</td><td align=\"center\">gp30 391–410</td><td align=\"left\"><monospace>-------------------L                </monospace></td><td align=\"left\">2,317</td><td align=\"center\">&gt; 90.00</td></tr><tr><td align=\"left\">P^BLV-L/A</td><td align=\"center\">gp30 391–419</td><td align=\"left\"><monospace>---A---------A--A--A---------       </monospace></td><td align=\"left\">3,236</td><td align=\"center\">45.00</td></tr><tr><td align=\"left\">P^BLV-L404/410A</td><td align=\"center\">gp30 391–419</td><td align=\"left\"><monospace>-------------A-----A---------       </monospace></td><td align=\"left\">3,321</td><td align=\"center\">&gt; 90.00</td></tr><tr><td align=\"left\">P^BLV-ΔCCF</td><td align=\"center\">gp30 394–419</td><td align=\"left\"><monospace>   L-------------------------       </monospace></td><td align=\"left\">3,052</td><td align=\"center\">11.25</td></tr><tr><td align=\"left\">P^BLV-R403A</td><td align=\"center\">gp30 391–419</td><td align=\"left\"><monospace>------------A----------------       </monospace></td><td align=\"left\">3,321</td><td align=\"center\">22.50</td></tr><tr><td align=\"left\">C34</td><td align=\"center\">gp41 627–661</td><td align=\"left\"><monospace>GWMEWDREINNYTSLLIHSLIEESQNQQEKNEQELL</monospace></td><td align=\"left\">4,418</td><td align=\"center\">&gt; 90.00</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>* Maximum solubility in aqueous solution determined by laser nephelometry.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1742-4690-5-70-1\"/>", "<graphic xlink:href=\"1742-4690-5-70-2\"/>", "<graphic xlink:href=\"1742-4690-5-70-3\"/>", "<graphic xlink:href=\"1742-4690-5-70-4\"/>", "<graphic xlink:href=\"1742-4690-5-70-5\"/>", "<graphic xlink:href=\"1742-4690-5-70-6\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Langerhans cells (LC) are specialized dendritic cells (DC) normally found in the epidermis and mucosal stratified epithelia ##REF##10837075##[1]##–##REF##9561375##[5]##. Contrary to myeloid and plasmacytoid DC, LC express the C-type lectin CD207 (Langerin), the major constituent of Birbeck granules, which represent the hallmark of LC ##REF##3534103##[6]##–##REF##10661407##[8]##. LC also express a characteristic set of cell-surface molecules, such as cutaneous lymphocyte-associated antigen (CLA), E-cadherin and the CC chemokine receptor 6 (CCR6) ##REF##9294137##[9]##–##REF##8867763##[13]##. As immature cells, their primary function is to sense the environment for danger signals and capture antigens; then LC undergo a process of functional and phenotypic maturation and migrate to the regional lymph nodes ##REF##9561375##[5]##; ##REF##15963754##[10]##; ##UREF##0##[14]##; ##REF##17048704##[15]##.</p>", "<p>Several evidences suggest that LC are of myeloid origin and can differentiate from monocytes or CD34⁺ precursors ##REF##1279441##[16]##–##REF##8759731##[20]##. While the generation of DC from monocytes requires GM-CSF and IL-4 only, the additional presence of TGFβ1 in the cytokine milieu appears to be essential for the development of LC ##REF##10725716##[17]##; ##REF##9500798##[18]##. Accordingly, TGFβ1-deficient mice display a severe defect in LC, but not in DC, development ##REF##8976197##[21]##. IL-15 is the only other cytokine known until now to skew monocyte differentiation toward LC-type DC; though, IL-15-derived LC are Langerin⁺ but lack Birbeck granules ##REF##11581322##[22]##.</p>", "<p>According to the current model of LC differentiation, in steady-state conditions LC are maintained locally by a stable renewable population present in the skin ##REF##12415265##[23]##–##REF##17170756##[25]##. The existence of skin-resident LC precursors was postulated by Lareggina et al who described dermal CD14⁺ cells that express Langerin and CCR6 and are able to acquire LC features when cultured in the presence of TGFβ1 ##REF##11702065##[26]##. When the skin is exposed to inflammatory stimuli (UV rays, infections, allergens) LC increase their expression of class II MHC and costimulatory molecules and migrate to regional lymph nodes. In this situation of accelerated turnover, LC are replaced by blood-borne precursors such as inflammatory Gr-1⁺ monocytes recruited through a CCR2-dependent mechanism ##REF##12415265##[23]##; ##REF##16444257##[27]##; ##REF##16482165##[28]##.</p>", "<p>The epidermal environment can contribute to the attraction of precursors and to their differentiation into LC. Keratinocyte production of MIP-3α/CCL20 and TGF may direct CCR6⁺ LC precursors to the epidermis and induce their entry into the LC pathway, respectively ##REF##9294137##[9]##; ##REF##9500798##[18]##; ##REF##8976197##[21]##. However, LC differentiation could also depend on the dermal cytokine environment once LC precursors have entered the skin to colonize the dermis, or are trafficking through the dermis to the overlying epidermis. Indeed, emerging evidence support the concept that in certain skin pathological conditions a conspicuous expansion of the LC population occurs within the dermis ##REF##17514517##[29]##–##REF##15714455##[32]## suggesting, that LC differentiation factors might also be produced within the dermal compartment.</p>", "<p>Activin A is a member of the TGFβ1 family initially identified for its ability to control the secretion of follicle-stimulating hormone. Activin A is presently also known for its activity on growth and differentiation of various cell types during organogenesis, and for its role in wound healing, inflammation and tumor progression ##REF##11815670##[33]##–##REF##15451577##[36]##. Activin A binds to specific transmembrane serine/threonine kinase receptors (ActRIB and ActRII) and to follistatin, a secreted protein that inhibits protein functions by sequestration ##REF##8622651##[37]##–##REF##8340384##[39]##. Activin A is strongly induced after skin injury, probably by serum growth factors released upon haemorrhage and by macrophage-derived pro-inflammatory cytokines ##REF##8606007##[40]##; ##REF##8892977##[41]##. Transgenic mice overexpressing Activin A in the epidermis show strong hyperthickened epidermis, accelerated wound healing and enhanced scarring ##REF##16127153##[42]##; ##REF##10508154##[43]##. Conversely, in transgenic mice overexpressing the antagonist follistatin, skin wound closure is delayed and scar formation reduced ##REF##11574468##[44]##. Curiously, LC are strongly reduced in the skin of follistatin transgenic mice, suggesting a role of Activin A in LC biology ##REF##16180311##[45]##.</p>", "<p>This study shows that Activin A induces the differentiation of LC in vitro and ex-vivo and candidates Activin A as a new differentiation pathway that might be relevant in conditions characterized by the local production of Activin A and accumulation of LC.</p>" ]

[ "<title>Materials and Methods</title>", "<p>The study was conducted in accordance with a protocol approved by the Spedali Civili of Brescia Institutional Ethical Board (Brescia, Italy) and the Board of the CTO Hospital (Turin, Italy); written informed consent was obtained from all patients.</p>", "<title>Cell cultures</title>", "<p>CD14⁺ monocytes were isolated from buffy coats (Centro Trasfusionale Brescia, Italy) by positive magnetic separation using CD14 immunomagnetic beads (Miltenyi Biotec, Auburn, CA) ##REF##16189275##[60]##. To generate LC, monocytes were cultured for 6 days in 6 wells tissue culture plates (Costar, Corning, Cambridge, MA) in 10% heat-inactivated FCS RPMI 1640 (Sigma, St. Louis, MO) supplemented with 100 U/mL penicillin, 100 µg/mL streptomycin, 2 mM L-glutamine. GM-CSF 100ng/ml and IL-4 10 ng/ml were added at day 0. 10ng/ml TGFβ1 or 100 ng/ml Activin A were also added at day 0. All cytokines were from Peprotech, Rocky Hill, NJ. Half the culture medium was replaced with fresh medium containing cytokines on day 2 and 4. LC maturation was induced by incubation with CD40L-transfected J558 cells (1∶4 ratio) for 24 hrs.</p>", "<title>Flow cytometric analysis</title>", "<p>Surface phenotype analysis was performed using the following antibodies: phycoerythrin (PE)-conjugated anti-CD1a (anti-CD1a-PE), anti-Langerin (CD207)-PE, anti-E-cadherin, and anti-CD83-PE (Immunotech, Marseille, France); anti-human leukocyte antigen (HLA)-ABC-PE and FITC-conjugated anti-CD1a (anti-CD1a-FITC; Dako, Glostrup, Denmark); anti-HLA-DR-PE, anti-CD80-PE, (BD PharMingen, San Diego, CA); anti-CD86-PE, anti-CD14-PE, anti-CC chemokine receptor 7 (CCR7)-PE, and anti-CCR6-FITC (R&amp;D Systems). Anti-CLA-FITC rat mAb (BD PharMingen) was also used. Mouse immunoglobulin G1 (IgG1)-PE, mouse IgG1-FITC, rat anti-mouse IgG1-FITC, and rat IgG2a-PE were from BD Pharmingen. Purified mouse IgG1 (R&amp;D Systems), rat IgM-FITC, mouse IgG2b-PE, mouse IgG2a-PE (BD PharMingen), or mouse IgG2b-FITC (Beckman Coulter, Hialeah, FL) were used as an isotype control. Cells were analyzed with a FACScan flow cytometer (Becton Dickinson, San Jose, CA) using CellQuest software.</p>", "<title>Transmission electron microscopy</title>", "<p>10⁶ cells were fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) and postfixed with 1% osmium tetroxide in 0.1 M cacodylate buffer (pH 7.4). Next, cells were dehydrated through a graded series of ethanol and embedded in araldite (Poersch, Frankfurt, Germany). Ultrathin sections were counterstained with uranyl acetate and lead citrate and were examined with a Zeiss electron microscope (EM 906, Zeiss, Oberkochen, Germany).</p>", "<title>Evaluation of DC functions</title>", "<p>Irradiated immature and CD40L-stimulated LC were added in triplicate in graded doses to 2×10⁵ purified allogeneic T cells in 96-well round-bottom plates. [³H]Thymidine incorporation was measured on day 5 after a 16-h pulse (5 Ci/µmol; Amersham Biosciences, Freiburg, Germany). Human IL-12p70, Activin A, TGFβ1, MDC/CCL22, TARC/CCL17, MIP-3α/CCL20 protein levels in the culture supernatants were measured by sandwich ELISA (R&amp;D Systems, Minneapolis, MN). DC migration was evaluated using a 48-well microchemotaxis chamber (Neuroprobe, Pleasanton, CA) with 5-µm pore size polyvinylpyrrolidone polycarbonate filters (Neuroprobe) as previously described ##REF##12646652##[61]##.</p>", "<title>RNA purification and real time RT-PCR analysis</title>", "<p>RNA samples, extracted by using TRIzol (Invitrogen, Carlsbad, CA), were treated with DNase (Invitrogen) and single-stranded complementary DNA (cDNA) was synthesized by reverse transcription of 2 µg total RNA using random hexamers and the Superscript First-Strand Synthesis System for RT-PCR (Invitrogen). The cDNAs were then amplified in duplicate by real-time PCR using the Platinum SYBR Green (Invitrogen) in a final volume of 25 µl and quantitative analysis was carried out as previously described ##REF##17576240##[47]##. The sequences of primers were as follows: Activin A (sense <named-content content-type=\"gene\">5′-GCA GAA ATG AAT GAA CTT ATG GA-3′</named-content>; antisense <named-content content-type=\"gene\">5′-GTC TTC CTG GCT GTT CCT GAC T-3′</named-content>), TGFβ1 (sense <named-content content-type=\"gene\">5′-GCG TGC TAA TGG TGG AAA-3′</named-content>; antisense <named-content content-type=\"gene\">5′-CGG TGA CAT CAA AGA TAA CCA C-3′</named-content>), β-actin (sense, <named-content content-type=\"gene\">5′-GTT GCT ATC CAG GCT GTG-3′</named-content>; antisense, <named-content content-type=\"gene\">5′-TGT CCA CGT CAC ACT TCA-3′</named-content>).</p>", "<title>Skin explant cultures</title>", "<p>Skin specimens were obtained from patients undergoing corrective breast or abdominal plastic surgery. Cytokines were injected into the dermis with a MicroFine insulin syringe (29 gauge needle) in the indicated amounts and in a total volume of 20 µl. At the site of injection, a ∼5-mm wheal appeared and a 6 mm punch biopsy was taken. For immunohistochemistry, skin biopsies were cultured at air-medium interface with the epidermis side up in a six-well culture plate (Costar) on sterilized stainless steel grids covered with a filter (Millipore, Bedford, MA; 45 µm), at 37° C in 5% CO₂ humidified air ##REF##16751366##[46]##; ##REF##12391253##[62]##. At the indicated times, the explants were harvested, snap frozen, and stored in liquid nitrogen. To study phenotypic development of epidermal and dermal DC separately, the epidermal and dermal layers were separated by dispase digestion for 1 h at 37°C (Dispase grade II, 50mg/ml; Roche). To obtain emigrated skin cells, the epidermal and dermal layers were placed directly in 1 ml culture medium (floating with epidermis side up) in a 48-well culture plate (Costar) with 100 ng/ml Activin A. After 7 h, migratory cells were collected, counted in hemocytometers using trypan blue exclusion and cultured in the presence of GM-CSF and Activin A for 6 days.</p>", "<title>Tissue specimens and staining procedures</title>", "<p>Paraffin embedded tissue blocks were taken from the archive of the Department of Pathology of the University of Brescia/Spedali Civili di Brescia; tissues included normal skin, oral (13 cases) and cutaneous (21 cases) lichen planus and granulosa cell tumors (3 cases). Four micron tissue sections were used for immunohistochemical staining using primary Abs to the following antigens: human Activin A (clone E4, dilution 1∶50, Serotec, Oxford, UK), human Langerin (clone 12D6, 1∶200, Vector Laboratories, Burlingame, CA), human Factor VIII-related antigen (rabbit polyclonal, 1∶100, Neomarkers, Westnghouse, CA). Upon antigen retrieval with microwave treatment (3 cycles of 5 min×750 W) or thermostatic bath (40′ at 89°) in EDTA buffer solution, reactivity was revealed using Real EnVision Mouse/Rabbit-HRP (DakoCytomation, Glostrup, Denmark) or SuperSensitive IHC Detection System (BioGenex, San Ramon, CA). For double-immunofluorescence staining anti-human Langerin and Factor VIII-related antigen were revealed respectively using a goat anti-mouse IgG2b (1∶75, Southern Biotek, Birmingham, AL) followed by Streptavidin Texas Red (1∶75, Southern Biotek) and a FITC-conjugated swine anti-rabbit (1∶30, Dako Cytomation). For immunohistochemical staining of Activin A, positive tissue control was represented by granulosa cell tumors that strongly expressed this protein ##REF##9196439##[63]##; no reactivity was observed when omission of primary Ab and irrelevant isotype matched primary antibody were used. Immunostained sections were independently examined by two pathologists (WV and FF); digital images taken using the Olympus BX60 microscope and a DP-70 Olympus digital camera were processed using Analysis Image Processing software.</p>" ]

[ "<title>Results</title>", "<title>Activin A induces the differentiation of circulating monocytes into DC with phenotypic and ultrastructural features of LC</title>", "<p>Highly purified monocytes were cultured for 6 days with Activin A in the presence of GM-CSF and IL-4, two cytokines that were shown to cooperate with TGFβ1 in the differentiation of monocytes to LC ##REF##1279441##[16]##; ##REF##9500798##[18]##. These cells (thereafter called Act A-LC to differentiate them from TGFβ1-LC) were CD14⁻ and expressed typical LC markers, such as CD1a, Langerin, E-caderin, CLA and CCR6. At day 6, Act A-LC presented an immature phenotype with a modest expression of CD80, CCR7 and CD83 (##FIG##0##Fig. 1A##). The presence of Birbeck granules, the hallmark of epidermal LC ##REF##11809842##[7]## was assessed by transmission electron microscopy on ultrathin-sections of Act A-LC cells (##FIG##0##Fig. 1B##). These cells displayed dendritic morphology with slightly off centred indented nuclei. The cytoplasm contained Birbeck granules, cytoplasmic organelles with rod like profile, electron-opaque central lamella and rounded ends. More open-ended tennis-racket shaped granules were also observed. Taken together this set of data indicates that Activin A, in the presence of GM-CSF and IL-4, induces the differentiation of circulating monocytes into LC. The possibility that the effect of Activin A could be due to the secondary induction of TGFβ, was investigated by real-time PCR. As shown in ##FIG##0##Figure 1C##, TGFβ1 mRNA was barely detectable in Act A-LC cultures and similar results were obtained for TGFβ2 and TGFβ3 (data not shown); TGFβ1 was also weakly induced in TGFβ1-LC. On the contrary, Activin A mRNA was strongly upregulated by both Activin A and TGFβ1, suggesting the existence of an amplificatory loop. In agreement with mRNA levels, TGFβ1 concentration was below the detection limits (sensitivity 4.61 pg/ml) in Act A-LC supernatants, whereas Activin A was induced at 1.05 ng/10⁶ cells (n = 5, n = 6) in TGFβ1-LC. Furthermore, Act A-LC generation was not blocked by the addition of anti-TGFβ1 (data not shown), supporting a TGFβ1-independent differentiation of Act A-LC. Finally, it was tested whether bone morphogenetic protein (BMP-6), another TGF family member protein, could also induce LC differentiation. ##FIG##0##Figure 1D## shows that BMP-6 did not sustain phenotypic LC differentiation, indicating that the ability to induce LC differentiation is not a general feature shared by all TGFβ family member proteins.</p>", "<title>Phenotypical and functional characterization of Act A-LC</title>", "<p>Membrane phenotype and functions were tested in immature and CD40L-mature Act A-LC. As shown in ##FIG##1##Figure 2A## CD40L-activated Act A-LC showed increased surface expression of CD80 and CD83 as well as the expression of CXCR4 and CCR7, two maturation-associated chemokine receptors; these data are consistent with the acquisition of a mature phenotype. Immature Act A-LC efficiently stimulated T cell proliferation and their allostimulatory capacity was further enhanced upon maturation (##FIG##1##Fig. 2B##). In agreement with the expression of CCR6 (##FIG##0##Fig. 1A## and data not shown), immature, but not mature Act A-LC migrated in response to CCL20 (##FIG##1##Fig. 2C##). On the contrary, CD40L-mature cells migrated in response to CCL19, one of the CCR7 ligands (data not shown). Finally, CD40L-activated Act A-DC secreted IL-12p70, TNF, CCL22 and CCL20 in a similar manner to TGFβ1-LC, with the exception of IL-12 which was consistently produced at higher levels by Act A-LC (##FIG##1##Fig. 2D##). Therefore, LC generated in the presence of Activin A have the capacity to undergo a full maturation process based on membrane phenotype, migration and functional properties.</p>", "<title>Activin A induces the generation of Langerin⁺ cells ex-vivo in human skin explants</title>", "<p>To investigate the ability of Activin A to induce LC differentiation within the skin milieu, skin explants were intradermally injected with Activin A and subsequently cultured in six-well culture plates at the air-medium interface with the epidermis side up ##REF##16751366##[46]##. Explants were subsequently removed and examined by immunohistochemistry. As expected, fresh, untreated skin explants revealed several Langerin⁺ cells within the epidermis (##FIG##2##Fig. 3##) and a similar picture was observed following the injection of medium. Instead, the inoculation of 100 ng Activin A led to a profound increase in the number of Langerin⁺ cells both in the epidermis and in the dermal layer, with a maximal induction observed 72 hrs after the injection. The increase of the number of Langerin⁺ cells was of about 2-fold and 10-fold (n = 6) in the epidermis and dermis, respectively, and was statistically significant with respect to control skin (p&lt;0.05, by Student's t-test; ##FIG##2##Fig. 3##). The number of CD1a⁺ cells also increased in parallel to Langerin expression (data not shown). These results show that Activin A is able to induce the differentiation of LC precursors resident within normal skin.</p>", "<title>Activin A promotes LC differentiation from precursors cells present in the dermal layer</title>", "<p>To exclude that the increased number of LC observed in the dermis following Activin A injection could be due to migration of LC from the epidermis, the dermis was separated by dispase digestion and thereafter injected with Activin A. Also under these experimental conditions, Activin A inoculation strongly increased the number of Langerin⁺ cells (##FIG##3##Fig. 4##). To better address the potential of dermal precursors to differentiate into LC under the influence of Activin A, skin migratory cells were recovered from dermal layers and subsequently cultured in the presence of Activin A. At day 6 of culture, a consistent number of cells, 15% and 20% were positive for CD1a and Langerin, respectively. Conversely, no Langerin⁺ cells could be detected in the absence of Activin A. Altogether, these data show that skin precursors, present within the dermis, can be induced to differentiate into LC by Activin A.</p>", "<title>Dermal accumulation of Langerhans cells in lichen planus is associated with abundant production of Activin A</title>", "<p>Although LC are predominantly confined to the epidermis, we and others have recently documented that in certain pathological conditions, such as lichen planus, LC are also abundantly found in the stromal compartment ##REF##17514517##[29]##; ##REF##11696204##[30]##; ##REF##15714455##[32]##. As shown in ##FIG##4##Fig. 5## (panel a), in normal skin and mucosa LC are regularly distributed within the epithelium in the basal and suprabasal layers and are easily recognized based on their dendritic morphology and expression of Langerin ##REF##11809842##[7]##. On the contrary, in the large majority of the lichen planus cases investigated (32/34) variable numbers of Langerin⁺ dendritic cells were identifiable in the stromal compartment (##FIG##4##Fig. 5##, panel b), distributed as sparse cells or clusters within the mononuclear infiltrate (##FIG##4##Fig. 5##, panel c). Of interest, many of these Langerin⁺ cells were detected in close proximity to blood vessels, as shown by double immunofluorescence for Langerin and Factor VIII-related antigen (##FIG##4##Fig. 5##, panel c, inset). In normal skin a weak reactivity for Activin A was detected in epidermal keratinocytes and rare spindle cells (likely representing dermal macrophages or dendritic cells); a stronger positivity was also observed in scattered mast cell ##REF##17576240##[47]## (##FIG##4##Fig. 5##, panel d). Compared to normal skin, lesional skin and mucosa from lichen planus biopsies showed strong induction of Activin A in the upper layers of the epidermis and, particularly, in the stromal compartment (##FIG##4##Figure 5##, panel panel e). In the latter, Activin A was mostly produced by non-lymphoid mononuclear cells, endothelial cells and mast cells (##FIG##4##Figure 5##, panel f)</p>" ]

[ "<title>Discussion</title>", "<p>This study reports that Activin A, a protein abundantly produced in the skin during normal and pathological wound healing ##REF##8606007##[40]##; ##REF##17576240##[47]## and inflammatory/autoimmune diseases (this study), induces the differentiation of human CD14⁺ monocytes in Langerin⁺, Birbeck granules⁺, E-cadherin⁺, CLA⁺ and CCR6⁺ cells. LC originate in vitro from CD34⁺ bone marrow precursors ##REF##1279441##[16]##; ##REF##10725716##[17]##; ##REF##8839846##[19]##; ##REF##8759731##[20]##. Monocytes also represent LC blood precursor cells in vitro and in vivo ##REF##9500798##[18]##; ##REF##11581322##[22]##; ##REF##16444257##[27]##; ##REF##16482165##[28]##. Indeed, CD14⁺ cells can be induced to differentiate into LC by a cytokine combination including GM-CSF, IL-4 and TGFβ1 ##REF##9500798##[18]##. The crucial role of TGFβ in LC differentiation has been clearly documented by the observation that TGFβ1 null mice are devoid of LC ##REF##8976197##[21]##. More recently, it was shown that the cytokine milieu present at the inflammatory site may favour LC differentiation through an alternative pathway ##REF##11581322##[22]##; ##REF##16482165##[28]##. Indeed, in vitro experiments have shown that IL-15, in cooperation with GM-CSF, induces the differentiation of monocytes into cells that express LC markers, such as E-cadherin, CCR6 and Langerin but lacking the expression of conventional Birbeck granules ##REF##11581322##[22]##. The present study adds Activin A to the limited list of cytokines that possess the potential to promote LC differentiation. Activin A is a member of the TGFβ family and shares some of the intra-cellular signalling pathways with this cytokine ##REF##10593858##[48]##. However, the ability to induce LC differentiation is not a general feature shared by all TGFβ family member proteins, as documented by the lack of activity of BMP-6 in our assay conditions. The selective action of Activin A versus BMP-6 is likely to be due to the usage of specific transmembrane receptors and the activation of different signalling pathways ##REF##14746809##[49]##; ##REF##9529613##[50]##.</p>", "<p>LC form a cellular network in the epidermis that constitutes the first immunological barrier against pathogens and dangerous insults. Following antigen capture, LC leave the epidermis by a mechanism that depends on the expression of chemotactic receptors, adhesion molecules and proteases ##REF##15963754##[10]##; ##UREF##0##[14]##; ##REF##8393477##[51]##. Emigrating skin cells enter lymphatic vessels located in the superficial dermis to finally reach draining lymph nodes where they present processed antigens to naïve T cells ##REF##2230654##[52]##. During this migratory process, LC acquire a mature phenotype that is associated with the expression of homing receptors, co-stimulatory molecules and the ability to release several cytokines ##REF##15963754##[10]##; ##REF##2230654##[52]##; ##REF##11012599##[53]##. The results presented in this study show that LC generated in the presence of Activin A are fully competent to undergo a maturation process, as evaluated by the expression of CCR7 and the downregulation of CCR6, the expression of CD80 and CD83, the ability to induce T cell proliferation and to secrete high levels of chemokines (i.e. CCL20 and CCL22) and cytokines (TNF-α, IL-12p70).</p>", "<p>LC are normally confined to the basal and suprabasal layer of the epidermis and stratified epithelia of mucosal surfaces. These cells are clearly distinct from dermal/interstitial DC which lack Birbeck granules and Langerin expression, but express DC-SIGN, Factor XIIIa and more rarely CD1a ##REF##2576222##[54]##–##REF##12352970##[58]##. The current view proposes that under steady-state conditions, dermal-resident CD14⁺ precursor cells have the potential to migrate to the epidermis in response to CCL20 and there, in the presence of keratinocyte-derived TGFβ, differentiate into immature resident LC characterized by a weak T cell stimulatory activity. In the presence of the cytokine rich milieu that characterizes many pathological conditions, migratory CD14⁺ cells further differentiate into more mature LC which possess a higher antigen presenting activity ##REF##17170756##[25]##; ##REF##11702065##[26]##; ##REF##16482165##[28]##. In order to evaluate the potential of skin resident precursor cells to differentiate into LC in response to Activin A, we performed experiments in which skin biopsies were inoculated ex-vivo with Activin A and further incubated in vitro ##REF##16751366##[46]##. The immunohistochemical evaluation of these skin explants clearly show that the injection of Activin A induced a strong increase in the number of Langerin⁺/CD1a⁺ cells in both the epidermal and dermal compartments in a time-dependent manner. Due to the experimental conditions employed, Langerin⁺ cells must have originated from skin-resident precursor cells. Further we show that Langerin⁺ cells can be induced to differentiate by Activin A within the dermis in the absence of epidermis. In agreement with these results, cells emigrated from skin explants could also be induced to differentiate into Langerin⁺/CD1a⁺ cells by the presence of Activin A in vitro. These findings are compatible with the description of CD14⁺, Langerin⁺ LC precursors located in the superficial and deep dermis, predominantly in perivascular areas ##REF##7504023##[24]##; ##REF##11702065##[26]##. Although the precise characterization of Activin A-responsive dermal LC precursors is beyond the aim of the present study, these results clearly document that Activin A can induce the local differentiation of dermal LC precursors. In this contest it is interesting to note that Stoitzner et al. reported that in mice overexpressing in the follistatin, the natural Activin A antagonist, the number of LC is reduced ##REF##16180311##[45]##.</p>", "<p>Although Activin A is very weakly expressed in normal skin, its expression was dramatically increased in lichen planus biopsies. In this condition, Activin A was expressed both in the superficial epidermis and in the dermis by stromal cells, infiltrating leukocytes, including mast cells and some blood vessels. As expected on the basis of previous work ##REF##17514517##[29]##–##REF##15714455##[32]##, lichen planus biopsies show a prominent increase in the number of LC which were present in the deep and superficial derma and in the epidermis. Of note, LC were often present in clusters localized around Factor VIII⁺ blood vessels, suggesting the involvement of newly vascular-recruited precursor cells. Although these data generated in a human disease do not allow formal conclusions, it is tempting to speculate that during certain pathological conditions characterized by the local expression of Activin A and inflammatory cytokines (such as GM-CSF and IL-4), dermal LC precursors, or newly recruited blood elements, can be induced to differentiate to LC within the stromal compartment. This model may help to explain the origin of LC localized in the deep dermal layers, away from the epidermis and from superficial lymphatic vessels.</p>", "<p>Lichen planus is an autoimmune disease characterized by a prominent cellular infiltrate mainly composed of DC, LC, cytotoxic T lymphocytes and NK cells, localized in close proximity of apoptotic keratinocytes ##REF##17514517##[29]##–##REF##15714455##[32]##. The involvement of the TGFβ family members in this pathology is suggested by several observations. First, BMP-4 is upregulated in the epithelium of lichen planus ##REF##12191961##[59]##. Second, several lines of evidence suggest that the TGFβ activation and/or signal transduction pathway might be defective in this disease. Indeed, lichen planus is associated with epithelial hyperproliferation, a situation that is usually negatively controlled by TGFβ, and this is consistent with the identification of TGFβ positive T cells in the sub-epithelial lymphocytic infiltrate but not within the epithelium itself. It is therefore of interest to note that this defective TGFβ pathway is associated with a high expression of Activin A (this study). In this context it is likely that Activin A may have a prominent role in LC differentiation in lichen planus. Furthermore, Activin A may contribute to the pathogenesis of lichen planus by favouring epithelial hyperplasia.</p>", "<p>In summary, this study presents a new model in which Activin A induces the differentiation of circulating CD14⁺ cells into LC. Since Activin A is abundantly produced during certain inflammatory conditions, we propose that this cytokine represents a new pathway, alternative to TGFβ, responsible for LC differentiation during inflammatory/autoimmune conditions.</p>" ]

[]

[ "<p>Conceived and designed the experiments: TM SS. Performed the experiments: TM SS RD MF CC DA IC MR LS. Analyzed the data: TM SS WV CC FF GG SS. Contributed reagents/materials/analysis tools: WV ARE MG GG. Wrote the paper: FF SS.</p>", "<p>Langerhans cells (LC) represent a well characterized subset of dendritic cells located in the epidermis of skin and mucosae. In vivo, they originate from resident and blood-borne precursors in the presence of keratinocyte-derived TGFβ. Ιn vitro, LC can be generated from monocytes in the presence of GM-CSF, IL-4 and TGFβ. However, the signals that induce LC during an inflammatory reaction are not fully investigated. Here we report that Activin A, a TGFβ family member induced by pro-inflammatory cytokines and involved in skin morphogenesis and wound healing, induces the differentiation of human monocytes into LC in the absence of TGFβ. Activin A-induced LC are Langerin⁺, Birbeck granules⁺, E-cadherin⁺, CLA⁺ and CCR6⁺ and possess typical APC functions. In human skin explants, intradermal injection of Activin A increased the number of CD1a⁺ and Langerin⁺ cells in both the epidermis and dermis by promoting the differentiation of resident precursor cells. High levels of Activin A were present in the upper epidermal layers and in the dermis of Lichen Planus biopsies in association with a marked infiltration of CD1a⁺ and Langerin⁺ cells. This study reports that Activin A induces the differentiation of circulating CD14⁺ cells into LC. Since Activin A is abundantly produced during inflammatory conditions which are also characterized by increased numbers of LC, we propose that this cytokine represents a new pathway, alternative to TGFβ, responsible for LC differentiation during inflammatory/autoimmune conditions.</p>" ]

[]

[ "<p>The authors are in debt with Drs. Bonetti, Marocolo and Ungari (Department of Pathology, Spedali Civili di Brescia, Dermatopathology Section) for their valuable work in the characterization of tissue biopsies. We greatly appreciate the contribution of Dr. Gavazzoni (Dermatologia, Clinica Città di Brescia), Dr. Majorana (Clinica Odontoiatrica, Spedali Civili di Brescia), Dr. Calzavara-Pinton (Dermatologia, Spedali Civili di Brescia) and Dr. Santoro (Chirurgia Plastica, Spedali Civili di Brescia) for their clinical expertise and help in the characterization of the patients.</p>" ]

[ "<fig id=\"pone-0003271-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003271.g001</object-id><label>Figure 1</label><caption><title>Activin A promotes Langerhans cell differentiation from human CD14⁺ monocytes.</title><p>(A) Phenotypic analysis of monocytes cultured for 6 days with GM-CSF and IL-4 in the presence of Activin A (Act A-LC) or TGFβ1 (TGFβ1-LC). Cells were stained with the indicated moAbs (filled histograms) or isotype-matched negative control moAbs (open histograms). Percentages of positive cells are shown in the upper right corner of each histogram. The figure shows one experiment representative of at least five independent cultures. (B) Electron microscopy analysis of Act A-LC. Act A-LC exhibited abundant dendritic membrane protrusions and lobulated or indented nuclei (left panel, 3,000X, bar 40 µm). Cytoplasm presented a rough endoplasmic reticulum, many multilamellar organelles and numerous electron-dense structures reminiscent of Birbeck granules (right panel, 12,000X, bar 1 µm). The inset shows rod-shaped Birbeck granules (200,000X, bar 20 µm). (C) TGFβ1 and Activin A mRNA expression in Act A-LC and TGFβ1-LC cultures. Monocytes were cultured in the presence of Act A or TGFβ1 for the indicated time and the expression of TGFβ1 and Activin A mRNA was determined by real-time PCR, relative to GAPDH mRNA used as internal control. The expression level in freshly isolated monocytes was assumed as the 1.0 value. Similar results were obtained in three different donors. (D) Effects of different TGF family members on LC differentiation. Monocytes were cultured for 6 days with GM-CSF in the presence of 10 ng/ml TGFβ1, 100 ng/ml Activin A, or 100 ng/ml BMP6 and analyzed for Langerin, E-caderin and CCR6 expression by flow cytometry analysis. Data are representative of at least four independent cultures.</p></caption></fig>", "<fig id=\"pone-0003271-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003271.g002</object-id><label>Figure 2</label><caption><title>Phenotypical and functional characterization of CD40L-activated Act A-LC</title><p>(A) Expression of maturation markers by Act A-LC. Act A-LC were incubated with CD40L-transfected fibroblasts for 40 hrs and stained with anti-CD80, CD83, CCR7 and CXCR4 moAbs (filled histograms) or isotype-matched negative control Abs (open histograms). Results obtained with TGFβ1-LC are also shown for comparison. The percentage of positive cells is reported in each panel. Data shown are representative of three independent experiments. (B) Allostimulatory capacity of Act A-LC. Irradiated immature or CD40L-matured Act A-LC (or TGFβ1-LC) were cultured with 2×10⁵ allogeneic purified T cells. Proliferation was assayed as uptake of [H³]thymidine added in the last 16 hrs of a 6-day culture assay. Results are expressed as mean counts per minute (cpm)±SD of one representative experiment performed in triplicate. Values are at the net of T cell proliferation in the absence of DC (3250±250 cpm). (C) Act A-LC migrate in response to CCL20. Immature or CD40L-mature Act A-LC or TGFβ1-LC were applied to the upper wells of the chemotaxis chamber. CCL20 was added to the lower level of the chamber. The number of cells migrated to the lower chamber was counted. Each assay was performed in triplicate and the results are expressed as the mean±SD number of migrated cells (representative of three experiments). (D) Cytokine release by Act A-LC. Immature or CD40L-mature Act A-LC or TGFβ1-LC were assessed for their ability to release the indicated cytokines by ELISA. Results are the average determination (±SD) of four independent experiments.</p></caption></fig>", "<fig id=\"pone-0003271-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003271.g003</object-id><label>Figure 3</label><caption><title>Intradermal injection of Activin A induces the differentiation of dermal and epidermal Langerhans cells in human skin explants.</title><p>Langerin expression was evaluated in the epidermis and dermis (full thickness skin explants) of skin explants, untreated and 72 hrs after i.d. injection of medium or 100 ng Activin A (magnification 100X, inset 400X) The number of Langerin⁺ cells were quantified in skin explants by evaluating six different skin sections (0.05 mm²/field; means±SD). * p&lt;0.05 by Student's t test vs. medium (lower right panel).</p></caption></fig>", "<fig id=\"pone-0003271-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003271.g004</object-id><label>Figure 4</label><caption><title>Activin A induces Langerhans cells differentiation in epidermis-depleted skin explants.</title><p>Langerin expression was evaluated in the dermal layer, separated from skin explants by dispase digestion and subsequently treated for 72 hrs after i.d. injection with 100 ng Activin A (magnification 100X, inset 400X.).</p></caption></fig>", "<fig id=\"pone-0003271-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003271.g005</object-id><label>Figure 5</label><caption><title>Dermal accumulation of Langerhans cells in lichen planus is associated to abundant production of Activin A.</title><p>Sections from normal skin (NS) (<italic>a</italic> and <italic>d</italic>) and lichen planus (LP) (<italic>b, c, e, f</italic>) biopsies were stained for Langerin (<italic>a</italic>–<italic>c</italic>) and Activin A (<italic>d–f</italic>). In normal skin, Langerin⁺ cells are regularly distributed in basal and suprabasal layers and show multiple fine dendrites; no positive cells are detectable in the dermis (panel <italic>a</italic>). In LP biopsies, in addition to intraepidermal LC, accumulation of Langerin⁺ cells is observed in the dermis within the dense monuclear cell infiltrate (panel <italic>b</italic>). At high power view, Langerin⁺ cells show an ovoidal/dendritic shape (panel <italic>c</italic>) and are found surrounding Factor VIII⁺ dermal blood vessels (arrow head, inset in <italic>c</italic>). Serial sections from the same tissue blocks were stained for Activin A. Normal skin (panel <italic>d</italic>) showed weak intraepithelial reactivity (red arrow head); in the dermis, mast cells and occasional spindle cells were positive for Activin A (black arrow heads). In LP, Activin A was strongly induced in the superficial layers of epidermis; in the dermis, a diffuse reactivity can be observed in numerous cells within the inflammatory infiltrate (panel <italic>e</italic>). This cell population includes endothelial cells and a mixture of non-lymphoid mononuclear cells (panel <italic>f</italic>). Magnification 100x (<italic>a, b, d, e</italic>; scale bar 200 micron) and 400x (<italic>c, f</italic>; scale bar 50 micron).</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This work was supported by AIRC (Associazione Italiana per la Ricerca sul Cancro); MIUR (Ministero dell'Istruzione Università e Ricerca; Cofin); Istituto Superiore di Sanità; NOBEL Project Cariplo “Genetic and functional genomics of myelomonocytic cells”; Fondazione Piemontese per gli Studi e le Ricerche sulle Ustioni; Compagnia di San Paolo, Fondazione CRT; Ricerca Sanitaria Finalizzata Regione Piemonte.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003271.g001\"/>", "<graphic xlink:href=\"pone.0003271.g002\"/>", "<graphic xlink:href=\"pone.0003271.g003\"/>", "<graphic xlink:href=\"pone.0003271.g004\"/>", "<graphic xlink:href=\"pone.0003271.g005\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>“Microbes run the world. It's that simple” ##UREF##0##[1]##. If that statement is true then it should be readily apparent that since viruses often control the microbes, perhaps it could also be said that viruses run the world. One undisputed fact is that viruses are the most abundant biological entities on the planet, with bacteriophages alone at an estimated total abundance of 10³⁰ viruses ##REF##16163346##[2]##.</p>", "<p>A great deal of study has gone into the investigation of viruses in aquatic systems since the discovery of their extraordinary abundance only 20 years ago ##REF##2755508##[3]##. In recent years there has been a growing appreciation of the diversity, complexity and importance of viral communities to ecosystem function [reviewed in ##REF##16163346##[2]##, ##REF##17853907##[4]##–##REF##10704475##[7]##]. However, most of this investigation has been focused on lytic viruses. In aquatic systems many viruses have the capability of forming a stable symbiosis with their host bacterium, otherwise known as lysogeny. During lysogeny, temperate phage genomes are maintained in their hosts as prophages, principally through integration into one of the cellular replicons or as an autonomous plasmid ##REF##12798228##[8]##. Compared to lytic viruses there is a paucity of information concerning lysogeny in the environment. Understanding lysogeny is imperative because of its impacts on host phenotype, community composition and perhaps most importantly, gene transfer processes [reviewed in ##UREF##2##[9]##]. It is becoming clear that viruses are vital repositories and vectors of not only viral genes, but host metabolic genes as well ##REF##18337718##[10]##–##REF##18213365##[12]##.</p>", "<p>Prophages are most commonly detected in natural environments and bacterial isolates by prophage induction. Many prophages maintain the ability to initiate a lytic cycle in response to physical or chemical manipulation, allowing them to exist as free bacteriophages and infect new hosts. This lytic cycle can often be artificially activated by metabolic or DNA damage to the host bacterium, commonly with a toxic substance such as Mitomycin C or ultraviolet radiation (UV). Using this or similar methodology, it has been estimated that roughly half of bacterial isolates contain prophages ##UREF##3##[13]##–##UREF##7##[17]##. In marine environments the calculated lysogenic fraction of the population can range from 0 to 100%. This frequency varies spatially and temporally with the reproducible trend of higher frequencies of lysogeny in less productive environments and extreme environments [reviewed in ##UREF##2##[9]##]. Seasonal cycles of lysogeny have also been documented in environments as divergent as sub-tropical Tampa Bay and saline Antarctic lakes ##UREF##8##[18]##, ##UREF##9##[19]##.</p>", "<p>Currently, analysis of environmental metagenomic samples is greatly expanding our understanding of the incredibly complex environmental bacterial and viral communities. The analysis of metagenomic sequence data for community structure and the taxonomic distribution of identified genes is an important first step, analogous to answering the question “who's there?” A seminal, early metagenomic work demonstrated that certain genes and metabolic pathways appear to be depth specific. This suggested depth-specific communities and metabolism ##REF##16439655##[20]##. This same work also observed large numbers of lytic viral sequences in the photic zone and a prevalence of prophage and transposase sequences at depth. This result provided support for the contention that lysogeny is more prevalent in deep marine environments.</p>", "<p>Deciphering ecosystem functioning, or “what are they doing,” by sequence analysis alone is inherently limited. Despite this limitation certain metagenomic studies have attempted to connect metagenomic sequence information to putative ecosystem functions ##REF##16439655##[20]##, ##REF##18223640##[21]##. Some recent metagenomic studies have also begun to examine the functioning of viral communities specifically. Bench <italic>et al</italic>. utilized a viral metagenome library to conclude that cyanophages were very important components of the Chesapeake Bay virioplankton and that the photosynthesis core gene <italic>psbA</italic> may be nearly universal in these cyanophage populations ##UREF##10##[22]##. Another recent metagenomic study went one step further and verified that viral <italic>psbA</italic> genes are not only present, but transcribed ##REF##18043651##[23]##.</p>", "<p>Metagenomic analysis has also begun to demonstrate that differing biogeochemical environments appear to select for specific metabolic functions. A recent large-scale comparative analysis of 42 microbial and 45 viral metagenomes, including the viral metagenome from this study, indicated that the metabolic profiles of the microbial communities were diagnostic of the environment of isolation ##REF##18337718##[10]##.</p>", "<p>This study presents a more fine-scale analysis of the induced viral metagenome representing two important milestones. Firstly, it provides an ecological context by connecting an important physiological parameter to expression of a gene function inferred from metagenomic sequence data. In addition, it is the first metagenomic study focused on lysogeny. We used a viral metagenome prepared from an induced natural microbial population to further our understanding of the functioning of lysogeny in natural systems. For this work our hypothesis was that genes specific to the process of lysogeny would be identifiable in the Tampa Bay viral metagenome and that expression of these genes would be detectable and concurrent with prophage induction in natural samples.</p>" ]

[ "<title>Materials and Methods</title>", "<p>To construct the induced viral metagenome a 200 liter sample of seawater was obtained from Tampa Bay on December 13, 2005. The sample was pre-filtered with 5 µm mesh to remove zooplankton and large phytoplankton. The microbial fraction in the sample was then concentrated using a Pro-Flux model M-12 tangential flow filtration device (Millipore, Bedford MA, U.S.A.) using a spiral-wound filter with a 30 kDa cutoff. The volume was reduced to 1160 ml. 1000 ml was treated with 1 µg ml⁻¹ Mitomycin C to stimulate prophage induction and 160 ml was left untreated as a control. After a 24 hour incubation in the dark, the number of viral particles was enumerated in both the treatment and control samples using SYBR Gold staining. The treatment concentrate contained 3.5×10⁹ virus-like particles (VLP's) ml⁻¹, which was a statistically significant 3-fold increase in virus abundance in comparison to the non-amended control. The treatment sample was centrifuged and 0.2 µm filtered to separate most of the remaining bacterial cells and debris from the viruses. The viruses were then concentrated by polyethylene glycol precipitation and purified by cesium chloride density gradient centrifugation ##UREF##14##[31]##. The viral capsids were disrupted by formamide and the nucleic acids precipitated with ethanol and purified by CTAB extraction ##UREF##14##[31]##. The nucleic acid from the purified viral particles was amplified using the GenomiPhi DNA amplification kit (G.E./Amersham Biosciences, <ext-link ext-link-type=\"uri\" xlink:href=\"http://www1.gelifesciences.com\">www1.gelifesciences.com</ext-link>). The resulting purified, phi29 amplified DNA from four separate reactions was pooled and sequenced using pyrosequencing technology by 454 Life Sciences ##REF##16056220##[32]##.</p>", "<p>This metagenome was included in a large-scale comparative analysis by Dinsdale et al, which encompassed all of the currently available microbial and viral metagenomes ##REF##18337718##[10]##. The sequences from this metagenome are freely available on the SEED platform (<ext-link ext-link-type=\"uri\" xlink:href=\"http://www.theseed.org\">http://www.theseed.org</ext-link>), under the accession number 4440102.3. The sequences are also being made accessible from the CAMERA database (<ext-link ext-link-type=\"uri\" xlink:href=\"http://www.camera.calit2.net\">www.camera.calit2.net</ext-link>) and from NCBI (<ext-link ext-link-type=\"uri\" xlink:href=\"http://www.ncbi.nlm.nih.gov\">www.ncbi.nlm.nih.gov</ext-link>), deposited in the short read archive under the genome project ID number 28619.</p>", "<p>The resulting sequences were then compared against the Gen Bank non-redundant database using the BLASTX algorithm. Batch blasts were performed using the Code Quest system (TimeLogic, Carlsbad CA, <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.timelogic.com\">www.timelogic.com</ext-link>). Sequences were considered “known” if they had a BLASTX similarity with an e-value ≤0.001. Completed blasts were interrogated for the presence of lysogeny related genes using a word search for related gene names or functions. Genes of interest were examined and re-blasted against the non-redundant database individually. Primers and probes for real-time PCR were designed based on the 4 specific selected integrase sequences (##SUPPL##0##Table S1##).</p>", "<p>Comparison of the viral metagenome hits to P-SSM2 was performed by extracting all sequences with top blast hits to P-SSM2. The stated nucleotide position of each hit was converted to amino-acid position. The reads were then sorted in ascending order and graphed in comparison to the complete nucleotide sequence of P-SSM2. Areas of high coverage are observed as high slope due to the density of sorted points in that area along the genome of P-SSM2. Gaps larger than 500 bp were examined on the fully sequenced genome of P-SSM2 in GenBank for determination of gene content. Attempts to assemble these sequences using P-SSM2 as the reference genome at the nucleotide level were unsuccessful.</p>", "<p>To estimate viral diversity and community structure in the library contig spectrum analysis was performed using mathematical rank abundance modeling with the online PHAge Communities from Contig Spectra (PHACCS) tool (<ext-link ext-link-type=\"uri\" xlink:href=\"http:biome.sdsu.edu/phaccs\">http:biome.sdsu.edu/phaccs</ext-link>) ##REF##15743531##[33]##. Average contig spectra were calculated using 20 assemblies of 20,000 randomly selected sequences from the library as performed using Circonspect (<ext-link ext-link-type=\"uri\" xlink:href=\"http://biome.sdsu.edu/circonspect/\">http://biome.sdsu.edu/circonspect/</ext-link>) with a 98% minimal match percentage and a 35 bp overlap. PHACCS modeling parameters included an average genome size of 50,000 bp. The diversity estimates were determined based on best-fit mathematical modeling, in this case to a logarithmic model.</p>", "<p>High-stringency assemblies of the viral metagenome were performed using the SeqMan program in the Lasergene software suite (<ext-link ext-link-type=\"uri\" xlink:href=\"http://www.dnastar.com/products/lasergene.php\">www.dnastar.com/products/lasergene.php</ext-link>) with a minimum overlap of 30 and minimum match percentage of 95%. There were 132 assembled contigs ≥400 bp which were batch-blasted against the GenBank non-redundant database. Forty-two percent of the consensus contig sequences had a significant match to the non-redundant database. Once the size of the contigs dropped below 400 bp the percentage of positive hits decreased precipitously and most of the contigs were observed to be trivial (i.e. containing only two or three sequences), so these contigs were not considered further. The assemblies were performed in order to more accurately assess the community composition since it has been observed that smaller read lengths are much less likely to be identified ##REF##18192407##[26]##.</p>", "<p>Phage proteomic tree analysis was performed on both the unassembled and assembled sequences from the metagenome ##REF##12142423##[27]##. The sequences were compared to the phage and prophage genome database using tBLASTx (E-value cutoff &lt;10⁻³). This database contains sequences from 510 completely sequenced prophage and phage genomes. A comparative phage proteomic tree was constructed using Phage Proteomic Tree version 4 (PPT, <ext-link ext-link-type=\"uri\" xlink:href=\"http://phage.sdsu.edu/rob/PhageTree/v4\">http://phage.sdsu.edu/rob/PhageTree/v4</ext-link>). The significant blast hits of both the assembled and unassembled sequences were then mapped on the tree using Bio-Metamapper, an online tool for metagenomic tree analysis (<ext-link ext-link-type=\"uri\" xlink:href=\"http://scums.sdsu.edu/Mapper\">http://scums.sdsu.edu/Mapper</ext-link>).</p>", "<p>Samples for the seasonal study of lysogenic gene expression and all other environmental parameters were collected from the same site as the metagenome sample, the St. Petersburg Pier. Two liter samples were collected bi-weekly from January 31, 2006 through January 24, 2007. Parameters measured included temperature, salinity, total monthly precipitation, bacterial and viral direct counts, ambient <italic>Synechococcus</italic>, and infectious cyanophage abundance, total primary productivity, secondary productivity and prophage induction with the total bacteria and <italic>Synechococcus</italic> populations.</p>", "<p>Bacterial and viral direct counts were performed using SYBR Gold staining as previously described ##REF##11157214##[34]##. <italic>Synechococcus</italic> cells were enumerated using their natural auto fluorescence and infectious cyanophage abundance was estimated using the MPN technique as previously described ##REF##15691939##[35]##. Ambient cell abundances and counts for prophage induction experimental samples were measured using the same methods.</p>", "<p>Primary productivity was estimated using the ¹⁴C-HCO₃ incorporation method ##UREF##15##[36]##. In brief, 100 ml⁻¹ samples were placed in duplicate 500ml acid washed sterilized, polycarbonate bottles and approximately 50 µCi of ¹⁴C-HCO₃ was added to each sample. Light and dark incubations were prepared for determination of the light dependent rate of carbon fixation. Sample bottles were incubated at ambient water temperature in a flowing ambient water incubator, covered with two layers of neutral density screening. Samples were filtered at T = 0 and T = 2 hours Radioactive counts were corrected for efficiency by use of a ¹⁴C and were counted in a scintillation counter as previously described ##UREF##15##[36]##.</p>", "<p>Bacterial productivity was measured by the method of Kirchman et al. ##REF##3994368##[37]##. Briefly, ambient water samples from each sample were split in two, a treatment and control. The control flask contained 5% trichloroacetic acid (TCA) to kill the bacteria. 4,5-³H-leucine was added to both flasks to a final concentration of 8 nM. Treatment and control flasks were wrapped in aluminum foil and incubated as for primary productivity. Sub-samples were taken at 60 and 120 minutes after T₌0 and were filtered as previously described.</p>", "<p>Real time PCR (qPCR) was initially performed on samples of the DNA extracted from the microbial fraction of Tampa Bay seawater to determine if the integrases from the viral metagenome were still present. The microbial fraction was collected by first passing the sample through 5 µm mesh to remove macro organisms. The microbes were then collected on a 0.2 µm polycarbonate cartridge filter (Sterivex, Millipore, Bedford MA, U.S.A.). The DNA was extracted using phenol chloroform extraction according to standard protocols ##UREF##14##[31]##. Real-time PCR was performed using PCR primers and FAM™/TAMRA™-labeled probes designed for each specific integrase sequence (##SUPPL##0##Table S1##). Reactions were prepared in 50 µl volumes using 1X TaqMan one-step PCR master mix (Applied Biosystems, <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.appliedbiosystems.com\">www.appliedbiosystems.com</ext-link>), 500 nm concentration of each primer and 125 nm concentration of probe. Reactions were analyzed in an Applied Biosystems 7700 Real-Time PCR system and quantified in comparison to a known copy number of artificially synthesized positive control DNA. Two of each type of initial positive amplicon (4 total) from this integrase assay were sequenced and found to be identical to the sequences of interest.</p>", "<p>Microbial fraction RNA samples were collected by filtering 200 ml of sample seawater onto 0.45 µm durapore filters (Millipore, <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.millipore.com\">www.millipore.com</ext-link>), placed in a 2 ml screw-cap tube with 750 µl of the first step lysis buffer from the RNeasy kit (RLT buffer, Qiagen, Valencia CA) with β-mercaptoethanol added as per manufacturers instructions and frozen at -80°C until sample extraction and processing. RNA extraction was continued utilizing the RNeasy kit protocol (Qiagen, Valencia CA) including on-column DNase digestion according to the manufacturer's instructions. The resulting RNA was purified using the Ambion MegaClear kit to provide the highly purified RNA recommended for amplification (Ambion Inc., Austin TX) and then amplified using the Ambion MessageAmp II RNA amplification kit (Ambion Inc., Austin TX). RNA was quantified and quality assessed using a NanoDrop ND1000 spectrophotometer (Thermo Fisher Scientific, <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.nanodrop.com\">www.nanodrop.com</ext-link>).</p>", "<p>Real-time RT-PCR was performed on amplified microbial fraction RNA collected on all samples as for the real-time PCR (as described above) with the addition of reverse transcriptase enzyme and a reverse transcription cycle of 45°C for 30 minutes. Positive control curves were prepared from in-vitro transcribed RNA prepared from the synthetic DNA positive controls.</p>", "<p>Standard statistical analyses including paired t-test to determine significance of prophage induction events and multiple correlations of all measured parameters were performed using MiniTab release 13.1. Multivariate statistical analyses were performed using Primer v.5.2.9 software (Primer-E Ltd., Plymouth Marine Laboratory, U.K. <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.primer-e.com\">www.primer-e.com</ext-link>). This method is based on multidimensional scaling of complex data sets in order to determine relationships with no assumptions of normal distribution. Initially, separate similarity matrices were constructed of all measured environmental variables, including prophage inductions in comparison to the expression pattern of both integrase genes. The sample obtained on 2/28/06 was excluded from the analysis because of loss of the prophage induction experiments. These similarity matrices were compared using the RELATE test (sample statistic ρ) to determine if the two matrices were significantly related. The matrix of gene expression was then transformed to presence/absence because of the uncertainty involved in the copy number calculation due to the amplification step. This was then compared to the distribution of all measured environmental parameters using the BIO-ENV test to determine which variables best explained the pattern of integrase gene expression.</p>", "<p>The number of genome equivalents in each microbial sample included in the Global Ocean Sampling series in the CAMERA database was determined by BLASTX searching each sample using the <italic>Escherichia coli recA</italic> gene as a query. For all searches the e-value cutoff for determining a significant hit was 10⁻³. This gene has been used to estimate the number of bacterial genotypes in metagenomic libraries because it is a well-conserved, known single-copy, core metabolic gene ##REF##18223640##[21]##. The sequences of both detected integrases were then used as a query in the same fashion and the observed frequencies of both used to calculate an environmental frequency distribution of the integrase genes. For estimation of the frequency of gene occurrence in the four viral metagenomes the number of gene hits was compared to the estimated genotype abundance previously calculated using contig spectrum analysis ##UREF##11##[24]##. This is a minimum estimate of prevalence since not all of those genotypes would be adequately sampled through metagenome sequencing.</p>" ]

[ "<title>Results and Discussion</title>", "<title>Metagenome characteristics</title>", "<p>The induced viral metagenome from Tampa Bay produced 294,068 reads with an average length of 104 bp for total of 29.1 Mb of sequence information. Initial BLAST (teraBLASTx) analysis of the metagenome demonstrated a percentage of significant (e-value &lt;10⁻³) blast hits to known sequences of 6.6%, which is comparable to other viral metagenomes prepared and analyzed in a similar fashion ##UREF##11##[24]##. Contig spectrum analysis of the community structure of the metagenome indicated that there were approximately 15,400 viral genotypes in the library with the most abundant genotype estimated at 4.43% of the community. For this library the best-fit model was the logarithmic model, consistent with other aquatic viral libraries prepared similarly ##REF##18311127##[25]##. The diversity measured by the Shannon-Wiener index was high at 9.13 nats. These values are in the median range of four other viral metagenomes analyzed similarly ##UREF##11##[24]##. Sequences were considered viral if the top blast hit was below the pre-determined e-value cutoff and was to a sequence annotated as viral. Interestingly, this metagenome was markedly higher in the percentage of recognizable viral sequences at 30.5% in comparison to an average of 6.6% for 13 other marine bacterial and viral metagenomes isolated from similar environments ##REF##18337718##[10]##. This result suggested that the Mitomycin C treatment of the original bacterial concentrate may have caused a shift in the composition of the viral community, presumably due to induction of prophages. However, this could only be demonstrated conclusively by comparison to a non-induced library from the same sample. Similar high frequencies of viral hits have also been observed in viral metagenomes from dissimilar environments such as hypersaline or coral-associated libraries ##REF##18337718##[10]##.</p>", "<p>The top fifteen most frequent virus identifications using BLASTx analysis from the GenBank nr database of unassembled sequences from the library are listed in ##TAB##0##Table 1##. Seven of the top 15 most frequent hits were to cyanophages, with the cyanophage P-SSM2 being the top contributor. This elevated occurrence of cyanophages has been observed in other metagenomic libraries and underscores the apparent importance of cyanophages in marine ecosystems ##REF##18213365##[12]##, ##REF##16439655##[20]##, ##UREF##10##[22]##, ##UREF##11##[24]##. A closer examination of the sequences similar to P-SSM2 demonstrated that this phage type appears to be important in this estuarine environment despite the fact that its <italic>Prochlorococcus</italic> host is found in oligotrophic oceanic environments, not estuaries like Tampa Bay. A comparison of the distribution of the amino acid sequences similar to P-SSM2 to the amino acid position on the complete P-SSM2 genome demonstrated high over all coverage of the putative viral functions (##FIG##0##Figure 1##). A similar analysis was performed on the Chesapeake Bay viral metagenome ##UREF##10##[22]##, however that analysis demonstrated a dissimilar pattern of P-SSM2 genome coverage. In contrast to the Chesapeake Bay sample which showed a high level of coverage in the areas of phage structural genes, the Tampa Bay viral metagenome showed an almost opposite pattern, with high coverage in the area of genes involved in replication and nucleotide metabolism and complete gaps in coverage in the area of known phage structural genes. These gaps were the most pronounced in areas of P-SSM2 containing identified tail fiber genes, which are likely involved in host specificity. We hypothesize that this is due to a high number of genetically distinct but functionally similar groups or consortia of cyanophages in Tampa Bay. This hypothesis is supported by the inability to assemble the same sequences to the P-SSM2 reference sequence on the nucleotide level. However, it should be noted that these differences may not be solely due to the fact that the Chesapeake Bay is a dissimilar environment to Tampa Bay. In addition to the contrasting environment, the Chesapeake viral library had a longer average read length. The longer read lengths have been demonstrated to yield higher detection of viral and microbial genes as well as recognition of more distal homologs ##REF##18192407##[26]##.</p>", "<p>The next level of analysis involved a high stringency assembly of the metagenome sequences. A low number of the total sequences were placed in non-trivial assemblies (greater than 2–3 sequences assembling). However, once the sequences in the metagenome were assembled and re-analyzed, a different picture of the viral community emerged (##TAB##1##Table 2##). In this case, the analysis was based on the contig size and number of sequences that were placed in assemblies. The contigs were grouped according to decreasing size and the top virus hit for each contig is shown in ##TAB##1##Table 2##. For the assembled sequences nine of the top 15 hits were to experimentally verified prophage or temperate phages. This analysis supported our hypothesis that the induced viral metagenome was enriched in prophages. The assembly process is only able to piece together portions of genomes present in relatively high abundances. The high number of prophages in the assembled sequences may have been due to a higher number of same sequences caused by the induction of a large number of phages with identical or similar genomes.</p>", "<p>An additional two top hits were to bacterial whole genomes, suggesting viral carriage of host genes or possibly unidentified prophage (##TAB##1##Table 2##). The first was to <italic>Pelobacter carbinolicus</italic>, a Fe(III) reducing, anaerobic aquatic sediment species. The second was <italic>Roseobacter</italic> sp. MED 193 from the Mediterranean sea. An examination of these genes indicated that they were found in prophage-like elements (i.e., were surrounded by phage genes and/or lysogeny genes; data not shown).</p>", "<p>The lytic cyanophage P-SSM2 was still present as the 15^th largest contig to be assembled, suggesting this virus was also in very high abundance in this sample. The assembly not only changed the observed distribution of top phage hits, but markedly decreased the e-values obtained because of the increased sequence length.</p>", "<p>Phage proteomic tree analysis was also performed and provided a visual representation of this shift in types of phage hits between the raw and unassembled sequences as shown in ##SUPPL##1##Figure S1##\n##REF##12142423##[27]##. However, it is important to note that the analysis of the assembled sequences was performed with a smaller data set, including only the non-trivial assemblies.</p>", "<title>Induced viral metagenome integrase genes</title>", "<p>The primary objective for obtaining the induced viral metagenome was to identify viral genes involved in lysogeny that were present in Tampa Bay lysogens. Initial searches of the data attempted to locate cI-type repressors. This was the initial target because repressors are known to be expressed constitutively, and were hypothesized to be more readily detectable in RNA-based gene expression experiments. However, no repressor-like genes were identified.</p>", "<p>Another gene known to be involved in the functioning of the lysogenic switch is the enzyme integrase, which catalyzes the integration of the temperate phage into the host genome. Interrogation of the known blast hits revealed 103 (de-replicated) phage integrase-like sequences from the metagenome. Of these sequences, each was examined individually and four were chosen with the lowest e-values (ranging from 10⁻⁴–10⁻⁹) as well as a full-length hit of the query sequence to a known phage-like integrase (i.e., greater than 30 amino acids in length). The four integrases were identified as a <italic>Vibrio</italic>-like phage integrase, <italic>Clostridium-</italic>like phage integrase, <italic>Roseovarius/Oceanicola</italic>-like phage integrase and <italic>Alkalimnicola</italic>-like phage integrase. General properties and the sequences of the four identified integrases are listed in ##SUPPL##0##Table S1##.</p>", "<p>The real-time PCR integrase assay was initially tested on a sample of concentrated bacterial fraction DNA from the same environment sampled for the metagenome. The total microbial fraction was used in order to detect putative integrated prophage rather than free phage. Two of the four integrases were detected in this initial screening. The <italic>Vibrio-</italic>like integrase was detected at an abundance of 1.1×10⁵ gene copies L⁻¹ of bay water. Based on ambient bacterial abundance at the time of sampling, this gene was present in 0.005% of the ambient population. The <italic>Oceanicola</italic>-like integrase was also detected in this sample, but just at the detection limit. In this case the estimated gene copy number was 1300 copies L⁻¹, or an estimated 0.00008% of the ambient population. The other two integrases were not detected in this initial trial. These results indicated that at least some of the genes identified by the metagenome sequencing were not ephemeral and if active, should be detectable as transcripts throughout the annual cycle.</p>", "<title>Seasonal sampling and integrase gene expression in environmental samples</title>", "<p>For the gene expression study all parameters were measured over an entire annual cycle at the same sampling site in Tampa Bay. The ambient parameters showed a similar pattern to previous seasonal observations in this environment ##UREF##9##[19]##. This estuary is typified by a spring and fall bloom in primary productivity and <italic>Synechococcus</italic> abundance (##SUPPL##2##Figure S2##). The total heterotrophic bacterial populations show a more subtle seasonal oscillation typified by high summer and low winter abundances (##SUPPL##2##Figure S2##).</p>", "<p>Prophage induction experiments were also performed on each sampling date for the total heterotrophic bacterial populations as well as the sub-population of <italic>Synechococcus</italic>. ##FIG##1##Figure 2## depicts the distribution of prophage induction events over the annual cycle. In the <italic>Synechococcus</italic> experiments positive prophage inductions are clustered in the winter samples, with no induction observed during the summer months. This is consistent with previous observations of seasonal patterns of <italic>Synechococcus</italic> prophage induction in Tampa Bay. For the total bacterial inductions, statistically significant inductions were more frequent throughout the annual cycle and only showed a period of no inductions during the autumn season. Interestingly, in this study the heterotrophic and <italic>Synechococcus</italic> inductions were somewhat temporally separated, with the <italic>Synechococcus</italic> inductions occurring earlier. In the 1999–2000 seasonal study the inductions temporally co-occurred ##UREF##9##[19]##, ##REF##12200280##[28]##.</p>", "<p>The most important observation was the distribution of the positive integrase gene expression depicted by the shading in ##FIG##1##Figure 2##. Expression of the <italic>Clostridium</italic>-like integrase was observed on six of the sampling dates and the <italic>Vibrio</italic>-like integrase was detected on three of the sampling dates, overlapping with positive <italic>Clostridium</italic>-like expression. The estimated copy numbers the <italic>Vibrio</italic>-like and <italic>Clostridium</italic>-like integrases were estimated as ranging from 2.4–1280 L⁻¹ and 37 to 265 L⁻¹, respectively. In all cases a detection of integrase gene expression was present in conjunction with statistically significant prophage induction. On no occasion was integrase expression present without prophage induction. In addition the detections of integrase expression were clustered in the “lysogenic season” with none in the “lytic season.” These findings supported our initial hypothesis.</p>", "<p>Multivariate statistical analysis of the distribution of integrase gene expression in comparison to all measured environmental variables indicated that the two parameters most correlated with gene expression were the combined parameters of cyanophage induction and light incubated total prophage induction (ρ = 0.608, p = 0.1). This is statistically significant at the 90% confidence interval, which is a good correlation considering the low number of data points available for comparison. One sampling date was excluded from the analysis due to the loss of the prophage induction experiments, leaving five sample dates with integrase gene expression for statistical analysis.</p>", "<title>Distribution of integrase genes in global environmental samples</title>", "<p>Our next investigation was to determine the frequency of both expressed integrases in diverse global environments by investigating their frequency within environmental microbial and viral metagenome libraries. Examination of the integrase gene frequency in microbial samples gives a first estimate of the distribution of similar genes as putative prophage. The frequency of the same genes in viral metagenomes gives a rough estimate of their prevalence in the free virus fraction.</p>", "<p>To query environmental microbial samples we estimated the integrase gene frequency in the microbial fraction samples from the global ocean sampling expedition by comparison to the frequency of the single-copy gene <italic>recA</italic> as a proxy for estimation of total genome equivalents ##UREF##12##[29]##. The <italic>recA</italic> gene was identified in 50 of the 56 samples listed on the CAMERA database with an average copy number of 149. The <italic>Vibrio</italic>-like integrase was observed in 39 of the 56 sites with an estimated average of 3.14% of the microbial population carrying a similar gene. In contrast, the <italic>Clostridium</italic>-like integrase was only found in 17 of the 56 sites with an estimated average frequency of 0.58%. There was no clear trend in the type of environment in which the genes were observed.</p>", "<p>In four viral metagenomes from widely separated geographical environments ##UREF##11##[24]##, genes significantly similar to the <italic>Clostridium</italic>-like integrase were not found. However, the <italic>Vibrio</italic>-like integrase was found in all four viromes with a varying frequency. There were two similar genes in the 416,456 sequence library from British Columbia, containing an estimated 129,000 viral genotypes. One similar gene was detected in both the Gulf of Mexico and Sargasso Sea libraries with 263,908 and 399,343 sequences in each and estimated viral genotype abundances of 15,400 and 5140, respectively. In contrast, in the Arctic library of 188,590 sequences, eleven similar genes were observed with a modeled population estimate of 532 genotypes. This is consistent with the previous finding of high numbers of identifiable prophage genes in this metagenome of Arctic free viruses ##UREF##11##[24]##. Both findings provide support for the contention that lysogeny is prevalent in environments less favorable to bacterial host growth ##UREF##8##[18]##, ##UREF##13##[30]##. It is important to note here that the estimated number viral of genotypes presented for each library are theoretical “best guesses” and may vary depending on the assembly criteria used for modeling.</p>", "<title>Concluding Remarks</title>", "<p>We have demonstrated the capacity of metagenomic sequence information to provide useful and relevant clues to environmental functions of interest. This study represents an important first connection between viral metagenomic sequence data and an ecologically relevant function within a natural environment. We have demonstrated that viral integrase genes are present and can be stable in the population over periods of time and that expression of these genes was detectable in natural samples. Expression was observed to be temporally variable and most importantly to co-occur with prophage induction. This provided strong experimental support for our initial hypothesis.</p>", "<p>Besides demonstrating temporal variability in expression, it appears that similar integrase genes vary in global distribution as well. Integrase-like genes were more frequently observed in microbial samples than free virus samples, which is not surprising since the main function of integrase is to integrate into a host's DNA. In the microbial fraction there was no observable association with a particular environment, but in the virus fraction libraries the gene was much more prevalent in the sample from the Arctic, suggesting greater activity of temperate viruses in colder environments.</p>", "<p>Metagenomics may also allow a new perspective on microbes at an ecosystem scale, particularly by highlighting the importance of a specific organism type to overall ecosystem function. This was illustrated in this study by the high functional coverage of the cyanophage P-SSM2. The amino-acid level genome analysis indicated an important functional role being performed in the Tampa Bay estuary by a diverse assemblage of similar phage. Once the raw sequences were assembled, blast hits of the large contigs did not demonstrate the same dominance of cyanophages, leading to the hypothesis that this may have been due to high prevalence of similar but not identical cyanophages in this sample.</p>" ]

[ "<title>Results and Discussion</title>", "<title>Metagenome characteristics</title>", "<p>The induced viral metagenome from Tampa Bay produced 294,068 reads with an average length of 104 bp for total of 29.1 Mb of sequence information. Initial BLAST (teraBLASTx) analysis of the metagenome demonstrated a percentage of significant (e-value &lt;10⁻³) blast hits to known sequences of 6.6%, which is comparable to other viral metagenomes prepared and analyzed in a similar fashion ##UREF##11##[24]##. Contig spectrum analysis of the community structure of the metagenome indicated that there were approximately 15,400 viral genotypes in the library with the most abundant genotype estimated at 4.43% of the community. For this library the best-fit model was the logarithmic model, consistent with other aquatic viral libraries prepared similarly ##REF##18311127##[25]##. The diversity measured by the Shannon-Wiener index was high at 9.13 nats. These values are in the median range of four other viral metagenomes analyzed similarly ##UREF##11##[24]##. Sequences were considered viral if the top blast hit was below the pre-determined e-value cutoff and was to a sequence annotated as viral. Interestingly, this metagenome was markedly higher in the percentage of recognizable viral sequences at 30.5% in comparison to an average of 6.6% for 13 other marine bacterial and viral metagenomes isolated from similar environments ##REF##18337718##[10]##. This result suggested that the Mitomycin C treatment of the original bacterial concentrate may have caused a shift in the composition of the viral community, presumably due to induction of prophages. However, this could only be demonstrated conclusively by comparison to a non-induced library from the same sample. Similar high frequencies of viral hits have also been observed in viral metagenomes from dissimilar environments such as hypersaline or coral-associated libraries ##REF##18337718##[10]##.</p>", "<p>The top fifteen most frequent virus identifications using BLASTx analysis from the GenBank nr database of unassembled sequences from the library are listed in ##TAB##0##Table 1##. Seven of the top 15 most frequent hits were to cyanophages, with the cyanophage P-SSM2 being the top contributor. This elevated occurrence of cyanophages has been observed in other metagenomic libraries and underscores the apparent importance of cyanophages in marine ecosystems ##REF##18213365##[12]##, ##REF##16439655##[20]##, ##UREF##10##[22]##, ##UREF##11##[24]##. A closer examination of the sequences similar to P-SSM2 demonstrated that this phage type appears to be important in this estuarine environment despite the fact that its <italic>Prochlorococcus</italic> host is found in oligotrophic oceanic environments, not estuaries like Tampa Bay. A comparison of the distribution of the amino acid sequences similar to P-SSM2 to the amino acid position on the complete P-SSM2 genome demonstrated high over all coverage of the putative viral functions (##FIG##0##Figure 1##). A similar analysis was performed on the Chesapeake Bay viral metagenome ##UREF##10##[22]##, however that analysis demonstrated a dissimilar pattern of P-SSM2 genome coverage. In contrast to the Chesapeake Bay sample which showed a high level of coverage in the areas of phage structural genes, the Tampa Bay viral metagenome showed an almost opposite pattern, with high coverage in the area of genes involved in replication and nucleotide metabolism and complete gaps in coverage in the area of known phage structural genes. These gaps were the most pronounced in areas of P-SSM2 containing identified tail fiber genes, which are likely involved in host specificity. We hypothesize that this is due to a high number of genetically distinct but functionally similar groups or consortia of cyanophages in Tampa Bay. This hypothesis is supported by the inability to assemble the same sequences to the P-SSM2 reference sequence on the nucleotide level. However, it should be noted that these differences may not be solely due to the fact that the Chesapeake Bay is a dissimilar environment to Tampa Bay. In addition to the contrasting environment, the Chesapeake viral library had a longer average read length. The longer read lengths have been demonstrated to yield higher detection of viral and microbial genes as well as recognition of more distal homologs ##REF##18192407##[26]##.</p>", "<p>The next level of analysis involved a high stringency assembly of the metagenome sequences. A low number of the total sequences were placed in non-trivial assemblies (greater than 2–3 sequences assembling). However, once the sequences in the metagenome were assembled and re-analyzed, a different picture of the viral community emerged (##TAB##1##Table 2##). In this case, the analysis was based on the contig size and number of sequences that were placed in assemblies. The contigs were grouped according to decreasing size and the top virus hit for each contig is shown in ##TAB##1##Table 2##. For the assembled sequences nine of the top 15 hits were to experimentally verified prophage or temperate phages. This analysis supported our hypothesis that the induced viral metagenome was enriched in prophages. The assembly process is only able to piece together portions of genomes present in relatively high abundances. The high number of prophages in the assembled sequences may have been due to a higher number of same sequences caused by the induction of a large number of phages with identical or similar genomes.</p>", "<p>An additional two top hits were to bacterial whole genomes, suggesting viral carriage of host genes or possibly unidentified prophage (##TAB##1##Table 2##). The first was to <italic>Pelobacter carbinolicus</italic>, a Fe(III) reducing, anaerobic aquatic sediment species. The second was <italic>Roseobacter</italic> sp. MED 193 from the Mediterranean sea. An examination of these genes indicated that they were found in prophage-like elements (i.e., were surrounded by phage genes and/or lysogeny genes; data not shown).</p>", "<p>The lytic cyanophage P-SSM2 was still present as the 15^th largest contig to be assembled, suggesting this virus was also in very high abundance in this sample. The assembly not only changed the observed distribution of top phage hits, but markedly decreased the e-values obtained because of the increased sequence length.</p>", "<p>Phage proteomic tree analysis was also performed and provided a visual representation of this shift in types of phage hits between the raw and unassembled sequences as shown in ##SUPPL##1##Figure S1##\n##REF##12142423##[27]##. However, it is important to note that the analysis of the assembled sequences was performed with a smaller data set, including only the non-trivial assemblies.</p>", "<title>Induced viral metagenome integrase genes</title>", "<p>The primary objective for obtaining the induced viral metagenome was to identify viral genes involved in lysogeny that were present in Tampa Bay lysogens. Initial searches of the data attempted to locate cI-type repressors. This was the initial target because repressors are known to be expressed constitutively, and were hypothesized to be more readily detectable in RNA-based gene expression experiments. However, no repressor-like genes were identified.</p>", "<p>Another gene known to be involved in the functioning of the lysogenic switch is the enzyme integrase, which catalyzes the integration of the temperate phage into the host genome. Interrogation of the known blast hits revealed 103 (de-replicated) phage integrase-like sequences from the metagenome. Of these sequences, each was examined individually and four were chosen with the lowest e-values (ranging from 10⁻⁴–10⁻⁹) as well as a full-length hit of the query sequence to a known phage-like integrase (i.e., greater than 30 amino acids in length). The four integrases were identified as a <italic>Vibrio</italic>-like phage integrase, <italic>Clostridium-</italic>like phage integrase, <italic>Roseovarius/Oceanicola</italic>-like phage integrase and <italic>Alkalimnicola</italic>-like phage integrase. General properties and the sequences of the four identified integrases are listed in ##SUPPL##0##Table S1##.</p>", "<p>The real-time PCR integrase assay was initially tested on a sample of concentrated bacterial fraction DNA from the same environment sampled for the metagenome. The total microbial fraction was used in order to detect putative integrated prophage rather than free phage. Two of the four integrases were detected in this initial screening. The <italic>Vibrio-</italic>like integrase was detected at an abundance of 1.1×10⁵ gene copies L⁻¹ of bay water. Based on ambient bacterial abundance at the time of sampling, this gene was present in 0.005% of the ambient population. The <italic>Oceanicola</italic>-like integrase was also detected in this sample, but just at the detection limit. In this case the estimated gene copy number was 1300 copies L⁻¹, or an estimated 0.00008% of the ambient population. The other two integrases were not detected in this initial trial. These results indicated that at least some of the genes identified by the metagenome sequencing were not ephemeral and if active, should be detectable as transcripts throughout the annual cycle.</p>", "<title>Seasonal sampling and integrase gene expression in environmental samples</title>", "<p>For the gene expression study all parameters were measured over an entire annual cycle at the same sampling site in Tampa Bay. The ambient parameters showed a similar pattern to previous seasonal observations in this environment ##UREF##9##[19]##. This estuary is typified by a spring and fall bloom in primary productivity and <italic>Synechococcus</italic> abundance (##SUPPL##2##Figure S2##). The total heterotrophic bacterial populations show a more subtle seasonal oscillation typified by high summer and low winter abundances (##SUPPL##2##Figure S2##).</p>", "<p>Prophage induction experiments were also performed on each sampling date for the total heterotrophic bacterial populations as well as the sub-population of <italic>Synechococcus</italic>. ##FIG##1##Figure 2## depicts the distribution of prophage induction events over the annual cycle. In the <italic>Synechococcus</italic> experiments positive prophage inductions are clustered in the winter samples, with no induction observed during the summer months. This is consistent with previous observations of seasonal patterns of <italic>Synechococcus</italic> prophage induction in Tampa Bay. For the total bacterial inductions, statistically significant inductions were more frequent throughout the annual cycle and only showed a period of no inductions during the autumn season. Interestingly, in this study the heterotrophic and <italic>Synechococcus</italic> inductions were somewhat temporally separated, with the <italic>Synechococcus</italic> inductions occurring earlier. In the 1999–2000 seasonal study the inductions temporally co-occurred ##UREF##9##[19]##, ##REF##12200280##[28]##.</p>", "<p>The most important observation was the distribution of the positive integrase gene expression depicted by the shading in ##FIG##1##Figure 2##. Expression of the <italic>Clostridium</italic>-like integrase was observed on six of the sampling dates and the <italic>Vibrio</italic>-like integrase was detected on three of the sampling dates, overlapping with positive <italic>Clostridium</italic>-like expression. The estimated copy numbers the <italic>Vibrio</italic>-like and <italic>Clostridium</italic>-like integrases were estimated as ranging from 2.4–1280 L⁻¹ and 37 to 265 L⁻¹, respectively. In all cases a detection of integrase gene expression was present in conjunction with statistically significant prophage induction. On no occasion was integrase expression present without prophage induction. In addition the detections of integrase expression were clustered in the “lysogenic season” with none in the “lytic season.” These findings supported our initial hypothesis.</p>", "<p>Multivariate statistical analysis of the distribution of integrase gene expression in comparison to all measured environmental variables indicated that the two parameters most correlated with gene expression were the combined parameters of cyanophage induction and light incubated total prophage induction (ρ = 0.608, p = 0.1). This is statistically significant at the 90% confidence interval, which is a good correlation considering the low number of data points available for comparison. One sampling date was excluded from the analysis due to the loss of the prophage induction experiments, leaving five sample dates with integrase gene expression for statistical analysis.</p>", "<title>Distribution of integrase genes in global environmental samples</title>", "<p>Our next investigation was to determine the frequency of both expressed integrases in diverse global environments by investigating their frequency within environmental microbial and viral metagenome libraries. Examination of the integrase gene frequency in microbial samples gives a first estimate of the distribution of similar genes as putative prophage. The frequency of the same genes in viral metagenomes gives a rough estimate of their prevalence in the free virus fraction.</p>", "<p>To query environmental microbial samples we estimated the integrase gene frequency in the microbial fraction samples from the global ocean sampling expedition by comparison to the frequency of the single-copy gene <italic>recA</italic> as a proxy for estimation of total genome equivalents ##UREF##12##[29]##. The <italic>recA</italic> gene was identified in 50 of the 56 samples listed on the CAMERA database with an average copy number of 149. The <italic>Vibrio</italic>-like integrase was observed in 39 of the 56 sites with an estimated average of 3.14% of the microbial population carrying a similar gene. In contrast, the <italic>Clostridium</italic>-like integrase was only found in 17 of the 56 sites with an estimated average frequency of 0.58%. There was no clear trend in the type of environment in which the genes were observed.</p>", "<p>In four viral metagenomes from widely separated geographical environments ##UREF##11##[24]##, genes significantly similar to the <italic>Clostridium</italic>-like integrase were not found. However, the <italic>Vibrio</italic>-like integrase was found in all four viromes with a varying frequency. There were two similar genes in the 416,456 sequence library from British Columbia, containing an estimated 129,000 viral genotypes. One similar gene was detected in both the Gulf of Mexico and Sargasso Sea libraries with 263,908 and 399,343 sequences in each and estimated viral genotype abundances of 15,400 and 5140, respectively. In contrast, in the Arctic library of 188,590 sequences, eleven similar genes were observed with a modeled population estimate of 532 genotypes. This is consistent with the previous finding of high numbers of identifiable prophage genes in this metagenome of Arctic free viruses ##UREF##11##[24]##. Both findings provide support for the contention that lysogeny is prevalent in environments less favorable to bacterial host growth ##UREF##8##[18]##, ##UREF##13##[30]##. It is important to note here that the estimated number viral of genotypes presented for each library are theoretical “best guesses” and may vary depending on the assembly criteria used for modeling.</p>", "<title>Concluding Remarks</title>", "<p>We have demonstrated the capacity of metagenomic sequence information to provide useful and relevant clues to environmental functions of interest. This study represents an important first connection between viral metagenomic sequence data and an ecologically relevant function within a natural environment. We have demonstrated that viral integrase genes are present and can be stable in the population over periods of time and that expression of these genes was detectable in natural samples. Expression was observed to be temporally variable and most importantly to co-occur with prophage induction. This provided strong experimental support for our initial hypothesis.</p>", "<p>Besides demonstrating temporal variability in expression, it appears that similar integrase genes vary in global distribution as well. Integrase-like genes were more frequently observed in microbial samples than free virus samples, which is not surprising since the main function of integrase is to integrate into a host's DNA. In the microbial fraction there was no observable association with a particular environment, but in the virus fraction libraries the gene was much more prevalent in the sample from the Arctic, suggesting greater activity of temperate viruses in colder environments.</p>", "<p>Metagenomics may also allow a new perspective on microbes at an ecosystem scale, particularly by highlighting the importance of a specific organism type to overall ecosystem function. This was illustrated in this study by the high functional coverage of the cyanophage P-SSM2. The amino-acid level genome analysis indicated an important functional role being performed in the Tampa Bay estuary by a diverse assemblage of similar phage. Once the raw sequences were assembled, blast hits of the large contigs did not demonstrate the same dominance of cyanophages, leading to the hypothesis that this may have been due to high prevalence of similar but not identical cyanophages in this sample.</p>" ]

[]

[ "<p>Conceived and designed the experiments: LDM FR JHP. Performed the experiments: LDM JM AL MH. Analyzed the data: LDM MB FR JHP. Wrote the paper: LDM.</p>", "<p>Phage integrase genes often play a role in the establishment of lysogeny in temperate phage by catalyzing the integration of the phage into one of the host's replicons. To investigate temperate phage gene expression, an induced viral metagenome from Tampa Bay was sequenced by 454/Pyrosequencing. The sequencing yielded 294,068 reads with 6.6% identifiable. One hundred-three sequences had significant similarity to integrases by BLASTX analysis (e≤0.001). Four sequences with strongest amino-acid level similarity to integrases were selected and real-time PCR primers and probes were designed. Initial testing with microbial fraction DNA from Tampa Bay revealed 1.9×10⁷, and 1300 gene copies of <italic>Vibrio</italic>-like integrase and <italic>Oceanicola</italic>-like integrase L⁻¹ respectively. The other two integrases were not detected. The integrase assay was then tested on microbial fraction RNA extracted from 200 ml of Tampa Bay water sampled biweekly over a 12 month time series. <italic>Vibrio</italic>-like integrase gene expression was detected in three samples, with estimated copy numbers of 2.4-1280 L⁻¹. <italic>Clostridium</italic>-like integrase gene expression was detected in 6 samples, with estimated copy numbers of 37 to 265 L⁻¹. In all cases, detection of integrase gene expression corresponded to the occurrence of lysogeny as detected by prophage induction. Investigation of the environmental distribution of the two expressed integrases in the Global Ocean Survey Database found the <italic>Vibrio</italic>-like integrase was present in genome equivalents of 3.14% of microbial libraries and all four viral metagenomes. There were two similar genes in the library from British Columbia and one similar gene was detected in both the Gulf of Mexico and Sargasso Sea libraries. In contrast, in the Arctic library eleven similar genes were observed. The <italic>Clostridium</italic>-like integrase was less prevalent, being found in 0.58% of the microbial and none of the viral libraries. These results underscore the value of metagenomic data in discovering signature genes that play important roles in the environment through their expression, as demonstrated by integrases in lysogeny.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>We are very grateful to Florent Angly for assistance with the contig spectrum analysis and Rob Edwards for assistance with some of the bioinformatic analyses. Many thanks also to Dana Hall for assistance with the Phage Proteomic Tree analysis.</p>" ]

[ "<fig id=\"pone-0003263-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003263.g001</object-id><label>Figure 1</label><caption><title>Amino acid coverage of cyanophage P-SSM2 genome by sequences from the Tampa Bay viral metagenome.</title><p>The slope of the line indicates the genome coverage with high slope indicating high coverage. Arrows identify gaps in coverage and the bracket indicates the area of highest coverage.</p></caption></fig>", "<fig id=\"pone-0003263-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003263.g002</object-id><label>Figure 2</label><caption><title>Prophage Induction and Integrase Gene Expression over an annual cycle.</title><p>The top panel depicts <italic>Synechococcus</italic> cyanophage induction (♦ = temperature, gray columns = cyanophage induction); the bottom panel shows heterotrophic prophage induction (white columns = heterotrophic light incubation, black columns = heterotrophic dark incubation). Light gray shading indicates expression of <italic>Clostridium</italic>-like integrase. Dark gray shading indicates expression of both <italic>Clostridium</italic>-like and <italic>Vibrio</italic>-like integrases. Asterisks denote statistically significant prophage induction. Note the breaks in the y axis in both panels denoted by breaks in the axis line and in the figure column.</p></caption></fig>" ]

[ "<table-wrap id=\"pone-0003263-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003263.t001</object-id><label>Table 1</label><caption><title>Top 15 Hits for Unassembled Tampa Bay Induced Viral Metagenome.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">number of hits</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Virus Identification</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Target Accession</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Category</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Virus/Organism Type</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">2572</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Cyanophage P-SSM2, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AY939844.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">T4-like <italic>Prochlorococcus</italic> cyanophage</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1184</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Cyanophage P-SSM4, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AY940168.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">T4-like <italic>Prochlorococcus</italic> cyanophage</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">901</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bacteriophage S-PM2, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AJ630128.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">T4-like “photosynthetic” <italic>Synechococcus</italic> cyanophage</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">524</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Roseophage SIO1, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AF189021.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Lytic bacteriophage of <italic>Roseobacter</italic>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">201</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Cyanophage P-SSP7, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AY939843.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">T7-like <italic>Prochlorococcus</italic> cyanophage, possibly temperate</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">174</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bacteriophage M6, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">DQ163916.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>Pseudomonas aeruginosa</italic> bacteriophage, siphovirus NOS</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">170</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bacteriophage PA11, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">DQ163915.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>Pseudomonas aeruginosa</italic> bacteriophage NOS</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">93</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bacteriophage Phi JL001, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AY576273.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">temperate phage associated with sponge bacterium</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">89</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Pseudomonas aeruginosa phage PaP3, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AY078382.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>Pseudomonas aeruginosa</italic> bacteriophage NOS</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">81</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bacteriophage S-RSM2, incomplete</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AJ628768.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Infective <italic>Synechococcus</italic> myovirus, genes ORF1, omp1 gene, psbA gene, psbD gene and tal gene</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">74</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Cyanophage P60, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AF338467.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Lytic cyanophage</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">68</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bacteriophage KVP40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AY283928.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Marine broad host-range T4-like vibriophage</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Chlamydia phage PhiCPG1, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">U41758.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Microvirus (ssDNA bacteriophage)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">61</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bordetella phage BPP-1, complete genome</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AY029185.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Podovirus, P22-T7 hybrid. Related to temperate <italic>Yersinia</italic> phage PY54</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">58</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Uncultured cyanophage clone BAC9D04</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AY456121.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">cyanophage sequence from environmental clone NOS. Genes include hypothetical protein, head-tail connector protein, capsid assembly protein, and PSII D1 protein (psbA) genes, complete cds; and unknown genes</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003263-t002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003263.t002</object-id><label>Table 2</label><caption><title>Top 15 Assembled Contigs with BLASTX hits, Tampa Bay Induced Viral Metagenome.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig number</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig size (bp)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Number of Sequences in Contig</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Top BLASTX Hit</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Target Accession</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">e-value</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Category</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Virus/Organism Type</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_13758</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3097</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">248</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">putative prophage terminase large subunit</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP 455525.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">e-112</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Prophage [Salmonella enterica subsp. enterica serovar Typhi str. CT18]</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_15615</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3004</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">290</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">orf13 [Haemophilus phage HP1], putative adenine-specific methylase</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP 043482.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.00E-13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">temperate phage HP1 of Haemophilis influenzae</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_13453</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">307</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">gp6 [Salmonella typhimurium bacteriophage ES18]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">YP 224144.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.00E-18</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">temperate, generalized transducing phage (dsDNA)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_18645</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1744</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">189</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">tail protein [Yersinia phage PY54]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP 892067.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.00E-14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">temperate phage of Yersinia enterolytica</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_16978</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1631</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">138</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">gp9 [Salmonella typhimurium bacteriophage ES18]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">YP 224147.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.00E-22</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">temperate, generalized transducing phage (dsDNA)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_18692</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1596</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">195</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">phage protein [Pseudomonas entomophila L48]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">YP 609639.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.00E-10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">prophage (phage protein in bacterial genome)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_14529</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1429</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">205</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Hypothetical protein CBG24242 [Caenorhabditis briggsae]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CAE56519.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.00E-07</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">eukaryote</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_14768</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1400</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">98</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">putative DNA primase [Pelobacter carbinolicus DSM 2380]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">YP 357233.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.00E-29</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">bacterial whole genome</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_14399</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1205</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">replication initiation protein [Banana bunchy top virus]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AAG44003.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.00E-06</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">lytic plant virus, multicomponent ssDNA</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_18753</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1128</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">89</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">hypothetical protein MED193_12573 [Roseobacter sp. MED193]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">ZP 01058432.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.00E-07</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">bacterial whole genome</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_18853</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">932</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">70</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">viral A-type inclusion protein, putative [Trichomonas vaginalis G3]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">XP 001330650</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.00E-04</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">viral protein in Trichomonad (eukaryotic pathogen)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_16209</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">923</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">41</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">hypothetical protein PputGB1DRAFT_4712 [Pseudomonas putida GB-1]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">ZP 01715014.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.00E-27</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">prophage (Phage-like protein in bacterial genome)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_18743</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">881</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">54</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">gp14 [Salmonella typhimurium bacteriophage ES18]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">YP 224152.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.00E-05</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">temperate, generalized transducing phage (dsDNA)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_18669</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">800</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">31</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">hypothetical protein ph57 [Staphylococcus phage PH15]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">YP 950719.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.00E-10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">P</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">temperate phage of Staphylococcus epidermidis</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Contig_14647</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">737</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">38</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">conserved hypothetical protein [Cyanophage P-SSM2]</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">YP 214416.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.00E-04</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">lytic T4-like cyanophage</td></tr></tbody></table></alternatives></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pone.0003263.s001\"><label>Table S1</label><caption><p>Integrase sequences from Tampa Bay Metagenome.</p><p>(0.03 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003263.s002\"><label>Figure S1</label><caption><p>Phage Proteomic Tree Analysis of Unassembled and Assembled Sequences From Induced Viral Metagenome. The left side of the figure (y-axis) is the tree constructed based on the phage and prophage database. The x-axis represents the abundance (number of blast hits) of each phage type normalized to the must abundant type, which is set at 1. The left column (blue) is the relative distribution of phage hits in the raw sequences. The right column (red) indicates the relative distribution of phage hits in the assembled sequences. The most abundant phage identifications in each column are labeled in the figure.</p><p>(5.90 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003263.s003\"><label>Figure S2</label><caption><p>Ambient parameters in Tampa Bay throughout the annual cycle. The top panel depicts primary productivity (green) and bacterial productivity (blue). The center panel depicts total viral abundance (black) and total bacterial abundance (brown). The bottom panel depicts the ambient Synechococcus abundance (red) and infectious cyanophage abundance (gray). Error bars indicate the standard deviation.</p><p>(11.10 MB TIF)</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"nt101\"><p>Category: C = cyanophage, P = prophage, - = other</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt102\"><p>Category: C = cyanophage, P = prophage, - = other</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>L.M. and J.P. were supported by grants OCE-0221763 and EF-0801593 from the National Science Foundation. M.B. was supported by grant MCB-0701984 from the National Science Foundation. M.B. also received grant support from the Alfred P. Sloan Foundation (BR-4772).</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003263.g001\"/>", "<graphic id=\"pone-0003263-t001-1\" xlink:href=\"pone.0003263.t001\"/>", "<graphic id=\"pone-0003263-t002-2\" xlink:href=\"pone.0003263.t002\"/>", "<graphic xlink:href=\"pone.0003263.g002\"/>" ]

[ "<media xlink:href=\"pone.0003263.s001.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003263.s002.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003263.s003.tif\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Langerhans cell histiocytosis (LCH, a.k.a. histiocytosis X), is a rare disease that affects mainly young children, and features granulomas consisting of Langerhans-like cells (LC), mixed with macrophages, eosinophiles, multinucleated giant cells, and lymphocytes, that can be found within various tissues ##REF##13057466##[1]##, ##REF##8813865##[2]##.</p>", "<p>The presence of LC in granuloma is a key diagnostic feature of LCH. LC are members of the dendritic cell (DC) family, that trigger and shape immune responses, and the pathophysiology of LCH is likely to involve immune mechanisms (reviewed in ##REF##9515204##[3]##, ##REF##12697451##[4]##). We have previously reported that LC found in LCH granuloma were phenotypically and functionally immature/semi-mature LC ##REF##11222366##[5]##. Immature/semi-mature DCs are believed to be prone to induce regulatory T cells, that inhibit polyclonal T cell responses and promote tolerance ##REF##11208869##[6]##, ##REF##14644138##[7]##, ##REF##11560993##[8]##, ##REF##11773639##[9]##. The accumulation of immature LC in LCH granulomas was associated with the expansion of FoxP3⁺ CD25 + CD4⁺ regulatory T cells both in granuloma and in the blood of patients ##REF##17696642##[10]##. Therefore local and general immunosuppression, which favors reactivation of herpes-virus infection, may be a feature of LCH.</p>", "<p>Environmental agents and viruses, in particular Epstein-Barr virus (EBV), or vaccination, have been proposed to trigger, or to play a role in the pathogenesis of the disease ##REF##15243808##[11]##, ##UREF##0##[12]##. Herpesviruses are DNA viruses responsible for persistent infection. EBV is the etiological agent of several malignancies ##REF##6101750##[13]##, ##REF##15510157##[14]##, and EBV &amp; Cytomegalovirus (CMV) are responsible for hemophagocytic syndromes in human with several inherited immunodeficiencies ##REF##15661029##[15]##, ##REF##17080092##[16]##. EBV has been reported to infect monocytes and Langerhans cells (LC), during the natural course of infection in human ##REF##17376918##[17]##, ##REF##17376908##[18]##. Infection with EBV has been reported to be associated with LCH, to represent a possible etiology, and/or to contribute to its pathophysiology in some studies ##REF##15243808##[11]##, ##UREF##0##[12]##, ##REF##15257550##[19]##. However, other studies failed to replicate these findings, and the possible causative role of EBV in LCH is debated ##REF##8042610##[20]##, ##REF##17027063##[21]##, ##UREF##1##[22]##. CMV can also infect DC and LC ##REF##12805456##[23]##, ##REF##16951359##[24]##, ##REF##14962896##[25]##, and one single study reported CMV detection in lesional LC in one third of 29 patients by immunohistochemistry, in situ hybridization, and PCR ##REF##10071244##[26]##. HHV-6 infects mainly T cells, but is also reported to infect myeloid cells ##REF##1646280##[27]##, ##REF##12116029##[28]## and HHV-6 DNA or immunoreactivity was detected in lesions of 50% to 75% of patient with LCH ##REF##8228322##[29]##, ##REF##14676546##[30]##, however, control studies performed by the same group concluded that the prevalence of HHV-6 in the tissue of LCH patients is the same as that found in tissue from individuals without disease ##REF##16479562##[31]##.</p>", "<p>Sero-epidemiological studies have been useful to demonstrate the role of EBV in Burkitt lymphoma and Hodgkin disease, when high antibody titers to EBV structural antigens (VCA) have been associated to the risk of developing Burkitt's lymphoma and Hodgkin diseases ##REF##210392##[32]##, ##REF##2537928##[33]##. However, to our knowledge no sero-epidemiological study have been conducted in LCH. The present study was therefore designed to investigate the role of EBV, CMV, and HHV-6 using two methods. First, we performed a case-controlled sero-epidemiological study to investigate a relationship between the onset of LCH in young children and the antibody response to infection with EBV, CMV, or HHV-6, and second we searched for the presence of viruses in the serum of patients, and in Langerhans cells in tumor samples, by PCR and, when positive, we investigated the cellular target of the viruses by immunolabeling and in situ hybridization. Results ruled out an epidemiological association between these herpes-virus and LCH.</p>" ]

[ "<title>Methods</title>", "<title>Patients &amp; controls</title>", "<p>The diagnostic and inclusion criteria as well as the definition of the organs involvement for the French nationwide LCH survey have been described elsewhere ##REF##8813865##[2]##. Briefly, the extension of the diseases has been classified in three groups according to histiocyte society criteria: group 1: single organ extension, without risk organs group 2 mulsystem organ without risk organs and group 3 patients with risk organs i.e. lung and/or liver and/or spleen and/or hematological dysfunction. According to French bioethics laws, informed consent was signed if the patients participated and the database was approved by the French computer watchdog commission (CNIL certificate n° 99.087). Clinical information, radiological findings and extension were recorded, together with treatments received. Data monitoring, based on medical charts, was done by a clinical research associate who visited each center. Involvement of at least one new organ, as described elsewhere 2, was considered to define an LCH episode. Serum specimens were obtained, after written witnessed informed consent was obtained from the parents of all patients, from 83 pediatric patients with LCH included in the French LCH registry, following a research protocol approved by the ethics committee of the Nantes University Hospital (France, EU). Biopsies samples were also obtained from 19 patients with LCH included in the French LCH registry after written witnessed informed consent was obtained. Control serum samples were obtained from the children admitted to the outpatient unit of the Grenoble University Hospital (France, EU), according to institutional guidelines, and were matched for age with patient samples for statistical analysis.</p>", "<title>Serology</title>", "<p>Qualitative and quantitative analysis of specific Ig against EBV, CMV and HHV-6 were performed on serum samples from 83 patients and 235 controls using microplate ELISA kits: IgG anti-VCA (ETI-VCA-G, DiaSorin®), IgG anti-EBNA (ETI-EBNA-G, DiaSorin®), IgM anti-VCA (DiaSorin®,ETI-EBV-M), IgG anti-CMV(ETI-CYTOK-G PLUS, DiaSorin®) and IgG anti-HHV-6 (HHV-6 IgG EIA, Biotrin®). Patients and controls were classified in three categories according to the detection of specific IgG i/above the upper limit of the grey zone for “infected” status, ii/in the grey zone for “unknown” status, iii/below the lower limit of the grey zone for “uninfected” status.</p>", "<title>DNA extraction from frozen biopsies and serum</title>", "<p>DNA was extracted from sliced frozen biopsies using ALLPrep DNA/RNA minikit (Qiagen, Hilden, Germany). Total DNA was eluted with 200 ul of water and amplification of the beta-globin gene by real time PCR was used to evaluate total cell number per sample (Beta-globine PCR Kit, Roche Diagnostics®). DNA was extracted from 200ul of serum using Blood DNA mini kit and eluted in a volume of 100ul.</p>", "<title>Quantitative PCR for the detection of EBV, CMV and HHV-6</title>", "<p>ten ul of DNA extract were used to detect and quantify viral genomes by real time PCR assays. All positive PCR were run a second time for confirmation.</p>", "<p>EBV tyrosine kinase gene (TK) was amplified from biopsies DNA with T1 (<named-content content-type=\"gene\">5′-GGGGCAAAATACTGTGTTAG-3′</named-content>)+T2 (<named-content content-type=\"gene\">5′-CGGGGGACACCATAGT-3′</named-content>) primers and LC1 (<named-content content-type=\"gene\">5′-ATGTTTCCTCCCTCGCTTCTTCAG-fluo-3′</named-content>)+LC2 (<named-content content-type=\"gene\">5′-ATGTTTCCTCCCTCGCTTCTTCAG-fluo-3′</named-content>) probes. PCR were run on a Light Cycler. The EBV-negative DG75 cell line was used as a negative control. The EBV-positive Burkitt's lymphoma cell line “Namalwa” ##REF##6300885##[45]## harboring 2 copies of viral genome per cell is used as a standard for quantification. Detection limit is 2 copies of viral genome per amplification ##REF##11793388##[46]##.</p>", "<p>CMV immediate early-1 gene (IE1) was amplified using forward <named-content content-type=\"gene\">5′-GCAGACTCTCAGAGGAT-3′</named-content>+reverse <named-content content-type=\"gene\">5′-AGCGCCGCATTGAGGA-3′</named-content> primers and a 6-carboxyfluoresceine (FAM) -<named-content content-type=\"gene\">5′ ATCTGCATGAAGGTCTTTGCCCAGTACATT-3′</named-content> carboxytetramethyl rhodamine (TAMRA) probe. PCR were run on a ABI 7300 real time PCR system (Applied Biosystem).Quantification was obtained using a plasmid and the detection limit is 20 copies of viral genome per amplification ##REF##12734246##[35]##.</p>", "<p>HHV-6 U65-U66 gene was amplified using forward <named-content content-type=\"gene\">5′- GACAATCACATGCCTGGATAATG-3′</named-content>+reverse <named-content content-type=\"gene\">5′-TGTAAGCGTGTGGTAATGGACTAA -3′</named-content> primers and a 6-carboxyfluoresceine (FAM) -<named-content content-type=\"gene\">5′ AGCAGCTGGCGAAAAGTGCTGTGC-3′</named-content> carboxytetramethyl rhodamine (TAMRA) probe. PCR were run on a ABI 7300 real time PCR system (Applied Biosystem).Quantification was obtained using a plasmid and detection limit is 25 copies of viral genome per amplification ##REF##11742650##[47]##.</p>", "<p>PCR results were expressed as the number of viral genome copies per million cells for biopsies and as the number of viral genome copies per ml of serum from peripheral blood.</p>", "<title>In situ hybridization &amp; immunohistochemistry</title>", "<p>Detection of EBV EBERs RNA by in situ hybridization was performed first and was followed by immunohistochemistry with antibodies against B-cell, T-cell and Langerhans cell antigens. Five micrometers paraffin-embedded sections were mounted onto glass slides and pretreated in 0.4% pepsine HCl 0.2M, then hybridized with EBER PNA probes (DAKO), following the manufacturer's instructions, and revealed with NBT/BCIP after APAAP amplification. Immunohistochemistry was then performed using a streptavidin-biotin peroxidase method (LSAB2, DAKO) after microwave heating, revealed with AEC or DAB. Antibodies used were directed against the following antigens : CD20 (L26, 1∶200, Dako,), CD79a (JCB117, 1∶50, Dako), CD3 (UCHT1, 1∶50 Dako), CD1a (MTB1, 1∶25,Novocastra), CD68 (KP1, 1∶200, Dako).</p>", "<title>Statistical methods</title>", "<p>Stata Software® version 8 was used for all statistical analyses. Categorical data were compared by using Fisher's exact test, and quantitative data (titer if seropositivity) by using Kruskal-Wallis non parametric test. All tests were two-tailed. P values of less than 0.05 were considered to indicate statistical significance unless otherwise stated.</p>" ]

[ "<title>Results</title>", "<p>Prevalence of EBV, CMV, and HHV-6 infection as a function of age and of clinical presentation of disease. The presence of antibodies against EBV, CMV and HHV-6 was investigated by ELISA in the serum of 78 children diagnosed with LCH and 206 age matched controls. To avoid contamination by maternal IgG, only children six month-old and older where studied for the presence of specific antibodies in serum. Twenty six children with LCH (33.3%) and 94 controls (47.1%) tested seropositive for EBV. No significant difference was observed between patient and control groups when matched for age (##TAB##0##Table 1##). Similarly, no significant difference was observed with the control group when patients were grouped by disease stage/clinical presentation (##TAB##1##Table 2##). Similar results were observed for CMV and HHV-6, and the prevalence of both infections in children with LCH were comparable to their prevalence among age-matched controls (##TAB##1##Tables 2##,and ##TAB##2##3##).</p>", "<p>Serum IgG titers directed against EBV, as well as CMV and HHV-6, did not differ between patients and controls. Titers of EBV antibodies have been shown to differ from controls in several diseases linked to EBV ##REF##15510157##[14]##, ##REF##16390946##[34]##. IgG VCA titers, and in some cases anti-Epstein-Barr nuclear antigen (EBNA)1 antibody, are consistently higher in patients with nasopharynx carcinoma or Hodgkin disease than in control populations ##REF##15510157##[14]##. Lower EBNA1-IgG antibody titer is also considered as a possible serological sign for a defective control of the persistent latent EBV carrier state. In the present study, both VCA IgG and EBNA IgG titers were found to be similar in patients with LCH in comparison with controls, when matched for age, or when grouped by disease stage/clinical presentation (##FIG##0##Figure 1A–E##). Anti HHV-6 or anti CMV IgG titers were also not different between patients and the control group (data not shown)</p>", "<p>Prevalence of detectable serum viral load for EBV, CMV, and HHV-6. Defective control of infection by Herpesviruses such as EBV, CMV and HHV-6 ultimately results in replicative infection and viremia ##REF##12734246##[35]##. Serum viral load during CMV infection is a sensitive technique, but in the present study, out of 83 patients and 236 age-matched controls, 0% of patient and 3% of controls (n = 7) had detectable CMV serum viral load (##TAB##3##table 4##) indicating the very low indidence of active infection. Serum viral load during EBV infection is considered to be less sensitive than during CMV infection ##REF##16206064##[36]##, nevertheless we found that only 1.2% of patients (n = 1) and a similar percentage (0.9% of controls, n = 2) had a detectable EBV viral load in serum (##TAB##3##Table 4##). These results are consistent with anti-VCA IgG titers, which were found similar in patients and controls (see ##FIG##0##figure 1##), since EBV viral load was shown to be correlated with anti-VCA IgG titer ##REF##16390946##[34]##. The significance of the detection of HHV-6 in the serum is still discussed ##REF##17229866##[37]##, ##REF##17339132##[38]##, but again only 3.6% of patients (n = 3) and 2.5% of controls (n = 6) had a detectable HHV-6 serum viral load, and there was no difference between groups as per fischer exact's test (##TAB##3##Table 4##).</p>", "<p>Detection of EBV, CMV, and HHV-6 DNA in biopsy samples from LCH granuloma. We then investigated whether Herpesviruses were present in LCH granuloma, in the absence of overt viremia. As shown in ##TAB##4##table 5##, EBV DNA was found in 15% of biopsy samples examined (3/19 patients, #1, 9, &amp; 15), and HHV-6 DNA in 26% of biopsy samples examined (5/19 patients). CMV DNA was not detected in this series. EBV and HHV-6 viral loads were low with a median of 400 copies/10⁶ cells except for patient #5. Patient #5 display a very high HHV-6 load in the biopsy and in the blood over time (data not shown) compatible with the detection of a chromosomally integrated HHV-6 DNA ##REF##17133548##[39]##.</p>", "<p>Therefore, our data support previous results showing the presence of EBV or HHV-6 in a subset of LCH granuloma. However, since HHV-6 and EBV are responsible for persistent infection of T cells and B cells respectively, the detection of virus DNA in LCH samples could be attributed to infection of bystander cells ##REF##17027063##[21]##.</p>", "<p>We therefore investigated whether EBV infects bystanders lymphocytes, macrophages, and Langerhans cells in samples from patients #1, 9, &amp; 15, which tested positive by PCR, using in situ hybridisation with a probe against EBV EBERs RNA on paraffin sections of LCH granuloma, followed by immunohistochemistry with antibodies against B-cells (CD20 &amp; CD79a), T-cells (CD3), macrophages (CD68), or Langerhans cell (CD1a) antigens on the same tissue sections. EBV positive cells were always found in areas rich in B cells, and were labeled with either CD20, or CD79a antibodies (##FIG##1##Figure 2##), while no EBER+ cell labelled with CD1a, CD3, or CD68 cell was observed. These data indicate that EBV infects bystander B cells, which are present in a subsets of LCH granuloma ##UREF##0##[12]##. Although positive controls were obtained, we did not observed immunoreactivity against HHV-6 in samples positive for HHV-6 DNA by PCR (data not shown), therefore the cellular target of HHV6 in LCH granuloma was not identified.</p>" ]

[ "<title>Discussion</title>", "<p>The present study is the largest case-controlled sero-epidemiological study performed in this disease. The results argue against an epidemiological association between LCH and EBV, CMV, or HHV-6 infection.</p>", "<p>Further investigation of the cellular localization of the viruses in LCH tissue samples from 19 patients indicated that, when present, EBV infected bystander lymphocytes and not Langerhans cells. We think that our results therefore resolve the much debated issue of the role of EBV in the pathogenesis of LCH ##REF##15257550##[19]##, ##REF##17027063##[21]##, ##UREF##1##[22]##, ##REF##14676546##[30]##, ##REF##16479562##[31]##. Because EBV infected B cells are present in the blood of healthy seropositive individuals ##REF##3030940##[40]## at a low frequency in the order of 10-⁶\n##REF##1333480##[41]## it is not unexpected to find a small number of EBV infected B cells in LCH granuloma, which contains B cells. In addition, because the growth of EBV infected B cells is under the control of CD8 cytotoxic T cells in healthy individuals ##REF##7565411##[42]##, the regulatory T cell-rich environment of the LCH granuloma ##REF##17696642##[10]##, may represent a sanctuary for such EBV infected B cells.</p>", "<p>Anti CMV IgG titers in patients and controls, and absence of detection of the virus by PCR in the serum and in biopsy samples strongly suggest that CMV is not associated with LCH.</p>", "<p>The case of HHV6 is more complicated. HHV-6 was detected by PCR in the serum of 3.8% of patients and 2.5% of controls with viral loads above 10⁴ copies/ml in 5 out of 9 detected (data not shown). The prevalence of HHV-6 in patients and controls were similar to the reported prevalence of chromosomally integrated HHV6 among blood donors ##REF##17133548##[39]##. Therefore HHV6 detection may be related to the presence of chromosomally integrated HHV-6. HHV6 DNA was also detected by PCR in biopsies from 5/19 patients (26%), suggesting that HHV6, like EBV can be found in LCH granuloma. However the viral load was very low, and the virus was not detectable by immunochemistry suggesting that HHV-6 infection is quiescent. Together with our serological data and serum PCR, and in the light of the study of Glotzbecker et al, ##REF##16479562##[31]## who found a similar frequency PCR positive samples in LCH granuloma and control tissues, we hypothetize that HHV6 do not productivey infect LC in vivo and may be carried by bystander lymphocytes which are present in the granuloma.</p>", "<p>Our study did not identify a subset of patients with ‘herpes-viruses associated’ LCH. We did not found an association between EBV, CMV, or HHV-6 infection and patients when they were stratified by age, or by clinical stage.</p>", "<p>Among other herpes viruses, HHV-1 (HSV-1), HHV-2 (HSV-2), HHV-3 (VZV) and HHV-7 infections were never reported to be associated with LCH, and were not investigated in this study. Among herpesviruses, HHV-8 has been initially reported to infect dendritic cells, but several studies have excluded its association with LCH ##REF##12210448##[43]##, ##REF##10911374##[44]##.</p>" ]

[]

[ "<p>Conceived and designed the experiments: BS TM JD FG. Performed the experiments: EJ BS FD MLV. Analyzed the data: EJ BS TM MLV IJ JD FG. Contributed reagents/materials/analysis tools: TM PM CG LM JG MD FJ JMS CT IJ. Wrote the paper: BS JD FG.</p>", "<title>Background</title>", "<p>Langerhans cell histiocytosis (LCH) is a rare disease that affects mainly young children, and which features granulomas containing Langerhans-type dendritic cells. The role of several human herpesviruses (HHV) in the pathogenesis of LCH was suggested by numerous reports but remains debated. Epstein-barr virus (EBV, HHV-4), &amp; Cytomegalovirus (CMV, HHV-5) can infect Langerhans cells, and EBV, CMV and HHV-6 have been proposed to be associated with LCH based on the detection of these viruses in clinical samples.</p>", "<title>Methodology</title>", "<p>We have investigated the prevalence of EBV, CMV and HHV-6 infection, the characters of antibody response and the plasma viral load in a cohort of 83 patients and 236 age-matched controls, and the presence and cellular localization of the viruses in LCH tissue samples from 19 patients.</p>", "<title>Principal Findings</title>", "<p>The results show that prevalence, serological titers, and viral load for EBV, CMV and HHV-6 did not differ between patients and controls. EBV was found by PCR in tumoral sample from 3/19 patients, however, EBV small RNAs EBERs –when positive-, were detected by in situ double staining in bystander B CD20⁺ CD79a⁺ lymphocytes and not in CD1a⁺ LC. HHV-6 genome was detected in the biopsies of 5/19 patients with low copy number and viral Ag could not be detected in biopsies. CMV was not detected by PCR in this series.</p>", "<title>Conclusions/Significance</title>", "<p>Therefore, our findings do not support the hypothesis of a role of EBV, CMV, or HHV-6 in the pathogenesis of LCH, and indicate that the frequent detection of Epstein-barr virus (EBV) in Langerhans cell histiocytosis is accounted for by the infection of bystander B lymphocytes in LCH granuloma. The latter observation can be attributed to the immunosuppressive micro environment found in LCH granuloma.</p>" ]

[]

[]

[ "<fig id=\"pone-0003262-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003262.g001</object-id><label>Figure 1</label><caption><title>VCA-IgG and EBNA-IgG titers (UA/ml) in patients and controls with past EBV infection.</title><p>VCA-IgG (A, B, C) and EBNA-IgG titers (D, E, F) (UA/ml) were determined as indicated in <xref ref-type=\"sec\" rid=\"s4\">methods</xref>, and patients were compared to controls without stratification (A, D), and after stratification based on age (B, E), or disease extension (C, F).</p></caption></fig>", "<fig id=\"pone-0003262-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003262.g002</object-id><label>Figure 2</label><caption><title>Detection of EBV-infected B cells by in situ hybridization combined with immunohistochemistry.</title><p>Granuloma serial sections were stained for CD1a (upper left) or for CD20 (middle left) by immunohistochemistry or for EBERs by in situ hybridisation (upper right, arrows indicate EBER positive cells). Combined detection of CD20 or CD79a with EBERs shows that EBV-infected cells are B cells (arrows).</p></caption></fig>" ]

[ "<table-wrap id=\"pone-0003262-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003262.t001</object-id><label>Table 1</label><caption><title>Prevalence of EBV, CMV, and HHV-6 infection in LCH patients</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Controls</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH3</td></tr></thead><tbody><tr><td colspan=\"7\" align=\"left\" rowspan=\"1\">EBV</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Non-infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">109</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">52</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">12</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">12</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">52.9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">66.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">63.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">75</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">26</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">41.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">33.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">34.9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">36.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">25</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">206</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">78</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">43</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Test Fisher's exact</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.475</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr><tr><td colspan=\"7\" align=\"left\" rowspan=\"1\">CMV</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Non-infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">131</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">58</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">32</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">75.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">76.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">68.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">81.2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">18</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">35.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">23.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">21.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">31.6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">18.8</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<xref ref-type=\"table-fn\" rid=\"nt101\">Unknown*</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">204</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">77</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">42</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Test Fisher's exact</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.282</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr><tr><td colspan=\"7\" align=\"left\" rowspan=\"1\">HHV-6</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Non-infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">33</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">11.9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">26.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">31.2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">161</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">62</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">37</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">11</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">80.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">80.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">88.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">68.7</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<xref ref-type=\"table-fn\" rid=\"nt101\">Unknown*</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">201</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">77</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">42</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Test Fisher's exact</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.424</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003262-t002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003262.t002</object-id><label>Table 2</label><caption><title>Prevalence of EBV, CMV, and HHV-6 infection as a function of age</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td colspan=\"2\" align=\"left\" rowspan=\"1\">Age 0.5 to 5</td><td colspan=\"2\" align=\"left\" rowspan=\"1\">Age 5 to 10</td><td colspan=\"2\" align=\"left\" rowspan=\"1\">Age 10 to 20</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Controls</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Controls</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Controls</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH</td></tr></thead><tbody><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">EBV</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Non-infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">79</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">35</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">71.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">85.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">35.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">45.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">25.6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">46.1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">31</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">32</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14.6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">54.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">74.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">53.8</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">110</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">41</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">53</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">24</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">43</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Test Fisher's exact</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.093</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.577</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.378</td></tr><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">CMV</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Non-infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">74</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">29</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">27</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">11</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">69.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">85</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">54.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">54.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">62.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">84.6</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">33</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">23</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">30.6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">43.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">41.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">37.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">15.4</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Unknown*</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">107</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">53</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">24</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">43</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Test Fisher's exact</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.062</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.82</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.186</td></tr><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">HHV-6</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Non-infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">18</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">35</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">23.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.7</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">84</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">26</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">47</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">24</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">30</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">12</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">79.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">88.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">71.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">92.3</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Unknown*</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">106</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">53</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">24</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">42</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Test Fisher's exact</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.037</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.217</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.356</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003262-t003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003262.t003</object-id><label>Table 3</label><caption><title>Prevalence of HHV-6 infection in children before the age of 5 years</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td colspan=\"2\" align=\"left\" rowspan=\"1\">Age 0.5 to 1</td><td colspan=\"2\" align=\"left\" rowspan=\"1\">Age 1 to 1.5</td><td colspan=\"2\" align=\"left\" rowspan=\"1\">Age 1.5 to 2</td><td colspan=\"2\" align=\"left\" rowspan=\"1\">Age 2 to 5</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">controls</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">controls</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">controls</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">controls</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Non-infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">35.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">66.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">33.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">30.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">21</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Infected</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">49</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">15</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">33.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">95.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">66.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">69.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">80.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">79</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Unknown*</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">12</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">18</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">61</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Test Fisher's exact</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.238</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.054</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.529</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.663</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003262-t004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003262.t004</object-id><label>Table 4</label><caption><title>Prevalence of detectable serum viral load for EBV, CMV, and HHV-6</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Patients</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Controls</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td></tr></thead><tbody><tr><td colspan=\"4\" align=\"left\" rowspan=\"1\">EBV</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Negative</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">82</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">234</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">316</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">98.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99.1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Positive</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.9</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">83</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">236</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">319</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td colspan=\"4\" align=\"left\" rowspan=\"1\">Fisher's exact = 1.000</td></tr><tr><td colspan=\"4\" align=\"left\" rowspan=\"1\">1-sided Fisher's exact = 0.596</td></tr><tr><td colspan=\"4\" align=\"left\" rowspan=\"1\">CMV</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Negative</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">83</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">229</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">312</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97.</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97.8</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Positive</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">83</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">236</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">319</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td colspan=\"4\" align=\"left\" rowspan=\"1\">Fisher's exact = 0.197</td></tr><tr><td colspan=\"4\" align=\"left\" rowspan=\"1\">1-sided Fisher's exact = 0.118</td></tr><tr><td colspan=\"4\" align=\"left\" rowspan=\"1\">HHV-6</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Negative</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">80</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">230</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">310</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">96.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97.2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Positive</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.8</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Total</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">83</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">236</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">319</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td colspan=\"4\" align=\"left\" rowspan=\"1\">Fisher's exact = 0.701</td></tr><tr><td colspan=\"4\" align=\"left\" rowspan=\"1\">1-sided Fisher's exact = 0.428</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003262-t005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003262.t005</object-id><label>Table 5</label><caption><title>Detection of EBV, CMV, and HHV-6 DNA in biopsy samples from LCH granuloma.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Patient</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">stage</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">sex</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">HHV-6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">EBV</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CMV</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6 10³\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8 10²\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6 10¹\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2 10⁶\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.7 10⁵\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">12</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4 10²\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">16</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4 10²\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">18</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">19</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LCH3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">M</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4 10²\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr></tbody></table></alternatives></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"nt101\"><p>Unknown^*: IgG value are in the ‘grey zone’ (see <xref ref-type=\"sec\" rid=\"s4\">methods</xref>)</p></fn><fn id=\"nt102\"><p>LCH1: single system, no risk organs</p></fn><fn id=\"nt103\"><p>LCH2: multisystem, no risk organs</p></fn><fn id=\"nt104\"><p>LCH3: risk organs</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt105\"><p>Positive results are expressed as the number of viral genome copies per million cells.</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>EJ was supported by the Fondation pour la Recherche Medicale (FRM). BS was the recipient of a PhD fellowship from INSERM. Work was supported by a grant from the Agence Nationale de la Recherche (EPILCH 2005) to JD, IJ and FG, by grants from the FRM, the City of Paris, and an EURopean Young Investigator (EURYI) award to FG. Sponsors or funders did not have any role in the design and conduct of the study, collection, analysis, and interpretation of the data, and in the preparation, review, or approval of the manuscript.</p></fn></fn-group>" ]

[ "<graphic id=\"pone-0003262-t001-1\" xlink:href=\"pone.0003262.t001\"/>", "<graphic id=\"pone-0003262-t002-2\" xlink:href=\"pone.0003262.t002\"/>", "<graphic id=\"pone-0003262-t003-3\" xlink:href=\"pone.0003262.t003\"/>", "<graphic xlink:href=\"pone.0003262.g001\"/>", "<graphic id=\"pone-0003262-t004-4\" xlink:href=\"pone.0003262.t004\"/>", "<graphic id=\"pone-0003262-t005-5\" xlink:href=\"pone.0003262.t005\"/>", "<graphic xlink:href=\"pone.0003262.g002\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>The rapid development of genomic databases, bioinformatics tools, laboratory robotics and enabling technologies such as cDNA and oligonucleotide microarrays have provided new insights and understanding into biological and disease processes thru the global analysis of gene expression patterns. Continued development of high-throughput platforms, such as protein microarray technologies, are essential to furthering our understanding of protein function, quantitative proteomics, molecular interactions and protein profiling ##REF##10866210##[1]##–##REF##15253650##[3]##. Unfortunately, inherent cost and technical limitations, including the required production of large libraries of purified proteins and long-term maintenance of array stability and integrity, have caused protein microarray development to lag behind that of DNA microarrays ##UREF##0##[2]##, ##REF##10606671##[4]##. Nevertheless, despite these limitations, several groups have demonstrated proof-of-concept and the potential of protein microarray technology ##REF##10606671##[4]##–##UREF##1##[9]##.</p>", "<p>In an effort to address these issues, Nord et al. developed an alternative platform, termed protein microbead display, wherein proteins are captured <italic>via</italic> antigen-antibody binding as they are synthesized ##REF##15232106##[10]##. This technology utilizes a biotin labeled PCR product (containing a T7 promoter and a FLAG epitope in-frame with two IgG binding domains) anchored onto microbeads through streptavidin-biotin affinity binding. Anti-FLAG antibody is then immobilized onto the same microbead. The beads are incubated in a coupled cell-free transcription-translation extract to produce the targeted protein. Newly synthesized proteins are trapped via Flag peptide (antigen)-Flag antibody interaction. More recently, Ramachandran et al. ##REF##9334221##[11]## applied a similar antibody mediated protein anchoring technology to a microarray format. This platform employs purified expression construct DNAs arrayed onto a microscope slide <italic>via</italic> biotin-avidin interaction. The encoded inserts are fused with GST protein to produce GST-fusion proteins. The slides are simultaneously printed with polyclonal GST antibody to capture the newly synthesized GST fusion-proteins following coupled cell-free transcription-translation on the surface of the microarray. In both cases, newly synthesized proteins are captured through protein-protein (antigen-antibody) interactions. Although both technologies have been successfully employed, they also have their limitations. First, both platforms require a second protein, the antibody, to capture the synthesized fusion-protein. This antibody needs to be purified which adds to both labor and cost. Second, given that proteins (i.e. the capture antibody) need to be arrayed with the expression construct, maintaining the stability and integrity of the microarrays for extended periods of time remains an issue. We have addressed both of these issues by eliminating the need for antibody or other capture reagent to immobilize the newly synthesized proteins onto the microarray surface. In our system, the expression vector DNA not only directs the synthesis of each protein, but also serves to capture the protein at it's designated location on the microarray surface. Since only plasmid DNA is printed, array fabrication is simple and array stability is not an issue. To accomplish this we have exploited the high-affinity binding (∼3–7×10 ⁻¹³ M) of <italic>E. coli</italic> Tus protein to Ter, a 20 bp DNA sequence involved in the regulation of <italic>E. coli</italic> DNA replication ##REF##11523786##[12]##, ##UREF##2##[13]##. In our system, each protein of interest is synthesized as a Tus fusion protein and each expression construct directing the protein synthesis contains embedded Ter DNA sequence. The embedded Ter sequence functions as a capture reagent for the newly synthesized Tus fusion protein.</p>" ]

[ "<title>Methods</title>", "<title>A. Construction of base microarray plasmid</title>", "<p>For convenience, a recombinational cloning system was used (Invitrogen, Carlsbad, CA). First, a destination vector was made using Tus as the carboxy fusion partner. A modified, Tus (E47Q) with higher affinity for the Ter DNA sequence ##REF##11523786##[12]## was amplified from plasmid DNA by standard procedure. Oligos used for Tus amplifications were:</p>", "<p>Forward- <named-content content-type=\"gene\">5′-ATT TTA <bold><underline>GCT AGC</underline></bold> GGA GGT GCG CGT TAC GAT CTC GTA GAC CGA CTC-3′</named-content> and Reverse <named-content content-type=\"gene\">5′-TATATT <bold><underline>CAA TTG</underline></bold> TTA atg atg gtg atg atg gtg ATC TGC AAC ATA CAG GTG CAG CCG TGG 3′</named-content>.</p>", "<p>Restriction sites NheI and MunI are indicated as bold and underlined. A six-histidine tag (small letters) was incorporated in reverse oligo so that Tus will be his-tagged for downstream identification.</p>", "<p>The PCR product was purified, digested with NheI and MunI, run on an agarose gel, and the fragment excised. The fragment was then cloned into a derivative pDest47 (Invitrogen, Carlsbad, CA) termed pDest472 that had been digested with the same enzymes to create pDest 472-Tus. Correct clones were selected by digestion and verified by sequencing.</p>", "<p>A Ter site (bold) was synthesized by annealing two complementary oligos:</p>", "<p>\n<named-content content-type=\"gene\">CCGGC <bold>CACTTTAGTTACAACATACTTATT</bold>AT</named-content>\n</p>", "<p>\n<named-content content-type=\"gene\">CGA<bold>TAATAAGTATGTTGTAACTAAAGTG</bold>G</named-content>\n</p>", "<p>Following annealing, these oligos form a double stranded Ter site with ClaI and NgoMIV overhangs. The annealed oligo was cloned in pDest 472-Tus digested with NgoMIV and ClaI to create pDest Microarray TT-1. The clone was verified by sequencing. This is the base plasmid to clone any protein of interest by recombinational cloning.</p>", "<p>In addition to a wild type Ter site, a mutant Ter site was also tested for Tus fusion capture. The mutant Ter site was obtained during the course of cloning the wild-type Ter site. The sequence of the mutant was found to be:</p>", "<p>\n<bold><named-content content-type=\"gene\">CACTTTAGTTACAACATA<underline>T</underline>TTATT</named-content></bold>\n</p>", "<p>The site of the mutation is underlined. It has been shown that mutation at this particular site will reduce binding affinity by almost 4-fold ##REF##9334221##[11]##. This position is equivalent to position 6 according to the nomenclature in the manuscript. The Ter sequence has been presented in reverse orientation ##REF##9334221##[11]##.</p>", "<title>B. Construction of GFP fusion plasmid</title>", "<p>As a proof-of-concept, we cloned GFP as a fusion with Tus. pEL100 contains eGFP gene in pDonr223 (Invitrogen, Carlsbad, CA). It contains a Kozak sequence upstream of ATG and no stop codon at the C-terminus. Thus, upon recombinational cloning into pDest Microarray TT-1, GFP will be fused in frame with Tus. Recombinational cloning was performed as per manufacturer's (Invitrogen, Carlsbad, CA) directions. Finally, the clone was sequenced to confirm correct insertion.</p>", "<title>C. Microarray Fabrication</title>", "<p>Microarray protein expression vectors were prepared in 3× standard saline citrate (SSC) in a 384-well plate (Genetix, Boston, MA) and arrayed on nitrocellulose coated “Fast Slides” (Schleicher &amp; Schuell BioScience, Keene, NH) using a Microgrid II microarray robot at 50% humidity. Microarray features were printed at a spacing of 0.55mm (center to center) with 1.2mm spacing between each sub array. After printing, microarrays were baked at 80°C for 30 min. Slides were blocked with 0.1% PVP/0.05% Tween 20 for 1 h prior to expression.</p>", "<p>To confirm uniform DNA spotting, a sample slide from each print set was stained for DNA content using the IDT (Coralville, IA) Cy3-SpotQC detector oligo (9mers) diluted in 0.1% PVP/0.05% Tween 20 buffer. Following incubation, slides were washed 1× for 3 minutes in 10×SSC, 0.2% Sarkosyl, followed by a second wash in 10×SSC for one minute. After a third wash in 2×SSC, slides were dried and scanned with an Axon GenePix 4000 scanner (##SUPPL##0##Figure S1##).</p>", "<title>D. <italic>In situ</italic> Expression of Proteins</title>", "<p>\n<italic>In situ</italic> expression was performed using a cell-free expression system (TNT Quick coupled transcription/translation system (Promega, Madison, WI)). In brief, 30 µl of rabbit reticulocyte lysate, supplemented with methionine, was added directly to the slide and incubated in a water bath. Expression and immobilization were carried out at 30°C for 1.5 hours followed by a 2 hour incubation at 15°C.</p>", "<title>E. Confirmation of Expression and Immobilization of Expressed Proteins</title>", "<p>Expression of GFP-Tus protein was confirmed with a Cy3–Cy5 labeled antibody to the poly-histidine (poly-his) tag. Prior to incubation with labeled antibody, slides were blocked for 1 hr with 0.1% PVP/0.05% Tween 20. Monoclonal antibodies to poly-his, GFP (Sigma-Aldrich, St Louis, MO), and beta-globin (Novus, Littleton, CO) were labeled with fluorescent dye N-hydroxysuccinimide (NHS) ester-linked Cyanine 3 (Cy3) and Cyanine 5 (Cy5) (Amersham, Piscataway, NJ). In brief, 90 µl of antibody diluted to the concentration of 0.55 mg/ml in 0.1 M sodium bicarbonate/carbonate buffer pH 9.0 was mixed with 20 µl of 60 µM of Cy3 or Cy5 in sodium bicarbonate/carbonate buffer and incubated on ice. After reaction had proceeded for 90 minutes, 8 µl of Blocking Buffer (BD Biosciences, San Jose, CA) was added to the solution to quench the reactions and the solutions were allowed to sit for another 30 min with additional mixing approximately every 10 min. Unbound dye was removed by passing each sample through a size-exclusion chromatography spin column (sephadex G-15 (Sigma Aldrich, St. Louis, MO). Molar concentration for labeled protein and dye were calculated. The Cy5- labeled anti-his was mixed with equal amount of the Cy3-labeled anti-his and diluted in microarray hybridization buffer (0.1% PVP/0.1% Tween 20). Hybridization to the array was performed in an incubation chamber at 4°C with gentle rocking for at least 12 h. After incubation, slides were washed 3× for 5 minutes each in 10× PBS/0.05% Tween 20, followed by one wash in 10× PBS for one minute. All washes were performed at 4°C. Slides were dried and scanned on an Axon GenePix 4000 scanner (Union City, CA), and fluorescence data were collected and evaluated with the GenePix Pro 5.0 software. For the microarray imaging, the Axon GenePix 4000 scanner was set at 100% laser power and 350 PMT gain.</p>", "<title>System Design</title>", "<p>a) Plasmid DNA encoding protein of interest (POI) was constructed such that the protein of interest (POI) is fused with an <italic>E. coli</italic> protein called Tus. b) The plasmid also contains one or more Tus binding sites termed Ter. Tus protein binds the Ter DNA sequence as a monomer with very high affinity, ∼3–7×10 ⁻¹³ M ##REF##9334221##[11]##. c) Plasmids are arrayed by a commercially available microarray printing robot. d) The POI-Tus fusion protein is synthesized on the microarray by coupled cell-free protein synthesis (either mammalian or prokaryotic). If the cocktail is derived from <italic>E. coli</italic>, it is made from a Tus minus strain. e) Affinity of the expressed POI-Tus fusion protein for the Ter sequence is significantly greater than the antigen-antibody affinity described by either Nord et al.##UREF##1##[9]## or Ramachandran et al ##REF##15232106##[10]##.</p>" ]

[ "<title>Results</title>", "<p>The design of the expression construct (pDest-Microarray TT-1) and basic concept of the array platform are shown in ##FIG##0##Figure 1## and ##SUPPL##0##Figure S1##. The basic expression vector is based on the Gateway (Invitrogen, Carlsbad, CA) recombinational cloning system making it easier to generate libraries of constructs. To validate the Tus-Ter DNA-binding protein system for the development of an <italic>in situ</italic> self assembling protein array, as well as demonstrate the specificity of Tus-Ter binding, we printed a microarray containing a set of clones in pDest-Microarray TT-1 encoding a GFP-Tus-His6 fusion-protein and a Ter site, an identical vector with a point mutation in the Ter site, and a third construct without a Ter site.</p>", "<p>Anti-his antibody was labeled with Cy3 and Cy5, mixed in equimolar amounts, and hybridized to this microarray. The results are shown in ##FIG##1##Figure 2##. We observed significant signal intensities (arbitrary units of 14000 and 20574 from Cy5 and Cy3 respectively) corresponding to the vector containing a wild-type Ter site (##FIG##1##Fig. 2##), confirming the expression and binding of the GFP-Tus fusion protein. In contrast for the vector without any Ter site, no significant signal was observed (arbitrary units of 0 and 2463 from Cy5 and Cy3 respectively; ##FIG##1##Fig. 2##). The no-Ter/wild-type Ter signal ratio (−TER/+TER) is 0 for Cy5 and 0.12 for Cy3, consistent with significantly higher binding of the fusion protein to plasmid containing wild-type Ter as compared to plasmid without any Ter. Similarly, the vector containing a point mutation in the Ter site showed low to moderate signal (arbitrary units of 4738 and 10920 from Cy5 and Cy3 respectively; ##FIG##1##Fig. 2##). The mutated-Ter/wild-type Ter signal ratio (mTER/+TER) is 0.34 for Cy5 and 0.53 for Cy3, indicating that a mutation in the Ter site results in reduced binding efficiency to the fusion protein. These data are in complete agreement with previous reports that Tus binds to the same mutated Ter with 4–6 fold lower efficiency \n##UREF##2##[12. 13]## as well as our own calculations of Tus: Ter and Tus: mutant Ter binding efficiencies and off rates (##SUPPL##2##Table S1##). These data demonstrate that an intact Ter sequence is necessary and sufficient for optimal binding to the Tus protein, allowing the effective binding of Tus-fusion proteins to Ter site(s) present in the vector and supporting the application of the Tus-Ter system for protein microarray fabrication.</p>", "<p>In a parallel set of experiments, we have extended these findings to another DNA-binding protein system (lacI/LacO) and demonstrate that they can also function to mediate protein microarray fabrication in a similar manner (DC, CB, KS, JH, and DM; data not shown).</p>", "<p>As a more direct test of this platform, expression plasmids encoding 14 different proteins were immobilized onto the surface of a microarray. Expression from each of the constructs was confirmed by probing with a labeled antibody directed against the His-tag engineered into each construct (##FIG##2##Fig. 3A##). As expected, although an equal amount of DNA was printed for each feature, the relative amount of protein produced and retained by each construct varied modestly, presumably due to characteristic differential transcription/translation efficiencies (##FIG##2##Fig. 3A## and ##SUPPL##1##Figure S2##). To validate that the individual targeted proteins were indeed expressed and captured at their designated location on the microarray, replicate arrays were probed with antibodies directed against the unique fusion partners specific for each construct (##FIG##2##Fig. 3B and 3C##). As shown, each of the target proteins was expressed and captured at a specific and designated location that was pre-determined by the insert encoded in the expression construct printed.</p>" ]

[ "<title>Discussion</title>", "<p>We have developed a simple and cost effective strategy for the rapid generation of protein microarrays. Because only DNA expression constructs are printed, the inherent cost, stability, and technical limitations most commonly associated with other protein microarray strategies are eliminated. As, shown in ##FIG##0##Figure 1## and ##SUPPL##0##Figure S1##, the microarray is fabricated by the printing of DNA expression constructs that function to not only direct the synthesis of the desired protein, but also as the ‘capture reagent’ for immobilization of the encoded protein onto the microarray surface. The ‘capture reagent’ function of the printed DNA expression constructs is mediated by the specific and high-affinity binding (∼3–7×10 ⁻¹³ M) of <italic>E. coli</italic> Tus protein to Ter, a 20 bp DNA sequence. In ##SUPPL##2##Table S1## we show that both the specificity and the high-affinity binding, characteristic of wild-type Tus protein for Ter, is also true for the cloned versions of Tus:Ter interaction is maintained in a coupled in vitro transcription/translation system within a microarray environment. Finally, in ##FIG##2##Figure 3## we demonstrate that not only are the designated proteins specifically expressed, but that the individual expression constructs encoding each individual protein exclusively captures it's encoded protein without detectable diffusion or ‘bleeding’ to adjacent features on the microarray.</p>", "<p>These data support the utility and effectiveness of this platform as a method for the high-throughput production of low-cost protein microarrays for the study of protein-protein, protein-nucleic acid, or protein-small molecule interaction. The open format of these arrays, together with their long ‘shelf-life’ and simple low-cost printing scheme, make this a cost-effective, versatile, production friendly platform amendable to a wide variety of uses and applications.</p>" ]

[]

[ "<p>Conceived and designed the experiments: DKC. Performed the experiments: KS CB TH. Analyzed the data: DKC JH TH DJM. Contributed reagents/materials/analysis tools: DKC TH DJM. Wrote the paper: DKC DJM. Designed the experiments: JH TH DJM.</p>", "<p>We describe a novel, simple and low-cost protein microarray strategy wherein the microarrays are generated by printing expression ready plasmid DNAs onto slides that can be converted into protein arrays on-demand. The printed expression plasmids serve dual purposes as they not only direct the synthesis of the protein of interest; they also serve to capture the newly synthesized proteins through a high affinity DNA-protein interaction. To accomplish this we have exploited the high-affinity binding (∼3–7×10 ⁻¹³ M) of <italic>E. coli</italic> Tus protein to Ter, a 20 bp DNA sequence involved in the regulation of <italic>E. coli</italic> DNA replication. In our system, each protein of interest is synthesized as a Tus fusion protein and each expression construct directing the protein synthesis contains embedded Ter DNA sequence. The embedded Ter sequence functions as a capture reagent for the newly synthesized Tus fusion protein. This “all DNA” microarray can be converted to a protein microarray on-demand without need for any additional capture reagent..</p>" ]

[ "<title>Supporting Information</title>" ]

[]

[ "<fig id=\"pone-0003265-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003265.g001</object-id><label>Figure 1</label><caption><title>Concept of protein microarray on demand.</title></caption></fig>", "<fig id=\"pone-0003265-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003265.g002</object-id><label>Figure 2</label><caption><title>Exploiting the specificity of Tus:Ter interaction.</title><p>Plasmid vectors encoding a green florescent protein (GFP)- TUS - poly-histidine fusion protein and a Ter sequence containing a point mutation (pMUT), a wild-type Ter sequence (pNOMut), and no Ter sequence were immobilized on the surface of a microarray, incubated in a cell-free rabbit reticulocyte transcription/translation extract, and hybridized with Cy-labeled anti-histidine antibody.</p></caption></fig>", "<fig id=\"pone-0003265-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003265.g003</object-id><label>Figure 3</label><caption><title>Microarray feature expression and capture specificity.</title><p>pNOMut variants encoding different proteins fused to TUS - poly-histidine were immobilized on the surface of a microarray and incubated in a cell-free rabbit reticulocyte transcription/translation extract. Duplicate microarrays were then hybridized with cy-labeled monoclonal antibodies specific for A) His, B) GFP, and C) human <italic>B</italic>-globin.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pone.0003265.s001\"><label>Figure S1</label><caption><p>Design of expression construct and basic microarray fabrication schema</p><p>(0.67 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003265.s002\"><label>Figure S2</label><caption><p>Validation of microarray printing. Different proteins fused to TUS - poly-histidine were immobilized on the surface of a microarray and stained for DNA content as described in <xref ref-type=\"sec\" rid=\"s2\">Materials and Methods</xref>.</p><p>(9.91 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003265.s003\"><label>Table S1</label><caption><p>(0.03 MB DOC)</p></caption></supplementary-material>" ]

[ "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under contract N01-CO-12400. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003265.g001\"/>", "<graphic xlink:href=\"pone.0003265.g002\"/>", "<graphic xlink:href=\"pone.0003265.g003\"/>" ]

[ "<media xlink:href=\"pone.0003265.s001.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003265.s002.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003265.s003.doc\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Depression in older populations can differ from depression in younger adult individuals and is often associated with chronic medical illness, disability, and cognitive impairment. It is also associated with aging-related brain changes including the presence of hyperintense lesions, bright areas occurring in the brain parenchyma as seen on T2-weighted magnetic resonance imaging (MRI). These changes are primarily ischemic in origin ##REF##12215077##[1]##, and although are observed in normal aging, are often more severe in older depressed individuals ##REF##15927454##[2]##.</p>", "<p>The relationship between these hyperintense lesions and antidepressant treatment outcomes is unclear. Several studies have concluded that greater hyperintense lesion severity is associated with poorer response to antidepressants ##REF##16297603##[3]##–##REF##14971633##[7]##. Greater hyperintensity severity is additionally associated with significantly more adverse drug reactions ##REF##8898812##[8]##, which may result in early drug discontinuation or inability to increase doses to therapeutic levels. In contrast, other studies have not found a relationship between cross-sectional lesion severity and acute antidepressant outcomes ##REF##9209723##[9]##–##REF##17586780##[16]##, although relationships may exist between longitudinal change in lesion severity and longer-term course of depression ##REF##14609884##[11]##.</p>", "<p>More recent work investigating late-life depression has utilized diffusion tensor imaging (DTI). DTI can quantify water diffusion, which in living tissue is constrained by neuronal integrity and modulated by myelin ##REF##12489094##[17]##. DTI measures include the apparent diffusion coefficient (ADC), a general measure of diffusion which may serve as a surrogate marker for fiber density ##REF##14643084##[18]##, and anisotropy, which measures the direction of water diffusion and has been proposed to be a surrogate marker for white matter orientation and organization. Similar to what is observed in old stroke regions, white matter hyperintense lesions increase ADC and decrease anisotropy ##REF##11513816##[19]##, and greater hyperintense lesion severity is associated with more widespread alterations in DTI measures even in normal appearing white matter ##REF##17515396##[20]##. Although several studies have reported that depressed elders exhibit reduced frontal and temporal anisotropy ##REF##16876144##[21]##–##REF##17346365##[24]##, this technique has not been used as extensively to study treatment response. One group has examined the relationship between DTI measures and antidepressant response in geriatric depression, associating failure to achieve remission with reduced fractional anisotropy in multiple regions, including the cingulate gyrus and dorsolateral prefrontal cortex ##REF##12411231##[25]##, ##REF##18172016##[26]##.</p>", "<p>The purpose of this study was to use DTI to examine if measures of frontal white matter microstructure were associated with acute 12-week response to sertraline. We hypothesized that lower fractional anisotropy (FA) and higher diffusivity (measured as ADC), a pattern of findings associated with both chronic stroke and hyperintense lesions ##REF##11513816##[19]##, ##REF##17515396##[20]##, would be associated with failure to achieve remission. In turn, we hypothesized that individuals who remitted would exhibit higher frontal FA measures.</p>" ]

[ "<title>Methods</title>", "<p>The protocol for this trial and supporting CONSORT checklist are available as supporting information; see ##SUPPL##0##Checklist S1## and ##SUPPL##1##Protocol S1##.</p>", "<title>Sample</title>", "<p>Subjects were recruited from advertisements and outpatient clinical referrals at Duke University Medical Center. To be eligible, subjects had to be age 60 years or older, meet DSM-IV criteria for Major Depressive Disorder without psychosis as assessed through clinical evaluation and the Structured Clinical Interview for DSM-IV (SCID) ##REF##1637252##[27]##, and exhibit a baseline Montgomery-Asberg Depression Rating Scale ##REF##444788##[28]## score of 18 or greater. Exclusion criteria included: 1) any MRI contraindications; 2) comorbid Axis I diagnosis, including current substance abuse or bipolar disorder; 3) active suicidality; 4) current psychotic symptoms, 5) severe or unstable medical conditions; 6) primary neurological disorders including dementia or stroke; 7) Mini-Mental State Examination (MMSE) ##REF##1202204##[29]## score&lt;24; 8) current depressive episode that failed to respond to adequate trials of two prior antidepressants administered for at least 6 weeks at therapeutic doses, or had failed an adequate trial of sertraline; 9) current psychotherapy.</p>", "<p>All study procedures were explained to each participant, and those who provided written informed consent were enrolled. The study was approved by the Duke University Health System Institutional Review Board.</p>", "<title>Study Design and Clinical Measures</title>", "<p>Baseline data acquisition included assessing demographic data, depression severity using the MADRS, and severity of medical burden using the Cumulative Illness Rating Scale (CIRS) modified for geriatric populations ##REF##1594710##[30]##. Age of depression onset was ascertained through the SCID and review of medical records.</p>", "<p>After screening, subjects taking antidepressant medications at doses approved to treat depression underwent a washout period of up to two weeks, while those taking lower doses had the antidepressant stopped without a washout. Eleven subjects had antidepressants discontinued after enrollment, while two had St. John's Wort discontinued. The study protocol allowed limited use of zolpidem or zaleplon for sleep, or lorazepam for anxiety. Subjects on other concomitant medications for medical indications, sleep, or anxiety could continue them provided doses remained stable. Eleven subjects required hypnotics for sleep, seven required benzodiazepines for anxiety, and seven subjects were on anticonvulsants or benzodiazepines for medical indications such as neuropathy or restless legs syndrome.</p>", "<p>After completing any needed washout, subjects received sertraline for twelve weeks. They were generally started at 25 mg for a few days to rule out drug sensitivity, and then increased to 50 mg daily. Subjects were assessed at 2, 4, 6, 8, and 12 weeks. A dose increase of 50 mg was allowed at each visit, up to a possible maximum dose of 200 mg daily, and the decision to titrate the dose was guided by the Clinical Global Impression – Severity scale (CGI-S) ##UREF##1##[31]##. Subjects with a CGI-S of 3 or greater had a dose increase unless there was a concern for tolerability. Subjects with a CGI-S of 2 could be titrated at investigator discretion, and a CGI-S of 1 was considered to be in remission. If a dose increase resulted in intolerable side effects, the dose could be reduced to the previously tolerated level. Depression severity was assessed at each visit using the MADRS.</p>", "<title>MRI Acquisition and Processing</title>", "<p>Subjects were imaged with a 1.5 T whole-body, research-dedicated MRI (Signa, GE Healthcare) using a standard head (volumetric) radiofrequency coil. The scanner alignment light was used to adjust head tilt and rotation so the axial plane light passed across the canthomeatal line and the sagittal lines were aligned with the center of the nose. A rapid sagittal localizer scan was acquired to confirm alignment. MRI was obtained within the first two weeks of study participation, typically at the first follow-up visit after initiating study drug.</p>", "<p>The diffusion tensor images were acquired using a single shot 2D diffusion tensor echo planar pulse sequence in the axial plane with a 32 cm field-of-view, 3 mm slice thickness (no gaps between slices), relaxation time (Tr) = 6000, excitation time (Te) = 100, 128 (phase)×128 (frequency), full imaging bandwidth = 180 KHz, 4 signal averages, 6 diffusion directions, each with a b-value of 1000 sec/mm² plus an acquisition with b = 0 using the Basser scheme ##REF##9621916##[32]##.</p>", "<p>Diffusion tensor images were processed using custom programs run on MATLAB software (version 7, The MathWorks, Natick, MA) that calculated the diffusion tensor eigenvalues in each voxel ##REF##8661285##[33]##. FA and ADC images were calculated, and regions-of-interest (ROIs) were placed by a single analyst (WDT) using Analyze software (version 6.5, Mayo Clinic). Scans were coded with no identifiable information, and the analyst was blinded to subject identity and treatment outcome.</p>", "<p>Oval ROIs of identical size (45.7 mm²) were used to measure FA and ADC values. The same ROI was used for both measures, and all ROIs were placed on axial slices. Hyperintense lesions were avoided for all ROI placements. This decision was made based on the observation that hyperintensities can have strong effects on DTI measures that is not representative of the DTI measures of surrounding white matter ##REF##11513816##[19]##. There were no cases where the presence of hyperintense lesions prevented appropriate ROI placement according to ROI placement procedures.</p>", "<p>ROI placement procedures have been previously described ##REF##17515396##[20]##, ##REF##16876144##[21]##. The superior (SFG) and middle frontal gyri (MFG) ROIs were placed halfway between the precentral sulcus and anterior boundary of the brain, on the most inferior slice where both gyri were visible as separate structures. The anterior cingulate cortex (ACC) ROIs were placed in the white matter lateral to the cingulated gyrus, on the most inferior slice where the anterior horns of the lateral ventricle were still visible. For the corpus callosum, two ROIs were placed to either side of midline, on the slice ventral to the slice where it was divided by the longitudinal fissure. Results from these two separate corpus callosum ROIs were averaged together to create a composite measure. We used the corpus callosum as a control region that we did not expect to be associated with antidepressant outcomes as it is a region of high anisotropy that was not different between depressed and nondepressed elders in previous studies ##REF##16876144##[21]##.</p>", "<p>Reliability was established by repeated measurements on multiple DTI scans. Intraclass correlation coefficients (ICCs) of FA measures were: left ACC, 0.910; right ACC, 0.992; left SFG, 0.991; right SFG, 0.985; left MFG, 0.975; right MFG, 0.983; left corpus callosum, 0.971; right corpus callosum, 0.934. ICCs for ADC measures (using the same ROIs as the FA measures) were: left ACC, 0.945; right ACC, 0.957; left SFG, 0.978; right SFG, 0.987; left MFG, 0.931; right MFG, 0.893; left corpus callosum, 0.751; right corpus callosum, 0.799.</p>", "<title>Statistical Analysis</title>", "<p>SAS software (version 9.1, SAS Institute, Cary, NC) was used for all statistical analyses. Remission was defined as achieving a MADRS of 10 or less at any assessment, as this cutoff has been shown to provide a valid definition ##REF##12200208##[34]##. Bivariate analyses of demographic and neuroimaging data were conducted between remitted and nonremitted groups, using the chi-square test for categorical variables and two-tailed t-test for continuous variables, or the Satterthwaite t-test if variances were not equal. Normality of neuroimaging data was assessed both graphically and by Wilks test.</p>", "<p>Primary analyses used the GENMOD procedure in SAS to develop generalized estimating equation (GEE) models ##UREF##2##[35]## using an exchangeable correlation structure which examined the relationship between the repeated outcome measure of failure to remit at each assessment point and the independent variables. For missing data, the GEE models assume missing completely at random (MCAR). The dependent measure was the repeated measure dichotomous variable of remitted or not remitted, and we modeled the probability of failing to remit. Independent variables included the DTI measure, baseline depression severity measured with the MADRS, age, sex, medical comorbidity measured using the CIRS, and a variable accounting for which assessment period was being evaluated (baseline, 2-, 4-, 6-, 8- or 12-week). These covariates were selected for the model as they were either significantly different between the remitting and non-remitting groups (age, baseline MADRS) or have been previously associated with antidepressant outcomes in late-life depression (medical comorbidity, ##REF##12028167##[36]##; sex, ##REF##15948303##[37]##). In an exploratory step, we also tested for a DTI measure – age interaction and a DTI measure – baseline MADRS interaction effect on remission. Multicolinearity between covariates was assessed in models by examining variance inflation factors (VIF); all VIFs were less than 30. Separate models were created to examine each DTI measure. We did not control for multiple comparisons.</p>", "<p>In secondary analyses, we sought to examine if DTI measures were associated with time to remission, an approach used by others ##REF##12411231##[25]##. This survival model used the PHREG procedures in SAS ##UREF##3##[38]##, examining the time to first remission or last assessment, at which point subjects were censored. Independent variables included DTI measures, age, baseline MADRS score, sex, and CIRS score.</p>" ]

[ "<title>Results</title>", "<p>101 depressed subjects signed consent and enrolled in the study between January 2002 and March 2006; 74 of those subjects are included in this analysis (##FIG##0##Figure 1##). 35 subjects were female and 39 were male, with a mean age of 68.1 years (range 60–95 years, standard deviation = SD = 7.3 years). The group's mean initial MADRS score was 25.4 (SD = 4.2, range 18–37); at study completion, the mean final MADRS score was 11.5 (SD = 7.9, range 0–42), with a mean sertraline daily dose of 102.0 mg (SD = 37.6 mg, range 25 mg–200 mg).</p>", "<p>Over the course of the twelve-week study, 37 subjects achieved a MADRS score of 10 or less and achieved remission, while 37 subjects did not. 67 subjects completed all 12 weeks of the study. Four subjects completed only 4 weeks, 1 completed only 6 weeks, and 2 completed only 8 weeks. All of the subjects who withdrew early were classified as not achieving remission, except one of the subjects who completed only 8 weeks who did remit. Remitted subjects were younger and less severely depressed at baseline (##TAB##0##Table 1##). We next tested for bivariate differences in unadjusted FA and ADC measures between subjects who did and did not remit (##TAB##1##Table 2##). In these analyses, the only measure significantly different between remitted and nonremitted groups was the FA value of the right anterior cingulate cortex.</p>", "<p>\n##TAB##2##Table 3## details the results of models examining the relationship between failure to remit and regional FA, while also controlling for baseline depression severity, age, sex, and medical comorbidity as measured with the CIRS. In these models, higher anisotropy of the white matter of the superior frontal gyri bilaterally and anterior cingulate cortices bilaterally was significantly associated with failure to remit. In these models where FA was significantly associated with lack of remission, higher baseline MADRS (<italic>p</italic>&lt;0.01 all models) and greater age (<italic>p</italic>&lt;0.05 all models) were also associated with nonremission, while neither gender nor CIRS score reached a level of statistical significance in any model. Similar models were created examining the relationship between ADC values and nonremission; however ADC was not associated with nonremission in any model (data not shown). When DTI-age and DTI-MADRS interactions were included, these terms did not reach statistical significance in any model.</p>", "<p>Finally, we used an approach similar to that found in a previous report ##REF##12411231##[25]## to determine if these DTI measures were associated with a time to first remission. The 37 subjects who remitted has a mean time to remission of 6.8 weeks (SD = 2.8 weeks, median = 6 weeks). In models controlling for baseline MADRS score, age, sex, and CIRS score, no DTI measure was significantly associated with time to remission (data not shown).</p>" ]

[ "<title>Discussion</title>", "<p>Contrary to our hypothesis, we found that higher FA measures, not lower, were associated with a failure to achieve remission. These measures were of the anterior cingulate and superior frontal gyrus, regions previously identified as exhibiting depression-related differences on DTI ##REF##16876144##[21]## and associated with antidepressant response in functional imaging studies ##REF##15110034##[39]##. These relationships were statistically significant after controlling for age, sex, baseline depression severity, and medical comorbidity. We found no association between ADC and remission.</p>", "<p>These results are discrepant with the two other published studies examining the relationship between antidepressant outcomes and white matter anisotropy ##REF##12411231##[25]##, ##REF##18172016##[26]##. These studies, conducted by the same group, found that individuals who failed to achieve remission exhibited lower anisotropy in multiple regions, including the anterior and posterior cingulate cortex and the dorsolateral prefrontal cortex. In contrast, we found lower likelihood of remission to be associated with increased frontal anisotropy. Despite examining smaller samples (the larger study included 48 evaluable subjects, while we examined 78), those studies had a similar 12-week study design using serotonin reuptake inhibitors and comparable demographic characteristics. One methodological difference is the use of a voxel-based analysis of the anisotropy data, while this study used a region-of-interest approach. The difference in conclusions with these studies may be related to the differences in sample size or methodology, but may also reflect heterogeneity in the pathophysiology of depression in older subjects.</p>", "<p>This discrepancy across studies raises questions about what biological factors most strongly contribute to DTI measures. ADC is an overall measure of water diffusion, and given the size of imaging voxels relative to the microstructural environment, includes intracellular and extracellular spaces ##REF##8108392##[40]##. The relationship between brain microstructure and FA values are likely multifactorial. Current hypotheses of what factors contribute to anisotropy posit that myelination and axonal sheaths may restrict diffusion perpendicular to the direction of the fiber, while integrity of axonal structures such as microtubules may be associated with diffusion parallel to the fiber ##REF##11352623##[41]##–##REF##15862213##[43]##, both of which contribute to the overall FA value. Additionally, there appears to be significant heterogeneity in white matter fiber tract location and shape ##REF##16236527##[44]##.</p>", "<p>Another methodological difference between the studies is that our analysis method avoided hyperintense lesions. Hyperintense lesions are associated with anisotropy changes within their boundaries ##REF##11513816##[19]##, but also are associated with widespread changes in normal appearing white matter ##REF##17515396##[20]##. Thus the differences in anisotropy observed in previous studies ##REF##18172016##[26]## may be related to clusters of hyperintensities, which would result in focal reductions in anisotropy. The same cannot be said of our current study, although the more widespread relationship between hyperintensity volume and DTI measures of normal appearing white matter ##REF##17515396##[20]## would be expected to be observed here.</p>", "<p>However, hyperintensities did not likely contribute to our current findings. As greater hyperintensity severity is associated with lower FA ##REF##11513816##[19]##, ##REF##17515396##[20]##, this would imply that greater hyperintensity severity is associated with remission. This conclusion does not appear likely as those published studies which did find a significant relationship between hyperintensity volume and treatment outcomes have all concluded that greater hyperintensity severity is associated with lower likelihood of remission ##REF##16297603##[3]##–##REF##14971633##[7]##.</p>", "<p>Other neuropathological differences may contribute to our findings. Reductions in FA may be related to multiple factors at the cellular level, including reduced axonal integrity or thinning myelin, although myelin may play only a modulatory role in anisotropy measures ##REF##12489094##[17]##, ##REF##16950152##[45]##. Depression has been associated with deficits in neuronal size and density in frontal regions ##REF##11386983##[46]##, ##REF##15953590##[47]##, which would hypothetically be related to axonal integrity. This implies that older individuals who respond more favorably to antidepressants may have neuronal loss and impaired axonal integrity in frontal regions, which could result in reduced anisotropy. This leads to the hypothesis that those who experience this neuronal loss and altered neural connectivity may be more responsive to antidepressant medication, either by the effect antidepressants have on neurotransmitters or through antidepresants' neurotrophic effects. In contrast, individuals with no structural connectivity deficits may be more likely to exhibit a poor response. Functional MRI studies have demonstrated that depression is characterized by impaired functional connectivity between cortical and limbic regions, while antidepressant response is accompanied by improved functional connectivity ##REF##15856081##[48]##, ##REF##11063978##[49]##. In some older individuals, this deficit in functional connectivity may be secondary to structural changes, and these individuals may respond to antidepressants better than individuals with functional connectivity deficits with intact structural connections. This is only a hypothesis, which can best be examined in the future by combining DTI and functional neuroimaging with an antidepressant trial.</p>", "<p>One study limitation is that we could have added further covariates to the model as antidepressant remission is associated with additional factors such as disability, depression history or length of current depressive episode, and subjective social support. However, we were also limited by the sample size in the number of covariates we could include. An additional statistical concern was that we did not control for multiple comparisons, which increases the risk of a Type I error. Had we adjusted for multiple comparisons, the observed differences would not have reached statistical significance. Finally, neuroimaging limitations include that neuroimaging was not obtained until after sertraline had been started. This approach is less optimal than completing MRI before administering any drug, but one would not expect to see any changes on structural neuroimaging within a two week time frame, although such studies have not been done to confirm this. Additionally, other image analysis techniques such as voxel-based morphometry could have been used. Although this method has the advantage of examining the entire brain, it may miss critical regions of interest and does not necessarily replace a region-of-interest approach ##REF##15721994##[50]##.</p>", "<p>Our finding that failure to remit to antidepressant treatment is associated with higher anisotropy in the anterior cingulate cortex and superior frontal gyrus is contrary to our hypothesis and to previously published studies. However, the finding appears robust despite the study's limitations. The discrepancy across studies may reflect limitations of DTI, but may also reflect population heterogeneity – either in the depressed geriatric population, or heterogeneity in white matter structure more broadly. As such, it requires replication and emphasizes the need for larger-scale studies examining neuroanatomic correlates of antidepressant response which are linked to functional imaging studies examining functional connectivity between frontal and limbic brain regions.</p>" ]

[]

[ "<p>Conceived and designed the experiments: WDT MEP JRM YIS KRK PMD. Performed the experiments: PMD. Analyzed the data: WDT MK. Contributed reagents/materials/analysis tools: MEP JRM. Wrote the paper: WDT.</p>", "<title>Introduction</title>", "<p>Neuroanatomic features associated with antidepressant treatment outcomes in older depressed individuals are not well established. This study used diffusion tensor imaging to examine frontal white matter structure in depressed subjects undergoing a 12-week trial of sertraline. We hypothesized that remission would be associated with higher frontal anisotropy measures, and failure to remit with lower anisotropy.</p>", "<title>Methods</title>", "<p>74 subjects with Major Depressive Disorder and age 60 years or older were enrolled in a twelve-week open-label trial of sertraline and completed clinical assessments and 1.5T magnetic resonance brain imaging. The apparent diffusion coefficient (ADC) and fractional anisotropy (FA) were measured in regions of interest placed in the white matter of the dorsolateral prefrontal cortex, anterior cingulate cortex, and corpus callosum. Differences in ADC and FA values between subjects who did and did not remit to treatment over the study period were assessed using generalized estimating equations, controlling for age, sex, medical comorbidity and baseline depression severity.</p>", "<title>Results</title>", "<p>Subjects who did not remit to sertraline exhibited higher FA values in the superior frontal gyri and anterior cingulate cortices bilaterally. There were no statistically significant associations between ADC measures and remission.</p>", "<title>Conclusions</title>", "<p>Failure to remit to sertraline is associated with higher frontal FA values. Functional imaging studies demonstrate that depression is characterized by functional disconnection between frontal and limbic regions. Those individuals where this disconnection is related to structural changes as detected by DTI may be more likely to respond to antidepressants.</p>", "<title>Trial Registration</title>", "<p>ClinicalTrials.gov <ext-link ext-link-type=\"uri\" xlink:href=\"http://clinicaltrials.gov/ct2/show/NCT00339066\">NCT00339066</ext-link>\n</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>We would like to acknowledge Brian D. Boyd B.S. and Cynthia R. Key for assistance in image preparation.</p>" ]

[ "<fig id=\"pone-0003267-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003267.g001</object-id><label>Figure 1</label><caption><title>Subject flow.</title><p>Details of subject enrollment and why excluded from the present study.</p></caption></fig>" ]

[ "<table-wrap id=\"pone-0003267-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003267.t001</object-id><label>Table 1</label><caption><title>Demographic characteristics between subjects who did and did not remit.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Remitted (N = 37)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Nonremitted (N = 37)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Test Result</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>df</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>p</italic> value</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Age</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65.8 (5.7)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">70.5 (8.0)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">t = 2.94</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64.9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0045</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sex (% Female)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">43.2% (16/37)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">48.6% (18/37)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">χ² = 0.49</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.4849</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Race (% Caucasian)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">94.6% (35/37)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">94.6% (35/37)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.0000</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Age of Onset</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">55.0 (15.8)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">55.0 (16.4)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">t = 0.00</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.0000</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">CIRS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.5 (2.8)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.4 (3.6)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">t = 1.26</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.2133</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">MMSE</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28.3 (1.7)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">27.6 (1.7)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">t = 1.74</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0857</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">MADRS, baseline</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">24.4 (3.8)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">26.4 (4.5)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">t = 2.05</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0438</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">MADRS, final</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.4 (2.9)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">17.6 (6.4)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">t = 10.51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">50.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">&lt;0.0001</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Final Sertraline Dose</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97.8 (36.5)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">106.1 (38.8)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">t = 0.93</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.3578</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003267-t002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003267.t002</object-id><label>Table 2</label><caption><title>Unadjusted differences in DTI measures between subjects who did and did not remit.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Remitted (N = 37)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Nonremitted (N = 37)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">t value</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>df</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>p</italic> value</td></tr></thead><tbody><tr><td colspan=\"6\" align=\"left\" rowspan=\"1\">FA values</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• SFG, left</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.408 (0.080)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.421 (0.082)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.80</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.4284</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• SFG, right</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.392 (0.069)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.402 (0.080)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.53</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.6007</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• MFG, left</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.283 (0.064)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.262 (0.046)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.54</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.1284</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• MFG, right</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.268 (0.052)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.264 (0.049)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.7351</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• ACC, left</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.365 (0.063)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.374 (0.076)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.56</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.5802</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• ACC, right</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.332 (0.055)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.368 (0.072)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.44</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0170</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• CC, genu</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.423 (0.070)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.454 (0.101)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">63.9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.1343</td></tr><tr><td colspan=\"6\" align=\"left\" rowspan=\"1\">ADC values</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• SFG, left</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">76.1 (5.8)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">77.4 (11.2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.63</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">53.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.5308</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• SFG, right</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">76.2 (5.2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">77.2 (11.4)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.48</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">50.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.6344</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• MFG, left</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">77.5 (7.3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">79.0 (6.1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.91</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.3643</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• MFG, right</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">79.6 (6.2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">80.1 (5.8)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.30</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.7636</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• ACC, left</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">83.2 (5.8)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">83.4 (5.0)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.21</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.8348</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• ACC, right</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">82.6 (4.9)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">84.0 (6.9)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.97</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.3342</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">• CC, genu</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">104.6 (14.4)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">101.7 (15.4)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.82</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.4122</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003267-t003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003267.t003</object-id><label>Table 3</label><caption><title>Relationship between FA values and remission in multivariate models.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Regional FA value</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Estimate</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SE</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Test Result (Z-score)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>p</italic> value</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">SFG, left</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0099</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0036</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0064</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">SFG, right</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0068</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0030</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.25</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0244</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">MFG, left</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0007</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0032</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.8405</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">MFG, right</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0050</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0043</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.16</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.2476</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">ACC, left</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0079</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0036</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.18</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0290</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">ACC, right</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0088</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0036</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.42</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0156</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">CC, genu</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0035</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0028</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.26</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.2080</td></tr></tbody></table></alternatives></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pone.0003267.s001\"><label>Checklist S1</label><caption><p>CONSORT Checklist</p><p>(0.06 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003267.s002\"><label>Protocol S1</label><caption><p>Trial Protocol</p><p>(0.21 MB PDF)</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"nt101\"><p>\n<italic>Age and age of onset presented in years. Final of sertraline is in milligrams. Continuous variables presented as mean (standard deviation). Two-tailed pooled t-tests used for continuous variables (Satterthwaite t-test for unequal variances), chi-square test for sex, and Fisher's exact test for race. CIRS = Cumulative Illness Rating Scale; MADRS = Montgomery-Asberg Depression Rating Scale.</italic>\n</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt102\"><p>\n<italic>FA = fractional anisotropy, ADC = apparent diffusion coefficient, SFG = superior frontal gyrus, MFG = middle frontal gyrus, ACC = anterior cingulate cortex, CC = corpus callosum. All results presented as mean (standard deviation). All tests reported using two-tailed pooled t-tests, except when unequal variances were present, in which case a Satterthwaite t-test was used.</italic>\n</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt103\"><p>\n<italic>FA = fractional anisotropy; SFG = superior frontal gyrus; MFG = middle frontal gyrus; ACC = anterior cingulate cortex; CC = corpus callosum. Models examined remission of depression as the dependent variable, with FA, age, sex, baseline depression severity and medical comorbidity as independent variables.</italic>\n</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>Drs. Taylor, Kuchibhatla, Payne and MacFall report no conflicts of interest. Dr. Sheline reports serving on the advisory board for Wyeth Pharmaceuticals and Cyberonics and serving on the Eli-Lilly Speakers Bureau. Dr. Krishnan reports serving as a consultant for Amgen, Bristol-Myer Squibb, CeNeRx, Corcept, GlaxoSmithKline, Johnson &amp; Johnson, Lundbeck, Merck, Organon, Pfizer, Sepracor, and Wyeth. Dr. Doraiswamy has received research grants and honoraria for consulting or speaking from several pharmaceutical companies and antidepressant manufacturers, and stock in Sonexa Therapeutics.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>The study was supported by NIMH grants K23 MH 65939, R01 MH 62158, and R01 MH 078216. Sertraline was provided by Pfizer. These funding sources had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003267.g001\"/>", "<graphic id=\"pone-0003267-t001-1\" xlink:href=\"pone.0003267.t001\"/>", "<graphic id=\"pone-0003267-t002-2\" xlink:href=\"pone.0003267.t002\"/>", "<graphic id=\"pone-0003267-t003-3\" xlink:href=\"pone.0003267.t003\"/>" ]

[ "<media xlink:href=\"pone.0003267.s001.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003267.s002.pdf\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Modern instrumental methods have been generating a significant advancement in biology research. Especially in the field of functional genomics, transcriptomics and proteomics measurements have provided fundamental insight in many biological processes. The missing link between these measurements and the phenotype is called metabolomics ##UREF##0##[1]##. This new field concerns the measurement of small biomolecules in body fluids, cells, tissues, etc. The type of data being generated in metabolomics studies is characterized by a very broad acquisition of semi-quantitative data of a large number of metabolites ##UREF##0##[1]##–##UREF##2##[4]##. This results in data sets of a very complex nature. Not only are these data sets high-dimensional, they also exhibit mixtures of types of variation introduced by the specific experimental setup ##REF##16762068##[5]##.</p>", "<p>Traditionally, a set of measurements is analyzed by postulating a model describing systematic variation and assuming the left-overs (residuals) as being random. Due to the complexity of metabolomics data, this concept breaks down. There are many sources of variation in the data <italic>non-informative</italic> for the underlying biological question. An example of this type of variation are metabolites which are not under tight regulatory control and are thus allowed to vary almost independently across the experiments ##UREF##3##[6]##. Such non-informative variation affects the data in a structured way and infiltrates the systematic or modeled part of the data. This results in poor interpretability and the failure to unearth subtle <italic>informative</italic> variation. In this paper, we propose a new conceptual framework for analyzing metabolomics data based on the idea to separate informative from non-informative variation. The informative variation should describe the systematic biological variation in relevant metabolites induced by underlying biological phenomena. What we are ultimately aiming for is to discover these biological phenomena.</p>", "<p>Our assumption is that the studied biological phenomena are not represented by all measured metabolites, but that simple structures (subsets of related metabolites) in (parts of) the data exist, each simple structure or component describing an underlying biological phenomenon. In the development of our discovery tool we are aiming for a method that fulfills the following requirements: i) being able to identify simple structures, in which just a limited number of metabolites are represented by the structure; ii) representing each simple structure by a model, the type of model depending on the data collected and driven by <italic>a priori</italic> biological knowledge; iii) assuming that a (large) part of the data will most probably not be informative. The last assumption is reasonable given the holistic nature of metabolomics, where the aim is to measure all metabolites present.</p>", "<p>We have called this new approach <italic>simplivariate</italic> models since they are in-between univariate and multivariate models and use simple building blocks (see ##FIG##0##Figure 1##). Univariate models look at one-metabolite-at-a-time; they are easy to interpret but lack an overall view on the data since no correlations between metabolite values are used. On the other extreme are multivariate models; they provide a full view but often lack good interpretation especially in high-dimensional data cases. Simplivariate models try to have the best of both worlds: simplicity, comprehensiveness and correlation.</p>", "<p>Although the simplivariate framework is general and can be used in exploratory analysis, regression analysis and discriminant analysis, in this paper we will focus on explorative methods. Usually in exploratory data analysis for metabolomics data, use is made of either of two types of techniques: projection (dimension reduction) methods or clustering methods. The first type of techniques (with Principal Components Analysis (PCA) as an example) searches for structures consisting of highly co-varying metabolites to construct new representations of the data ##UREF##4##[7]##. Clustering techniques can roughly be divided into two categories: hierarchical clustering (based on linking objects or variables on dissimilarity measures), leading to a set of nested clusterings, and partitioning algorithms, where the result is just one partitioning, and a model is defined to represent the clusters. Both types of techniques do not fulfill the criteria i) to iii) of simplivariate models explained above, e.g., both PCA and hierarchical clustering do not look for components using a limited set of metabolites.</p>", "<p>First, the simplivariate modeling framework will be presented in its full generality. Next, two techniques that fit into that framework will be discussed using real-life metabolomics data. Finally, shortcomings of these methods will be discussed and suggestions of improvement will be given.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Simplivariate models</title>", "<p>A flexible framework is built by defining a <italic>simplivariate</italic> model that describes the partitioning of a data matrix <bold>X</bold> (<italic>I</italic> objects (e.g. experiments)×<italic>J</italic> variables (e.g. metabolites)) in components containing subsets of related variables (e.g. metabolites):\n</p>", "<p>In which every element <italic>x_ij</italic> of matrix <bold>X</bold> can be written as a sum of contributions from different components. These components <italic>ϕ_ijk</italic> describe the <italic>informative</italic> parts of the data and can be very diverse in nature. The variation of <italic>x_ij</italic> that is not included in factors <italic>ϕ_ijk</italic>- <italic>non-informative variation</italic> - is indicated by <italic>e_ij</italic>. Although the symbol <italic>e_ij</italic> is commonly used to indicate random variation, it has a very different meaning here. The non-informative part is certainly non-random in the strict senses of randomness. To introduce the concept of simplicity not all variables are included in the factors <italic>ϕ_ijk</italic>.\n</p>", "<p>Here <italic>δ_jk</italic> indicates the presence of variable <italic>j</italic> in component <italic>k</italic> and <italic>γ_ik</italic> indicates the presence of an object <italic>i</italic> in component <italic>k</italic> (δ_jk = 1 if variable <italic>j</italic> is present in group <italic>k</italic>, 0 otherwise and γ_ik = 1 if object <italic>i</italic> is present in group <italic>k</italic>, 0 otherwise).</p>", "<p>For simplicity we have used the same symbol <italic>ϕ_ijk</italic> in equations (1) and (2), but their difference is clear from those equations.</p>", "<p>When decomposing <bold>X</bold> into simple components, the idea is that interpretation will be easier, since not all original variables are included in those components. Only variables that are closely related will be used. In the case of metabolomics data, metabolites that are functionally related (e.g. part of the same pathway) may form a simple model.</p>", "<title>Simple structures</title>", "<p>The components <italic>ϕ_ijk</italic> can be very diverse in nature, and represent the relations between objects and variables in each of the subsets. Three examples of such component <italic>ϕ_ijk</italic> are:representing simple component <italic>k</italic> by a constant. If this would be an exhaustive partitioning of all variables and objects this would resemble two-mode clustering ##UREF##5##[8]##. Another simple model iswhich is a purely additive model for simple component <italic>k</italic>, that resembles a two-way ANOVA decomposition of a data matrix ##UREF##5##[8]##. The next model to consider iswhich is a purely multiplicative simple component <italic>k</italic>, equivalent to a rank-one component PCA decomposition of a data matrix.</p>", "<p>Combinations of representations Eq. 4 and 5 are also possible resulting in mixed models:The choice for one of these types of models should be based on information on the structure of the data and on <italic>a priori</italic> biological knowledge.</p>", "<p>In equation (2) <italic>δ_jk</italic> and <italic>γ_ik</italic> indicate the presence of element <italic>ϕ_ijk</italic> in factor <italic>k</italic>. For illustrative purposes, for the moment we will assume that all objects are present in every factor <italic>k</italic>, so γ_ik is always 1:\n</p>", "<title>Influence of preprocessing</title>", "<p>The type of preprocessing applied to the data is influencing the outcome of an analysis ##REF##16762068##[5]##, ##UREF##6##[9]##. In the case of only searching for structures in the variables (so all objects are a member of all substructures, as is the case in for instance PCA), it is well-known ##UREF##6##[9]## that the mixed models as mentioned in equation (6) can be treated as pure multiplicative models by first removing any sample or variable means by column or row centering. Apart from centring the data, also scaling can be applied to assure that less abundant metabolites (variables) have the same <italic>a priori</italic> chance to be important in the final model as more abundant metabolites. In our case, we do not partition in the sample direction. Hence, centering across the samples and scaling each variable to standard deviation one seems reasonable.</p>", "<title>Existing algorithms for simple models</title>", "<p>There are several algorithms described in literature that can create simple models according to our definition in the previous sections. In this paper, we have chosen two algorithms, both representing both the multiplicative and additive model classes. In the following section, a short explanation of both methods will be given.</p>", "<title>Interpretable dimension reduction (IDR)</title>", "<p>IDR ##UREF##7##[10]## uses the PCA solution as starting point for creating simple models. By reducing and summarizing the number of non-zero elements of the loading vector, the loadings are simpler to interpret. IDR uses two constraints for obtaining simpler loadings of which the homogeneity constraint is used and discussed in this paper. This homogeneity constraint is applied to a loading that is obtained by PCA. Each loading value is rounded off to the nearest ±1. To increase the interpretability, zeros are introduced into the loadings, starting by replacing the absolute smallest loading values with zeros and continuing until the largest loading value is left over. Modified loadings are normalized. Each time after introducing another zero in the loadings, the angle to the original loadings is determined. The optimal number of inserted zeros is given by the lowest angle to the original variables and this one will be chosen. This method can either be used on a complete set of (PCA) loadings or in an iterative way simplifying one loading at a time. We use this method in a iterative way, deflating one simple component before starting with the next one. Step 1 to 8 of IDR with the homogeneity is as follows:</p>", "<p>Set the <italic>k</italic> values of PCA loading vector <italic>α</italic> to , matching the sign with the original value.</p>", "<p>Look for the absolute lowest non zero value of <italic>α</italic>, and set it to zero.</p>", "<p>Calculate the inner product the original loadings vector <italic>α</italic> and the simplified <italic>α</italic>.</p>", "<p>Convert the inner product to an angle with the inverse cosine.</p>", "<p>Repeat steps 2–5 until only the largest absolute value is left over.</p>", "<p>The simplified <italic>α</italic> that has the lowest angle is the optimal new IDR component.</p>", "<p>Calculate scores () with optimal IDR component (): Subtract the IDR component from the original data: \n</p>", "<p>Repeat this procedure of all IDR components.</p>", "<p>The final IDR model has the form:Here <italic>t_ik</italic> are the scores and <italic>p_jk</italic> the loadings originating from PCA for component <italic>k</italic>. Many values of <italic>p</italic>\n_jk are zero. This can be made explicit by writingwhere the symbol <italic>δ_jk</italic> is the same as before and the nonzero values of <italic>p_jk</italic> are either 1 or −1. Clearly, eq (9) is a special case of eqns (2) and (5) showing that IDR fits into the simplivariate framework.</p>", "<title>Plaid models</title>", "<p>Plaid ##UREF##8##[11]##–##UREF##10##[13]## is a form of two mode clustering that allows for overlapping clusters. By iteratively searching the data, plaid tries to find patches in the data that can be modeled by an ANOVA##UREF##4##[7]## decomposition. Objects or variables can be in more than one cluster or in no cluster at all. Plaid has originally been devised for micro-array data, but can be extended to other types of data.</p>", "<p>The plaid model consists of a series of additive layers intended to capture the underlying structure of matrix <bold>X</bold>. The plaid model also includes the possibility of a background layer containing all variables and objects. Plaid models each cluster with standard 2-way Anova decomposition for each layer <italic>k</italic>:where a <italic>μ</italic>\n_k is introduced to serve as a general mean (model (4) is essentially the same as model (10) ##UREF##5##[8]##). This gives Eq. 11: the decomposition of matrix <bold>X</bold> into <italic>K</italic>+<italic>1</italic> plaid models assuming that all samples contribute to the plaid (as before):\n</p>", "<p>Here, <italic>ϕ_ijk</italic> is the plaid contribution for element x<italic>_ij</italic> from plaid model <italic>k</italic> and <italic>ϕ_ij</italic>\n₀ is the background layer model for entire the entire data matrix <bold>X</bold> (<italic>I</italic>×<italic>J</italic>). It can be seen that Eq. (11) is a special case of Eq. <bold>(7)</bold>. The background layer is especially important when dealing with micro-array data and can be used to model the background signal. This layer will be omitted from our analysis, because it has no meaning for metabolomics data. Instead the proper preprocessing will be used to correct for offsets and scale differences. An algorithmic overview of the plaid algorithm is shown below:</p>", "<p>Choose starting values for and (indicating cluster membership)</p>", "<p>Update layer effects using plaid cluster estimate <italic>X_k</italic> using ANOVA decomposition. s indicates iteration number.\n</p>", "<p>Update cluster membership\n</p>", "<p>repeat step 2–3 for s iterations</p>", "<p>Compute final layer effects as in step 2</p>", "<p>Prune plaid cluster to remove ill fitting metabolites.</p>", "<p>Test <italic>X_k</italic> for significance, stop procedure if <italic>X_k</italic> is not significant otherwise accept</p>", "<p>Subtract <italic>X_k</italic> from X</p>", "<p>Apply backfitting for each obtained plaid cluster</p>", "<p>Apply pruning to remove ill fitting metabolites and continue at step 2</p>", "<p>The above algorithm is the original Plaid algorithm. We used it with some adaptations to our circumstances:</p>", "<p>we did not apply significance testing but selected 6 plaids for illustration.</p>", "<p>we applied a one step backfitting procedure</p>", "<p>we did used γ_j = 1 throughout and, hence, did not have to optimize those values.</p>", "<p>When residuals of selected metabolites after the plaid fit are larger than the prune fraction (0.70, see ##TAB##0##Table 1##), metabolites will be excluded from that plaid cluster. This mechanism ensures small and tight clusters in which the feature of the plaid cluster is clear in all members of the plaid cluster ##UREF##9##[12]##.</p>", "<title>Background of the dataset</title>", "<p>\n<italic>E. coli</italic> NST 74, a phenylalanine overproducing strain and <italic>E. coli</italic> W3110, a wild type strain were grown in batch fermentations at 30°C in a Bioflow II (New Brunswick Scientific) bioreactor as previously described ##REF##16223263##[14]##. In short, samples were grown on MMT12 medium with glucose as carbon source, a constant pH and a constant oxygen tension of 30%. Samples were taken at 16, 24, 40 and 48 hours and analyzed by GC-MS and LC-MS. Peaks related to the substrates used for growth (glucose and succinate) were removed from the data. Deliberate variations in the default protocol resulted in the experimental design that can be found in ##REF##16223263##[14]##. The resulting data set consisted of 28 measurements and 188 metabolites. Extensive details on experimental setup, GC-MS and LC-MS analysis and subsequent preprocessing can be found in ##REF##16223263##[14]##.</p>", "<p>Plaid and IDR were programmed in Matlab 7.1 ##UREF##11##[15]## and are available on the internet at <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.bdagroup.nl/downloads/bda_downloads.html\">http://www.bdagroup.nl/downloads/bda_downloads.html</ext-link>. All computations were performed on an Intel Xeon 3.4 GHz computer with 3.25 GB of memory.</p>" ]

[ "<title>Results and Discussion</title>", "<p>Metabolomics data is highly dynamic in range. Metabolites can have very different and very large concentration ranges. Some metabolites will be zero since their concentrations will be too low to detect under some experimental conditions. This indicates that metabolomics data is not pure multiplicative in nature and can benefit from removing column means.</p>", "<p>For illustrative purposes, some metabolite measurements are plotted in ##FIG##1##Figure 2##. The upper part of ##FIG##1##Figure 2## shows the concentration range of 10 metabolites (dotted black line; left) together with an additive fit (red line; middle) and the multiplicative fit (dashed blue line; right) for this set of 10 metabolites. The lower part of ##FIG##1##Figure 2## shows the same fit, but after auto scaling the data. It can be seen from ##FIG##1##Figure 2## that the range of an additive fit is the same for all metabolites and is given by the range of the α_i's values. It is clear that an additive model has large difficulties modeling data with highly varying ranges for the metabolites. This justifies scaling of the data. The offsets of these ranges are determined by the values of the β_j's. The range of a multiplicative fit can be more dynamic since it is determined by a multiplication of the values α_i's and β_j's. Additive and multiplicative simple components have clearly a different behavior.</p>", "<p>\n##FIG##2##Figure 3## shows the percentage of the original data set captured by PCA and IDR components. As expected, the PCA components explain a larger part of the data, since IDR components are constrained PCA components and thus explain less variance. IDR components &gt;18 explain more than the original PCA components. This is easily explained, since the first 18 PCA components have almost explained the total variation in the data set, while the IDR components still capture variance that was left out by earlier IDR components. For the remainder of this paper we will focus on the first six components. They describe the largest effects in the data set and give us a clear understanding of IDR and plaid.</p>", "<p>\n##FIG##3##Figure 4## shows the loadings of the PCA solution for six components in a gray-scale fashion. This figure clearly shows the problem of PCA for interpreting the solution: all components have contributions from all metabolites. This point exactly illustrates the reason for developing simplivariate models.</p>", "<p>\n##FIG##4##Figure 5## shows the determination of the optimal number of zeros in the first IDR simple component loading. The minimum is indicated by the dotted line and an asterisk. Each IDR component has a different number of zeros that results in a minimal angle between simple IDR component and original loadings. For the first IDR component, the optimal angle is 26.4 degrees and a total of 110 zeros is introduced in this simple component loading, while 78 loadings are non-zero.</p>", "<p>\n##FIG##5##Figure 6## shows the IDR simple loading vectors for the first six loadings. There is a clear distinct pattern of metabolite concentrations entering the loading (either 1 or −1, indicated by black and grey and metabolite concentrations not entering the loading (being zero, indicated by white). ##FIG##6##Figure 7## shows the first six plaid models. In ##FIG##3##Figure 4##, ##FIG##5##6## and ##FIG##6##7## all metabolites have been ordered in such a way that metabolites are grouped as much as possible according to the different plaid clusters. Since the plaid models are only created in the variable mode (which is always the case for IDR), the object mode is not shown. One difference between plaid clusters and IDR components is striking: plaid clusters contain less metabolites and are easier to interpret. The intrinsic mechanism to lower the number of selected metabolites in IDR is automatic and cannot be intervened with. The number of zeros introduced in IDR is regulated by the optimization criterion (see step 6 of IDR algorithm) and artificially lowering the number of metabolites would yield a threshold PCA, which basically cuts of loadings values above a certain value. Hence, the interpretability can therefore not be increased. Initially, plaid also selects (too) many metabolites, however the pruning mechanism (present in the original algorithm; see <xref ref-type=\"sec\" rid=\"s2\">Materials and Methods</xref>) is able to remove ill-fitting metabolites (see ##TAB##0##Table 1## for the settings that have been used in the plaid algorithm).</p>", "<p>Although IDR and plaid have different underlying models, multiplicative or additive, there are similarities between the IDR components in ##FIG##5##Figure 6## and the plaid models in ##FIG##6##Figure 7##. Many of the metabolites that are selected by IDR are also selected by the plaid models. One phenomenon is strikingly present in ##FIG##5##Figures 6## and ##FIG##6##7##. In plaid component 1, only metabolites are present that have a positive IDR value (black in ##FIG##6##Figure 7##). In plaid component 2 only metabolites are present that show an IDR value of −1 in IDR component 2. Plaid components 3 and 4 are even more illustrative, since they are both represented by IDR component 3: plaid component 3 corresponds to IDR values of −1 and plaid component 4 corresponds to IDR values of +1. The reason for this phenomenon is that the additive plaid models can only represent positively correlated metabolites, missing an important part of the relationships in the data. This idea is illustrated by ##FIG##7##Figure 8## where the correlations are shown between the metabolites in IDR component 1 and between the metabolites in plaid cluster 1. What we clearly see, also in the distributions of the correlation coefficients, is that the plaid cluster contains (almost) no negatively correlated metabolites, while metabolites in IDR component 1 can be positively and negatively correlated. The differences between IDR and plaid become larger for higher components/plaid models.</p>", "<title>Biological interpretation</title>", "<p>There are too many metabolites present in each IDR components to come to a meaningful analysis of the IDR results. However, the plaid clusters are relatively simple and contain biological meaningful metabolite clusters. For instance, the first plaid cluster contains all intermediates of the Krebs cycle whose concentration is above the detection limit in this data set, i.e. fumarate, malate; 2-ketoglutarate, and citrate (##FIG##6##Fig. 7##). Moreover, three metabolites which are just one enzymatic step removed from these TCA cycle intermediates, i.e. 2-hydroxyglutarate, glutamate and aspartate are also present in this first plaid cluster.</p>", "<p>Another example is plaid cluster 4 that contains many intermediates of the phenylalanine biosynthesis pathway, i.e. erythrose-4-phosphate, 3-dehydroquinate, shikimate-3-phosphate, chorismate, phenylpyruvate, and phenylalanine itself, and several compounds which are side routes of this pathway, i.e. 3-phenyllactate, and tyrosine. Interestingly, prephenate, an intermediate at the splitting point of the phenylalanine and tyrosine biosynthesis routes, is not clustered in plaid cluster 4 but in plaid cluster 3. In contrast, when analyzing this data set by IDR, all the phenylalanine-related intermediates described above, including prephenate, end up in the same IDR component, i.e. IDR component 3 (##FIG##5##Fig. 6##). However, prephenate shows a negative loading while all other intermediates have a positive loading. One of the enzymes catalyzing the formation of prephenate (chorismate mutase encoded by pheA) is controlled by feedback inhibition by phenylalanine and also the two enzymes catalyzing its conversion (prephenate dehydratase and prephenate dehydrogenase) are controlled by feedback inhibition by phenylalanine and tyrosine, respectively. This might very well explain why this intermediate (prephenate) shows a negative correlation with the other phenylalanine intermediates (IDR analysis) and thus ends up in a different plaid cluster. Remarkably, shikimate, another phenylalanine biosynthesis intermediate, is neither clustered in plaid cluster 4 (##FIG##6##Fig. 7##) nor in IDR component 3 (##FIG##5##Fig. 6##). Interestingly, ppGpp, a major regulator of cellular metabolism, is present in plaid cluster 4/IDR component 3 indicating a link between phenylalanine biosynthesis and the stringent response in <italic>E. coli</italic>.</p>", "<p>The most useful results are obtained with plaid which models (patches of) data with an additive model while IDR uses a multiplicative model. It is possible to mix both models to obtain a mixed model representation (see section on simple structures, model number 4). Mixed models might also help to further strengthen the plaid clusters. Additive plaid models can only contain positively correlated metabolite concentrations, while metabolites that are negatively correlated can still be part of the same biochemical process.</p>", "<title>Conclusions</title>", "<p>The presented framework provides a good basis for simplivariate data analysis models. The two presented methods IDR and Plaid fit well in this framework. IDR suffers from too many selected metabolites which makes it rather ineffective for creating more interpretable models. This selection is intrinsic for the method and cannot be tuned. Plaid, on the other hand, was shown to be very effective in creating clusters with distinct biochemical meanings. This shows that the concept of simplivariate models is valuable.</p>", "<p>The Plaid models also have shortcomings, notably, their inability to model metabolites belonging to the same processes having either positive or negative correlations. This can possibly be overcome by using simple components with a mixed-model structure. Moreover, the pruning mechanism present in plaid that prevents that too many metabolites are selected in a plaid cluster, remains a crude way of cleaning up a solution. It is inefficient to first create large plaid clusters (at a certain computational cost) and decreasing them after they are finished. By more carefully optimizing a plaid cluster this should be prevented. This will be subject of further research.</p>", "<p>The framework allows for any simple component structure to include in the simplivariate model. When some of the metabolites are known to be linked in certain experiments by interlinked pathways and/or co-regulation, then these can be forced in one simple component with a structure reflecting these pathways/ this co-regulation. Also metabolic network information can be used to choose simple component structures. All these extensions are the subject of a follow-up paper.</p>", "<title>Notation</title>", "<p>Matrix <bold>X</bold> (boldface), vector <bold>x</bold> (boldface), scalar <italic>x</italic> (italic).</p>", "<p>Sizes: <bold>X</bold> (<italic>I</italic> objects×<italic>J</italic> variables), objects, <italic>i</italic> = 1,…,<italic>I</italic>; variables <italic>j</italic> = 1,…,<italic>J</italic>; groups <italic>k</italic> = 1,…,<italic>K</italic>; Each <italic>k</italic> represents a simple component that are used to described the data.</p>", "<p>Group memberships: δ_jk = indicator for group membership of variable <italic>j</italic> in group <italic>k</italic> (δ_jk = 1 if variable <italic>j</italic> is present in group <italic>k</italic>, 0 otherwise); γ_ik = indicator for group membership of object <italic>i</italic> in group <italic>k</italic> (γ_ik = 1 if object <italic>i</italic> is present in group <italic>k</italic>, 0 otherwise).</p>", "<p>PCA-scores: <bold>T</bold> (<italic>I</italic>×<italic>R</italic>), <bold>t</bold>\n_r (<italic>r</italic> = 1,…<italic>R</italic>), <italic>t</italic>\n_ir. (<italic>R</italic> = number of principal components used)</p>", "<p>PCA-loadings: <bold>P</bold> (<italic>J</italic>×<italic>R</italic>), <bold>p</bold>\n_r, <italic>p</italic>\n_jr.</p>" ]

[ "<title>Results and Discussion</title>", "<p>Metabolomics data is highly dynamic in range. Metabolites can have very different and very large concentration ranges. Some metabolites will be zero since their concentrations will be too low to detect under some experimental conditions. This indicates that metabolomics data is not pure multiplicative in nature and can benefit from removing column means.</p>", "<p>For illustrative purposes, some metabolite measurements are plotted in ##FIG##1##Figure 2##. The upper part of ##FIG##1##Figure 2## shows the concentration range of 10 metabolites (dotted black line; left) together with an additive fit (red line; middle) and the multiplicative fit (dashed blue line; right) for this set of 10 metabolites. The lower part of ##FIG##1##Figure 2## shows the same fit, but after auto scaling the data. It can be seen from ##FIG##1##Figure 2## that the range of an additive fit is the same for all metabolites and is given by the range of the α_i's values. It is clear that an additive model has large difficulties modeling data with highly varying ranges for the metabolites. This justifies scaling of the data. The offsets of these ranges are determined by the values of the β_j's. The range of a multiplicative fit can be more dynamic since it is determined by a multiplication of the values α_i's and β_j's. Additive and multiplicative simple components have clearly a different behavior.</p>", "<p>\n##FIG##2##Figure 3## shows the percentage of the original data set captured by PCA and IDR components. As expected, the PCA components explain a larger part of the data, since IDR components are constrained PCA components and thus explain less variance. IDR components &gt;18 explain more than the original PCA components. This is easily explained, since the first 18 PCA components have almost explained the total variation in the data set, while the IDR components still capture variance that was left out by earlier IDR components. For the remainder of this paper we will focus on the first six components. They describe the largest effects in the data set and give us a clear understanding of IDR and plaid.</p>", "<p>\n##FIG##3##Figure 4## shows the loadings of the PCA solution for six components in a gray-scale fashion. This figure clearly shows the problem of PCA for interpreting the solution: all components have contributions from all metabolites. This point exactly illustrates the reason for developing simplivariate models.</p>", "<p>\n##FIG##4##Figure 5## shows the determination of the optimal number of zeros in the first IDR simple component loading. The minimum is indicated by the dotted line and an asterisk. Each IDR component has a different number of zeros that results in a minimal angle between simple IDR component and original loadings. For the first IDR component, the optimal angle is 26.4 degrees and a total of 110 zeros is introduced in this simple component loading, while 78 loadings are non-zero.</p>", "<p>\n##FIG##5##Figure 6## shows the IDR simple loading vectors for the first six loadings. There is a clear distinct pattern of metabolite concentrations entering the loading (either 1 or −1, indicated by black and grey and metabolite concentrations not entering the loading (being zero, indicated by white). ##FIG##6##Figure 7## shows the first six plaid models. In ##FIG##3##Figure 4##, ##FIG##5##6## and ##FIG##6##7## all metabolites have been ordered in such a way that metabolites are grouped as much as possible according to the different plaid clusters. Since the plaid models are only created in the variable mode (which is always the case for IDR), the object mode is not shown. One difference between plaid clusters and IDR components is striking: plaid clusters contain less metabolites and are easier to interpret. The intrinsic mechanism to lower the number of selected metabolites in IDR is automatic and cannot be intervened with. The number of zeros introduced in IDR is regulated by the optimization criterion (see step 6 of IDR algorithm) and artificially lowering the number of metabolites would yield a threshold PCA, which basically cuts of loadings values above a certain value. Hence, the interpretability can therefore not be increased. Initially, plaid also selects (too) many metabolites, however the pruning mechanism (present in the original algorithm; see <xref ref-type=\"sec\" rid=\"s2\">Materials and Methods</xref>) is able to remove ill-fitting metabolites (see ##TAB##0##Table 1## for the settings that have been used in the plaid algorithm).</p>", "<p>Although IDR and plaid have different underlying models, multiplicative or additive, there are similarities between the IDR components in ##FIG##5##Figure 6## and the plaid models in ##FIG##6##Figure 7##. Many of the metabolites that are selected by IDR are also selected by the plaid models. One phenomenon is strikingly present in ##FIG##5##Figures 6## and ##FIG##6##7##. In plaid component 1, only metabolites are present that have a positive IDR value (black in ##FIG##6##Figure 7##). In plaid component 2 only metabolites are present that show an IDR value of −1 in IDR component 2. Plaid components 3 and 4 are even more illustrative, since they are both represented by IDR component 3: plaid component 3 corresponds to IDR values of −1 and plaid component 4 corresponds to IDR values of +1. The reason for this phenomenon is that the additive plaid models can only represent positively correlated metabolites, missing an important part of the relationships in the data. This idea is illustrated by ##FIG##7##Figure 8## where the correlations are shown between the metabolites in IDR component 1 and between the metabolites in plaid cluster 1. What we clearly see, also in the distributions of the correlation coefficients, is that the plaid cluster contains (almost) no negatively correlated metabolites, while metabolites in IDR component 1 can be positively and negatively correlated. The differences between IDR and plaid become larger for higher components/plaid models.</p>", "<title>Biological interpretation</title>", "<p>There are too many metabolites present in each IDR components to come to a meaningful analysis of the IDR results. However, the plaid clusters are relatively simple and contain biological meaningful metabolite clusters. For instance, the first plaid cluster contains all intermediates of the Krebs cycle whose concentration is above the detection limit in this data set, i.e. fumarate, malate; 2-ketoglutarate, and citrate (##FIG##6##Fig. 7##). Moreover, three metabolites which are just one enzymatic step removed from these TCA cycle intermediates, i.e. 2-hydroxyglutarate, glutamate and aspartate are also present in this first plaid cluster.</p>", "<p>Another example is plaid cluster 4 that contains many intermediates of the phenylalanine biosynthesis pathway, i.e. erythrose-4-phosphate, 3-dehydroquinate, shikimate-3-phosphate, chorismate, phenylpyruvate, and phenylalanine itself, and several compounds which are side routes of this pathway, i.e. 3-phenyllactate, and tyrosine. Interestingly, prephenate, an intermediate at the splitting point of the phenylalanine and tyrosine biosynthesis routes, is not clustered in plaid cluster 4 but in plaid cluster 3. In contrast, when analyzing this data set by IDR, all the phenylalanine-related intermediates described above, including prephenate, end up in the same IDR component, i.e. IDR component 3 (##FIG##5##Fig. 6##). However, prephenate shows a negative loading while all other intermediates have a positive loading. One of the enzymes catalyzing the formation of prephenate (chorismate mutase encoded by pheA) is controlled by feedback inhibition by phenylalanine and also the two enzymes catalyzing its conversion (prephenate dehydratase and prephenate dehydrogenase) are controlled by feedback inhibition by phenylalanine and tyrosine, respectively. This might very well explain why this intermediate (prephenate) shows a negative correlation with the other phenylalanine intermediates (IDR analysis) and thus ends up in a different plaid cluster. Remarkably, shikimate, another phenylalanine biosynthesis intermediate, is neither clustered in plaid cluster 4 (##FIG##6##Fig. 7##) nor in IDR component 3 (##FIG##5##Fig. 6##). Interestingly, ppGpp, a major regulator of cellular metabolism, is present in plaid cluster 4/IDR component 3 indicating a link between phenylalanine biosynthesis and the stringent response in <italic>E. coli</italic>.</p>", "<p>The most useful results are obtained with plaid which models (patches of) data with an additive model while IDR uses a multiplicative model. It is possible to mix both models to obtain a mixed model representation (see section on simple structures, model number 4). Mixed models might also help to further strengthen the plaid clusters. Additive plaid models can only contain positively correlated metabolite concentrations, while metabolites that are negatively correlated can still be part of the same biochemical process.</p>", "<title>Conclusions</title>", "<p>The presented framework provides a good basis for simplivariate data analysis models. The two presented methods IDR and Plaid fit well in this framework. IDR suffers from too many selected metabolites which makes it rather ineffective for creating more interpretable models. This selection is intrinsic for the method and cannot be tuned. Plaid, on the other hand, was shown to be very effective in creating clusters with distinct biochemical meanings. This shows that the concept of simplivariate models is valuable.</p>", "<p>The Plaid models also have shortcomings, notably, their inability to model metabolites belonging to the same processes having either positive or negative correlations. This can possibly be overcome by using simple components with a mixed-model structure. Moreover, the pruning mechanism present in plaid that prevents that too many metabolites are selected in a plaid cluster, remains a crude way of cleaning up a solution. It is inefficient to first create large plaid clusters (at a certain computational cost) and decreasing them after they are finished. By more carefully optimizing a plaid cluster this should be prevented. This will be subject of further research.</p>", "<p>The framework allows for any simple component structure to include in the simplivariate model. When some of the metabolites are known to be linked in certain experiments by interlinked pathways and/or co-regulation, then these can be forced in one simple component with a structure reflecting these pathways/ this co-regulation. Also metabolic network information can be used to choose simple component structures. All these extensions are the subject of a follow-up paper.</p>", "<title>Notation</title>", "<p>Matrix <bold>X</bold> (boldface), vector <bold>x</bold> (boldface), scalar <italic>x</italic> (italic).</p>", "<p>Sizes: <bold>X</bold> (<italic>I</italic> objects×<italic>J</italic> variables), objects, <italic>i</italic> = 1,…,<italic>I</italic>; variables <italic>j</italic> = 1,…,<italic>J</italic>; groups <italic>k</italic> = 1,…,<italic>K</italic>; Each <italic>k</italic> represents a simple component that are used to described the data.</p>", "<p>Group memberships: δ_jk = indicator for group membership of variable <italic>j</italic> in group <italic>k</italic> (δ_jk = 1 if variable <italic>j</italic> is present in group <italic>k</italic>, 0 otherwise); γ_ik = indicator for group membership of object <italic>i</italic> in group <italic>k</italic> (γ_ik = 1 if object <italic>i</italic> is present in group <italic>k</italic>, 0 otherwise).</p>", "<p>PCA-scores: <bold>T</bold> (<italic>I</italic>×<italic>R</italic>), <bold>t</bold>\n_r (<italic>r</italic> = 1,…<italic>R</italic>), <italic>t</italic>\n_ir. (<italic>R</italic> = number of principal components used)</p>", "<p>PCA-loadings: <bold>P</bold> (<italic>J</italic>×<italic>R</italic>), <bold>p</bold>\n_r, <italic>p</italic>\n_jr.</p>" ]

[ "<title>Conclusions</title>", "<p>The presented framework provides a good basis for simplivariate data analysis models. The two presented methods IDR and Plaid fit well in this framework. IDR suffers from too many selected metabolites which makes it rather ineffective for creating more interpretable models. This selection is intrinsic for the method and cannot be tuned. Plaid, on the other hand, was shown to be very effective in creating clusters with distinct biochemical meanings. This shows that the concept of simplivariate models is valuable.</p>", "<p>The Plaid models also have shortcomings, notably, their inability to model metabolites belonging to the same processes having either positive or negative correlations. This can possibly be overcome by using simple components with a mixed-model structure. Moreover, the pruning mechanism present in plaid that prevents that too many metabolites are selected in a plaid cluster, remains a crude way of cleaning up a solution. It is inefficient to first create large plaid clusters (at a certain computational cost) and decreasing them after they are finished. By more carefully optimizing a plaid cluster this should be prevented. This will be subject of further research.</p>", "<p>The framework allows for any simple component structure to include in the simplivariate model. When some of the metabolites are known to be linked in certain experiments by interlinked pathways and/or co-regulation, then these can be forced in one simple component with a structure reflecting these pathways/ this co-regulation. Also metabolic network information can be used to choose simple component structures. All these extensions are the subject of a follow-up paper.</p>" ]

[ "<p>Conceived and designed the experiments: JAH MMWBH JAW AS. Performed the experiments: MJvdW. Analyzed the data: JAH. Contributed reagents/materials/analysis tools: MJvdW. Wrote the paper: JAH MMWBH JAW AS. Group Leader: AS RB MJvdW.</p>", "<p>Current address: Biometris, Wageningen University, Wageningen, The Netherlands</p>", "<p>One of the new expanding areas in functional genomics is metabolomics: measuring the metabolome of an organism. Data being generated in metabolomics studies are very diverse in nature depending on the design underlying the experiment. Traditionally, variation in measurements is conceptually broken down in systematic variation and noise where the latter contains, e.g. technical variation. There is increasing evidence that this distinction does not hold (or is too simple) for metabolomics data. A more useful distinction is in terms of informative and non-informative variation where informative relates to the problem being studied. In most common methods for analyzing metabolomics (or any other high-dimensional x-omics) data this distinction is ignored thereby severely hampering the results of the analysis. This leads to poorly interpretable models and may even obscure the relevant biological information. We developed a framework from first data analysis principles by explicitly formulating the problem of analyzing metabolomics data in terms of informative and non-informative parts. This framework allows for flexible interactions with the biologists involved in formulating prior knowledge of underlying structures. The basic idea is that the informative parts of the complex metabolomics data are approximated by simple components with a biological meaning, e.g. in terms of metabolic pathways or their regulation. Hence, we termed the framework ‘simplivariate models’ which constitutes a new way of looking at metabolomics data. The framework is given in its full generality and exemplified with two methods, IDR analysis and plaid modeling, that fit into the framework. Using this strategy of ‘divide and conquer’, we show that meaningful simplivariate models can be obtained using a real-life microbial metabolomics data set. For instance, one of the simple components contained all the measured intermediates of the Krebs cycle of <italic>E. coli</italic>. Moreover, these simplivariate models were able to uncover regulatory mechanisms present in the phenylalanine biosynthesis route of <italic>E. coli</italic>.</p>" ]

[]

[]

[ "<fig id=\"pone-0003259-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003259.g001</object-id><label>Figure 1</label><caption><title>Graphical representation of the three different approaches to the analysis of multivariate data.</title><p>From left to right: the univariate, simplivariate and multivariate approach.</p></caption></fig>", "<fig id=\"pone-0003259-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003259.g002</object-id><label>Figure 2</label><caption><title>Concentration ranges for 10 metabolites before (top figure) and after (lower figure) autoscaling.</title><p>Data is taken from E. Coli data as used in the remained of this paper. The whiskers indicate the total concentration range for each of the 10 metabolites. Each metabolite is represented three times. The left black lines for each metabolite are the actual concentrations. The middle red line indicates the fit/model with an additive model. The right blue lines indicate the fit/model with a blue multiplicative model.</p></caption></fig>", "<fig id=\"pone-0003259-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003259.g003</object-id><label>Figure 3</label><caption><title>Percentage explained of original dataset given a certain number of components.</title><p>Solid line represents IDR components, dotted line represents PCA components. See text for explanation.</p></caption></fig>", "<fig id=\"pone-0003259-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003259.g004</object-id><label>Figure 4</label><caption><title>PCA solution.</title><p>The values of the loadings are indicated by a grayscale color as indicated by the colorbar. The grouping of metabolites is identical to the grouping of the plaid solution for clarity.</p></caption></fig>", "<fig id=\"pone-0003259-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003259.g005</object-id><label>Figure 5</label><caption><title>Determination of the optimal number of zeros for the first IDR component.</title><p>The optimum is chosen where the angle between the simple component and principal components is minimal. This is indicated by a dotted line and an asterisk.</p></caption></fig>", "<fig id=\"pone-0003259-g006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003259.g006</object-id><label>Figure 6</label><caption><title>The first 6 IDR components obtained with deflation.</title><p>Black squares indicate a +1, white indicates a zero, grey indicates a −1. The grouping of metabolites is identical to the grouping of the plaid solution for clarity.</p></caption></fig>", "<fig id=\"pone-0003259-g007\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003259.g007</object-id><label>Figure 7</label><caption><title>The first 6 plaid components.</title><p>Black squares indicate a +1, white indicates a zero. Results have been grouped as much as possible for clarity.</p></caption></fig>", "<fig id=\"pone-0003259-g008\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003259.g008</object-id><label>Figure 8</label><caption><title>Correlations between metabolites present in IDR component 1 (top part) and plaid component 1 (bottom part).</title><p>Positive correlations are indicated by a white square, negative correlations are indicated by a black square.</p></caption></fig>" ]

[ "<table-wrap id=\"pone-0003259-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003259.t001</object-id><label>Table 1</label><caption><title>Settings for the plaid algorithm.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Setting</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Value</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Maximum iterations</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">50</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Number of permutation in significant testing</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">25</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Backfitting</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">one step</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Maximum number of layers</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Prunefraction<xref ref-type=\"table-fn\" rid=\"nt101\">*</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.7</td></tr></tbody></table></alternatives></table-wrap>" ]

[ "<disp-formula><label>(1)</label></disp-formula>", "<disp-formula><label>(2)</label></disp-formula>", "<disp-formula><label>(3)</label></disp-formula>", "<disp-formula><label>(4)</label></disp-formula>", "<disp-formula><label>(5)</label></disp-formula>", "<disp-formula><label>(6)</label></disp-formula>", "<disp-formula><label>(7)</label></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula><label>(8)</label></disp-formula>", "<disp-formula><label>(9)</label></disp-formula>", "<disp-formula><label>(10)</label></disp-formula>", "<disp-formula><label>(11)</label></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"nt101\"><label>*</label><p>Minimum of proportional reduction in residual sum of squares required for cluster membership.</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This work was part of the BioRange programme of the Netherlands Bioinformatics Centre (NBIC), which is supported by a BSIK grant through the Netherlands Genomics Initiative (NGI). The manuscript was reviewed by all members of the SP2.2.3 project group of NBIC.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003259.g001\"/>", "<graphic xlink:href=\"pone.0003259.e001.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003259.e002.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003259.e003.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003259.e004.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003259.e005.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003259.e006.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003259.e007.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pone.0003259.e008.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003259.e009.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003259.e010.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003259.e011.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003259.e012.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003259.e013.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003259.e014.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003259.e015.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003259.e016.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pone.0003259.e017.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003259.e018.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003259.e019.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003259.e020.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic id=\"pone-0003259-t001-1\" xlink:href=\"pone.0003259.t001\"/>", "<graphic xlink:href=\"pone.0003259.g002\"/>", "<graphic xlink:href=\"pone.0003259.g003\"/>", "<graphic xlink:href=\"pone.0003259.g004\"/>", "<graphic xlink:href=\"pone.0003259.g005\"/>", "<graphic xlink:href=\"pone.0003259.g006\"/>", "<graphic xlink:href=\"pone.0003259.g007\"/>", "<graphic xlink:href=\"pone.0003259.g008\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>T-cells epitopes are short linear peptides generated by cleavage of antigenic proteins. The identification of T-cell epitopes in protein sequences is important for understanding disease pathogenesis, identifying potential autoantigens, and designing vaccines and immune-based cancer therapies. A major step in identifying potential T-cell epitopes involves identifying the peptides that bind to a target major histocompatibility complex (MHC) molecule. Because of the high cost of experimental identification of such peptides, there is an urgent need for reliable computational methods for predicting MHC binding peptides ##REF##16846250##[1]##.</p>", "<p>There are two major classes of MHC molecules: MHC class I (MHC-I) molecules characterized by short binding peptides, usually consisting of nine residues; and MHC class II (MHC-II) molecules with binding peptides that range from 11 to 30 residues in length, although shorter and longer peptide lengths are not uncommon ##REF##7890324##[2]##. The binding groove of MHC-II molecules is open at both ends, allowing peptides longer than 9-mers to bind. However, it has been reported that a 9-mer core region is essential for MHC-II binding ##REF##7890324##[2]##, ##REF##7612235##[3]##. Because the precise location of the 9-mer core region of MHC-II binding peptides is unknown, predicting MHC-II binding peptides tends to be more challenging than predicting MHC-I binding peptides.</p>", "<p>Despite the high degree of variability in the length of MHC-II binding peptides, most existing computational methods for predicting MHC-II binding peptides focus on identifying a 9-mer core peptide. Computational approaches available for predicting MHC-II binding peptides from amino acid sequences include: (i) Motif-based methods such as methods that use a position weight matrix (PWM) to model an ungapped multiple sequence alignment of MHC binding peptides ##REF##11751237##[4]##–##REF##18031584##[8]##, and a statistical approach based on Hidden Markov Models (HMMs) ##REF##9849933##[9]##, ##REF##16233301##[10]##; (ii) Machine learning methods based on Artificial Neural Networks (ANN) ##REF##14962912##[6]##, ##REF##12717023##[11]##–##UREF##0##[13]## and Support Vector Machines (SVMs) ##REF##16844990##[14]##–##REF##17105666##[17]##; (iii) Semi-supervised machine learning methods ##REF##16351712##[18]##, ##UREF##1##[19]##.</p>", "<p>The choice of one method over another for MHC-II binding peptide prediction requires reliable assessment of their performance relative to each other. Such assessments usually rely on estimates of their performance on standard benchmark datasets (typically obtained using cross-validation). Several studies ##REF##15349703##[5]##, ##REF##14960470##[15]##–##REF##17105666##[17]##, ##UREF##1##[19]## have reported the performance of MHC-II binding peptide prediction methods using datasets of <italic>unique</italic> peptides. Such datasets can in fact contain peptide sequences that share a high degree of sequence similarity with other peptide sequences in the dataset. Hence, several authors ##REF##14962912##[6]##, ##REF##17608956##[7]##, ##REF##16233301##[10]##, ##UREF##2##[20]## have proposed methods for eliminating <italic>redundant</italic> sequences. However, because MHC-II peptides have lengths that vary over a broad range, similarity reduction of MHC-II peptides is not a straightforward task ##REF##17608956##[7]##. Consequently, standard cross-validation based estimates of performance obtained using such datasets are likely to be overly optimistic because the test set is likely to contain sequences that share significant sequence similarity with one or more sequences in the training set.</p>", "<p>In order to obtain more realistic estimates of performance of MHC-II binding peptide prediction methods, we explored several methods for constructing <italic>similarity-reduced</italic> MHC-II datasets. We constructed <italic>similarity-reduced</italic> MHC-II benchmark datasets, derived from MHCPEP ##UREF##3##[21]##, MHCBN ##REF##12651731##[22]##, and IEDB ##REF##15760272##[23]## databases, using several approaches to reduce the degree of pair-wise sequence similarity shared by sequences in the resulting datasets. The similarity reduction procedures were applied separately to binders and non-binders. Details of the similarity reduction methods are provided in the Materials and <xref ref-type=\"sec\" rid=\"s4\">Methods</xref> Section. Specifically, we generated:</p>", "<p>Datasets of unique peptides MHCPEP-UPDS, MHCBN-UPDS, and IEDB-UPDS extracted from MHCPEP, MHCBN, and IEDB, respectively.</p>", "<p>Datasets of <italic>similarity-reduced</italic> peptides, MHCPEP-SRDS1, MHCBN-SRDS1, and IEDB-SRDS1 derived from the corresponding UPDS datasets using a similarity reduction procedure which ensures that no two peptides in the resulting dataset share a 9-mer subsequence.</p>", "<p>Datasets of <italic>similarity-reduced</italic> peptides, MHCPEP-SRDS2, MHCBN-SRDS2, and IEDB-SRDS2, extracted MHCPEP-SRDS1, MHCBN-SRDS1, and IEDB-SRDS1 respectively by filtering the binders and non-binders in SRDS1 such that the sequence identity between any pair of peptides is less than 80%.</p>", "<p>Datasets of <italic>similarity-reduced</italic> peptides, MHCPEP-SRDS3, MHCBN-SRDS3, and IEDB-SRDS3, derived from the corresponding UPDS datasets using the similarity reduction procedure introduced by Raghava and previously used to construct the MHCBench dataset ##UREF##2##[20]##.</p>", "<p>Datasets of <italic>weighted</italic> unique peptides, MHCPEP-WUPDS, MHCBN-WUPDS, and IEDB-WUPDS, derived from the corresponding UPDS datasets (where the weight assigned to a peptide is inversely proportional to the number of peptides that are similar to it).</p>", "<p>We then used the resulting <italic>similarity-reduced</italic> benchmark datasets to explore the effect of similarity reduction on the performance of different MHC-II binding peptide prediction methods and, more importantly, to rigorously compare the performance of the different prediction methods.</p>", "<p>Our experiments focused on two state-of-the-art methods for training MHC-II binding peptide predictors using variable-length MHC-II peptides and a third method that is designed to exploit the sequence similarity between a test peptide sequence and the peptide sequences in the training set (and is hence likely to perform well on <italic>non similarity-reduced</italic> datasets but poorly on the <italic>similarity-reduced</italic> datasets).</p>", "<p>Specifically, we compared: (i) An approach ##REF##16806474##[16]## that maps each variable-length peptide into a fixed-length feature vector (the so-called composition-transition distribution or CTD) consisting of sequence-derived structural features and physicochemical properties of the input peptide sequence; (ii) An approach ##REF##17105666##[17]## that uses a local alignment (LA) kernel that defines the similarity between two variable-length peptides as the average of all possible local alignments between the two peptides; (iii) An approach that uses the k-spectrum kernel ##UREF##4##[24]## with <italic>k</italic> = 5.</p>", "<p>Because neither the programs used to calculate secondary structure and solvent accessibility of peptides used for generating the CTD representation ##REF##16806474##[16]## nor the precise choices of parameters used for training the LA kernel based classifier ##REF##17105666##[17]## were available to us, we used in our experiments, our own implementations of the corresponding methods. Hence, the results of our experiments should not be viewed as providing direct assessment of performance of the exact implementations of the CTD and LA methods developed by the original authors and used in studies reported in ##REF##16806474##[16]##, ##REF##17105666##[17]##. However, it is worth noting that, the broad conclusions of our study are largely independent of the specific machine learning methods or data transformations.</p>", "<p>Our results demonstrate that, regardless of the similarity reduction method employed, a substantial drop in performance of classifiers is observed compared to their reported performance on benchmark datasets of <italic>unique</italic> peptide sequences. Our results also demonstrate that conclusions regarding the superiority of one prediction method over another can be misleading when they are based on evaluations using benchmark datasets with a high degree of sequence similarity (e.g., the benchmark dataset of <italic>unique</italic> peptide sequences). These results underscore the importance of using <italic>similarity-reduced</italic> datasets in evaluating and comparing alternative MHC-II peptide prediction methods.</p>" ]

[ "<title>Materials and Methods</title>", "<p>The datasets used in this study are derived from MHCPEP ##UREF##3##[21]##, MHCBN ##REF##12651731##[22]##, and IEDB ##REF##15760272##[23]##, which are manually curated repositories of MHC binding peptides reported in the literature.</p>", "<p>We extracted 22 MHC-II allele datasets (each with at least 100 binders) from the MHCPEP database. Because MHCPEP contains only MHC-II binding peptides (“positive examples”), for each allele, we generated an equal number of non-binders (“negative examples”) by randomly extracting protein fragments from SwissProt ##REF##10592178##[35]## protein sequences such that: (i) The length distribution of negative examples is identical to that of the positive examples; (ii) None of the non-binding peptides appear in the set of binders.</p>", "<p>Unlike MHCPEP, MHCBN is a database of binding and non-binding MHC peptides. MHCBN version 4.0 has 35 MHC-II alleles with at least 100 binders. Out of these 35 alleles, only eight alleles have at least 100 non-binders. We extracted the MHCBN benchmark dataset used in this study from the alleles for which at least 100 binders and non-binders peptides are available in MHCBN.</p>", "<p>The Immune Epitope Database and Analysis Resource (IEDB) ##REF##15760272##[23]## is a rich resource of MHC binding data curated from the literature or submitted by immunologists. For each reported peptide, IEDB provides qualitative (i.e., Negative or Positive) and quantitative (i.e., IC50) measurements whenever available. We used both qualitative and quantitative measurements for constructing 12 HLA binary labeled datasets as follows:</p>", "<p>Peptides with no reported quantitative measurements are discarded.</p>", "<p>Peptides with “Positive” qualitative measurement and quantitative measurement less than 500 nM are classified as binders.</p>", "<p>Peptides with “Positive” qualitative measurement and quantitative measurement greater than or equal 500 nM are classified as non-binders.</p>", "<p>Peptides with “Negative” qualitative measurement and quantitative measurement greater than or equal 500 nM are classified as non-binders.</p>", "<p>Peptides with “Negative” qualitative measurement and quantitative measurement less than 500 nM are discarded.</p>", "<p>The reported MHC binding sites are typically identified using truncation, substitution, or mutations in a base peptide ##REF##1704034##[36]##. Because different reported MHC-II binding peptides might actually correspond to experimental manipulation of the same MHC-II binding region using different experimental techniques or different choices of amino acids targeted for truncation, substitution, or mutation, it is not surprising that that MHC databases contain a significant number of highly similar peptides. Hence, we used several similarity reduction methods to extract several different versions of the dataset from each set of sequences.</p>", "<p>It should be noted that the existence of highly similar peptides belonging to the same category may result in an over-optimistic estimation of the classifier performance. Therefore, we applied the similarity reduction procedures separately to the set of binders and non-binders in each dataset. The following sections describe the similarity reduction procedures and the resulting <italic>similarity-reduced</italic> datasets.</p>", "<title>Similarity reduction procedures</title>", "<p>An example of two different types of similar peptides that frequently occur in MHC peptides databases is shown in ##FIG##4##Figure 5##. In type I, two peptides differ from each other in terms of only one or two amino acids (see ##FIG##4##Figure 5A##). Such highly similar peptides are likely to have come from different mutation experiments targeting different sites of the same MHC-II binding peptide. For example, Garcia et al. ##REF##17306766##[37]## report an HLA-DRB1*0401 binding peptide (WGENDTDVFVLNNTR) and 12 additional binding peptides derived from that peptide by replacing one of the amino acid in (WGENDTDVFVLNNTR) sequence with Glycine and experimentally determining the binding affinity of the new peptide. In type II, we find that a shorter peptide in one allele dataset corresponds to a sub-sequence of a longer one that is also in the allele dataset (see ##FIG##4##Figure 5B##).</p>", "<p>Standard approaches to identifying similar peptide sequences rely on the use of a sequence similarity threshold. Sequences that are within a certain predetermined similarity threshold relative to a target sequence are eliminated from the dataset. However, the use of such a simple approach to obtaining a similarity reduced dataset is complicated by the high degree of variability in the length of MHC-II peptides. Using a single fixed similarity cutoff value (e.g. 80%) might not be effective in eliminating type II similar peptides. On the other hand, an attempt to eliminate one of the two such similar sequences by using of a more stringent similarity threshold could result in elimination of most of the dataset.</p>", "<p>To address this problem, we used a two-step similarity reduction procedure to eliminate similar peptides of types I and II:</p>", "<p>Step 1 eliminates similar peptides based on a criterion proposed by Nielsen et al. ##REF##17608956##[7]##. Two peptides are considered similar if they share a 9-mer subsequence. This step will eliminate all similar peptides of type II but is not guaranteed to remove all similar peptides of Type I. For example, this method will not eliminate one of the two peptides in ##FIG##4##Figure 5A## although they share 84.6% sequence similarity.</p>", "<p>Step 2 filters the dataset using an 80% similarity threshold to eliminate any sequence that has a similarity of 80% or greater with one or more sequences in the dataset.</p>", "<p>In addition, we also used a procedure proposed by Raghava ##UREF##2##[20]## for similarity reduction of MHCBench benchmark datasets. Briefly, given two peptides <italic>p</italic>\n₁ and <italic>p</italic>\n₂ of lengths <italic>l</italic>\n₁ and <italic>l</italic>\n₂ such that <italic>l</italic>\n₁≤<italic>l</italic>\n₂, we compare <italic>p</italic>\n₁ with each <italic>l</italic>\n₁-length subpeptide in <italic>p</italic>\n₂. If the percent identity (PID) between <italic>p</italic>\n₁ and any subpeptide in <italic>p</italic>\n₂ is greater than 80%, then the two peptides are deemed to be similar. For example, to compute the PID between (ACDEFGHIKLMNPQRST) and (DEFGGIKLMN), we compare (DEFGGIKLMN) with (ACDEFGHIKL), (CDEFGHIKLM), …, (IKLMNPQRST). The PID between (DEFGGIKLMN) and (DEFGHIKLMN) is 90% since nine out of 10 residues are identical.</p>", "<p>Finally, we explored a method for assigning weights to similar peptides as opposed to eliminating similar peptides from the dataset. Specifically, the peptides within the binders category that are similar to each other (i.e., share a 9-mer subsequence or have sequence similarity of 80% or greater) are clustered together. Each peptide that is assigned to a cluster is similar to at least one other peptide within the cluster, and no two similar peptides are assigned to different clusters. Each peptide in a cluster is assigned a weight of , where <italic>n</italic> is the number of peptides assigned to the cluster. The process is repeated with peptides in the non-binders category. The result is a dataset of <italic>weighted</italic> instances.</p>", "<p>Thus, from each MHC-II benchmark dataset, we generated five versions summarized below:</p>", "<p>Three datasets of <italic>unique</italic> peptides, MHCPEP-UPDS, MHCBN-UPDS, and IEDB-UPDS extracted from MHCPEP, MHCBN, and IEDB, respectively after eliminating short peptides consisting of fewer than 9 residues, unnatural peptides, peptides with greater than 75% Alanine residues, and duplicated peptides.</p>", "<p>Three datasets of <italic>similarity-reduced</italic> peptides, MHCPEP-SRDS1, MHCBN-SRDS1, and IEDB-SRDS1 derived from the corresponding UPDS datasets described above using only step 1 of the two-step similarity reduction procedure described above which ensures that no two peptides in the resulting datasets of binders or non binders share a 9-mer subsequence.</p>", "<p>Three datasets of <italic>similarity-reduced</italic> peptides, MHCPEP-SRDS2, MHCBN-SRDS2, and IEDB-SRDS2, extracted MHCPEP-SRDS1, MHCBN-SRDS1, and IEDB-SRDS1 respectively by filtering the binders and non-binders in SRDS1 such that the sequence identity between any pair of peptides in the binders category or in the non-binders category is less than 80%.</p>", "<p>Three datasets of <italic>similarity-reduced</italic> peptides, MHCPEP-SRDS3, MHCBN-SRDS3, and IEDB-SRDS3, derived from the corresponding UPDS datasets by applying the similarity reduction procedure introduced by Raghava which has been used to construct the MHCBench dataset ##UREF##2##[20]##.</p>", "<p>Three <italic>weighted</italic> unique peptide datasets, MHCPEP-WUPDS, MHCBN-WUPDS, and IEDB-WUPDS, derived from the corresponding UPDS datasets by applying the peptide weighting method described above.</p>", "<p>The procedure used to generate the five different versions of each allele-specific dataset using the different similarity reduction methods and the peptide weighting method described above is shown in ##FIG##5##Figure 6##. Note that UPDS can contain similar peptides of both types I and II; SRDS1 can contain similar peptides of type I; SRDS2 is free from both type I and type II similar peptides; SRDS3 simulates <italic>similarity-reduced</italic> datasets using the method employed with MHCBench; WUPDS is a <italic>weighted</italic> version of the UPDS dataset where similar peptides are grouped into disjoint clusters and the weight of each peptide is set to one over the size of its cluster.</p>", "<title>Summary of the datasets</title>", "<title>Datasets derived from MHCPEP</title>", "<p>\n##TAB##0##Table 1## summarizes the number of binders in each <italic>unique</italic> peptides dataset, MHCPEP-UPDS, and the corresponding three <italic>similarity-reduced</italic> datasets, MHCPEP-SRDS1, MHCPEP-SRDS2, and MHCPEP-SRDS3. Note that on average, the number of binders in the <italic>similarity-reduced</italic> datasets, MHCPEP-SRDS1, MHCPEP-SRDS2, and MHCPEP-SRDS3, is reduced to 48%, 33%, and 39%, respectively, of the number of binders in MHCPEP-UPDS datasets.</p>", "<title>Datasets derived from MHCBN</title>", "<p>\n##TAB##1##Table 2## summarizes the number of binders and non-binders in MHCBN-UPDS, MHCBN-SRDS1, MHCBN-SRDS2 and MHCBN-SRDS3 datasets derived for each of the eight MHCBN alleles satisfying our selection criteria. Note that the average number of binders in <italic>similarity-reduced</italic> datasets, MHCBN-SRDS1, MHCBN-SRDS2, and MHCBN-SRDS3, is reduced to 55.48%, 45.46%, and 61.39%, respectively, of the number of binders in MHCBN-UPDS datasets. Similarly, the average number of non-binders in <italic>similarity-reduced</italic> datasets, MHCBN-SRDS1, MHCBN-SRDS2, and MHCBN-SRDS3, is reduced to 67.55%, 64.24%, and 87.47%, respectively, of the number of non-binders in MHCBN-UPDS datasets.</p>", "<title>Datasets derived from IEDB</title>", "<p>\n##TAB##2##Table 3## summarizes the number of binders and non-binders in IEDB-UPDS, IEDB-SRDS1, IEDB-SRDS2, and IEDB-SRDS3 datasets derived for 12 HLA alleles. We observed that the average number of binders in <italic>similarity-reduced</italic> datasets, IEDB-SRDS1, IEDB-SRDS2, and IEDB-SRDS3, is reduced to 51.17%, 47.66%, and 63.5%, respectively, of the number of binders in MHCBN-UPDS datasets. Similarly, the average number of non-binders in <italic>similarity-reduced</italic> datasets, IEDB-SRDS1, IEDB-SRDS2, and IEDB-SRDS3, is reduced to 60.86%, 59.38%, and 82.9%, respectively, of the number of non-binders in MHCBN-UPDS datasets.</p>", "<title>Independent blind set</title>", "<p>Recently, Wang et al. ##UREF##9##[30]## introduced a comprehensive dataset of previously unpublished MHC-II peptide binding affinities and utilized it to assessing the performance of nine publicly available MHC-II prediction methods. The dataset covers 14 HLA alleles and two Mouse alleles. Out of the 14 HLA allele-specific datasets, five datasets are used in our experiments as independent blind test data to evaluate the performance of the classifiers trained using the corresponding MHCBN allele-specific datasets. ##TAB##18##Table 19## shows the number of test peptides in each allele-specific dataset and the number of binders and non-binders obtained using an IC50 cutoff of 500 nM employed to categorize peptides into binders and non-binders ##REF##17608956##[7]##.</p>", "<title>Prediction methods</title>", "<p>Our experiments focused on two approaches for training MHC-II binding peptide predictors from variable-length MHC-II peptides have been recently proposed in ##REF##16806474##[16]##, ##REF##17105666##[17]## and a method based on <italic>k</italic>-spectrum kernel ##UREF##4##[24]## that is designed to rely on the presence of high degree of sequence similarity between training and test peptides (and hence is expected to perform well on redundant datasets but poorly on <italic>similarity-reduced</italic> datasets). We implemented the three methods in java using Weka machine learning workbench ##UREF##11##[38]##. Brief descriptions of each of the three prediction methods are included below.</p>", "<title>Composition-Transition-Distribution (CTD)</title>", "<p>The basic idea of this approach is to map each variable-length peptide into a fixed-length feature vector such that standard machine learning algorithms are applicable. This method was used and explained in details in ##REF##16806474##[16]##, ##REF##12824396##[39]##. 21 features are extracted from each peptide sequence as follows:</p>", "<p>First, each peptide sequence <italic>p</italic> is mapped into a string <italic>s_p</italic> defined over an alphabet of three symbols, {1,2,3}. The mapping is performed by grouping amino acids into three groups using a physico-chemical property of amino acids (see ##TAB##19##Table 20##). For example the peptide (AIRHIPRRIR) is mapped into (2312321131) using the hydrophobicity division of amino acids into three groups (see ##TAB##19##Table 20##).</p>", "<p>Second, for each peptide string <italic>s_p</italic>, three descriptors are derived as follows:</p>", "<p>Composition (C): three features representing the percent frequency of the symbols, {1, 2, 3}, in the mapped peptide sequence.</p>", "<p>Transition (T): three features representing the percent frequency of <italic>i</italic> followed by <italic>j</italic> or <italic>j</italic> followed by <italic>i</italic>, for <italic>i</italic>, <italic>j</italic>∈{1,2,3}.</p>", "<p>Distribution (D): five features per symbol representing the fractions of the entire sequence where the first, 25, 50, 75, and 100% of the candidate symbol are contained in <italic>s_p</italic>. A total of 15 features are derived from each peptide.</p>", "<p>\n##TAB##19##Table 20## shows division of the 20 amino acids into three groups based on hydrophobicity, polarizability, polarity, and Van der Waal's volume properties. Using these four properties, we derived 84 CTD features from each peptide sequence. In our experiments, we trained SVM classifiers using RBF kernel and peptide sequences represented using their amino acid sequence composition (20 features) and CTD descriptors (84 features).</p>", "<title>Local alignment (LA) kernel</title>", "<p>Local alignment (LA) kernel ##REF##14988126##[40]## is a string kernel designed for biological sequence classification problems. The LA kernel measures the similarity between two sequences by adding up the scores obtained from local alignments with gaps of the sequences. This kernel has several parameters: the gap opening and extension penalty parameters <italic>d</italic> and <italic>e</italic>, the amino acid mutation matrix <italic>s</italic>, and the factor <italic>β</italic> which controls the influence of suboptimal alignments in the kernel value. Saigo et al. ##REF##14988126##[40]## used the BLOSUM62 substitution matrix, gap opening and extending parameters equal 11 and 1, respectively, and <italic>β</italic> ranges from 0.2 to 0.5. In our experiments, we tried a range of values for gap opening/extension and <italic>β</italic> parameters and got the best performance out of LA kernel using BLOSUM62 substitution matrix, gap opening and extending parameters equal 10 and 1, respectively, and <italic>β</italic> = 0.5. Detailed formulation of the LA kernel and a dynamic programming implementation of the kernel are provided in ##REF##14988126##[40]##.</p>", "<title>\n<italic>k</italic>-spectrum kernel</title>", "<p>Intuitively, a <italic>k</italic>-spectrum kernel ##UREF##4##[24]## captures a simple notion of string similarity: two strings are deemed similar (i.e., have a high <italic>k</italic>-spectrum kernel value) if they share many of the same <italic>k</italic>-mer substrings. We used the <italic>k</italic>-spectrum with relatively large <italic>k</italic> value, <italic>k</italic> = 5. As noted earlier, the choice of a relatively large value for <italic>k</italic> was motivated by the desire to construct a predictor that is expected to perform well in settings where the peptides in the test set share significant similarity with one or more peptides in the training set.</p>", "<title>Performance evaluation</title>", "<p>The prediction accuracy (ACC), sensitivity (Sn), specificity (Sp), and correlation coefficient (CC) are often used to evaluate prediction algorithms ##REF##10871264##[26]##. The CC measure has a value in the range from −1 to +1 and the closer the value to +1, the better the predictor. The Sn and Sp summarize the accuracies of the positive and negative predictions respectively. ACC, Sn, Sp, and CC are defined as follows:where <italic>TP</italic>, <italic>FP</italic>, <italic>TN</italic>, <italic>and FN</italic> are the numbers of true positives, false positives, true negatives, and false negatives respectively.</p>", "<p>Although these metrics are widely used to assess the performance of machine learning methods, they all suffer from an important limitation of being threshold-dependent. Threshold-dependent metrics describe the classifier performance at a specific threshold value. It is often possible to increase the number of true positives (equivalently, the sensitivity) of the classifier at the expense of an increase in false positives (equivalently, the false alarm rate). The ROC (Receiver Operating Characteristic) curve shows the performance of the classifier over all possible thresholds. The ROC curve is obtained by plotting the true positive rate as a function of the false positive rate or, equivalently, sensitivity versus (1-specificity) as the discrimination threshold of the binary classifier is varied. Each point on the ROC curve describes the classifier at a certain threshold value and, hence, a particular choice of tradeoff between true positive rate and false negative rate. The area under ROC curve (AUC) is a useful summary statistic for comparing two ROC curves. The AUC is defined as the probability that a randomly chosen positive example will be ranked higher than a randomly chosen negative example. An ideal classifier will have an AUC = 1, while a classifier performs no better than random will have an AUC = 0.5, any classifier performing better than random will have an AUC value that lies between these two extremes.</p>", "<title>Implementation and SVM parameter optimization</title>", "<p>We used the Weka machine learning workbench ##UREF##11##[38]## for implementing the spectrum, and LA kernels (RBF kernel is already implemented in Weka). For the SVM classifier, we used the weka implementation of the SMO algorithm ##UREF##12##[41]##. For <italic>k</italic>-spectrum and LA kernels, the default value of the cost parameter, <italic>C</italic> = 1, was used for the SMO classifier. For the RBF kernel, we found that tuning the SMO cost parameter <italic>C</italic> and the RBF kernel parameter <italic>γ</italic> is necessary to obtain satisfactory performance. We tuned these parameters using a two dimensional grid search over the range <italic>C</italic> = 2⁻⁵,2⁻³,…,2³, <italic>γ</italic> = 2⁻¹⁵,2⁻¹³,…,2³.</p>" ]

[ "<title>Results</title>", "<title>Limitations of the <italic>unique</italic> peptides MHC-II data</title>", "<p>\n##TAB##0##Tables 1##–\n##TAB##2##3## show that MHC-II datasets derived from MHCPEP, MHCBN, and IEDB databases have a large number of highly similar peptides: the number of peptides in the <italic>similarity-reduced</italic> versions in the three benchmark datasets is ≈50% of the original number. In each case, the estimated performance of the prediction methods evaluated on <italic>similarity-reduced</italic> datasets is substantially worse than that estimated using the datasets of unique peptides. This finding is especially significant in light of the fact that MHCPEP and MHCBN datasets have been used for comparing alternative MHC-II peptide prediction methods in most of the published studies ##REF##15349703##[5]##, ##REF##14962912##[6]##, ##REF##14960470##[15]##–##UREF##1##[19]##, ##UREF##5##[25]##.</p>", "<p>For the sake of brevity, we focus discussion here on the results of two representative examples of datasets extracted from the MHCPEP and MHCBN benchmarks and provide the complete set of results in the supplementary materials (##SUPPL##0##Data S1##).</p>", "<p>As shown in ##TAB##3##Table 4##, for the MHCPEP benchmark, we focus on the results on the data for HLA-DR4, which has the largest number of <italic>unique</italic> binders. On the MHCPEP-UPDS version of the HLA-DR4 dataset, the 5-spectrum kernel outperforms the other two prediction methods and CTD outperforms the LA kernel. We notice a substantial drop in the observed performance of the three prediction methods on the <italic>similarity-reduced</italic> and <italic>weighted</italic> datasets relative to that on their UPDS counterpart.</p>", "<p>In the case of the MHCBN benchmark, we focus on the results on the HLA-DRB1*0301 data (##TAB##4##Table 5##) because it has been used in a number of recent studies of MHC-II binding peptide prediction methods ##REF##16806474##[16]##, ##REF##17105666##[17]##, ##UREF##5##[25]##. Most MHCBN allele-specific datasets are unbalanced, i.e., the numbers of binding peptides in the datasets are larger (typically by a factor of 2 to 4) than the corresponding numbers of non-binding peptides (see ##TAB##1##Table 2##). On such unbalanced datasets, classification accuracy can be misleading in terms of providing a reliable and useful assessment of the performance of the classifier. A classifier that simply returns the label of the majority class as the predicted label for each instance to be classified can achieve a rather high accuracy; However such a classifier is rather useless in reliably identifying members of the minority class. Hence, in the case of unbalanced datasets, the correlation coefficient (CC) or the area under the Receiver Operating Characteristic (ROC) curve (AUC) provide more useful measures than accuracy in assessing the performance of the classifiers ##REF##10871264##[26]##. As shown in ##TAB##4##Table 5##, the observed performance of the three prediction methods on HLA-DRB1*0301 MHCBN-UPDS version of this dataset appears to be overly optimistic relative to that on its <italic>similarity-reduced</italic> and <italic>weighted</italic> counterparts. Interestingly, the 5-spectrum kernel is competitive with CTD and LA on the MHCBN-UPDS dataset, whereas its performance on MHCBN-SRDS1 and MHCBN-SRDS2 is much worse than that of the CTD and the LA classifiers.</p>", "<p>Our results also demonstrate that conclusions of superior performance of one method relative to another that are based on estimates of performance obtained using UPDS versions of MHC-II benchmark datasets can be misleading. For example, from results shown in ##TAB##3##Tables 4## and ##TAB##4##5##, one might be tempted to conclude that predictors that use the 5-spectrum kernel are competitive with those that use CTD representation and the LA kernel. However, the 5-spectrum kernel is outperformed by CTD and LA on the <italic>similarity-reduced</italic> datasets. Similarly, conclusions drawn from experiments using the UPDS datasets (##TAB##3##Tables 4## and ##TAB##4##5##) regarding the performance of the CTD and the LA kernel classifiers are contradicted by the their observed performance on the corresponding <italic>similarity-reduced</italic> datasets SRDS1 and SRDS2.</p>", "<title>Limitations of the MHCBench benchmark data</title>", "<p>Comparison of SRDS1, SRDS2, and SRDS3 versions of the datasets used in this study reveals an important limitation of the MHCBench dataset which is a widely used benchmark for comparing MHC-II binding peptide prediction methods.</p>", "<p>Recall that the SRDS3 versions of our datasets are derived using the same procedure that was used in MHCBench to generate <italic>similarity-reduced</italic> datasets. It is clear from the data summarized in ##TAB##0##Tables 1##–\n##TAB##2##3## that the size of a SRDS3 version of a dataset is: often larger than the size of its SRDS2 counterpart, and sometimes larger than the size of its SRDS1 counterpart. Closer examination of the peptides in SRDS3 datasets reveals that SRDS3 datasets may contain several highly similar peptides (e.g., peptides with more than 80% sequence similarity). This is illustrated by the example shown in ##FIG##0##Figure 1##: the two peptides in the SRDS3 version of the HLA-DRB1*0301 dataset share overall sequence similarity of 85.71%. However, the procedure used to construct <italic>similarity-reduced</italic> MHCBench dataset will keep both of these peptides in the resulting dataset because the computed percent identity (PID) between the two peptides is only 7.7%, well below the threshold of 80% PID used to identify similar peptides in MHCBench ##UREF##2##[20]##. Thus, the similarity reduction procedure used in MHCBench dataset (which relies on a strict gapless alignment) may not eliminate all highly similar peptides.</p>", "<p>The preceding observation explains why the number of peptides in the SRDS3 versions of the datasets is usually greater than that in SRDS1 and SRDS2 datasets (see ##TAB##0##Tables 1##–\n##TAB##2##3##). More importantly, because of the presence of a number of highly similar peptides in some SRDS3 datasets, the observed performance of the three prediction methods on the SRDS3 datasets may be overly optimistic relative to that estimated from their SRDS1 and SRDS2 counterparts. Because the classifier using the 5-spectrum kernel in fact relies on the degree of (gapless) match between a sequence pattern present in one or more training sequences and a test sequence, it benefits from the presence of a high degree of similarity between a test sequence and one or more training sequences in ways that the other two classifiers do not. Consequently classifiers that use the 5-spectrum kernel can appear to be competitive with, and perhaps even outperform those that use the CTD representation or the LA kernel when their performance is compared using SRDS3 datasets (and for similar reasons, the MHCBench benchmark data).</p>", "<title>Comparison of the CTD, LA, and the k-spectrum kernel methods</title>", "<p>In machine learning and bioinformatics literature, claims of superiority of one method over another are often based on the outcome of suitable statistical tests. Hence it is interesting to examine the differences in the conclusions obtained when statistical tests are used to compare the performance of prediction methods based on the empirical estimates of their performance on the UPDS, SRDS1, SRDS2, SRDS3, and WPDS versions of the datasets.</p>", "<p>Several non-parametric statistical tests ##UREF##6##[27]##, ##UREF##7##[28]## have been recently recommended for comparing different classifiers on multiple datasets (accounting for the effects of multiple comparisons) ##UREF##8##[29]##. In our analysis, we apply a three-step procedure proposed by Demšar ##UREF##8##[29]##. First, the classifiers to be compared are ranked on the basis of their observed performance (e.g., AUC) on each dataset. Second, the Friedman test is applied to determine whether the measured average ranks are significantly different from the mean rank under the null hypothesis. Third, if the null hypothesis can be rejected at a significance level of 0.05, the Nemenyi test is used to determine whether significant differences exist between any given pair of classifiers.</p>", "<title>Statistical analysis of results on the MHCPEP datasets</title>", "<p>\n##TAB##5##Tables 6##–\n\n\n##TAB##9##10## compare the AUC of the three prediction methods on the five versions of the MHCPEP datasets. For each dataset, the rank of each classifier is shown in parentheses. The last row in each table summarizes the average AUC and rank for each classifier. Demšar ##UREF##8##[29]## has suggested that the average ranks by themselves provide a reasonably fair comparison of classifiers. Interestingly, the LA kernel has the <italic>worst</italic> rank among the three methods when the comparison is based on the observed performance on the UPDS datasets, whereas it has the <italic>best</italic> rank among the three methods when the comparison is based on the <italic>similarity-reduced</italic> or the <italic>weighted</italic> datasets. ##TAB##5##Tables 6##–\n\n\n##TAB##9##10## also show that the rank of the 5-spectrum kernel is competitive with that of CTD on UPDS and SRDS3. This observation is consistent with the presence of a number of highly similar sequences in SRDS3 datasets.</p>", "<p>To determine whether the differences in average ranks are statistically significant, we applied the Friedman test ##UREF##8##[29]## to the rank data in ##TAB##5##Tables 6##–\n\n\n##TAB##9##10##. At significance level of 0.05, the Friedman test did not indicate a statistically significant difference between the methods on the UPDS and WUPDS datasets. However, in the case of the <italic>similarity-reduced</italic> datasets, the Friedman test indicated statistically significant differences between the methods being compared. Thus, we conclude that the three methods are competitive with each other on the UPDS and WUPDS datasets, and that there is at least one pair of classifiers with significant difference in performance on the three versions of <italic>similarity-reduced</italic> datasets. Furthermore, for each version of MHCPEP <italic>similarity-reduced</italic> datasets, the Nemenyi test was applied to determine whether significant differences exist between any given pair of classifiers. ##FIG##1##Figure 2## summarizes the results of the pair-wise comparisons performed using the Nemenyi test. We find that on the SRDS1 versions of the datasets, both the LA and the CTD methods significantly outperform the 5-spectrum kernel and that there are no statistically significant differences between the LA kernel and the CTD classifier. On SRDS2 datasets, we find that, the performance of each of the three methods is significantly different from that of the other two methods, with the LA and the CTD methods ranked first and second, respectively. On SRDS3 datasets, we observe that the performance of the LA kernel is significantly better than that of the CTD and the 5-spectrum classifiers, with no significant differences between the CTD and the 5-spectrum classifiers.</p>", "<title>Statistical analysis of results on the MHCBN and the IEDB datasets</title>", "<p>We summarize the results of applying Demšar's three-step procedure to the results obtained on the five versions of MHCBN and IEDB datasets, respectively. In the case of the MHCBN datasets, ##TAB##10##Tables 11##–\n\n\n##TAB##14##15## show the estimated AUC and rank of each classifier on each dataset. The results of the Freidman test (at a significance level of 0.05) applied to the results in each table did not indicate significant differences in performance among the CTD, the LA, and the 5-spectrum kernel classifiers on the UPDS dataset. However, the test indicated statistically significant differences among the methods in the case of the SRDS1, SRDS2, SRDS3, and the WUPDS datasets. ##FIG##2##Figure 3## summarizes the results of the pair-wise comparisons using the Nemenyi test. In the case of the SRDS1 and the SRDS2 datasets, we find that the performance of both the LA kernel and the CTD classifiers is significantly better than that of the 5-spectrum kernel classifier and that there are no significant differences between the LA kernel and the CTD classifiers. In the case of the SRDS3 datasets, we find that the performance of the LA kernel classifier is significantly better than that of the CTD and the 5-spectrum classifiers, and that no significant differences exist between the CTD and the 5-spectrum classifiers. In the case of the WUPDS datasets, we find that the LA kernel classifier significantly outperforms the 5-spectrun kernel and that there are no significant differences between the LA and the CTD and between the CTD and the 5-spectrum classifiers.</p>", "<p>Results of Demšar's statistical test applied to the IEDB datasets are shown in Tables S46–S50 (##SUPPL##0##Data S1## in supporting information) and ##FIG##3##Figure 4##. As in the case of MHCPEP and MHCBN, we see no significant differences in the performance of different classifiers on IEDB-UPDS datasets. However, in the case of the other datasets, we find at least one pair of classifiers with significant differences in performance. As shown in ##FIG##3##Figure 4##, both the LA and the CTD classifiers significantly outperform the 5-spectrum classifier on the SRDS1 and the SRDS2 versions of the IEDB datasets. However, no significant differences are observed between the CTD and the 5-spectrum methods on the SRDS3 and WUPDS versions of the IEDB datasets.</p>", "<title>Performance on the blind test set</title>", "<p>The results summarized above underscore the importance of <italic>similarity-reduced</italic> MHC-II datasets for obtaining a realistic estimation of the classifier performance and avoiding misleading conclusions. However, one might argue that in practice, when developers of MHC-II binding peptide prediction methods make an implementation of their methods publicly available (e.g., as an online web server or as a web service), it might be better to utilize as much of the available data as possible to train the predictor. Hence, it is interesting to explore whether the UPDS datasets should be preferred over the <italic>similarity-reduced</italic> counterparts to avoid any potential loss of useful information due to the elimination of highly similar peptides in a setting where the goal is to optimize the predictive performance of the classifier on novel peptides. In what follows, we attempt to answer this question using five allele-specific blind test sets ##UREF##9##[30]## to evaluate the performance of the three prediction methods trained on the <italic>unique</italic>, <italic>similarity-reduced</italic>, and <italic>weighted</italic> versions of the MHCBN data for the corresponding alleles.</p>", "<p>\n##TAB##15##Table 16## shows that the 5-spectrum kernel classifier consistently performs poorly (<italic>AUC</italic>≈0.5) on the allele-specific blind test sets regardless of the version of the MHCBN dataset used for training the classifier. This finding is consistent with the cross-validation performance estimates obtained on the MHCBN SRDS1 and SRDS2 datasets (see ##TAB##11##Tables 12## and ##TAB##12##13##).</p>", "<p>\n##TAB##16##Table 17## shows the performance on the blind test sets of the CTD classifiers trained on different versions of MHCBN datasets. Interestingly, the CTD classifiers appear to be relatively insensitive to the choice of the specific version of the MHCBN dataset on which they were trained, with an average AUC≈0.66 in each case.</p>", "<p>Finally, ##TAB##17##Table 18## summarizes the performance on the blind test sets of the LA classifiers trained on the different versions of MHCBN datasets. Interestingly, the best performance (on four out of the five allele-specific blind test sets) is observed in the case of the LA classifiers trained on the SRDS2 versions of the corresponding allele-specific datasets.</p>", "<p>In summary, our results show that MHC-II predictors trained on the similarity reduced versions of the dataset generally outperform those trained on the UPDS dataset. This suggests that similarity reduction contributes to improved generalization on blind dataset.</p>" ]

[ "<title>Discussion</title>", "<title>Related work</title>", "<p>Several previous studies have considered the importance of similarity reduction in datasets of MHC-II peptides. MHCBench ##UREF##2##[20]## is a benchmark of eight HLA-DRB1*0401 datasets representing a set of <italic>unique</italic> peptides (Set1), a dataset of natural peptides (Set2, derived from Set1 by removing peptides with &gt;75% Alanine residues), two non-redundant datasets (Set3a and Set3b derived from Set1 and Set2, respectively), two balanced datasets (Set4a and Set4b derived from Set1 and Set2 by randomly selecting equal numbers of binding and non-binding peptides), and two recent datasets of ligands (Set5a and Set5b, derived from Set1 and Set2 by considering only the most recently reported peptides). However, this benchmark considers only a single MHC-II allele, namely, HLA-DR4 (B1*0401). More importantly, as shown by our analysis of SRDS3 datasets, the similarity reduction procedure used in MHCBench is not stringent enough to ensure elimination of highly similar peptides.</p>", "<p>Nielsen et al. ##REF##14962912##[6]## and Murugan et al. ##REF##16351712##[18]## trained their classifiers using data extracted from MHCPEP and SYFPETHI databases and evaluated the classifiers using ten test sets, from which peptides similar to peptides in the training datasets had been removed. Recently, Nielsen et al. ##REF##17608956##[7]## presented an MHC-II benchmarking dataset for regression tasks: each peptide is labeled with a real value indicating the binding affinity of the peptide. In this benchmark dataset, each set of allele-specific data had been partitioned into five subsets with minimal sequence overlap. However, neither of these studies explicitly examined the limitations of widely used benchmark datasets or the full implications of using MHC-II datasets of unique peptides in evaluating alternative methods.</p>", "<p>Mallios ##REF##12941539##[31]## compared three HLA-DRB1*0101 and HLA-DRB1*0401 prediction tools using an independent test set of two proteins. A consensus approach combining the predictions of the three methods was shown to be superior to the three methods. However, the significance of this result is limited by the small dataset utilized in this study.</p>", "<p>Two recent studies ##UREF##9##[30]##, ##REF##18034454##[32]## have pointed out some of the limitations of existing MHC-II prediction methods in identifying potential MHC-II binding peptides. Gowthaman et al. ##REF##18034454##[32]## used 179 peptides derived from eight antigens and covering seven MHC-II alleles to evaluate the performance of six commonly used MHC-II prediction methods and concluded that none of these methods can reliably identify potential MHC-II binding peptides. Wang et al. ##UREF##9##[30]## introduced a large benchmark dataset of previously unpublished peptides and used it to assess the performance of nine publicly available MHC-II binding peptide prediction methods. Both studies showed that the predictive performance of existing MHC-II prediction tools on independent blind test sets is substantially worse than the performance of these tools reported by their developers. Our work complements these studies by providing a plausible explanation of this result.</p>", "<p>We have shown that the previously reported similarity reduction methods may not eliminate highly similar peptides, i.e., peptides that share &gt;80% sequence identity still <italic>pass</italic> the similarity test. We have proposed a two-step similarity reduction procedure that is much more stringent than those currently in use for similarity reduction with MHC-II benchmark datasets. We have used the similarity reduction method used in MHCBench, as well as our proposed 2-stage method to derive <italic>similarity-reduced</italic> MHC-II benchmark datasets based on peptides retrieved from MHCPEP and MHCBN databases. Comparison of the <italic>similarity-reduced</italic> versions of MHCPEP, MHCBN, and IEDB datasets with their original UPDS counterparts showed that nearly 50% of the peptides in the UPDS datasets are, in fact, highly similar.</p>", "<title>Extensions to multi-class and multi-label prediction problems</title>", "<p>Our description of the proposed similarity reduction procedure assumes a 2-class prediction problem. However, our proposed approach can easily be adapted to multi-class prediction (wherein an instance has associated with one of several mutually exclusive labels). One can simply apply the similarity reduction procedure separately to data from each class.</p>", "<p>A more interesting setting is that of multi-label prediction (wherein each instance is associated with a subset of a set of candidate labels). Consider for example, the problem of predicting promiscuous MHC binding peptides ##REF##17303158##[33]##, where each peptide can bind to multiple HLA molecules. Current methods for multi-label prediction typically reduce the multi-label prediction task to a collection of binary prediction tasks ##UREF##10##[34]##. Hence, the similarity reduction methods proposed in this paper can be directly applied to the binary labeled datasets resulting from such a reduction.</p>", "<title>Implications for rigorous assessment of MHC-II binding peptide prediction methods</title>", "<p>The results of our study show that the observed performance of some of the methods (e.g., the CTD and the LA kernels) on benchmark datasets of <italic>unique</italic> peptides can be rather optimistic relative to the performance of the same methods on <italic>similarity-reduced</italic> counterparts of the same datasets or on blind test sets. This suggests that the performance of existing MHC-II prediction methods, when applied to novel peptide sequences, may turn out to be less satisfactory than one might have been led to believe based on the reported performance of such methods on some of the widely used benchmark. Moreover, the conclusions based on observed performance on datasets of <italic>unique</italic> peptides regarding the superior performance of one method relative to another can be highly unreliable in more realistic settings e.g., predictions of novel peptides.</p>", "<p>These results underscore the importance of rigorous comparative evaluation of a broad range of existing methods for MHC-II binding peptides prediction methods using <italic>similarity-reduced</italic> datasets. We expect that such studies are likely to show much greater room for improvement over the state-of-the-art MHC-II prediction tools than one might be led to believe based on reported performance on the widely-used benchmark datasets and motivate the research community to develop improved methods for this important task. We hope that such comparisons will be facilitated by the availability of the <italic>similarity-reduced</italic> versions of MHCPEP, MHCBN, and IEDB datasets used in our experiments. These datasets (##SUPPL##1##Datasets S1##, ##SUPPL##2##S2## and ##SUPPL##3##S3##), Java source code implementation of the similarity reduction and weighting procedures (##SUPPL##4##Code S1##), and the supplementary materials (##SUPPL##0##Data S1##) have been made freely available (see Supporting Information).</p>" ]

[]

[ "<p>Conceived and designed the experiments: YEM. Performed the experiments: YEM. Analyzed the data: YEM DD VH. Contributed reagents/materials/analysis tools: YEM DD VH. Wrote the paper: YEM DD VH.</p>", "<p>Choice of one method over another for MHC-II binding peptide prediction is typically based on published reports of their estimated performance on standard benchmark datasets. We show that several standard benchmark datasets of <italic>unique peptides</italic> used in such studies contain a substantial number of peptides that share a high degree of sequence identity with one or more other peptide sequences in the same dataset. Thus, in a standard cross-validation setup, the test set and the training set are likely to contain sequences that share a high degree of sequence identity with each other, leading to overly optimistic estimates of performance. Hence, to more rigorously assess the relative performance of different prediction methods, we explore the use of <italic>similarity-reduced</italic> datasets. We introduce three <italic>similarity-reduced</italic> MHC-II benchmark datasets derived from MHCPEP, MHCBN, and IEDB databases. The results of our comparison of the performance of three MHC-II binding peptide prediction methods estimated using datasets of <italic>unique</italic> peptides with that obtained using their <italic>similarity-reduced</italic> counterparts shows that the former can be rather optimistic relative to the performance of the same methods on <italic>similarity-reduced</italic> counterparts of the same datasets. Furthermore, our results demonstrate that conclusions regarding the superiority of one method over another drawn on the basis of performance estimates obtained using commonly used datasets of <italic>unique</italic> peptides are often contradicted by the observed performance of the methods on the <italic>similarity-reduced</italic> versions of the same datasets. These results underscore the importance of using <italic>similarity-reduced</italic> datasets in rigorously comparing the performance of alternative MHC-II peptide prediction methods.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>We thank anonymous reviewers for their thoughtful and constructive reviews. We thank Fadi Towfic for helpful comments on the manuscript.</p>" ]

[ "<fig id=\"pone-0003268-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.g001</object-id><label>Figure 1</label><caption><title>Example of two peptides from MHCBN-SRDS3 HLA-DRB1*0301 dataset.</title><p>Although the two peptides share 85.71% sequence similarity, the computed percent identity (PID) used to define the similarity between these two peptides in MHCBench benchmark is only 7.7%.</p></caption></fig>", "<fig id=\"pone-0003268-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.g002</object-id><label>Figure 2</label><caption><title>Pair-wise comparisons of classifiers with the Nemenyi test applied to results on a) MHCPEP-SRDS1, b) MHCPEP-SRDS2, and c) MHCPEP-SRDS3.</title><p>Classifiers that are not significantly different (at p-value = 0.05) are connected.</p></caption></fig>", "<fig id=\"pone-0003268-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.g003</object-id><label>Figure 3</label><caption><title>Pair-wise comparisons of classifiers with the Nemenyi test applied to results on a) MHCBN-SRDS1, b) MHCBN-SRDS2, c) MHCBN-SRDS3, and d) MHCBN-WUPDS.</title><p>Classifiers that are not significantly different (at p-value = 0.05) are connected.</p></caption></fig>", "<fig id=\"pone-0003268-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.g004</object-id><label>Figure 4</label><caption><title>Pair-wise comparisons of classifiers with the Nemenyi test applied to results on a) IEDB-SRDS1, b) IEDB-SRDS2, c) IEDB-SRDS3, and d) IEDB-WUPDS.</title><p>Classifiers that are not significantly different (at p-value = 0.05) are connected.</p></caption></fig>", "<fig id=\"pone-0003268-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.g005</object-id><label>Figure 5</label><caption><title>Two types of similar peptides that frequently appear in MHC databases.</title></caption></fig>", "<fig id=\"pone-0003268-g006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.g006</object-id><label>Figure 6</label><caption><title>An overview of the process used for generating five different versions of each allele dataset using the different similarity-reduction methods described in the text.</title></caption></fig>" ]

[ "<table-wrap id=\"pone-0003268-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t001</object-id><label>Table 1</label><caption><title>Number of binding peptides in MHCPEP benchmark dataset.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">UPDS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS3</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">113</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">67</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">32</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">39</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">84</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">79</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">82</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">135</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">73</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">75</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">703</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">336</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">242</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">278</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">315</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">148</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">104</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">134</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">192</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">81</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">69</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">73</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1085</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">439</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">298</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">353</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">189</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">92</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">61</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">75</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">341</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">137</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">87</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">101</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">125</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">47</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">46</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">54</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">94</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">41</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">37</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">473</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">160</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">103</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">121</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">68</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">36</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">121</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">48</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">36</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">49</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">158</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">82</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">45</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">115</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">45</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">39</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">55</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ab</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">136</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">62</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">61</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ad</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">415</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">168</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">135</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ag7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">157</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">62</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">53</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">81</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ak</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">254</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">96</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">67</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">85</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ed</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">294</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">188</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">68</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">76</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ek</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">334</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">204</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">78</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t002</object-id><label>Table 2</label><caption><title>Number of binding/non-binding peptides in MHCBN benchmark dataset.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">UPDS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS3</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">636/180</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">328/111</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">223/106</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">259/130</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">416/168</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">197/124</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">153/123</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">232/149</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">218/173</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">111/131</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">80/129</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100/154</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">531/127</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">325/88</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">279/76</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">390/112</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">261/230</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">137/150</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">127/145</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">175/215</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">805/201</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">471/136</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">404/119</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">543/174</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">292/107</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">179/68</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">152/66</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">213/92</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB11101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">352/137</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">213/87</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">182/87</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">239/131</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t003</object-id><label>Table 3</label><caption><title>Number of binding/non-binding peptides in IEDB benchmark dataset.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">UPDS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS3</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1105/432</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">645/268</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">623/261</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">938/365</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">135/556</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">78/292</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">69/276</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">81/396</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">317/412</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">197/262</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">176/255</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">215/340</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0404</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">113/132</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">69/100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">62/98</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">74/109</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0405</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">113/119</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">74/85</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">70/84</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">81/89</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">228/302</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">147/203</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">137/202</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">173/274</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0802</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65/120</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">46/101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">46/100</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">49/108</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-1101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">197/411</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">122/218</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">111/212</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">139/328</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-1302</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">152/103</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">105/81</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97/81</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">110/92</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-1501</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">269/283</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">165/176</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">142/174</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">185/260</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB4-0101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">92/215</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64/120</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">63/119</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">85/200</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB5-0101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">215/377</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">123/201</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">113/194</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">147/309</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t004</object-id><label>Table 4</label><caption><title>Performance of prediction methods on MHCPEP HLA-DR4 <italic>unique</italic>, <italic>similarity-reduced</italic>, and <italic>weighted</italic> datasets using 5-fold cross-validation test.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Dataset</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Method</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">ACC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sn</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sp</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AUC</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">UPDS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">86.59</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">73.36</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99.82</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.759</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.906</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">77.10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">71.71</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">82.49</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.545</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.862</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">90.55</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">81.29</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99.82</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.825</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.917</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">69.93</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">66.06</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">73.80</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.400</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.723</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">68.56</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">63.78</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">73.35</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.373</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.751</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">70.96</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">43.28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">98.63</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.503</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.710</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64.77</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">60.40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">69.13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.296</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.692</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64.43</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65.10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">63.76</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.289</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.711</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">56.04</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">33.22</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">78.86</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.136</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.578</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65.01</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">61.47</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">68.56</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.695</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64.02</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">62.89</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65.16</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.281</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.717</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">68.56</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">38.81</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">98.30</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.462</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.679</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">WUPDS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">85.41</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">31.98</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99.91</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.516</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.730</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">79.38</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">22.50</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">94.81</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.249</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.723</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">87.14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">42.14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99.35</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.580</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.723</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t005</object-id><label>Table 5</label><caption><title>Performance of prediction methods on MHCBN HLA-DRB1*0301 <italic>unique</italic>, <italic>similarity-reduced</italic>, and <italic>weighted</italic> datasets using 5-fold cross-validation test.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Dataset</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Method</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">ACC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sn</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sp</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AUC</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">UPDS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72.51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">73.95</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">70.87</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.448</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.787</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">71.89</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">73.56</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">70.00</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.436</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.795</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">70.26</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">82.76</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">56.09</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.405</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.770</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">63.41</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64.23</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">62.67</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.269</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.661</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">58.54</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">59.85</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">57.33</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.172</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.617</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">42.16</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">63.50</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">22.67</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.152</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.323</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">59.93</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">59.06</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">60.69</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.197</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.628</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">55.88</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">54.33</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">57.24</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.116</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.563</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">35.29</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">37.01</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">33.79</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.292</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.273</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64.62</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">60.57</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">67.91</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.285</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.675</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">67.18</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">61.14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72.09</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.334</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.736</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">63.08</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">49.71</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">73.95</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.244</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.678</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">WUPDS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65.27</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">61.04</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">68.88</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.300</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.678</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">66.66</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64.47</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">68.53</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.330</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.710</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">59.97</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">58.70</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">61.04</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.197</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.648</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t006</object-id><label>Table 6</label><caption><title>AUC values for the three methods evaluated on MHCPEP-UPDS datasets.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.908(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.905(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.939(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.628(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.903(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.934(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.856(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.860(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.853(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.883(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.872(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.863(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.884(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.866(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.829(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.854(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.869(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.862(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.917(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.862(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.906(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.905(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.864(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.887(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.916(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.858(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.904(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.894(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.896(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.903(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.836(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.880(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.913(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.910(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.938(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.958(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.875(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.905(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.920(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.887(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.829(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.867(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.907(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.907(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.934(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.924(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.891(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.886(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ab</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.865(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.855(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.875(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ad</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.942(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.898(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.902(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ag7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.916(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.896(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.887(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ak</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.909(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.872(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.881(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ed</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.918(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.921(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.936(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ek</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.934(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.940(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.951(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.885(1.91)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.886(2.27)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.900(1.77)</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t007\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t007</object-id><label>Table 7</label><caption><title>AUC values for the three methods evaluated on MHCPEP-SRDS1 datasets.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.789(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.852(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.853(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.544(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.854(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.881(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.677(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.799(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.726(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.662(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.801(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.744(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.694(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.795(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.781(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.603(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.678(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.572(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.710(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.751(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.723(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.691(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.776(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.784(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.721(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.702(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.732(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.552(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.625(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.694(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.620(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.746(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.721(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.703(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.912(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.890(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.746(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.827(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.837(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.711(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.718(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.667(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.789(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.806(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.876(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.651(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.788(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.603(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ab</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.620(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.705(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.680(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ad</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.787(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.818(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.804(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ag7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.718(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.778(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.702(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ak</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.761(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.800(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.796(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ed</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.826(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.903(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.932(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ek</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.874(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.913(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.941(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.702(2.77)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.789(1.45)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.770(1.77)</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t008\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t008</object-id><label>Table 8</label><caption><title>AUC values for the three methods evaluated on MHCPEP-SRDS2 datasets.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.566(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.678(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.573(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.590(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.954(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.817(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.616(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.713(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.709(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.562(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.715(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.711(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.548(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.633(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.614(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.514(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.602(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.572(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.578(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.711(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.692(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.583(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.622(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.625(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.562(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.622(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.599(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.526(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.717(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.680(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.488(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.754(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.690(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.528(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.810(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.792(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.518(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.827(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.587(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.592(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.698(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.689(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.568(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.612(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.550(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.578(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.664(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.595(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ab</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.570(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.624(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.638(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ad</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.623(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.700(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.618(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ag7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.713(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.756(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.632(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ak</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.586(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.664(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.661(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ed</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.645(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.760(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.744(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ek</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.606(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.756(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.703(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.575(2.86)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.709(1.09)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.659(2.05)</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t009\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t009</object-id><label>Table 9</label><caption><title>AUC values for the three methods evaluated on MHCPEP-SRDS3 datasets.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.663(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.655(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.754(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.608(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.900(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.900(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.699(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.757(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.706(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.676(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.747(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.720(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.724(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.736(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.686(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.623(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.657(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.532(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.679(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.717(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.695(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.719(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.723(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.617(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.631(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.765(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.613(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.608(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.732(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.714(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.520(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.779(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.792(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.544(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.854(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.850(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.563(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.623(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.630(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.805(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.713(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.663(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.629(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.769(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.682(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.800(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.780(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.672(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ab</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.606(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.611(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.618(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ad</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.821(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.785(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.676(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ag7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.823(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.804(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.757(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ak</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.768(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.766(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.691(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ed</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.828(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.852(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.787(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ek</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.714(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.789(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.699(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.684(2.23)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.751(1.45)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.702(2.27)</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t010\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t010</object-id><label>Table 10</label><caption><title>AUC values for the three methods evaluated on MHCPEP-WUPDS datasets.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.717(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.738(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.772(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.543(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.882(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.925(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DQ7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.716(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.786(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.812(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.696(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.710(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.699(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.682(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.688(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.617(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.612(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.678(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.614(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.723(2.5)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.723(2.5)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.730(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.765(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.709(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.733(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.714(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.599(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.632(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.796(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.810(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.802(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.806(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.819(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.738(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.612(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.798(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.830(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.620(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.714(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.605(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.760(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.627(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.587(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.747(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.760(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.679(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.838(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.786(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.718(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ab</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.650(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.669(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.636(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ad</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.815(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.740(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.707(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ag7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.820(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.797(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.700(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ak</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.778(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.684(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.680(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ed</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.742(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.760(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.805(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">I-Ek</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.734(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.824(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.805(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.722(2.11)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.741(1.7)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.719(2.18)</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t011\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t011</object-id><label>Table 11</label><caption><title>AUC values for the three methods evaluated on MHCBN-UPDS datasets.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.747(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.768(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.789(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.806(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.771(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.786(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.743(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.748(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.752(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.758(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.799(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.804(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.770(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.795(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.787(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.705(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.780(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.721(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.778(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.842(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.732(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB11101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.754(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.874(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.832(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.758(2.63)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.797(1.63)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.775(1.75)</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t012\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t012</object-id><label>Table 12</label><caption><title>AUC values for the three methods evaluated on MHCBN-SRDS1 datasets.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.545(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.784(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.738(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.456(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.707(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.750(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.533(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.657(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.692(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.456(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.690(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.748(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.323(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.617(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.661(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.381(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.676(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.655(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.424(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.665(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.748(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB11101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.493(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.776(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.759(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.451(3.00)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.697(1.63)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.719(1.38)</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t013\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t013</object-id><label>Table 13</label><caption><title>AUC values for the three methods evaluated on MHCBN-SRDS2 datasets.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.448(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.717(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.698(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.374(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.665(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.716(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.369(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.459(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.588(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.351(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.705(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.683(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.273(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.563(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.628(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.261(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.658(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.620(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.414(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.617(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.696(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB11101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.386(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.705(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.757(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.360(3.00)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.636(1.63)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.673(1.38)</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t014\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t014</object-id><label>Table 14</label><caption><title>AUC values for the three methods evaluated on MHCBN-SRDS3 datasets.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.685(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.768(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.743(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.709(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.741(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.719(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.557(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.616(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.608(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.691(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.819(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.725(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.678(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.736(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.675(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.624(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.760(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.710(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.737(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.794(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.671(3)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB11101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.755(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.816(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.775(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.680(2.75)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.756(1.00)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.703(2.25)</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t015\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t015</object-id><label>Table 15</label><caption><title>AUC values for the three methods evaluated on MHCBN-WUPDS datasets.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5-spectrum</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CTD</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.655(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.747(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.732(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.636(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.717(2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.740(1)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DR5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.518(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.594(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.543(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.535(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.672(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.666(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.648(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.710(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.678(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.536(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.757(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.701(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB10701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.667(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.724(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.702(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB11101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.676(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.820(1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.789(2)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.609(3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.718(1.13)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.694(1.88)</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t016\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t016</object-id><label>Table 16</label><caption><title>AUC values for 5-spectrum based classifiers trained using MHCBN- UPDS, SRDS1, SRDS2, SRDS3, and WUPDS datasets and evaluated on the blind test sets of Wang et al. ##UREF##9##[30]##.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">UPDS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">WUPDS</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.505</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.503</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.504</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.506</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.505</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.518</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.515</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.515</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.516</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.518</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.504</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.501</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.487</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.496</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-1101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.496</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.505</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.504</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.504</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.505</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.500</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t017\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t017</object-id><label>Table 17</label><caption><title>AUC values for CTD classifiers trained using MHCBN- UPDS, SRDS1, SRDS2, SRDS3, and WUPDS datasets and evaluated on the blind test sets of Wang et al. ##UREF##9##[30]##.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">UPDS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">WUPDS</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.689</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.707</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.684</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.714</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.629</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.595</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.589</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.597</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.596</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.585</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.605</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.584</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.611</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.633</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.601</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.675</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.711</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.699</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.684</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.694</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-1101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.732</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.719</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.713</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.735</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.659</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.658</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.662</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.668</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.649</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t018\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t018</object-id><label>Table 18</label><caption><title>AUC values for LA classifiers trained using MHCBN- UPDS, SRDS1, SRDS2, SRDS3, and WUPDS datasets and evaluated on the blind test sets of Wang et al. ##UREF##9##[30]##.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">UPDS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SRDS3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">WUPDS</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.675</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.650</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.756</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.736</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.703</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.604</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.647</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.651</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.637</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.604</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.554</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.548</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.610</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.595</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.573</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.627</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.692</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.692</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.677</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.627</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-1101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.775</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.722</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.730</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.775</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.647</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.652</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.682</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.675</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.656</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t019\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t019</object-id><label>Table 19</label><caption><title>Five allele-specific blind test set obtained ##UREF##9##[30]##.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Allele</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">peptides</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">binders</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">non-binders</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3882</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2579</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1303</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0301</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">502</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">209</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">293</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">512</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">286</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">226</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-0701</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">505</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">358</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">147</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">HLA-DRB1-1101</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">520</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">317</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">203</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003268-t020\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003268.t020</object-id><label>Table 20</label><caption><title>Categorization of amino acids into three groups for a number of physicochemical properties ##UREF##13##[42]##.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Property</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Group 1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Group 2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Group 3</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Hydrophobicity</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">RKEDQN</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">GASTPHY</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CVLIMFW</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Polarizability</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">GASCTPD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NVEQIL</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">MHKFRYW</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Polarity</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">LIFWCMVY</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">PATGS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">HQRKNED</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Van der Waal's volume</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">GASDT</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CPNVEQIL</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">KMHFRYW</td></tr></tbody></table></alternatives></table-wrap>" ]

[ "<inline-formula></inline-formula>", "<disp-formula></disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pone.0003268.s001\"><label>Data S1</label><caption><p>Detailed results on MHCPEP, MHCBN, and IEDB datasets</p><p>(0.12 MB PDF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003268.s002\"><label>Dataset S1</label><caption><p>Datasets derived from MHCPEP database</p><p>(0.36 MB ZIP)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003268.s003\"><label>Dataset S2</label><caption><p>Datasets derived from MHCBN database</p><p>(0.17 MB ZIP)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003268.s004\"><label>Dataset S3</label><caption><p>Datasets derived from IEDB database</p><p>(0.26 MB ZIP)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003268.s005\"><label>Code S1</label><caption><p>Java programs implementing the similarity-reduction and peptide weighting methods</p><p>(0.01 MB ZIP)</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"nt101\"><p>UPDS refers to datasets of non-redundant peptides. The last three columns refer to <italic>similarity-reduced</italic> datasets (see text for details).</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt102\"><p>UPDS refers to datasets of non-redundant peptides. The last three columns refer to <italic>similarity-reduced</italic> datasets (see text for details).</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt103\"><p>UPDS refers to datasets of non-redundant peptides. The last three columns refer to <italic>similarity-reduced</italic> datasets (see text for details).</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt104\"><p>For each dataset, the rank of each classifier is shown in parentheses.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt105\"><p>For each dataset, the rank of each classifier is shown in parentheses.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt106\"><p>For each dataset, the rank of each classifier is shown in parentheses.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt107\"><p>For each dataset, the rank of each classifier is shown in parentheses.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt108\"><p>For each dataset, the rank of each classifier is shown in parentheses.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt109\"><p>For each dataset, the rank of each classifier is shown in parentheses.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt110\"><p>For each dataset, the rank of each classifier is shown in parentheses.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt111\"><p>For each dataset, the rank of each classifier is shown in parentheses.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt112\"><p>For each dataset, the rank of each classifier is shown in parentheses.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt113\"><p>For each dataset, the rank of each classifier is shown in parentheses.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt114\"><p>Peptides are categorized into binders and non-binders using an IC50 cutoff 500 nM.</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This research was supported in part by a doctoral fellowship from the Egyptian Government to Yasser EL-Manzalawy and a grant from the National Institutes of Health (GM066387) to Vasant Honavar.</p></fn></fn-group>" ]

[ "<graphic id=\"pone-0003268-t001-1\" xlink:href=\"pone.0003268.t001\"/>", "<graphic id=\"pone-0003268-t002-2\" xlink:href=\"pone.0003268.t002\"/>", "<graphic id=\"pone-0003268-t003-3\" xlink:href=\"pone.0003268.t003\"/>", "<graphic id=\"pone-0003268-t004-4\" xlink:href=\"pone.0003268.t004\"/>", "<graphic id=\"pone-0003268-t005-5\" xlink:href=\"pone.0003268.t005\"/>", "<graphic xlink:href=\"pone.0003268.g001\"/>", "<graphic id=\"pone-0003268-t006-6\" xlink:href=\"pone.0003268.t006\"/>", "<graphic id=\"pone-0003268-t007-7\" xlink:href=\"pone.0003268.t007\"/>", "<graphic id=\"pone-0003268-t008-8\" xlink:href=\"pone.0003268.t008\"/>", "<graphic id=\"pone-0003268-t009-9\" xlink:href=\"pone.0003268.t009\"/>", "<graphic id=\"pone-0003268-t010-10\" xlink:href=\"pone.0003268.t010\"/>", "<graphic xlink:href=\"pone.0003268.g002\"/>", "<graphic xlink:href=\"pone.0003268.g003\"/>", "<graphic id=\"pone-0003268-t011-11\" xlink:href=\"pone.0003268.t011\"/>", "<graphic id=\"pone-0003268-t012-12\" xlink:href=\"pone.0003268.t012\"/>", "<graphic id=\"pone-0003268-t013-13\" xlink:href=\"pone.0003268.t013\"/>", "<graphic id=\"pone-0003268-t014-14\" xlink:href=\"pone.0003268.t014\"/>", "<graphic id=\"pone-0003268-t015-15\" xlink:href=\"pone.0003268.t015\"/>", "<graphic xlink:href=\"pone.0003268.g004\"/>", "<graphic id=\"pone-0003268-t016-16\" xlink:href=\"pone.0003268.t016\"/>", "<graphic id=\"pone-0003268-t017-17\" xlink:href=\"pone.0003268.t017\"/>", "<graphic id=\"pone-0003268-t018-18\" xlink:href=\"pone.0003268.t018\"/>", "<graphic xlink:href=\"pone.0003268.g005\"/>", "<inline-graphic xlink:href=\"pone.0003268.e001.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003268.g006\"/>", "<graphic id=\"pone-0003268-t019-19\" xlink:href=\"pone.0003268.t019\"/>", "<graphic id=\"pone-0003268-t020-20\" xlink:href=\"pone.0003268.t020\"/>", "<graphic xlink:href=\"pone.0003268.e002.jpg\" mimetype=\"image\" position=\"float\"/>" ]

[ "<media xlink:href=\"pone.0003268.s001.pdf\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003268.s002.zip\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003268.s003.zip\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003268.s004.zip\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003268.s005.zip\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Jevons found that he could estimate the number of beans in a box without error when there were four or fewer, but became increasingly inaccurate as the number of beans increased beyond four ##UREF##0##[1]##. Subsequent studies have confirmed his findings, and it is now generally assumed that the immediate and accurate apprehension of the numerosity of collections of four or fewer objects uses a process separate from enumerating larger collections ##REF##8473843##[2]##–##UREF##1##[6]##. Following Kaufman and colleagues, this process is called “subitizing” ##REF##15392567##[7]##.</p>", "<p>The current basis for this distinction has come from a discontinuity in the slope of the curve that relates enumeration time to the number of items to be enumerated. Enumeration in the “subitizing range” (1 to 3 or 4 items) typically yields a shallow slope whereas the slope for 5 items and above (the “counting range”) is considerable steeper. This pattern has traditionally been fitted with a bilinear function and two functionally separate enumeration mechanisms have been inferred (see ##REF##8473843##[2]## for a review). Furthermore, by analogy with classical studies of visual search ##REF##7351125##[8]##, a parallel and pre-attentive process has been inferred from the shallow slope for subitizing (equivalent to pop-out search) and a serial and attentive process from the steeper slope (equivalent to conjunction search) for counting ##REF##8768186##[3]##, ##REF##6535985##[9]##.</p>", "<p>Support for this distinction has come from brain imaging studies that show quantitative differences in parietal lobe activity for the counting range as compared with the subitizing range ##REF##9950713##[4]##, ##UREF##1##[6]##. More specific evidence for a pre-attentive subitizing mechanism has come from a neuropsychological study of neglect patients ##REF##10825363##[10]##. Neglect patients with extinction, who cannot report items in the contra-lesional field due to their inability to attend to this side of space, can nevertheless enumerate up to four objects when two of them are in the neglected field ##REF##10825363##[10]##.</p>", "<p>However, one brain-imaging study has failed to distinguish between the neural substrates of subitizing and counting, and found instead that human parietal cortex activation increased linearly with the number of items ##REF##11798277##[11]##. Balakrishnan and Ashby questioned the basis of the initial inference of two mechanisms from the performance data by demonstrating that a bilinear fit is unjustified and a continuous model of enumeration is equally supported by the performance data ##REF##1780203##[12]##, ##REF##1620801##[13]##.</p>", "<p>Moreover, the strong notion of pre-attentive/attentive dichotomy has been regarded as an oversimplified account in the attention literature (e.g. ##REF##2756067##[14]##, ##REF##11245840##[15]##) and particularly the hypothesis of attention-free perceptual processing has been questioned ##REF##11561922##[16]##, ##UREF##2##[17]##. Indeed there is evidence that even the simplest forms of feature detection (e.g. orientation detection), which had previously been thought of as occurring pre-attentively, depend on the availability of attentional resources in a dual-task situation ##REF##9194560##[18]##.</p>", "<p>In this study, we investigated how the judgement of both small and large numerosities is affected by a withdrawal of attentional resources, and more specifically, we tested the hypothesis that subitizing is a pre-attentive process. We reasoned that if subitizing is pre-attentive, it should be unaffected by experimental manipulations such as dual-task paradigms that reduce the availability of attentional resources ##REF##11561922##[16]##, ##REF##9194560##[18]##. In addition to imposing an additional task onto a numerosity judgement task, we employed the framework of load theory ##REF##7790827##[19]##, ##REF##15668100##[20]##. Load theory states that in a dual task situation, processing of secondary task stimuli depends on the attentional requirements of the primary task. Under high attentional load, processing capacity is entirely dedicated to the primary task leading to reduced (and sometimes eliminated) processing of the secondary task. Under low attentional load, however, the capacity limit is not reached and attentional resources “spill over” to perform the secondary task.</p>", "<p>In this experiment, we combined a secondary numerosity judgement task with a primary task with two levels of attentional load (low and high load). We predicted that if subitizing is a pre-attentive process, it should not be affected by dual versus single task manipulations and, more importantly, subitizing should not be affected by attentional load. However, if subitizing is constrained by attentional capacity, it should be compromised by both experimental manipulations.</p>" ]

[ "<title>Methods</title>", "<title>Subjects</title>", "<p>14 subjects (mean age: 23.1, 10 females) with normal or corrected-to-normal vision participated. All gave written informed consent and were paid for their participation. The study was approved by the ethics committee of the Dept. of Psychology at UCL.</p>", "<title>Visual stimulus</title>", "<p>The visual stimulus consisted of: (i) a central diamond shape (4° of visual angle) comprising 4 coloured triangles and (ii) a circle of gabor patches (10°) on a grey background (see example stimulus in ##FIG##0##Fig. 1a##). Eight different colour combinations were used for the central diamond shape (##FIG##0##Fig. 1b##). The gabor patches (2° each) in the circle were either vertically oriented high-contrast (100%) targets or horizontally orientated low-contrast (50%) distractors. The distance between patches was equal, patches occupied a different position in the circle in each trial and positions of targets and distractors within the circle were randomly assigned. The grey value of the background was adjusted to mid-grey and gamma corrected for output luminance (as was the gabor value). Stimuli were generated using the Cogent toolbox (<ext-link ext-link-type=\"uri\" xlink:href=\"http://www.vislab.ucl.ac.uk/Cogent/\">www.vislab.ucl.ac.uk/Cogent/</ext-link>) for MATLAB (Mathworks, Inc).</p>", "<title>Task and experimental procedure</title>", "<p>We employed a dual task paradigm. The primary task was a speeded target detection task at fovea which implemented the manipulation of attentional load. Under low load, subjects detected a simple feature (the colour red, independent of spatial arrangement), whereas under high load, subjects detected specific conjunctions of colour and spatial arrangement: either two green triangles aligned along the right-tilted diagonal or two yellow triangles aligned along the left-tilted diagonal (see ##FIG##0##Fig 1b##). Importantly, subjects were instructed not to respond to the opposite combinations. Both low and high load condition consisted of the same set of stimuli, only the task instructions changed.</p>", "<p>As a secondary task, subjects judged the number of targets ranging from 1 to 8. Total number of items in the circle ranged from 9 to13, counterbalanced for each target number and load condition. Distractors were used to de-correlate task difficulty from the overall processing effort required for multiple stimuli. The number of distractors did not co-vary with the number of targets. Therefore, numerosity judgement could be made neither on the basis of the total number of items present nor on the basis of the number of distractors. As distractors were equally luminous than targets, numerosity could not be judged based on overall luminance either.</p>", "<p>After a fixation cross (1s), the stimulus was displayed for 200 ms, followed by a mask which stayed on the screen until subjects responded (##FIG##0##Fig. 1c##). Inter-trial intervals varied randomly between 1 and 2 seconds. Note that short stimulus durations prevented verbal counting.</p>", "<p>Subjects always responded first to the primary task and subsequently to the secondary task, ensuring that attentional resources were manipulated by the processing requirements of the primary task and not by the number of items in the secondary task. Subjects responded with their right hand on two adjacent keys to the primary task and with their left hand to the secondary task using number keys 1–8.</p>", "<p>Overall, accuracy was emphasised over speed. Subjects were given practice trials before each block and had the opportunity to take breaks. The testing session lasted 1h.</p>", "<title>Experimental design</title>", "<p>Each colour combination of the primary task was combined once with each target numerosity of the secondary task, resulting in 64 trials per block.</p>", "<p>Subjects first performed 2 blocks of each task under single task condition (1 block low load, 1 block high load). Subjects were therefore well trained in each of the two tasks before being tested under dual task conditions. 4 blocks of dual task were performed (2 low and 2 high load in the order ABBA or BAAB, counterbalanced across subjects). Each subject performed 16 trials per target number per experimental condition (512 trials for the whole experiment).</p>" ]

[ "<title>Results</title>", "<title>Primary task–Load manipulation (##FIG##1##Fig. 2##)</title>", "<p>Reaction time and accuracy data of the primary task were compared using a repeated measures ANOVA with within subject factors “load condition” (low load vs. high load) and “task” (single task vs. dual task). As expected, subjects responded more slowly under high attentional load compared to low attentional load (<italic>F</italic> (1,13) = 114.57, <italic>p</italic>&lt;.001) and significantly less accurately (<italic>F</italic> (1,13) = 20.26, <italic>p</italic> = .001 ). Subjects were also slower under dual-task conditions (<italic>F</italic>(1,13) = 97.77, <italic>p</italic>&lt;.001) and less accurate (<italic>F</italic>(1,13) = 37.01, <italic>p</italic>&lt;.001) compared to single task conditions. These results confirm that our manipulation of attentional load was effective.</p>", "<title>Secondary task–Numerosity judgement</title>", "<title>Accuracy (##FIG##2##Fig. 3a##)</title>", "<p>Due to the sequential key responses to the primary and secondary task, reaction time data of the secondary (numerosity) task was not very meaningful and is not reported here.</p>", "<p>Overall, enumeration accuracy declined steadily with increasing numerosity forming a sigmoidal performance curve (##FIG##2##Fig. 3a##). We employed a repeated measures ANOVA with within-subject factors “experimental condition” (3 levels: single task, low load, high load) and “target number” (8 levels). There was a significant main effect of condition (<italic>F</italic>(2,26) = 42.49, <italic>p</italic>&lt;.001), with post-hoc comparisons (Bonferroni corrected) showing all three experimental conditions differing significantly from each other. Enumeration accuracy under both dual task conditions was reduced compared to single task condition (single task versus low load: <italic>p</italic> = .004, single task versus high load: <italic>p</italic>&lt;.001). More importantly, enumeration accuracy under high load was more severely impaired than under low load ( <italic>p</italic> = .002).</p>", "<p>As expected, enumeration accuracy decreased with increasing target number (<italic>F</italic>(7,91) = 92.65, <italic>p</italic>&lt;.001). There was also a significant interaction between target number and condition (<italic>F</italic>(14,182) = 5.62, <italic>p</italic>&lt;.001), indicating that our attentional manipulation affected subitizing and estimation ranges differently. Therefore, we conducted separate analyses on the subitizing (target number 1–4) and estimation range (target number 5–8). As this study was designed to prevent verbal counting, we refer to larger numerosity judgement as estimation rather than counting.</p>", "<title>Subitizing range</title>", "<p>The main effect of condition was particularly pronounced in the subitizing range. Accuracy dropped from single task conditions to dual task conditions and particularly between low load and high load conditions (main effect: <italic>F</italic>(2,26) = 58.88, <italic>p</italic>&lt;.001; post-hoc comparisons: single task versus low load: <italic>p</italic> = .003, single task versus high load: <italic>p</italic>&lt;.001, low load versus high load: <italic>p</italic>&lt;.001). There was a main effect of target number (<italic>F</italic>(3,39) = 15.94, <italic>p</italic>&lt;.001), but no interaction of target number with condition (<italic>F</italic> (6,78) = 1.29, <italic>p</italic>&gt;.05).</p>", "<title>Estimation range</title>", "<p>Overall, accuracy was low in the estimation range and differences between experimental conditions were less pronounced. Nevertheless, there was a main effect of condition (<italic>F</italic>(2,26) = 5.55, <italic>p</italic> = .010), mainly due to a significant difference between single task and high load condition (post-hoc comparison <italic>p</italic> = .007, all other comparisons: <italic>p</italic>&gt;.05.). As performance reached chance level (12.5%) for numerosities 7 and 8, we repeated the analysis with numerosity range 5–6. There was still a main effect of condition (<italic>F</italic>(2,26) = 3.55, <italic>p = </italic>.043) but post-hoc comparisons did not reach significance.</p>", "<title>Mean responses, response standard deviation and Weber fraction</title>", "<p>Accuracy reflects subject's behaviour only in a binary manner (whether subjects hit exactly the right numerosity or not), but does not consider trials with near misses. We therefore analysed the mean responses given for each target number and their standard deviation as a measure of deviation from the correct response and the distribution of responses. As a measure of discriminability, we adopted the notion of Weber fraction which we define as response standard deviation divided by target number.</p>", "<title>Mean responses (##FIG##2##Fig. 3b##)</title>", "<p>Comparison of the mean responses to the respective correct target number in each experimental condition (using one-sample t-tests) showed that subjects overestimated the small numerosities (numerosity 1: single task: <italic>t</italic>(13) = 2.48, <italic>p</italic> = .028, low load: <italic>t</italic>(13) = 3.51, <italic>p</italic> = .004, high load: <italic>t</italic>(13) = 4.71, <italic>p</italic>&lt;.001, numerosity 2: high load: <italic>t</italic>(13) = 3.23, <italic>p</italic> = .007, other conditions <italic>p</italic>&gt;.05) and underestimated the larger numerosities from 4 onwards (<italic>t</italic>(13)≤−2.7, <italic>p</italic>≤.018 in all experimental conditions).</p>", "<p>A repeated measures ANOVA showed a significant main effect of condition (<italic>F</italic>(2,26) = 9.59, <italic>p</italic> = .001), due to a difference between the single task and the high load condition (post-hoc comparison: <italic>p</italic> = .003, other comparisons: <italic>p</italic>&gt;.05). There was also a significant interaction of condition with target number (<italic>F</italic>(14,182) = 26.67, <italic>p</italic>&lt;.001). Thus, overestimation in the low numerosities and underestimation in the high numerosities occurred more strongly in the high load condition than in all other conditions.</p>", "<title>Response standard deviation (##FIG##2##Fig. 3c##)</title>", "<p>Response standard deviation increased significantly from single task to low load to high load conditions (main effect: <italic>F</italic>(2,26) = 98.43, <italic>p</italic>&lt;.001, all post-hoc comparisons: <italic>p</italic>&lt;.001). Standard deviations also increased with numerosity (<italic>F</italic>(7,91) = 24.57, <italic>p</italic>&lt;.001) and this effect interacted with the effect of condition (<italic>F</italic>(14,182) = 4.57, <italic>p</italic>&lt;.001).</p>", "<title>Weber fraction (##FIG##2##Fig. 3d##)</title>", "<p>Weber fraction was consistently higher under dual than under single task conditions, and again higher under high than under low attentional load (main effect: <italic>F</italic>(2,26) = 84.48, <italic>p</italic>&lt;.001; all post-hoc comparisons <italic>p</italic>&lt;.001). All effects replicated when subitizing and estimation ranges were analysed separately (all <italic>p</italic>&lt;.001).</p>", "<p>In the single task condition, Weber fraction did not differ across numerosities (<italic>F</italic>(7,91) = 1.46, <italic>p</italic>&gt;.05) consistent with the findings of Ross ##REF##12974571##[25]##. Under dual task conditions, however, Weber fraction was highest in the low numerosities and decreased towards higher numerosities (<italic>F</italic>(7,91) = 30.14, <italic>p</italic>&lt;.001). Considering all three conditions, the effect of target number interacted with condition (<italic>F</italic>(14,182) 27.01, <italic>p</italic>&lt;.001). This result confirmed the detrimental effect of attentional load, particularly in the subitizing range, as observed in the accuracy data.</p>", "<p>Taken together, these additional analyses showed a clear effect of attentional load also in the higher numerosities. High attentional load resulted in an increase of underestimation, response standard deviation and Weber fraction in the estimation range.</p>" ]

[ "<title>Discussion</title>", "<p>The idea of pre-attentive processing implies that some features of a visual scene are analysed in a privileged manner: unconstrained by a perceptual capacity limit, independent of the number of items to be processed and with the ability to consider the entire visual scene at once ##REF##7351125##[8]##, ##UREF##3##[21]##. Sagi and Julesz proposed such privileged processing stage for the case of subitizing ##REF##6535985##[9]##. Based on this approach, we tested subitizing ability under conditions of reduced attentional resources. We predicted that subitizing should not be affected by dual-task conditions nor by attentional load if it was a truly pre-attentive task. Our results clearly fail to support this prediction. Subitizing accuracy was impaired under dual-task conditions compared to single task conditions, even if the additional task comprised only the detection of a single salient feature (the colour red). More crucially, however, subitizing was even more severely impaired when the additional task required a judgement of high attentional load (a conjunction detection). Thus, the more attentional processing resources were taken away from the numerosity judgement task, the more subitizing ability deteriorated. Weber fractions strikingly mirror the accuracy data, indicating that discrimination ability decreased dramatically under high attentional load particularly in the subitizing range. These results challenge the traditional notion of a pre-attentive subitizing mechanism as proposed by many previous studies ##REF##8768186##[3]##, ##REF##9950713##[4]##, ##UREF##1##[6]##, ##REF##6535985##[9]##.</p>", "<p>Our results are in line with recent works demonstrating an impairment of subitizing performance in the attentional blink ##UREF##4##[22]##, ##UREF##5##[23]## and under conditions of inattentional blindness ##REF##17923120##[24]##. In addition to these studies, however, we demonstrate a differential effect on subitizing performance depending on the amount of attentional resources that are drawn away from the enumeration task.</p>", "<p>Furthermore, we also found a clear effect of dual task conditions and attentional load in the estimation range (numerosities 5–8), apparent as an increase in the degree of underestimation and response standard deviation. These findings suggest that the withdrawal of attentional resources affect numerosity judgement in a systematic manner: the more processing resources are taken away and the more difficult numerosity judgement becomes at higher numerosities, the more performance deviates from an unaffected distribution.</p>", "<p>The fact that both the enumeration of small and large quantities is equally affected by the manipulation of attentional resources (in proportion to their respective difficulty) could be interpreted as evidence against a functional dichotomy between subitizing and counting. Our results suggest that both small and large numerosity judgment reflect stages on a single, continuous enumeration mechanism. However, this study was not designed to investigate the nature of these mechanisms, and more specific studies are needed to address this issue. Nevertheless, our results render one of the main arguments for such a dichotomy unlikely: that subitizing is parallel and pre-attentive and might therefore be different from an attentive counting or estimation stage. In support of a continuous enumeration mechanism, Ross showed that Weber fractions are consistently around 25% across a wide range of numerosities, which implies that numerosity judgements in the subitizing range always fall within the performance limit set by this Weber fraction ##REF##12974571##[25]##. This finding provides a simple explanation for why subitizing appears relatively effortless and further strengthens the idea that numerosity judgement is subserved by a single mechanism rather than two functionally separate ones. Our results confirm those of Ross: in the single task condition, Weber fraction was constant across all numerosities and attentional load affected Weber fractions without any sharp discontinuity between the subitizing and the estimation range. Thus, our findings and those of Ross ##REF##12974571##[25]## raise the possibility that previous reports of preserved subitizing ability in neglect patients ##REF##10825363##[10]## as well as differential brain activations in healthy subjects ##REF##9950713##[4]##, ##UREF##1##[6]## might reflect a quantitative rather than qualitative difference between enumerating small and large numerosities.</p>" ]

[]

[ "<p>Conceived and designed the experiments: PV BB. Performed the experiments: PV. Analyzed the data: PV. Wrote the paper: PV BB BB.</p>", "<p>Traditionally, the visual enumeration of a small number of items (1 to about 4), referred to as subitizing, has been thought of as a parallel and pre-attentive process and functionally different from the serial attentive enumeration of larger numerosities. We tested this hypothesis by employing a dual task paradigm that systematically manipulated the attentional resources available to an enumeration task. Enumeration accuracy for small numerosities was severely decreased as more attentional resources were taken away from the numerical task, challenging the traditionally held notion of subitizing as a pre-attentive, capacity-independent process. Judgement of larger numerosities was also affected by dual task conditions and attentional load. These results challenge the proposal that small numerosities are enumerated by a mechanism separate from large numerosities and support the idea of a single, attention-demanding enumeration mechanism.</p>" ]

[]

[ "<p>We thank Silvia Pagano for excellent help with testing subjects and Marco Zorzi for provision of facilities.</p>" ]

[ "<fig id=\"pone-0003269-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003269.g001</object-id><label>Figure 1</label><caption><title>Stimuli and Experimental Procedure.</title><p>(a) Stimulus example. As primary task, subjects detected a certain colour target at fovea. As secondary task, subjects judged the numerosity of high-contrast gabor patches (1 up to 8) amongst low-contrast distractors. (b) Colour combinations of the primary task. Under low attentional load, detection of a single feature was required (the colour red). Under high attentional load, subjects detected specific conjunctions of colour and spatial arrangement (either green triangles aligned along the right-tilted diagonal or yellow triangles along the left-tilted diagonal). (c) Experimental procedure. Under dual task conditions, subjects responded first to the primary task and subsequently to the secondary task. Under single task conditions, subjects responded only to one of the tasks and ignored the stimuli of the other task.</p></caption></fig>", "<fig id=\"pone-0003269-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003269.g002</object-id><label>Figure 2</label><caption><title>Results of the Colour Detection Task.</title><p>(a) Mean accuracy (proportion correct) and (b) mean reaction times (ms) of the primary (colour detection) task under single task and dual task conditions (low load: green bars, high load: yellow bars). Error bars indicate one standard error of the mean.</p></caption></fig>", "<fig id=\"pone-0003269-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003269.g003</object-id><label>Figure 3</label><caption><title>Results of the Numerosity Task.</title><p>Performance of the numerosity task under the three experimental conditions (single task (black), low load (blue) and high load (red)). Error bars indicate one standard error of the mean. (a) Mean accuracy (proportion correct). (b) Mean responses. The dotted diagonal indicates perfect performance, values above the line represent overestimation, values below underestimation. (c) Response standard deviation. (d) Weber fraction (response STD/target number).</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This work was supported by a European Marie Curie Research Training Network grant to Butterworth, a Marie Curie Early Stage Research Fellowship to Vetter and a UCL Graduate School Research Scholarship and an Overseas Research Scholarship to Bahrami.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003269.g001\"/>", "<graphic xlink:href=\"pone.0003269.g002\"/>", "<graphic xlink:href=\"pone.0003269.g003\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Infectious diseases are the result of a dynamic interplay between the pathogen and the host. The host responds to an infection by activating diverse innate and adaptive immune defence mechanisms, whereas the disease causing pathogen produces an arsenal of virulence factors and evades the activated immune system by a great variety of adaptive responses. Today, the treatment of infectious diseases poses not only problems due to multiresistance but also problems due to the emergence of persisting bacteria that survive within biofilm structures in the human host ##REF##14527295##[1]##.</p>", "<p>Free-floating bacteria usually adapt to distinct environmental conditions with a characteristic change in their gene expression pattern. Different stresses may induce diverse but often overlapping stress responses. However, although many of adaptation strategies operate at the individual cell level, others operate at the population level. In the last decade, proof of bacterial interactiveness has accumulated from the observation of complex multicellular development phenomena in fruiting bodies, sporulation and biofilm formation ##REF##12073654##[2]##–##REF##12448714##[6]##. Living in populations provides a species with additional mechanisms of adaptation. Collectively and coordinately, bacteria act far more efficiently than they could as single cells. Thus, multicellularity was proposed as a general bacterial trait that has been increasingly witnessed and that affects multiple phenotypes. The social multicellular behavior in prokaryotes is supported by the identification of signaling molecules that mediate cooperative traits and coordinated behavior ##REF##15639629##[7]##. A coordinated expression of genes involved in virulence and persistence is crucial in host-pathogen interactions. During chronic infections many bacteria adopt a biofilm mode of growth with bacteria embedded in an exopolymeric matrix. This mode of growth acts as a protective niche and helps the bacteria to evade the host immune response and to withstand antimicrobial therapy ##REF##18001153##[8]##. In <italic>Pseudomonas aeruginosa</italic>, biofilm formation and the production of various virulence determinants are regulated via the action of a hierarchical quorum-sensing system mediated by the two chemically distinct classes of signal molecules, the <italic>N</italic>-acylhomoserine lactones ##REF##8288518##[9]## and the 4-quinolones ##REF##11724939##[10]##–##REF##10500159##[14]##. The latter group consists of more than 50 compounds ##REF##14739337##[15]## and includes the most active signal molecule 2-heptyl-3-hydroxy-4-quinolone which is commonly referred to as the pseudomonas quinolone signal (PQS) ##REF##10500159##[14]##. PQS regulates its own production by controlling the expression of the 4-quinolone biosynthetic genes ##REF##15968046##[16]## and its own package into membrane vesicles that deliver antimicrobials and toxins ##REF##16163359##[17]##, transport DNA and might be involved in enhanced survival upon challenge with stressing agents ##REF##17163978##[18]##.</p>", "<p>Stress responses involve considerable investment of cellular resources, often under conditions when cellular resources are limiting and diminishing, a process that may be unsustainable. Thus, a mechanism of uneven distribution of resources in the population, to the benefit of the fittest, may become inevitable. Phenotypically, this would mean that perception of the same stress by the different cells of a population would have very different outcomes, either cell damage, or an increase in tolerance of the stress and an increased fitness and probability of survival. In this study we provide evidence that, although PQS sensitizes the bacteria towards exogenous stresses, PQS also catalyses an efficient bacterial anti-stress response. Our findings add a new dimension to bacterial multicellular behaviour and identifies PQS as an essential factor to shape the population structure and to aid bacterial population adaptation to conditions of increased stress that may significantly contribute to bacterial persistence during chronic infectious diseases.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Bacterial strains and culture conditions</title>", "<p>PAO1 and isogenic transposon mutants with an insertion within the <italic>pqsA, pqsH</italic> and <italic>pqsL</italic> genes were obtained from the Washington Genome Center. The <italic>P. aeruginosa</italic> strains were routinely cultured at 37°C on Luria-Bertani (LB) agar or LB broth. To restore 4-quinolone production in the <italic>pqsA</italic> mutant the bacteria were transformed with the plasmid pLG10 (kindly provided by Colin Manoil) harboring the <italic>pqsA-E</italic> operon. To monitor a decrease in bioluminescence after exposure to the antibiotic ciprofloxacin, a <italic>luxCDABE</italic> operon was introduced in trans into the <italic>P. aeruginosa</italic> strains. Plasmid containing bacteria were cultured in LB under the addition of 500 µg/ml carbenicillin.</p>", "<title>Killing curves</title>", "<p>To monitor viability after antibiotic exposure, the <italic>P. aeruginosa</italic> PAO1 wild-type and the respective mutants were grown in LB broth to the same OD₆₀₀ in mid logarithmic phase before the bacteriocidal agents were added. Ciprofloxacin was used at a concentration of 5 µg/ml, imipenem and gentamicin at a concentration of 7.5 µg/ml. OD₆₀₀ values were obtained at the indicated time intervals and CFU counts were determined by plating appropriate dilutions of the bacterial suspension on LB agar plates.</p>", "<title>Determination of extracellular DNA</title>", "<p>For the determination of the release of DNA by <italic>P. aeruginosa</italic> cultures, the bacteria were grown in LB medium for up to 36 h, the cultures were centrifuged at 13.000 rpm for 10 min and the supernatant was mixed with loading buffer. The probes were subjected to agarose gel electrophoresis using a 1% agarose gel stained with GelStar Nucleic Acid Gel Stain (Lonza, USA) and photographed over UV-light.</p>", "<title>Growth curves</title>", "<p>To monitor the dependency of bacterial growth on PQS we exposed <italic>pqsA</italic> PAO1 mutant cultures to different concentrations of synthesized PQS ##REF##16872396##[21]##. Synthesized 2-heptyl-4-hydroxy-quinoline (HHQ) and 2,2′-dipyridyl served as controls. As opposed to the PAO1 wild-type, the <italic>pqsA</italic> mutant did not form clumps nor did the mutant produce pyocyanin so that the optical density could be monitored for prolonged incubation periods. Bacteria were cultured in LB medium in flasks shaken at 180 rmp.</p>", "<title>Flow cytometry</title>", "<p>Fluorescence-activated cell sorter (FACS) cytometry analysis was performed using the H₂O₂-activated green fluorescent dye redox dye dihydro-dichloro-fluorescein diacetate (H₂DCFDA, Molecular Probes). This dye is readily taken up by cells, cleaved by cellular esterases to produce non-cell permeant H₂DCF, which can be activated by ROS giving green fluorescent DCF as an indication of intracellular levels of oxidative stress.</p>", "<p>The bacteria were cultured overnight in LB, diluted to an OD₆₀₀ of 0.05 and grown to mid-exponential phase with and without the addition of various PQS concentrations. The 4-quinolone 2-heptyl-4-hydroxy-quinoline (HHQ) at equimolar concentrations and the iron chelator dipyridyl served as controls. Cells suspensions were adjusted to an OD₆₀₀ of 0.5, pelleted by centrifugation, and resuspended in phosphate buffered saline (PBS) containing 20 µM H₂DCFDA. The suspension was incubated for 4 h, diluted 1∶100 in PBS and the fluorescence levels of 50,000 cells were recorded using a FACScalibur cytometer (BD Biosciences). Summit software (Dako Colorado) was used for data analysis.</p>", "<title>Antioxidant activity</title>", "<p>To measure the radical-scavenging activity of PQS we utilized the stable 2,2-diphenyl-2-picrylhydrazyl (DPPH) radical. The odd electron in the DPPH free radical gives a strong absorption maximum at 517 nm and is purple in color. The color turns from purple to yellow as the molar adsorptivity of the DPPH radical at 517 nm reduces from 9660 to 1640 when the odd electron of DPPH radical becomes paired with a hydrogen from a free radical scavenging antioxidant to form the reduced DPPH-H. 1 mg/ml DPPH in methanol/water (90/10; vol/vol) was incubated for 15 min at room temperature and the absorbance changes were measured at 517 nm. The 4-quinolone HHQ and ascorbate served as controls.</p>", "<title>H₂O₂ sensitivity assay</title>", "<p>The H₂O₂ sensitivity disk assay was adapted from Hassett and colleagues ##REF##10594832##[39]##. Briefly, PAO1 strains were grown at 37°C in BHI medium. One hundred micoliters of the bacterial culture was suspended in 3 ml of LB soft agar at 40°C (0.6% (w/v) agar) mixed and poured on LB agar plates. Sterile filter paper disks were placed on the soft solid agar and the disks were spotted with 8 µl of 30% H₂O₂. Plates were incubated for 24 h and the diameter of the zone of growth inhibition was measured. All experiments were performed at least in triplicates.</p>", "<title>UV irradiation sensitivity assay</title>", "<p>200 µl of overnight LB-grown bacteria were diluted 1∶10 in LB medium, grown for another 4 h, with or without the addition of 100 µM PQS, and finally diluted 1∶10 in DeMoss medium ##REF##13664660##[40]##. The bacterial suspension was transferred into a glass petri dish at room temperature. Irradiation was performed with a UV lamp (LTF Labortechnik, VL-208G) placed 11 cm above the cells. Samples were removed at 0, 10, 20, 30, 40, 50 and 60 sec of UV irradiation, serially diluted and spotted on LB agar to determine the CFU counts after overnight incubation.</p>" ]

[ "<title>Results</title>", "<title>The pseudomonas quinolone signal (PQS) increases bacterial sensitivity to stressful conditions</title>", "<p>Considering the finding that autolysis in aging <italic>P. aeruginosa</italic> colonies has been observed to correlate with the level of PQS produced by the bacteria ##REF##12426335##[19]##, we hypothesized that PQS may act as an endogenous stress factor, and that this endogenous stressor may be deleterious above a threshold level (such as in PQS over-producing colonies that develop an autolytic phenotype) or if the cells encounter additional exogenous stresses.</p>", "<p>In order to further elucidate PQS specific effects and to differentiate them from the effects of other 4-quinolones, we measured the sensitivity of wild-type <italic>P. aeruginosa</italic> PAO1, an isogenic 4-quinolone non-producing <italic>pqsA</italic> mutant, a 4-quinolone positive but PQS negative <italic>pqsH</italic> mutant ##REF##12426334##[20]##, and a PQS overproducing <italic>pqsL</italic> mutant ##REF##12426335##[19]## towards the activity of ciprofloxacin and hydrogen peroxide. Whereas the PAO1 wild-type exhibited a steep fall in viability, measured as a decrease in optical density (##FIG##0##Figure 1A##) and colony forming units (##FIG##0##Figure 1B##) following exposure to ciprofloxacin, the <italic>pqsA</italic> and the <italic>pqsH</italic> mutant exhibited measurably increased tolerance as the fall in optical density and surviving bacteria exhibited a significant delay. The <italic>pqsL</italic> mutant was found to be most sensitive against the activity of ciproflocxacin. Growth of the wild-type, the mutants and the complemented strain without the addition of the antibiotic was shown to be comparable (##SUPPL##0##Figure S1##). Complementation of the <italic>pqsA</italic> mutant, through provision of the <italic>pqsA-E</italic> operon <italic>in trans</italic> (on plasmid pLG10 ##REF##12426334##[20]##, kindly provided by Colin Manoil), eliminated this tolerance (##SUPPL##1##Figure S2A##). The <italic>pqsA</italic> mutant strain also exhibited an enhanced antibiotic tolerance towards imipenem and gentamicin (##SUPPL##1##Figures S2B and C##).</p>", "<p>We have previously shown that the addition of exogenous PQS enhanced the susceptibility of <italic>P. aeruginosa</italic> towards the bactericidal activity of hydrogen peroxide ##REF##16872396##[21]##. In agreement with this observation, we demonstrate here that both the <italic>pqsH</italic> and the <italic>pqsA</italic> mutant exhibited an increased resistance towards the activity of hydrogen peroxide as compared to the PAO1 wild-type, whereas the complemented <italic>pqsA</italic> mutant as well as the PQS overproducing <italic>pqsL</italic> mutant exhibited an increased sensitivity (##FIG##1##Figure 2##). These results clearly demonstrate that the 4-quinolone PQS sensitizes the bacteria towards the bactericidal activity of antibiotics and oxidative stress.</p>", "<p>If the mediation of an endogenous stress that predisposes the bacteria to developing exogenous stresses is a significant contribution of endogenously-produced PQS, what is the situation for exogenously produced PQS, since a key role of PQS is thought to be interbacterial signaling?</p>", "<p>To test this, we added synthetic PQS and the biologically less active PQS precursor 4-quinolone 2-heptyl-4-hydroxy-quinoline (HHQ) to the PAO1 wild-type harboring a <italic>luxCDABE</italic> operon <italic>in trans</italic>, and measured the decrease in bioluminescence after exposure to the antibiotic ciprofloxacin. In contrast to the addition of HHQ, killing by cirpofloxacin was significantly enhanced by the simultaneous addition of PQS (##FIG##2##Figure 3##). These results implicate that PQS but not HHQ increases bacterial sensitivity to exogenous stresses.</p>", "<title>PQS synchronizes the entry of the whole bacterial population into stationary phase</title>", "<p>We furthermore noticed that PQS affects bacterial growth. To rule out that PQS induced the production of extracellular factors that might impede on bacterial growth, we added increasing concentrations of PQS to exponentially-growing cultures of a 4-quinolone non-producing and non-responding <italic>pqsA</italic> PAO1 transposon mutant. As depicted in ##FIG##3##Figure 4##, the addition of increasing PQS concentrations led to decreasing growth rates and decreasing maximum OD₆₀₀ values (##FIG##3##Figure 4## and ##REF##14507361##[11]##), whereas the addition of equimolar concentrations of HHQ or the iron chelator dipyridyl (PQS has been demonstrated to be an iron chelator ##REF##16872396##[21]##,##REF##17254955##[22]##) had no effect. This result led us to consider the possibility that, although PQS exhibits an inhibitory activity, its principal physiological role may be to elicit an efficient adaptive response.</p>", "<title>PQS reduces the intracellular level of reactive oxygen species</title>", "<p>Since PQS obviously sensitized the bacteria towards oxidative stress, albeit it was previously shown to induce the transcription of oxidative stress response genes ##REF##16872396##[21]##, we decided to monitor the intracellular levels of reactive oxygen species (ROS) in mid-exponential phase grown cultures by employing the redox dye dihydro-dichloro-fluorescein diacetate (H₂DCFDA). As shown in ##FIG##4##Figure 5##, exogenous addition of PQS reduced ROS levels in both wild-type and <italic>pqsA</italic> mutant cells in mid-log phase cultures in a dose-dependent manner. The anti-oxidative effect of PQS was even observed when PQS was added immediately prior to FACS analysis (data not shown). In contrast to PQS, the addition of at least equimolar concentrations of HHQ or the iron chelator dipyridyl did not show a clear reduction of the ROS levels (data not shown).</p>", "<title>PQS is a primary anti-oxidant</title>", "<p>To determine whether PQS is a primary anti-oxidant, we measured the absorbance changes of the stable 2,2-diphenyl-2-picrylhydrazyl (DPPH) radical in the presence of PQS. HHQ and ascorbate served as controls. As depicted in ##FIG##5##Figure 6##, PQS is an anti-oxidant that exhibited an anti-oxidant activity comparable to that of ascorbate, whereas the HHQ molecule did not show any antioxidant activity. This observation is explicable by the much stronger electron-donating potential of PQS as compared to HHQ. From a structural perspective, this is due to the bisoxygenated aromatic system of PQS, which may readily be oxidized to the corresponding ortho-quinoid system. On the contrary, this process is much less pronounced for HHQ, which lacks the 3-OH group and may therefore not form directly a quinoid system (##FIG##6##Figure 7##).</p>", "<title>PQS is a pro-oxidant</title>", "<p>Antioxidants, such as vitamin C may also initiate pro-oxidative processes. For example, while the primary effect of vitamin C is an antioxidant activity by reduction of metal ions, this may lead to the generation of free radicals through the Fenton reaction, which themselves cause oxidative stress. Thus, molecules can act as either antioxidants, or pro-oxidants, depending on the specific set of conditions. Some of the conditions that are important include the concentration of the chemical and if oxygen or transition metals are present. Since PQS was shown to exhibit an antioxidant activity, but at the same time enhanced the susceptibility of the bacteria against external stresses, we wondered whether PQS also exhibits a pro-oxidant activity. We therefore tested the potential of PQS to enhance oxidative DNA damage. As depicted in ##FIG##7##Figure 8## the addition of PQS or ascorbate but not HHQ significantly enhanced DNA fragmentation in a cell free assay under the presence of iron ions. To demonstrate that this PQS induced oxidative DNA damage also plays a role <italic>in vivo</italic>, we monitored DNA release and fragmentation in growing <italic>P. aeruginosa</italic> cultures. We found that DNA was released to a significantly lower extent in the <italic>pqsA</italic> and a <italic>pqsH</italic> mutant as compared to the wild-type and the <italic>pqsL</italic> mutant cultures grown to stationary phase. Expression of the PQS biosynthetic operon <italic>in trans</italic> restored DNA release in the <italic>pqsA</italic> mutant (##FIG##8##Figure 9##).</p>", "<p>Accordingly, PQS has a very similar effect as vitamin C, i.e. a primary antioxidant effect by its strong reducing potential forming a quinoid system, and an indirect second pro-oxidative effect, by initiation of radical mediated processes, presumably in a similar fashion as vitamin C by Fe(III)/Fe(II) mediated radical initiated oxidative processes (##FIG##6##Figure 7##). However, it is expected that this tendency may be even more pronounced for PQS due its strong tendency to coordinate Fe(III). This renders the further evaluation and study of this interesting system a potentially rewarding research goal.</p>", "<title>PQS is involved in resistance to UV irradiation</title>", "<p>Since we could show in this study that PQS can induce oxidative DNA damage we wondered whether this DNA damage may trigger DNA repair and thus induce UV-tolerance. Therefore we applied UV-stress and monitored bacterial survival (##FIG##9##Figure 10##). Wild-type cells showed killing patterns, characterized by roughly 7 log units of killing over 60 s of UV irradiation. The <italic>pqsH</italic> and the <italic>pqsA</italic> mutant were hypersensitive to UV and were killed almost completely within 20–30 s of irradiation. The exogenous addition of 100 µM PQS and the expression of the <italic>pqsA-E</italic> operon <italic>in trans</italic> partially restored UV tolerance in the <italic>pqsA</italic> mutant.</p>" ]

[ "<title>Discussion</title>", "<p>Cellular autolytic systems are well known to participate in a drug-induced lyses of gram-positive bacteria ##REF##11165930##[23]##–##REF##3732271##[25]## and mutants that lack these systems have been shown to become antibiotic tolerant ##REF##4393335##[26]##. Autolysis might be viewed as a maladaptive disbalance caused by the inhibition of cell growth, however it has also been speculated that autolysis represents a programmed response to stressful conditions ##REF##15870304##[27]##–##REF##10974124##[29]##. Programmed cell death is defined as an active process that results in cell suicide and is recognized as an essential mechanism in multicellular organisms where it is required for eliminating damaged or potentially harmful cells, and may have evolved to aid the development of structured populations for the benefit of the population as a whole ##REF##10974124##[29]##,##REF##14617136##[30]##. In this study we provide evidence that a participation of a self-induced cell death in the drug-induced disintegration applies also to the gram-negative organism <italic>P. aeruginosa</italic>. We demonstrate that PQS mediates an endogenous stress that predisposes bacteria to developing exogenous stresses, compliant with the previous finding that overproduction of PQS leads to a highly autolytic phenotype that can be reversed by mutations in the PQS biosynthetic genes ##REF##12426335##[19]##. Most importantly, simultaneously, PQS induces the entry of undamaged bacteria into a less metabolically active, less susceptible state and is shown to be a potent antioxidant capable to reduce the intrabacterial ROS levels.</p>", "<p>One of the most challenging adaptive responses in bacteria is to provide an oxidative stress response that allows the bacteria to optimally cope with reactive oxygen species (ROS) that are unavoidably generated during aerobic metabolism and to respond rapidly to exogenous ROS. Polymorphonuclear cells are the major effector cells responsible for the clearance of <italic>P. aeruginosa</italic> from the site of infection and an essential bacterial defence strategy of <italic>P. aeruginosa</italic> is the ability to withstand the oxidative stress that is induced during phagocytosis, when the bacteria are confronted with very high levels of reactive oxygen intermediates from the respiratory burst of human phagocytes.</p>", "<p>The anti-oxidant activity of PQS is very similar to that of vitamin C which has previously been shown to exhibit both anti-oxidant as well as pro-oxidant activities ##REF##8619018##[31]##–##REF##10336883##[33]##. The latter is due to the fact that ascorbate can reduce metal ions which leads to the generation of free radicals through the Fenton reaction. Addition of vitamin C to purified DNA in the presence of redox active metal ions has been shown to result in single-strand brakes and base modifications ##REF##8855437##[34]##–##REF##9212351##[36]##. This is thought to be due to binding of the metal ion to the DNA and resultant site-specific hydroxyl radical production and oxidative damage ##REF##8855437##[34]##. In the absence of added metal ions, however, vitamin C inhibits the formation of base modifications in purified DNA exposed to peroxynitrite or UV light ##REF##1516838##[35]##,##REF##8841522##[37]##,##REF##9667739##[38]##. In this study we have clearly demonstrated that not only ascorbate but also PQS strongly enhances DNA fragmentation in the presence of oxygen and metal ions <italic>in vitro</italic>, and monitoring of DNA release in stationary <italic>P. aeruginosa</italic> cultures <italic>in vivo</italic> revealed that PQS significantly enhanced DNA release and fragmentation. However, most importantly, at the same time PQS provides the bacteria with an increased fitness under UV radiation. It seems that damaged DNA may serve as a sensor for stressful conditions and thus may trigger DNA repair systems essential for the development of UV tolerance.</p>", "<p>We suggest that the identification of the interbacterial <italic>Pseudomonas</italic> signaling molecule, PQS, as a factor that exhibits both pro- and anti-oxidant activities points to PQS as a cellular “trainer” whose role is to mediate selection of the fittest through a “make or break” mechanism. PQS seems to issue decisions on life and death in <italic>P. aeruginosa</italic> populations, and thereby might be essential to shape the population structure, contribute to multi-cellular development processes in bacterial biofilms, and thus help to maintain discrete, ordered spatial structures. It is interesting to speculate that fragmentation of cellular populations may be a widespread stratagem for clonal survival under stressful conditions; it further raises the prospect that the factors involved in this process might ultimately be exploited as novel targets for anti-bacterial interventions that may even be effective in the treatment of chronic biofilm infections.</p>" ]

[]

[ "<p>Conceived and designed the experiments: SH. Performed the experiments: TB. Analyzed the data: SH TB. Wrote the paper: SH.</p>", "<p>When environmental conditions deteriorate and become inhospitable, generic survival strategies for populations of bacteria may be to enter a dormant state that slows down metabolism, to develop a general tolerance to hostile parameters that characterize the habitat, and to impose a regime to eliminate damaged members. Here, we provide evidence that the pseudomonas quinolone signal (PQS) mediates induction of all of these phenotypes. For individual cells, PQS, an interbacterial signaling molecule of <italic>Pseudomonas aeruginosa</italic>, has both deleterious and beneficial activities: on the one hand, it acts as a pro-oxidant and sensitizes the bacteria towards oxidative and other stresses and, on the other, it efficiently induces a protective anti-oxidative stress response. We propose that this dual function fragments populations into less and more stress tolerant members which respond differentially to developing stresses in deteriorating habitats. This suggests that a little poison may be generically beneficial to populations, in promoting survival of the fittest, and in contributing to bacterial multi-cellular behavior. It further identifies PQS as an essential mediator of the shaping of the population structure of <italic>Pseudomonas</italic> and of its response to and survival in hostile environmental conditions.</p>", "<title>Author Summary</title>", "<p>A little poison may be generically beneficial to bacterial populations. By eliminating damaged members and by promoting survival of the fittest, selective poisoning may significantly contribute to multi-cellular bacterial behavior. Here, we report that the pseudomonas quinolone signal (PQS) exhibits both beneficial and deleterious activities and propose that pro- and anti-oxidant effects fragment the population into less and more stress tolerant members. We suggest our findings point to PQS as a cellular “trainer” whose role is to mediate selection of the fittest through a “make or break” mechanism. PQS thereby issues decisions on life and death in <italic>Pseudomonas aeruginosa</italic> populations, and may be essential in shaping the population structure, contributing to multi-cellular development processes in bacterial biofilms, and helping to maintain discrete, ordered spatial structures.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>I am very grateful to Ken Timmis, Florian Bredenbruch, Dirk Menche and colleagues of the Chronic Pseudomonas Infection group at the Helmholtz Centre for Infection Research for providing comments that have improved the presentation.</p>" ]

[ "<fig id=\"ppat-1000166-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000166.g001</object-id><label>Figure 1</label><caption><title>Antibiotic tolerance in the absence of PQS production.</title><p>Killing curves of log phase grown <italic>P. aeruginosa</italic> PAO1 cultures treated with 5 µg/ml ciprofloxacin were recorded. Killing was significantly delayed in the <italic>pqsA</italic> and <italic>pqsH</italic> mutant compared to the wild-type and the <italic>pqsL</italic> mutant as determined by OD₆₀₀ determinations (A) and CFU counts after 3 h of exposure (B). Error bars mark the standard deviation of three independent experiments. CFU counts of the PAO1 wild-type and the <italic>pqsL</italic> mutant were significantly different (p&lt;0.05, as determined by t-test) as compared to the <italic>pqsA</italic> and <italic>pqsH</italic> mutants.</p></caption></fig>", "<fig id=\"ppat-1000166-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000166.g002</object-id><label>Figure 2</label><caption><title>Growth inhibition by H₂O₂ as determined by agar diffusion assays.</title><p>PQS non-producing strains (<italic>pqsA</italic> and <italic>pqsH</italic> mutant) exhibited an increased resistance towards H₂O₂ as compared to the wild-type (paired t-test; p-value &lt;0.005). The diameter of the inhibition zone is given as the mean+SD of a triplicate.</p></caption></fig>", "<fig id=\"ppat-1000166-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000166.g003</object-id><label>Figure 3</label><caption><title>Loss of bioluminescence of <italic>P. aeruginosa</italic> PAO1 wild-type after exposure to the antibiotic ciprofloxacin.</title><p>Killing of the bacteria by 5 µg/ml ciprofloxacin was significantly enhanced under the addition of PQS as opposed to HHQ (paired t-test; p value &lt;0.05).</p></caption></fig>", "<fig id=\"ppat-1000166-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000166.g004</object-id><label>Figure 4</label><caption><title>PQS induces bacteriostasis in a concentration dependent manner.</title><p>The addition of PQS to bacterial cultures of a <italic>pqsA</italic> mutant is characterized by a slope reduction in log phase of growth and lower maximum OD₆₀₀ values as compared to the cultures grown without added PQS. Bacteriostasis seems to be a PQS-specific effect because both the addition of equimolar concentrations of the biologically less active PQS precursor heptyl-4-hydroxy-quinoline (HHQ) or the iron chelator dipyridyl did not have the same effect. This is a representative experiment for at least three independent measurements.</p></caption></fig>", "<fig id=\"ppat-1000166-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000166.g005</object-id><label>Figure 5</label><caption><title>PQS reduces the intracellular ROI levels.</title><p>The median of fluorescence intensity (MFI) of 50,000 log phase grown cells treated with 20 µM H₂DCFDA is given. PQS was added to log phase grown PAO1 wild-type and <italic>pqsA</italic> mutant cultures at concentrations of 0, 50 and 100 µM PQS, respectively followed by an 1 h incubation period prior to FACS analysis. In contrast to PQS addition, the addition of at least equimolar concentrations of the biologically less active PQS precursor heptyl-4-hydroxy-quinoline (HHQ) or the iron chelator dipyridyl did not show a clear reduction of the ROI levels (data not shown). The assays were performed three times with independent cultures, the addition of PQS reduced the MFI significantly as determined by t-test (p&lt;0.05).</p></caption></fig>", "<fig id=\"ppat-1000166-g006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000166.g006</object-id><label>Figure 6</label><caption><title>PQS is a primary anti-oxidant.</title><p>The absorbance changes of the stable 2,2-diphenyl-2-picrylhydrazyl (DPPH) radical was measured in three independent experiments at the OD of 517 nm in the presence of PQS, HHQ and ascorbate. Error bars mark the standard deviation of three independent experiments. The decolorization is stoichiometric with respect to the number of electrons captured.</p></caption></fig>", "<fig id=\"ppat-1000166-g007\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000166.g007</object-id><label>Figure 7</label><caption><title>Proposed mechanism of the primary anti-oxidant effect and the secondary pro-oxidant effect of PQS.</title></caption></fig>", "<fig id=\"ppat-1000166-g008\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000166.g008</object-id><label>Figure 8</label><caption><title>PQS exhibits pro-oxidants activities.</title><p>The Fe2⁺ mediated DNA fragmentation process is strongly enhanced by the addition of PQS or vitamin C but not by the addition of the biological inactive precursor substance HHQ. Lane 1: chromosomal <italic>P. aeruginosa</italic> DNA under the addition of vitamin C and Fe2⁺, lane 2: DNA+vitamin C, lane 3: DNA , Fe2⁺ and PQS, lane 4: DNA and PQS, lane 5: DNA, Fe2⁺ and HHQ, lane 6: DNA and HHQ. The DNA was pre-incubated with PQS, HHQ or vitamin C, respectively, at room temperature for 45 min before exposure of the probes to 2.5 mM Fe2⁺ for 15 min and separation on an agarose gel.</p></caption></fig>", "<fig id=\"ppat-1000166-g009\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000166.g009</object-id><label>Figure 9</label><caption><title>PQS dependent release of extracellular DNA.</title><p>PAO1 wild-type, the <italic>pqsA</italic>, the <italic>pqsH</italic> and the <italic>pqsL</italic> mutant were grown in LB medium. Supernatant samples were collected after 8 h (lane 1, 3, 5, 7, 9) and 36 h (lane 2, 4, 6, 8, 10), respectively. Increasing amounts of fragmented DNA were found in the supernatant of the PAO1 wild-type cultures under prolonged incubation (lanes 1, 2). DNA release and fragmentation was delayed in the PAO1 <italic>pqsA</italic> and <italic>pqsH</italic> mutants (lane 3, 4 and lane 7, 8). The DNA release and fragmentation pattern was most pronounced in the <italic>pqsL</italic> mutant (lanes 5, 6) and the complemented <italic>pqsA</italic> mutant (<italic>pqsA-E</italic> operon on plasmid pLG10) (lanes 9, 10).</p></caption></fig>", "<fig id=\"ppat-1000166-g010\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000166.g010</object-id><label>Figure 10</label><caption><title>Killing curves upon exposure to UV irradiation.</title><p>Wild-type PAO1, the <italic>pqsA</italic> mutant, the <italic>pqsH</italic> mutant and the <italic>pqsA</italic> mutant complemented with pLG10 were grown overnight in LB, and diluted in DeMoss medium ##REF##13664660##[40]##. The cells were UV irradiated while shaking, and aliquots were removed at 10s intervals. The exogenous addition of 100 µM PQS to the <italic>pqsA</italic> mutant cultures 4 h before exposure to UV radiation restored UV tolerance, whereas PQS addition immediately before the UV stress did not show any effect. Serial dilutions were plated on LB agar to determine the viable cell count. The UV killing assays were performed three times with independent cultures, and the outcome of one representative experiment is shown.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"ppat.1000166.s001\"><label>Figure S1</label><caption><p>Growth behavior of PAO1 wild-type and mutant strains. Growth of the PAO1 wild-type, the <italic>pqsA</italic>, <italic>pqsH</italic>, <italic>pqsL</italic> mutants and the complemented <italic>pqsA</italic> mutant were comparable.</p><p>(0.08 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"ppat.1000166.s002\"><label>Figure S2</label><caption><p>Antibiotic tolerance in the absence of PQS production. Killing curves of log phase grown <italic>P. aeruginosa</italic> PAO1 cultures treated with 5 µg/ml ciprofloxacin (A). Killing was significantly delayed in the <italic>pqsA</italic> mutant compared to the wild-type as determined by OD₆₀₀ determinations and CFU counts after 3 h of exposure and could be reversed by the introduction of the <italic>pqsA-E</italic> operon <italic>in trans</italic>. Killing curves of bacterial cultures treated with 7.5 µg/ml gentamicin (B), and 7.5 µg/ml imipenem (C) were also recorded. Error bars mark the standard deviation of three independent experiments. CFU counts of the PAO1 wild-type and the <italic>pqsA</italic> mutant were significantly different (p&lt;0.05, as determined by t-test).</p><p>(0.24 MB TIF)</p></caption></supplementary-material>" ]

[ "<fn-group><fn fn-type=\"COI-statement\"><p>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p>Funding from the Helmholtz Gemeinschaft is gratefully acknowledged.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"ppat.1000166.g001\"/>", "<graphic xlink:href=\"ppat.1000166.g002\"/>", "<graphic xlink:href=\"ppat.1000166.g003\"/>", "<graphic xlink:href=\"ppat.1000166.g004\"/>", "<graphic xlink:href=\"ppat.1000166.g005\"/>", "<graphic xlink:href=\"ppat.1000166.g006\"/>", "<graphic xlink:href=\"ppat.1000166.g007\"/>", "<graphic xlink:href=\"ppat.1000166.g008\"/>", "<graphic xlink:href=\"ppat.1000166.g009\"/>", "<graphic xlink:href=\"ppat.1000166.g010\"/>" ]

[ "<media xlink:href=\"ppat.1000166.s001.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"ppat.1000166.s002.tif\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Newly introduced T cell based interferon gamma release assays (IGRA) are of a great diagnostic importance in distinguishing with high specificity the persons who are infected with <italic>Mycobacterium tuberculosis</italic>\n##REF##17339619##[1]##. These assays, however, do not discriminate between disease and latent tuberculosis infection (LTBI) ##REF##15818179##[2]##. This limitation lowers their clinical utility e.g. in aging persons originating from a country with a previously high tuberculosis burden who present with pulmonary infiltrates or other manifestations suggestive of tuberculosis reactivation. Therefore, more promising antigens or diagnostic algorithms are continuously being searched to improve the current armament for tuberculosis (TB) immunodiagnostics. <italic>Mycobacterium tuberculosis</italic> heparin binding hemagglutinin (HBHA) is a virulence factor that promotes bacterial aggregation, adhesion to the heparan sulphate proteoglycans of nonphagocytic cells, and dissemination of tubercle bacilli from the lungs to other tissues in patients suffering from tuberculosis ##REF##16510310##[3]##, ##REF##11449276##[4]##. Locht et al. ##REF##16510310##[3]## found that latently infected humans mount a strong Th1-type immune response to HBHA, whereas patients with active disease do not. Moreover, patients with active tuberculosis may develop a strong humoral response to native methylated HBHA ##REF##11865404##[5]##, ##REF##16148186##[6]##. The diagnostic utility of HBHA-based interferon gamma release assays (IGRAs) has been further evaluated in Belgium ##UREF##0##[7]##, a country with low tuberculosis (TB) incidence where bacille Calmette et Guérin (BCG) vaccinations are rarely used. In that work, HBHA-based IGRA was not influenced by prior BCG vaccination and was significantly more sensitive than Early Secretory Antigen Target-6 (ESAT-6)-based IGRA.</p>", "<p>Because some earlier studies ##REF##11865404##[5]##, ##UREF##0##[7]## concluded that previous BCG vaccination does not influence the assay performance we wondered whether HBHA-based methods would retain their diagnostic potential also in Finland. Our country currently has a low incidence of TB (6,1/100 000 ##UREF##1##[8]##) but our population bore a high TB burden as recently as four decades ago. Moreover, since early fifties till September 2006, 98% of the Finnish population have been vaccinated. The rationale for this exploratory investigation was to compare immunological responses to the recombinant methylated HBHA produced in <italic>Mycobacterium smegmatis</italic>\n##REF##15451098##[9]##, methylated synthetic peptides from HBHA, Purified Protein Derivative (PPD), and the proteins with the documented high specificity for <italic>M. tuberculosis</italic> infection ie. ESAT-6 and Culture Filtrate Protein Derivative (CFP-10). For this pilot study we enrolled a few tuberculosis (TB) patients with a variable degree of disease activity and healthy young and middle-age Finnish vaccinees who were practically free from any previous contact with <italic>M. tuberculosis</italic> and thus served as a valid control group.</p>" ]

[ "<title>Methods</title>", "<title>Human subjects and disease definitions</title>", "<p>For this study we enrolled the subjects as follows:</p>", "<title>a) Active TB group</title>", "<p>Those subjects in whom the symptoms of the disease started within one month, who were hospitalized, anti-tuberculosis treatment was commenced within 2–3 weeks and who had a recent positive bacteriological verification. In this group four presented with pulmonary tuberculosis, one with tuberculosis of gastrointestinal tract and one patient had tuberculosis of cervical vertebra. In these subjects the blood samples were obtained within the routine laboratory investigation.</p>", "<title>b) Inactive TB group</title>", "<p>This group consisted of outpatients who presented with no or mild symptoms and who were either partially treated for TB with PAS, streptomycin, and INH or thoracotomy in forties and fifties, in whom chest X-ray findings indicated lack of an active process. Patients with vertebral TB had destructive processes of their vertebra at early childhood. At adulthood their TB was confirmed by radiology and the thick needle aspiration biopsy that ruled out other diseases. At the time of investigation these patients did not receive anti-tuberculosis treatment. Culture or nucleic acid amplification analysis (NAA) were either negative or bacteriological analysis was not performed.</p>", "<title>c) Disease control group</title>", "<p>Four patients with pulmonary manifestations in whom non-tuberculous mycobacterial (NTM) infection was diagnosed by a positive culture isolation. At the time of investigation these subjects were outpatients and their symptoms were mild (mainly cough).</p>", "<title>d) Healthy control group</title>", "<p>Finnish-born University students (mean age 25 years) who on the interview did not admit any previous contact with a TB patient and thus served as an ideal negative control, and Finnish-born laboratory personnel (mean age 50 years), were enrolled. All subjects of this group were born after the routine BCG vaccination campaign was implemented in Finland. Repeated vaccinations were discontunued in Finland in 1990.</p>", "<title>Ethical considerations</title>", "<p>From the subjects of the group <italic>a)</italic> a verbal informed consent was obtained. The verbal informed consent was sufficient because the patient–doctor relationship was earlier established (Dr. Pusa), the patients attended medical care for the routine doctor's check-up and venipuncture was a part of their current medical treatment. From the groups <italic>b)</italic>, <italic>c)</italic> and <italic>d)</italic> a written informed consent was obtained before venipuncture. This study was approved by the Ethical Committee of the University Hospital of Helsinki, Internal Disease Department (Drno. 232/E5/07).</p>", "<title>Sample processing</title>", "<p>Whole blood was withdrawn and divided into portions. From one portion sera were prepared by conventional methods and frozen at −20°C until use. From another portion peripheral blood mononuclear cells (PBMC) were isolated by Ficoll Paque gradient (Amersham Biosciences Inc., Piscataway, USA) and frozen in the CTL (Cellular Technology Ltd., Cleveland, USA) media in a liquid nitrogen until use. The characteristics of the individuals enrolled in this study are presented in ##TAB##0##Table 1##.</p>", "<title>Synthetic peptides</title>", "<p>Twenty three 15-mer sequential peptides overlapping by nine amino acids and spanning the genomic sequence of the <italic>M. tuberculosis</italic> (H37Rv) HBHA-protein starting from 37th amino acid were synthesized (Proimmune, Oxford, UK; Alta Bioscience, Birmingham, UK). The average purity of the peptides was ∼73%. Three peptides from the protein C-terminus were chemically methylated at their lysine residues (see Legend of ##FIG##2##Fig. 3##). The methylation was done on the PEPscreen synthesis platform. Fmoc-Lys(Me,Boc)-OH were pre-dissolved at a 0.5 M concentration, placed on the deck, coupled and deprotected in the same way as standard amino acids. Additionally, the peptides for serology were biotinylated at their N-terminus. The quality control for all peptides was accomplished by mass spectrometry analysis. The peptides were dissolved in sterile asetonitrile and stored at −20°C in aliquots of 1–4 mg/ml with ∼30% glycerol for serology and in PBS for the T-cell assays.</p>", "<title>Recombinant HBHA antigen</title>", "<p>Recombinant methylated HBHA was expressed in <italic>M. smegmatis</italic> (rMtb-HBHA). The pMV3-38 plasmid that contained the full-length HBHA open reading frame was kindly provided by Dr.G. Delogu (University of Sassari, Sassari, Italy) ##REF##15451098##[9]##. The protein from the bacterial extract was primarily purified by phosphocellulose chromatography, i.e., cation exchange, because most mycobacterial proteins do not bind to this resin at pH 7.0, and then further purified by Ni-NTA chromatography ##REF##16893986##[10]##. This two-step purification of rMtb-HBHA was highly effective. Contaminating bands on PAGE gels stained with Coomassie blue were not observed when 5–10 µg of the purified protein was loaded. The purified protein was stored at −20°C in aliquots until use in serology and T cell analysis.</p>", "<title>T ELISPOT analysis</title>", "<p>T-cell reactivity to ESAT-6, CFP-10 (the synthetic peptides were from T-SPOT ®-<italic>TB</italic> kit, Oxford Immunotec, Oxford, UK), PPD (Statens Seruminstitut, Copenhagen, Denmark) and rMtb-HBHA was assessed by Enzyme-Linked ImmunoSpot assay (Mabtech Inc. Cincinnati, USA). Peripheral blood mononuclear cells (PBMC) were purified from fresh blood samples by density gradient centrifugation and stored with CryoABC Kit Freezing media (Cellular Technology Ltd.) in liquid nitrogen until use. After cells were washed with RPMI (HaartBio, Helsinki, Finland) and resuspensed in the CTL Test Media (Cellular Technology Ltd.), the cell count was performed by blood count-analyzer (ADVIA-60 Closed Tube Automated Haematology System, Bayer, Germany). The cells were diluted at 2.5×10⁶ PBMC/ml in CTL Test Media (Cellular Technology Ltd.). 250 000 PBMCs/well were stimulated in ELISPOT-plates in the presence of synthetic peptide-pool (10 µg/ml each), PPD (10 µg/ml), rMtb-HBHA (25 µg/ml). For viability test 50 000 PBMC/well were stimulated with phytohaemagglutinin (PHA) (Oxford Immunotech). CTL Test Media was used as a negative control. The plates were incubated at +37°C with 5% CO₂ for 48 h. Thereafter the plates were washed and the analysis was performed according to the manufacturer's instruction. The spots were counted the ELISPOT-reader (Biosys, Lyngby, Denmark) and the net values were calculated by subtracting the readings of the media control. Spot-sizes and the cytokine values were examined with the ELISPOT-reader, AID EliSpot Software Version 4.0 (AID GmbH, Strasburg, Germany). The viability control test for the HBHA synthetic peptides was performed by exposing PBMC to the CFP-10 peptide mixture with asetonitrile added at concentrations of 15% and 30% (vs. only 2.5% in the final peptide solution). The addition of asetonitrile did not violate the reactivity to the CFP-10 antigen (data not shown).</p>", "<title>IgG and IgM determinations</title>", "<p>The presence of IgG- and IgM-class antibodies to synthetic peptides, rMtb-HBHA and PPD were determined by enzyme-linked immunosorbent assay (ELISA). Synthetic peptides from <italic>Borrelia burgdorferi</italic> VlsE-protein IR6 region and positive serum samples from Lyme disease patients were used as the method control ##REF##17234451##[11]##. 96-well microplates (Microlon high binding, Greiner, Frickenhausen, Germany) were coated either with streptavidin (Roche, Mannheim, Germany) in PBS, pH 7.5 (100 ng/well), or PPD in PBS, pH 7.5 (1 mg/well), or with rMtb-HBHA in 0.1 M bicarbonate buffer, pH 9.5 (250 ng/well) overnight at +4°C. The plates were blocked with 0.25% Human serum albumin (HSA) (Finnish Red Cross, Helsinki, Finland) in PBS for 1 h at +37°C and washed four times with PBS containing 0.05% Tween20. Biotinylated synthetic peptides (500 ng/well) and IR6 (200 ng/well) in PBS, pH 7.5 were incubated for 2 h at room temperature. Plates were washed as above and the human sera diluted 1∶100 in PBS containing 0.5% HSA, 10% FCS and 0.1% Tween20 and were incubated for 2 h at room temperature. After wash Alkaline phosphatase-conjugated anti-Human IgG or IgM antibody (Jackson ImmunoResearch, W. Baltimore, USA) was added at 1∶5 000 dilution in PBS containing 0.5% HSA and 0.1% Tween20. After 2 h of incubation at room temperature the plates were washed four times. 4-nitrophenylphosphate (Boehringer Mannheim, Germany), 1 mg/ml in 0.1 M diethanolamine-0.5 mM MgCl₂ was added to each well and the reaction was stopped after 15 minutes with 0.1 M NaOH. The plates were read at 405 nm with iEMS Reader MF (Labsystems, Helsinki, Finland). IgM rheumatoid factor was controlled for all serum samples by immunonephelometric method at the HUSLAB, Unit of Immunology.</p>", "<title>Statistical analysis</title>", "<p>The correlation was analyzed by the non-parametric Spearman test. The Receiver operating characteristic (ROC) curve analysis and the Area under the curve (AUC) with respective 95% confidence intervals (CI) for each antigen were calculated with the GraphPad Prism version 4.0 (GraphPad Software, Inc. San Diego, CA).</p>", "<p>The amino acid sequences of the HBHA from <italic>M. tuberculosis</italic> and <italic>M. bovis</italic> were searched with the Entrez Protein search engine and then the sequences were aligned with the Needle program, EBLOSUM62-Matrix (EMBOSS, ##UREF##2##[12]##).</p>" ]

[ "<title>Results</title>", "<title>Cell-mediated immunity</title>", "<p>Employing both techniques for immune reactivity study of rMtb-HBHA, we observed considerable interindividual variation in all studied groups (##FIG##0##Fig. 1A–C## and ##FIG##2##Fig. 3 A–B; D–F##). These observations hold true irrespective of the measured parameters of immunoreactivity; neither the numbers of reactive cells, nor the amount of the IFNγ released, nor the optical densities in the IgG and IgM EIAs were discriminatory when comparing the groups. For example, the highest values of 430, 387, and 360 reactive cells/10⁶ lymphocytes were observed in the groups of active TB, inactive TB and healthy vaccinees, respectively. Strikingly, half of the vaccinated persons reacted strongly against rMtb-HBHA and the reactivities were equally high as those of the TB-patients (##FIG##0##Fig. 1A##). When the comparisons between the groups were done by semiquantitative calculation of the amount of the produced cytokine, ie. the size and the intensity of the spot, the highest median value of the cytokine production was found in the vaccinated group (##FIG##0##Fig. 1B##). Persons with positive environmental mycobacteria culture result reacted similarly as the persons from other groups with a high interindividual variation (##FIG##0##Fig. 1A and B##). Positive correlation was observed then the reactivities to rMtb-HBHA and PPD, an indicator of vaccination or infection, were compared for all the tested samples (##FIG##0##Fig. 1D##). One healthy 60-year-old individual who was not vaccinated in his childhood showed humoral and cell-mediated immune responses to rMtb-HBHA as measured by both EIAs and ELISPOT assay, indicating immunization with so-called atypical mycobacteria. Noteworthy, this person had a negative tuberculin skin test and his lymphocytes did not recognise ESAT-6 and CFP-10 peptide mixtures in ELISPOT (data not shown). Furthermore, the obtained reactivities were reproducible and the measurements did not exceed the expected between-run imprecisions of CV% ≤15 and ≤40% for EIAs and ELISPOT, respectively. When the frequencies of reactive cells were compared between persons of different ages, no differences were noticed either (##FIG##0##Fig. 1C##), indicating that there was no waning of immunological memory towards HBHA with age. In other words, ELISPOT analysis produced similar patterns of reactivities between the studied groups with 2 out of 5; 3 out of 6; and 7 out of 15 being strong reactors in the groups of inactive TB, active TB, and vaccinees, respectively (##FIG##0##Fig. 1A##).</p>", "<p>As expected, when tested concurrently with a mixture of specific peptides derived from ESAT-6 and CFP-10 of <italic>M. tuberculosis</italic>, samples from patients with active and inactive TB did react in the ELISPOT assay, whereas samples from all healthy individuals, did not (##FIG##1##Fig. 2 A and B##).</p>", "<title>Humoral immunity to HBHA</title>", "<p>High interindividual variations were observed in serology, ie. 3 out of 9 (Inactive TB) and 2 out of 4 (Active TB) were strong reactors in the IgG and IgM EIA whereas almost all vaccinees showed moderate to high reactivities in IgG and IgM EIAs (##FIG##2##Fig. 3A–B, D–F##). Noteworthy, the median value of IgM antibodies to rMtb-HBHA was the highest in the group of healthy BCG-vaccinated individuals. Synthetic peptides were not recognised by any of the tested sera in the IgG EIA (data not shown). When tested for IgM rheumatoid factor it was detectable only in one person in the TB-patients group. Interestingly, IgM-class antibodies were detectable in almost all of the individuals in the study, with an exception of two non-vaccinated infants (data not shown). The IgM-class antibodies recognised not only rMtb-HBHA but also the three 15-mer linear methylated peptides from the HBHA C-terminus (##FIG##2##Fig. 3D–F##). We believe that IgM-class antibodies alone reacting to the C-terminal peptides of HBHA probably indicate a non-specific reaction arising from heterophilic antibodies, for example, or immunological cross-reactivity with conserved sequences of environmental bacteria (e.g., <italic>rhodococcus</italic>, <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.ncbi.nlm.nih.gov/blast/Blast.cg\">http://www.ncbi.nlm.nih.gov/blast/Blast.cg</ext-link>).</p>", "<title>Comparison of the discriminatory power of rMtb-HBHA, PPD, ESAT-6 and CFP-10</title>", "<p>ROC curves were constructed for all tested antigens and the AUCs with respective 95% CI were compared for each tested antigen (##FIG##3##Fig. 4A–D##). As expected, ESAT-6 and CFP-10 produced ROC curves acceptable for diagnostics with AUCs ranging from 0.947 to 0.972 and a narrow 95% CI (0.84–1.052) (##FIG##3##Fig. 4C–D##). On the contrary, the AUC for rMtb-HBHA (0.636; 95% CI 0.391–0.886) was comparable to the one for PPD (0.736; 95%CI 0.531–0.941) (##FIG##3##Fig. 4A–B##) indicating in practice no discriminatory power between healthy vaccinees and persons with a TB infection. As a consequence of high interindividual variation in rMtb-HBHA immunoassays, the confidence interval for the respective AUC was wide.</p>" ]

[ "<title>Discussion</title>", "<p>The potential of HBHA for diagnostics have been recently reported ##REF##16510310##[3]##, ##REF##11865404##[5]##, ##REF##16148186##[6]##, ##UREF##0##[7]##, ##REF##16893986##[10]##, ##UREF##3##[13]##. In this pilot investigation we attempted to have an insight into the usefulness of this antigen for diagnostics in our fully vaccinated population. We aimed to compare immunoreactivity to rMtb-HBHA and HBHA methylated synthetic peptides with the immunoreactivity to the conventional PPD that have been used for decades with poor success because of inability to discriminate TB infection from immunity caused by vaccination. For comparison, we used two other secreted antigens, namely ESAT-6 and CFP-10. These antigens have been proven highly immunogenic and specific for <italic>M. tuberculosis</italic> infection and are absent from the majority of non-tuberculous mycobacteria and BCG substrain ##REF##17339619##[for review see 1]##. The major objective was to investigate whether immunological responses in persons with a proven contact with <italic>M. tuberculosis</italic> (Active and Inactive TB groups) would be quantitatively and qualitatively different from vaccinated individuals. In other words, we were interested to study the specificity of immune response to HBHA as a nominator of infection. The minor objective was to see whether immunological responses would differ in TB patients with a different degree of disease severity. We did not enrol a group of patients with a proven LTBI which may be considered as a limitation of our study. On the contrary, we enrolled young Finnish vaccinees that were unambiguously interpreted as free from LTBI and some patients with proven NTM infection. In our opinion, these two groups is a strength of our investigation. For analysis we used ELISPOT and EIA techniques.</p>", "<p>The rMtb-HBHA antigen used in this study was prepared as described by Delogu et al ##REF##15451098##[9]##. Using a limited clinical material we were not able to affirm the discriminatory power of rMtb-HBHA in serology. Zanetti et al ##REF##16148186##[6]## who referred to the same method of the recombinant HBHA expressed in <italic>M. smegmatis</italic>\n##REF##15451098##[9]##, showed that only forty-four out of 111 sera with active TB produced optical densities over the presumed cut-off level of 0.5 OD_(405nm). In their study, surprisingly, the combined group of vaccinated people and patients with presumed latent tuberculosis infection did not produce IgG reactivities above the cut-off level. Using a larger cohort size than we did, Zanetti et al. ##REF##16148186##[6]## demonstrated a trend towards higher frequencies of responders in IgG serology in patients with active TB, however only 1/3 of the tested subjects were classified as serology positive. It seems however, that this frequency of positive results would not satisfy diagnostic needs. In our investigation we were able to detect IgM antibodies that recognized rMtb-HBHA and synthetic peptides of the C-terminus practically from all the tested individuals and also from one 60-year-old person who has never been vaccinated nor has a LTBI (data not shown). In the study of Shin et al. ##REF##16893986##[10]## who used immunoblot and EIA techniques to investigate IgM reactivity to rMtb-HBHA, the antibodies were detectable in early and chronic TB patients whereas healthy students, the controls, were non-responders. In their study, however, the vaccination status of the controls was not reported, therefore we are left with uncertainty how would this antigen be recognised in BCG vaccinees.</p>", "<p>Masungi et al. ##REF##11865404##[5]## used HBHA that was purified from a BCG substrain. As others, they also reported that negative control subjects and the BCG vaccinees did not produce anti-HBHA antibodies at detectable level and the reactivity towards HBHA in lymphocyte stimulation assay was non-significant. On the contrary, when lymphocytes were stimulated with PPD, the controls and the BCG vaccinees reacted strongly ##REF##11865404##[5]##. In the most recent study of Hougardy et al. ##UREF##0##[7]## HBHA was also purified from a BCG substrain and the cell-mediated immunity was studied as Masungi et al. by lymphocyte stimulation and a subsequent measurement of secreted interferon gamma (IFNγ). In their study the authors used control subjects who had no history of a TB contact but in whom half were BCG vaccinated. Quite opposite to our findings they observed the steepest ROC when lymphocytes were stimulated with HBHA while the stimulation with ESAT-6 produced the least discriminatory ROC.</p>", "<p>Methylation of HBHA is crucial for effective T-cell immunity. Thus, native HBHA purified from BCG or <italic>M. tuberculosis</italic> H37Ra evoked a much stronger CD8+ and CD4+ mediated immune responses than the recombinant non-methylated HBHA produced in <italic>E. coli</italic>\n##UREF##3##[13]##. However, as indicated, recombinant <italic>M. smegmatis</italic> expression host was able to produce HBHA with a methylation pattern very similar to that of the native HBHA, as was evidenced by mass spectrometry analysis, amino acid sequencing, electrophoresis and recognition by the same monoclonal antibody ##REF##12149464##[14]##. There is however a possibility that folding of native HBHA produced by BCG or <italic>M. tuberculosis</italic> and of recombinant HBHA produced in <italic>M. smegmatis</italic> may be different which would cause minor differences in their immunoreactivities. The protein used in our study, rMtb-HBHA, was expressed in <italic>M. smegmatis</italic> and all attempts were made to avoid potential contamination that might have blurred the results. While it could be argued that minor <italic>M. smegmatis</italic> genome coded contaminant antigens might have caused the observed T cell responses in our study, the antibody responses (IgM) to the synthetic peptides support the view of weak specificity of the rMtb-HBHA responses in BCG vaccinated persons. In view of a longevity of immunological memory even without boosting ##UREF##4##[15]##, ##REF##12208874##[16]## it is not unexpected that our healthy vaccinees recognised rMtb-HBHA that has a 95.5% amino acid sequence homology to BCG HBHA ##UREF##2##[12]##.</p>", "<p>Results of immunological studies with a battery of synthetic peptides comprising mono- and dimethylated peptides of the C- terminus of HBHA were very disappointing. In this study IgM-class reactivities were detectable but in a very unpredictable fashion while IgG-class antibodies did not recognize peptides even methylated ones in any of the patient groups. The peptides did not evoke any T cell mediated response. Noteworthy, our study protocol confirmed that the non-responsiveness to the peptides was not attributed to the toxicity of asetonitrile used in the peptides solution. Temmerman et al ##UREF##3##[13]## studied T cell immunity with a peptide scan analysis first with non-methylated peptides and then probing a methylated peptide. Only one non-methylated peptide produced some IFNγ in a portion of LTBI patients. The reproducibility of this finding was not, however, reported. In their hands the methylated peptide induced also some IFNγ production but only in combination with the recombinant protein. The authors speculated that the methylated peptide might need a protein carrier. We too have no solid explanation why synthetic peptides behaved so differently in cell mediated and humoral immunity studies compared to the recombinant protein. In fact, we know yet little about how the HBHA is chopped in the lysosome and how the methylated antigens are presented by MHC II molecules.</p>", "<p>Because our cohorts were too small to be able to pick-up minor differences in reactivity between the studied subjects we chose to use ELISPOT, the most sensitive and functional technique to study cell-mediated immunity. We assessed not only the frequencies of reactive lymphocytes (##FIG##0##Fig. 1A##) but the production of IFNγ as well (##FIG##0##Fig. 1B##). In this way we tried to do the analysis as close as feasible to that of Masungi et al. and Hougardy et al. However, in neither of the analyses could we prove HBHA superiority over the well-established antigens for diagnostics, namely ESAT-6 and CFP-10. As expected, these latter antigens possessed sufficient discriminatory power to separate infected individuals from vaccinated persons (##FIG##1##Fig. 2A–B##). The ROC analysis with a narrow 95% CI confirmed this conclusion (##FIG##3##Fig. 4A–D##). In fact, recent study by Chee et al. ##UREF##5##[17]## showed that the two commercial methods for TB immunodiagnostics produced a poor agreement (κ = 0.257) when testing 270 patients with pulmonary TB. One was an ELISPOT- and the other was an EIA-based method. The authors speculated that the differences in results may be attributed to heterophilic antibody effects, non-specific IFNγ in the blood samples and lack of standardised lymphocyte counts in EIA-based technology compared to ELISPOT. The factors mentioned by Chee et al. ##UREF##5##[17]## may also contribute to controversy between our pilot study and earlier studies ##REF##11865404##[5]##, ##UREF##0##[7]## where EIA-based methods have been applied.</p>", "<p>This pilot study casts some concerns about the possibility to generalize about diagnostic potential of HBHA. We have observed that samples from healthy individuals in a country with almost complete vaccination coverage and even from individuals who have never been vaccinated may exhibit immune reactivities to HBHA, indicating that natural immunization to this protein or to cross-reactive peptides may occur. Therefore, on the basis of <italic>i)</italic> the presence of closely related antigen in a BCG substrain; <italic>ii)</italic> observed good correlation with PPD in serology and cell-mediated assays, <italic>iii)</italic> comparable ROC analysis for HBHA and PPD; <italic>iv)</italic> evidence of reactivities of vaccinees without any previous risk of contraction of <italic>M. tuberculosis</italic> infection, HBHA by no means is superior to ESAT-6 and CFP-10. In our environment HBHA practically does not add to tuberculosis immunodiagnostics. In conclusion, our results emphasize that the search for new more promising antigens for TB diagnostics should continue.</p>" ]

[]

[ "<p>Designed the study: TT IS. Drafted the article and revised it critically: TT IS LS. Carried out EIA and ELISPOT analysis and data processing: LS. Enrolled all the TB patients and studied and collected data from the medical records: LP. Produced recombinant HBHA: HJK. Designed synthetic peptides and processed clinical specimens: HS.</p>", "<title>Background</title>", "<p>In recent years T cell based interferon gamma release assays (IGRA) have been developed for immunodiagnosis of <italic>M. tuberculosis</italic> infection. At present these assays do not discriminate between disease and latency. Therefore, more promising antigens and diagnostic tools are continuously being searched for tuberculosis immunodiagnostics. The heparin binding hemagglutinin (HBHA) is a surface protein of <italic>M. tuberculosis</italic> which promotes bacterial aggregation and adhesion to non-phagocytic cells. It has been previously assumed that native, methylated form of this protein would be a promising antigen to discriminate latent from active infection.</p>", "<title>Methodology and Principal Findings</title>", "<p>We performed a pilot investigation to study humoral and T-cell mediated immunological responses to recombinant HBHA produced in <italic>M. smegmatis</italic> or to synthetic peptides in patients with recent or past tuberculosis, with atypical mycobacteriosis, or in healthy vaccinated individuals. The T cell reactivities to HBHA were compared to the respective reactivities towards Purified Protein Derivative (PPD) and two surface secreted proteins, ie. Early Secretory Antigen Target-6 (ESAT-6) and Culture Filtrate Protein-10 (CFP-10). Our pilot results indicate that methylated recombinant HBHA induced a strong T cell mediated immune response and the production of IgG and IgM-class antibodies in all patient groups, most surprisingly in young Finnish vaccinees, as well. We observed a positive correlation between the reactivities to HBHA and non-specific PPD among all studied subjects. As expected, ESAT-6 and CFP-10 were the most powerful antigens to discriminate disease from immunity caused by vaccination.</p>", "<title>Conclusions</title>", "<p>On the basis of results of this exploratory investigation we raise concerns that in countries like Finland, where BCG vaccination was routinely used, HBHA utility might not be sufficient for diagnostics because of inability to explicitly discriminate tuberculosis infection from immunoreactivity caused by previous BCG vaccination.</p>" ]

[]

[]

[ "<fig id=\"pone-0003272-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003272.g001</object-id><label>Figure 1</label><caption><title>The ability of rMtb-HBHA and PPD to induce the production of IFNγ was tested in the ELISPOT technique.</title><p>The groups of patients with inactive TB (n = 5); active TB (n = 6), healthy young (n = 7) and middle-age (n = 8) vaccinated subjects and patients with isolation of so-called atypical mycobacteria (n = 4) and were enrolled. Comparison of the cell-mediated immunological responses to rMtb-HBHA was performed as the determination of the number of reactive cells per10⁶ lymphocytes (A) and as the measurement of IFNγ production activities expressed as arbitrary units (B). Cell-mediated immunological responses in healthy vaccinated individuals was studied by division by age into two groups (C). The correlation of rMtb-HBHA with the PPD ELISPOT reactivities expressed as the frequencies of reactive cells per10⁶ is presented in (D). Data are shown as individual reactivities; the horizontal bars represent arithmetic median values.</p></caption></fig>", "<fig id=\"pone-0003272-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003272.g002</object-id><label>Figure 2</label><caption><title>Production of IFNγ by lymphocytes stimulated with ESAT-6 (Panel A) and CFP-10 (Panel B) was tested in the ELISPOT technique.</title><p>The groups of patients with inactive TB (n = 5), active TB (n = 6) and healthy vaccinated subjects (n = 13) were tested. Cell-mediated immunological responses were determined as the number of reactive cells per 10⁶ lymphocytes. The lower dotted line is the level of positivity suggested by the manufacturer, the upper dotted line is the level of positivity adopted at the HUSLAB diagnostic laboratory. The area between the two dotted lines represents the so-called grey-zone, an area of uncertainty for interpretation that was calculated based on assay imprecision (data not published).</p></caption></fig>", "<fig id=\"pone-0003272-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003272.g003</object-id><label>Figure 3</label><caption><title>The serological reactivities to synthetic peptides, rMtb-HBHA and PPD were tested in IgG and IgM ELISAs.</title><p>The groups of patients with inactive TB (n = 9), active TB (n = 4) and healthy vaccinated individuals (n = 4) were enrolled. Serological responses to rMtb-HBHA measured by the IgG EIA (A), IgM EIA (B), and IgG correlation in the PPD and rMtb-HBHA ELISAs (C). IgM responses to methylated 15-mer linear peptides from the C terminus of HBHA: IELPKKAAPA[KMe]KAAP (D), AAPAKKAAPA[KMe]KAAA (E), and AAPAKKAAPA[KMe][KMe]AAA (F). Individual responses are presented as optical densities (OD₄₀₅). The horizontal bars represent arithmetic median values.</p></caption></fig>", "<fig id=\"pone-0003272-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003272.g004</object-id><label>Figure 4</label><caption><title>ROC curves were constructed to compare ELISPOT results when the cells were stimulated with rMtb-HBHA (A), PPD (B), and peptide mixtures of ESAT-6 and CFP-10 (C–D).</title><p>The curves were established for infected (Active and Inactive TB) and the healthy control group. The calculated AUC and the respective confidence intervals (in brackets) are shown for each tested antigen.</p></caption></fig>" ]

[ "<table-wrap id=\"pone-0003272-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003272.t001</object-id><label>Table 1</label><caption><title>Characteristics of the individuals enrolled in the study.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Patient groups</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">ELISPOT (n)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">EIA (n)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Ethnicity</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Age range (yrs)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Gender</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bacteriology</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Diagnosis</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Active TB</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Finns, n = 3 Persons from endemic area, n = 3 (2 with extrap. TB)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">31–82</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Female 2/6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">All bacterilogically confirmed: culture+ n = 5, of those, acid faststaining+ n = 3 NAA+ n = 1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Tuberculosis: Pulmonary n = 4 GI channel n = 1 Cervical vertebra n = 1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Inactive TB</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Finns, n = 8 One person from an endemic area (extrap. TB)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">41–83</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Female 4/9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">At the time of examination culture and NAA negative or bacteriological analysis was not performed</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Partially treated for pulmonary TB n = 7 Partially treated/untreated TB of lumbar vertebra diagnosed radiologically n = 2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Atypical mycobacteria</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">All Finns</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">66–75</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">All Female</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>M. avium</italic> n = 2 <italic>M. intracellulare</italic> n = 1 <italic>M. abscessus</italic> n = 1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">co-morbidity with asthma, COPD</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Vaccinated subjects students laboratory personnel</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7 8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">All Finns</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">22–59 mean 25 mean 50</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Female 6/7 8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr></tbody></table></alternatives></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>We wish to thank the following private Finnish foundations for financial support: Finnish Lung Health Association (Filha ry), Pulmonary Association Heli, and the Tuberculosis Association of the University of Tampere. MD Raimo Tuuminen is acknowledged for help in data analysis.</p></fn></fn-group>" ]

[ "<graphic id=\"pone-0003272-t001-1\" xlink:href=\"pone.0003272.t001\"/>", "<graphic xlink:href=\"pone.0003272.g001\"/>", "<graphic xlink:href=\"pone.0003272.g002\"/>", "<graphic xlink:href=\"pone.0003272.g003\"/>", "<graphic xlink:href=\"pone.0003272.g004\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>The discovery that cells produce small RNAs called interfering RNAs (iRNAs) that are able to inhibit in a sequence-specific way gene expression at a post-transcriptional level, a phenomenon called RNA interference, is a recent breakthrough in biology. This discovery has already had and will have many consequences on our knowledge of cell physiology and pathology, on our ability to regulate protein production at the laboratory, using tailored small iRNAs (siRNAs), and soon in medicine. Cohabitation of iRNAs physiologically produced by human cells, called microRNAs (miRNAs), and viral RNAs in the same cell results in interactions that may have an impact on virus replication, host cell physiology, and anti-microbial immune response. In this article, we will describe briefly the miRNA machinery, review the various connections existing between iRNAs and HIV RNAs, evaluate their consequences for both actors, and consider how we could interfere with these connections in order to regulate virus infection.</p>" ]

[]

[]

[]

[ "<title>Conclusion</title>", "<p>The miRNA pathway appears to be a new branch of natural antiviral immunity. Understanding the complex relationship that exists between the miRNA pathway and HIV will enlighten the physiopathology of the infection and offer new therapeutic strategies. As miRNA might be involved in HIV replication and latency, pathogenicity, and immune response, manipulating miRNA might enable us to modulate all of these aspects of the infection. Much remains to be discovered with surprises on the way. For instance, the eventuality that miRNA might also regulate DNA transcription in mammal cells ##REF##15297624##[49]## by inducing DNA methylation ##REF##15963934##[50]## and/or other mechanisms ##UREF##2##[51]##,##REF##12193640##[52]## could offer additional means to inhibit HIV replication, as suggested by recent data ##REF##17663774##[23]##,##REF##19771207##[53]##.</p>" ]

[ "<p>In this review, a quick presentation of what interfering RNA (iRNA) are—small RNA able to exert an inhibition on gene expression at a posttranscriptional level, based on sequence homology between the iRNA and the mRNA—will be given. The many faces of the interrelations between iRNA and viruses, particularly HIV, will be reviewed. Four kinds of interactions have been described: i) iRNA of viral origin blocking viral RNA, ii) iRNA of viral origin downregulating cellular mRNA, iii) iRNA of cellular origin (microRNA) targeting viral RNA, and iv) microRNA downregulating cellular mRNA encoding cell proteins used by the virus for its replication. Next, HIV strategies to manipulate these interrelations will be considered: suppression of iRNA biosynthesis by Tat, trapping by the HIV TAR sequence of a cell component, TRBP, necessary for iRNA production and action, and induction by the virus of some microRNA together with suppression of others. Then, we will discuss the putative effects of these mutual influences on viral replication as well as on viral latency, immune response, and viral cytopathogenicity. Finally, the potential consequences on the human infection of genetic polymorphisms in microRNA genes and the therapeutic potential of iRNA will be presented.</p>" ]

[ "<title>Human Cellular miRNAs</title>", "<p>miRNAs are transcribed by intergene regions, introns, and even exons, either as polycistronic transcripts if they are clustered or as monocistronic transcripts if they are not ##UREF##0##[1]##–##REF##12624257##[3]##. These transcripts, called pri-miRNAs, are imperfect RNA hairpins of hundreds to thousands of base pairs ##REF##12198168##[4]##. They are processed in the nucleus by the Rnase III endonuclease Drosha to a stem loop structure of about 60 base pairs, the pre-miRNA (##FIG##0##Figure 1##) ##REF##14508493##[5]##. The transfer of pre-miRNA from the nucleus to the cytoplasma is mediated by the nuclear export factor exportin 5 ##REF##14681208##[6]##. There, pre-miRNAs are cleaved by another Rnase III endonuclease, Dicer ##REF##11452083##[7]##. Dicer delivers an approximately 22–base pair duplex. One strand of this duplex, the mature miRNA, still bound to Dicer, is driven towards ribosome-free compartments of the cytoplasma called P-bodies (processing bodies) by an association of molecules called RISC (RNA-induced silencing complex), which includes the endonuclease Argonaute-2 (Ago-2) ##REF##15284456##[8]##. The targets of the miRNA-loaded RISC are the RNAs presenting with sequence homology with nucleotides 2–7 in the 5′ portion of the miRNA. Most of the time, the consequence is an inhibition of translation of the mRNA, and sometimes, particularly if the match between the mRNA and the miRNA is perfect, the mRNA is cleaved. Of note, TAR RNA–binding protein (TRBP), a cell protein initially discovered for its capacity to bind to the TAR sequence of HIV RNA, also binds to Dicer and Ago-2, and is necessary for the maturation of pre-miRNA into miRNA as well as for interfering RNA function ##REF##15973356##[9]##,##REF##16142218##[10]##.</p>", "<p>A single miRNA targets at least 100 transcripts from various genes, and one mRNA may be targeted, at its 3′ end, by different miRNAs ##UREF##1##[11]##. Thus, miRNAs, whose number has been estimated to be 340 ##REF##17604727##[12]##, and more recently over 600 (<ext-link ext-link-type=\"uri\" xlink:href=\"http://microrna.sanger.ac.uk/cgi-bin/sequences/mirna_summary.pl?org=hsa\">http://microrna.sanger.ac.uk/cgi-bin/sequences/mirna_summary.pl?org=hsa</ext-link>), could regulate at least one-third of all human genes ##REF##15652477##[13]##.</p>", "<p>Researchers have taken advantage of this pathway to induce the specific destruction of mRNAs in order to silence genes of choice. They do so either by directly transfecting siRNAs of approximately 21 base pairs, or by delivering transgenes encoding hairpin RNAs, small hairpin RNAs (shRNAs) processed by Dicer to siRNAs, with a stretch perfectly complementary with the target mRNA. Now, it happens that some viral RNAs adopt a stem loop conformation recognized by Drosha or Dicer as a suitable substrate and are processed to iRNAs, and we will call these viral iRNAs (viRNAs) to distinguish them from the endogenous cellular miRNAs. As the characteristics of Drosha and Dicer substrates are not fully defined, the fact that a viral RNA will or will not be processed to an iRNA cannot at present be accurately predicted. Likewise, the characteristics that make an mRNA a target for a given iRNA, including its own structure ##REF##15687388##[14]##,##REF##14762201##[15]##, its accessibility, i.e., its localization in the cell and its association with other components of the cell ##REF##12915586##[16]##, are not absolutely established. Consequently, any bioinformatic study predicting that a viral RNA will be processed into viRNA or that an mRNA will be silenced by an iRNA has to be confirmed experimentally. It must also be kept in mind that many studies analyze the effect on viral RNA of iRNAs at high concentrations obtained after transfection, and that such results must be confirmed in the course of the infection at physiological concentrations of iRNAs.</p>", "<title>RNA Interferences in the Cell/Virus Complex</title>", "<p>Viral RNAs and the miRNA machinery may interfere in various ways (##FIG##1##Figure 2##). First, the viRNAs generated by the cell from virus RNAs may target back viral RNAs (pathway 1), but also cell mRNAs that happen to share some sequence homology with them (pathway 2). Second, cellular miRNAs may recognize viral RNAs and silence them (pathway 3). Finally, the cell may produce miRNAs that control the expression of a cellular protein necessary for the virus life cycle (pathway 4). We will review these various possibilities, comparing examples of what is already known for other viruses with what is known for HIV.</p>", "<title/>", "<title>Autosilencing: viRNA against viral RNA (pathway 1)</title>", "<p>Obviously, some viRNAs will match exactly with some viral RNAs issued from the same genomic sequence, and induce their destruction. This is the case, for instance, in the course of simian virus 40 (SV40) infection. The virus encodes two RNAs with a stem loop structure that are transformed by the host cell into miR-S1s that are complementary to mRNA coding for the viral T antigen. Consequently, T antigen mRNA is degraded ##REF##15931223##[17]##. In this scenario, host miRNA machinery turns viral RNA against viral RNA. The same phenomenon has been looked for in HIV infection. A precursor ##REF##15782219##[18]## and a recent ##REF##17855543##[19]## study by two major groups in the field failed to identify iRNA of HIV origin in infected cells. Yet, a nef- and LTR-specific HIV miRNA able to inhibit LTR-driven transcription has been evidenced by another group ##REF##15601474##[20]##,##REF##15722536##[21]##. Moreover, a fourth group has reported that a stem loop HIV RNA can be processed by Dicer to a viRNA able to target env mRNA and that transfection of the corresponding shRNA inhibited over 80% of env mRNA production ##REF##15894278##[22]##, a result that was challenged later ##REF##17855543##[19]##. Finally, Klase et al. ##REF##17663774##[23]## and Ouellet et al. ##REF##18299284##[24]## have shown that TAR is a source of viRNAs: Dicer interacts with TAR, and cleaves it to produce TAR-derived viRNAs able to exert gene downregulatory effects. The level of anti-HIV activity of this pathway, which is controversial, remains to be evaluated in the context of the infection. Anyway, the antiviral effect of HIV viRNAs is at best partial.</p>", "<title>viRNA against cell mRNA (pathway 2)</title>", "<p>It may happen that the host cell transforms a viral RNA into a viRNA with some degree of homology with a cell mRNA. In this case, the viRNA will inhibit the expression of the corresponding cellular gene. Such a possibility has been recently described during human cytomegalovirus (HCMV) infection. HCMV encodes an RNA transformed by the infected cell into a viRNA, hcmv-miR-UL112, that blocks the translation of major histocompatibility complex class I–related chain B (MICB). Of note, MICB is a stress-induced molecule expressed on HCMV-infected cells that is recognized by the natural killer (NK) cell activating receptor NKG2D ##REF##17641203##[25]##. Likewise, are there viRNAs of HIV origin able to target cell mRNAs? Computer-directed analyses have evidenced regions of base complementarity between HIV-1 sequences and human genes involved in HIV infection, e.g., <italic>CD4, CD28, CD40L, IL-2, IL-3, IL-12,</italic> and <italic>TNFβ</italic>\n##REF##15862587##[26]##. Moreover, Bennasser et al. have predicted the existence of five HIV RNAs that could be transformed by Dicer into five viRNAs able to target various cellular mRNAs ##REF##15601472##[27]##. If these predictions are experimentally confirmed, HIV could also manipulate cellular biology and host immune response via viRNAs that were produced by the infected cell.</p>", "<title>Cellular miRNA against virus RNA (pathway 3)</title>", "<p>Sometimes the cell does not borrow viral RNAs to produce viRNAs with antiviral activity, but rather uses its own miRNAs. A striking example of this strategy is represented by the cell miR-32 that targets the open reading frame 2 of the primate foamy virus type 1, thereby inhibiting virus translation ##REF##15845854##[28]##. In this way, a cellular miRNA may directly target a virus RNA and block virus production. Interestingly, the same process results in the opposite effect for hepatits C virus (HCV). The endogenous miR-122, which is specifically produced by liver cells, binds to the 5′ noncoding region of HCV, but, unusually, this results in an increase in viral RNA replication through a mechanism that remains to be unraveled ##REF##16141076##[29]##. Interactions between cellular miRNAs and HIV RNAs may also exist, in so far as an in silico study predicted that five human miRNAs, expressed in T cells, might target nef, vpr, vif, and env RNAs ##REF##16236258##[30]##. Recently, Huang et al. have shown that miR-28, -125b, -150, -223, and -382, cellular miRNAs that are overexpressed in quiescent T4 lymphocytes, target sequences in the 3′ end of HIV-1 RNA, silencing thus almost all viral messengers ##REF##17906637##[31]##. Neutralizing these cellular miRNAs by transfecting specific antagonists into nonactivated T4 cells from patients with HIV under highly active antiretroviral therapy increased by 10-fold the in vitro efficiency of virus isolation. These observations strongly argue for a role of cellular miRNAs in HIV latency. This hypothesis, together with the suppression exerted by HIV on miRNA biosynthesis, might partly explain why HIV has not mutated to escape from the inhibitory effect of iRNAs of viral or cellular origin.</p>", "<title>Cellular miRNA targeting cell mRNA encoding proteins involved in virus replication (pathway 4)</title>", "<p>Finally, a fourth situation, where host miRNAs limit viral proliferation by acting on cellular mRNA, has been recently proposed for HIV ##REF##17322031##[32]##. Triboulet et al. have shown that the cellular miRNAs miR-17-5p and miR-20 silence the mRNA encoding the histone acetylase PCAF. PCAF has been previously presented as a host cofactor for Tat transactivation of HIV LTR, and as being recruited by Tat and remodeling the histone architecture in the vicinity of the LTR, promoting thereby HIV gene expression (##FIG##2##Figure 3A##). This is to say that human cells seem to permanently downregulate HIV-1 replication by depriving the infected cell of an endogenous enzyme that could be necessary for virus gene expression.</p>", "<title>Regulation of the miRNA Pathway by the Virus</title>", "<p>The few examples we have just reviewed suggest that, most of the time, the miRNA machinery works against the virus. It is not surprising then, that many viruses have elaborated strategies to subvert this machinery. And HIV is not the last one.</p>", "<title/>", "<title>Inhibition of miRNA production: viral suppressors of RNA silencing</title>", "<p>A strategy adopted by many viruses involves binding to a component of the miRNA machinery in order to block it. For instance, the adenoviral protein VA1 inhibits the nucleo-cytoplasmic transport of pre-miRNA by forming a complex with exportin 5 ##REF##15542639##[33]##. HIV seems to target two other factors of RNA silencing, Dicer and TRBP (##FIG##3##Figure 4##). Bennasser et al. have shown that purified Tat protein inhibits the capacity of Dicer to process double-stranded RNA to short iRNAs in vitro ##REF##15894278##[22]##. Moreover, the TAR and RRE sequences of HIV RNA, by recruiting TRBP, could competitively inhibit the effect of TRBP on pre-miRNA processing and on miRNA function ##REF##16253139##[34]##,##REF##17360756##[35]##. Yet, the capacity of Tat to hinder miRNA biogenesis has been recently challenged ##REF##17855543##[19]##, and the impact on HIV infection of these inhibitions of miRNA expression remains to be quantified. Anyhow, if HIV encodes suppressors of RNA silencing, their effect is obviously incomplete.</p>", "<p>This double mechanism of suppression of RNA silencing strongly suggests that the interest of HIV is to counteract RNA interference; that is to say, the effect of the miRNA machinery is globally harmful for the virus. In support of this notion, a point mutation in Tat, which abrogates its inhibitory effect on Dicer, but not its transactivation effect on HIV LTR, results in a reduction in virus replication ##REF##15894278##[22]##. Likewise, Haasnoot et al. have shown that a Tat protein able to block Dicer activity is necessary for efficient HIV production ##REF##17590081##[36]##. Moreover, Triboulet et al. have reported that the inhibition of Drosha or Dicer using specific siRNA increases HIV replication in peripheral blood mononuclear cells from individuals with HIV ##REF##17322031##[32]##.</p>", "<title>Viral up- and downregulation of miRNA concentrations</title>", "<p>In addition to globally diminishing miRNA production, viruses may also specifically regulate, positively or negatively, the level of expression of some miRNAs. This is the case for tobacco mosaic virus, where infection results in an increase in miR-156, -160, -164, -166, -169, and -171 in <italic>Nicotiana tabacum</italic>\n##REF##17615233##[37]##. HIV infection also results in such a phenomenon. Triboulet et al. have reported that the infection of a lymphoid cell line by HIV-1 downregulates six miRNAs, including miR-17-5p and miR-20, and upregulates eleven others, including miR-122, -297, -370, and -373 ##REF##17322031##[32]##. The mechanism responsible for this selective regulation by HIV remains to be unveiled. During tobacco mosaic virus infection, it is the accumulation of two viral proteins, the movement protein and the coat protein, that is responsible for the change in miRNAs concentrations ##REF##17615233##[37]##. Of note, two of the miRNAs downregulated during HIV infection are precisely miR-17-5p and miR-20, which target the Tat cofactor PCAF. Thus, by reducing the amount of miR-17-5p and miR-20 available, the virus alleviates the negative control exerted by the cell on PCAF, thereby facilitating its own transcription (##FIG##2##Figure 3B##).</p>", "<title>Consequences of the Interactions between iRNAs and the Other RNAs during Infection</title>", "<p>Thus, viruses and the miRNA machinery of the host cell interact in various ways. These interactions may have consequences on the replication and the pathogenicity of the virus, but also on the immune response of the host.</p>", "<title/>", "<title>Consequences for virus replication/latency</title>", "<p>Interactions between iRNAs and other RNAs in the infected cell may have consequences on the virus life cycle, mostly negative, with one exception, HCV. For HIV, viRNA complementary to env, nef, and/or LTR sequences ##REF##15601474##[20]##–##REF##15894278##[22]## and cellular miRNAs targeting the 3′ end of HIV RNAs ##REF##17906637##[31]## could inhibit virus replication. Moreover, the cellular miR-17-5p and miR-20, through their repression of PCAF expression (##FIG##2##Figure 3B##), seem to exert the same effect ##REF##17322031##[32]##. In these various ways, RNA interference might downregulate HIV production. Logically, the virus tries to counter this effect by blocking Dicer activity and by hijacking TRBP.</p>", "<p>The global impact on in vivo infection of the interplay between virus and cell RNAs, and the relative importance of each interaction, remain to be determined. This interplay could influence the efficiency of HIV replication, and thereby the rate of disease progression, but it could also be involved in HIV latency and in the constitution of the viral reservoir. Finally, differences in the anti-HIV efficiency of RNA inteference could exist between individuals. These differences might be due to genetic polymorphisms in sequences regulating miRNA gene transcription or stability, as well as polymorphisms in cellular miRNA sequences resulting in variations in processing or targeting as already described for miR-138-2 and miR-30c-2 ##REF##15883035##[38]##.</p>", "<title>Consequences on the immune response</title>", "<p>Inhibition of virus gene expression by RNA interference may downregulate virus replication in vitro, but in vivo, this downregulation might help the virus to escape from the immune response. This is the case for the above-mentioned inhibition exerted by the SV40 miRNAs miR-S1s on the T antigen (see “Autosilencing: viRNA against virus RNA (pathway 1)”). This inhibition occurs late enough in the virus life cycle not to hinder SV40 replication, but early enough to reduce T lymphocyte cytotoxicity and interferon-γ production triggered by the presence of the viral T antigen ##REF##15931223##[17]##. The targeting of cell mRNA by viRNA (pathway 2) may also result in the reduction of immune response. This possibility is illustrated by the inhibition of MICB, a ligand for NK cell activating receptor, by an HCMV miRNA (“viRNA against cell mRNA (pathway 2)”). This targeting is not innocent, since it results in a decrease in antiviral NK activity. Some cellular miRNAs are involved in innate immunity, e.g., miR-155 induced by Toll-like receptor ligands or interferon-β (IFN-β) ##REF##17242365##[39]## and miRNA let-7a mediating IL-6-induced cell survival ##REF##17220301##[40]##. Some cellular miRNAs are involved in specific immunity, e.g., miR-155 regulating genes driving functions of lymphocytes ##REF##17463290##[41]##,##REF##17463289##[42]##, miR-150 controlling the differenciation of B cells via c-Myb ##REF##17923094##[43]##, and miR-181a modulating the antigen sensitivity of T cells ##REF##17382377##[44]##. A striking example of the role played by miRNAs in immunity has been lately given by Pedersen et al. This group has reported that some of the anti-HCV effect of IFN-β is mediated by five cellular miRNAs, induced by IFN-β, that target HCV RNA ##REF##17943132##[45]##. Consequently, changes in miRNA expression induced by HIV might also alter the immune response.</p>", "<title>Consequences on the pathogenicity</title>", "<p>These virus/cell, iRNA/RNA interactions could also have consequences on the pathogenicity of the infection. Because cellular miRNAs regulate major mechanisms of cell physiology, such as proliferation, differentiation, apoptosis, or tumorogenicity ##REF##15372042##[46]##, targeting of cell mRNAs by viRNAs (pathway 2) and/or specific or global virus-induced disturbance of the cellular miRNA production might have pathogenic effects. The extent to which the modifications of host biology observed in individuals with HIV are mediated by RNA interference remains to be addressed, however.</p>", "<title>Therapeutic Opportunities</title>", "<p>Besides strategies using synthetic interfering RNAs (siRNAs or shRNAs) to target mRNA encoding cell proteins necessary for virus replication or directly target HIV RNA, the knowledge of HIV RNA/RNA silencing connections might pave the way for new therapeutic approaches. The fact that siRNAs or siRNAs antagonists modified chemically for stability and conjugated to cholesterol or liposomes have a specific, long-lasting, and ubiquitous effect when administrated intravenously ##REF##16258535##[47]##,##REF##16565705##[48]## is particularly encouraging for this kind of approach.</p>", "<p>As interfering RNAs disturb HIV replication, the possibility to act on the viral infection through miRNAs must be considered. At least two opposite strategies may be proposed. First, the miRNA machinery could be manipulated in order to hinder virus replication. Second, because of the involvement of miRNA in HIV latency, the miRNA machinery could be manipulated in order to provoke virus replication in the reservoir cells with the aim of lysing them. This second approach has the advantage of only requiring a brief treatment that is sufficient to induce HIV expression in infected, non-productive quiescent cells. To reach this goal, a possibility is to target host miRNA. For instance, the neutralization of miR-17 and miR-20 by specific inhibitors should release the inhibition exerted by these miRNAs on PCAF and hopefully trigger HIV replication in reservoir cells, resulting in the eradication of this population. A drawback of such a strategy is that the inhibition of a given miRNA will increase the level of expression of all of its mRNA targets with possible side effects. As our knowledge on the impact of the disturbance induced by HIV on host immunity and cell physiology increases, new therapeutic strategies may arise.</p>", "<p>Last, the manipulation of miRNAs in order to boost HIV production in vitro may ameliorate the efficiency of virus recovery from patients' cells ##REF##17906637##[31]##,##REF##17322031##[32]##, and of virus mass production.</p>" ]

[ "<p>I am grateful to Kuan-Teh Jeang and Eamonn Fahy for critical reading of the manuscript.</p>" ]

[ "<fig id=\"ppat-1000162-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000162.g001</object-id><label>Figure 1</label><caption><title>Biosynthesis and Activity of iRNAs</title></caption></fig>", "<fig id=\"ppat-1000162-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000162.g002</object-id><label>Figure 2</label><caption><title>Interactions between iRNAs, Cellular mRNAs, and HIV RNAs</title></caption></fig>", "<fig id=\"ppat-1000162-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000162.g003</object-id><label>Figure 3</label><caption><title>Inhibition of HIV Expression by the Silencing Exerted by miR-17-5p and miR-20 on PCAF (A), and Downregulation of This Inhibition by HIV (B)</title></caption></fig>", "<fig id=\"ppat-1000162-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000162.g004</object-id><label>Figure 4</label><caption><title>Regulation of the miRNA Machinery by HIV</title></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<fn-group><fn fn-type=\"COI-statement\"><p>The author has declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p>The author received no specific funding for this work.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"ppat.1000162.g001\"/>", "<graphic xlink:href=\"ppat.1000162.g002\"/>", "<graphic xlink:href=\"ppat.1000162.g003\"/>", "<graphic xlink:href=\"ppat.1000162.g004\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[]

[]

[]

[]

[]

[ "<p>The behavior of chromosomes during meiosis has been likened to a middle school dance, where partners find one another, form couples that move about and trade information, and then separate to opposite sides of the dance hall. With chromosomes, as with the dancers, forming exclusive couples often is difficult—individuals can be attracted to more than one partner or find themselves trapped behind or between other couples—and, failing to form a couple effectively, end up on the wrong side of the dance hall. For chromosomes, this failure of pairing and segregation leads to an unbalanced chromosome complement (aneuploidy), with its attendant problems of sterility and genetic disease. Two papers in this issue of <italic>PLoS Genetics</italic>\n##REF##18818742##[1]##,##REF##18818741##[2]## demonstrate that telomere-promoted movements influence nearly every step in chromosome pairing and meiotic recombination, opening a new avenue to address questions that have intrigued biologists and vexed clinicians for over a hundred years.</p>", "<p>Chromosome movement is implicit in the classically recognized stages of meiotic prophase, but descriptions of directly visualized movements have been rare (see ##REF##9928494##[3]##). Early in prophase, chromosomes transition from having their centromeres clustered near the spindle pole (the Rabl orientation) to having their telomeres clustered at the nuclear periphery adjacent to the spindle pole (the bouquet orientation; see ##FIG##0##Figure 1##). The bouquet stage ends with dispersal of telomeres across the inner nuclear envelope as chromosomes finalize their intimate pairing by forming synaptonemal complexes (SCs) that link chromosome pairs closely along their lengths. The formation of these intimate, exclusive partnerships would seem to finish the task at hand and to end the need for active, whole-chromosome movements, but this turns out not to be the case.</p>", "<p>Movements that persist throughout meiotic prophase were first described by Hiraoka's group in the fission yeast <italic>Schizosaccharomyces pombe</italic> where, following bouquet formation, telomeres remain at the spindle pole while it leads the nucleus along microtubules, back and forth through the cell, until just before the first meiotic division ##REF##8146661##[4]##. Meiotic prophase is noncanonical in <italic>S. pombe</italic> in that synaptonemal complexes are not formed and recombination is not regulated to avoid forming crossovers near one another (i.e., there is no positive crossover interference). This has led some to question the generality of persistent movements. Recently, however, similarly persistent rapid prophase movements (RPMs) have been described in the budding yeast <italic>Saccharomyces cerevisiae</italic>\n##REF##16027219##[5]##–##REF##18585353##[7]##. Although these movements are of individual chromosomes rather than of the whole genomic complement, and although they appear to be promoted by actin rather than by microtubules, each system involves SUN domain–containing proteins that are known to mediate transnuclear envelope linkages, in the present case tethering telomeres to the cytoskeleton ##REF##18585352##[8]##–##REF##17495028##[10]##. Such linkages also are present in mammalian meiotic nuclei ##REF##17452644##[11]##,##REF##17543860##[12]##, indicating a widely conserved mechanism and suggesting conserved function(s).</p>", "<p>Before its role in these movements was recognized, the budding yeast Ndj1 protein was known to promote bouquet formation, the normal kinetics of SC formation, and the usual pattern of meiotic recombination; to maintain low levels of ectopic recombination (genetic exchanges between homologous DNA sequences in nonallelic locations); and, ultimately, to reduce the frequency of aneuploidy ##REF##9242487##[13]##–##REF##16648465##[17]##. Ndj1 also plays a role in anchoring telomeres to the inner nuclear envelope ##REF##11018056##[15]##, apparently by stabilizing the association of telomeres with the transmembrane SUN protein, Mps3 ##REF##17495028##[10]##. Reports that the meiosis-specific budding yeast protein Csm4 is similar to Ndj1 in being required to prevent aneuploidy ##REF##11470404##[18]## led the authors of the two current papers to ask, in remarkable molecular detail, whether the meiotic requirements for Csm4 are similar to those for Mps3 and Ndj1. The simple answer is “Yes,” but the angel is in the details.</p>", "<p>The authors find that Csm4, unlike Mps3 and Ndj1, is <italic>not</italic> required to anchor telomeres to the nuclear envelope but is required for telomeres to engage in the RPMs (see ##REF##18585353##[7]##,##REF##18585352##[8]##). Nevertheless, the impact on the progress of recombination, in all its currently understood molecular intricacies, is similar—delays in the appearances of recombination intermediates begin very early in prophase and persist or lengthen as prophase progresses. The implication of these observations is that the RPMs are the critical factor rather than telomere tethering to the nuclear envelope per se. So then, what is the role of the RPMs? Here, the authors diverge somewhat in their answers. The paper from the Shinohara lab proposes that RPMs promote the biochemistry of recombination more or less directly, perhaps by affecting chromosome structure ##REF##18818742##[1]##. The paper from the Alani and Kleckner labs proposes that RPMs function during an early phase when the cell determines which early recombination intermediates will become crossovers. They suggest that delays in this phase, perhaps due to a requirement for Ndj1 and Csm4 to resolve chromosome interlocks (at zygotene, see ##FIG##0##Figure 1##), generates the subsequent defects ##REF##18818741##[2]##. Tests of these hypotheses will require considerable ingenuity in experimental design.</p>", "<p>A simple and striking conclusion from these papers is that mechanical energy, pumped into the nucleus via the telomeres, contributes critically to the work of genetic recombination. Identification of the MNC complex (Mps3, Ndj1, Csm4) in <italic>Sa. cerevisiae</italic> and of related structures and pathways in other organisms is only the beginning to understanding how these transnuclear envelope tethers are constructed and regulated. Understanding how these connections and the movements they foster contribute to the faithful segregation of chromosomes in meiosis will be challenging and rewarding, like the middle school dance.</p>" ]

[]

[ "<fig id=\"pgen-1000210-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pgen.1000210.g001</object-id><label>Figure 1</label><caption><title>Chromosomes Pass through Distinct Organizational Phases as They Negotiate Meiotic Prophase.</title><p>The Rabl orientation is established in the prior division, where centromeres (green balls) are pulled to the spindle pole (dark blue structure with emanating microtubules, anchored in the nuclear membrane), with the telomeres (red balls) trailing. Early in prophase, centromeres move away from the pole while telomeres attach to the nuclear membrane and move to a small area adjacent to the spindle pole, forming the bouquet. Chromosomes move out of the bouquet as the synaptonemal complex forms (red ladder-like structure), marking entry into zygotene and generating interlocks, where chromosomes are trapped between synapsing pairs. Completion of synapsis and resolution of interlocks marks pachytene, where chromosome pairs appear well separated.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"pgen.1000210.g001\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Understanding an organism's metabolism at a system level and obtaining quantitative predictions for the different metabolic variables requires the identification and modeling of the physicochemical and regulatory constraints that are relevant at physiological growth conditions. Recently, there has been a surge of interest on how macromolecular crowding, i.e., the crowding of the cytoplasm by various molecular components, affects cellular function, including cell metabolism ##REF##11590012##[1]##,##REF##16825427##[2]##.</p>", "<p>At the local scale it is well known that molecular crowding affects the rate of biochemical reactions, diffusion, protein folding and protein-protein association/dissociation ##REF##16825427##[2]##,##REF##16740119##[3]##. More recently, we have shown that macromolecular crowding acts also at a global scale by imposing a limited solvent capacity. Specifically, we have shown that a flux balance modeling framework that incorporates the limited solvent capacity is successful in predicting the maximum growth rate, the sequence of substrate uptake from a complex medium and, to an extent, the changes in intracellular flux rates upon varying growth rate of the bacterium, <italic>Escherichia coli</italic>\n##REF##17652176##[4]##,##UREF##0##[5]##. Yet, these studies were limited by the absence of a full kinetic model of <italic>E. coli</italic> cell metabolism, hindering our ability to investigate the impact of the solvent capacity constraint on in vivo metabolite concentrations and enzyme activities.</p>", "<p>During cellular metabolism the concentration of enzymes and metabolites are continuously adjusted in order to achieve specific metabolic demands. It is highly likely that during evolution global metabolic regulation has evolved such as to achieve a given metabolic demand with an optimal use of intracellular resources. However, the size of enzymes and intermediate metabolites are dramatically different. Enzymes are macromolecules that occupy a relatively large amount of space within a cell's crowded cytoplasm, while metabolites are much smaller. This implies that metabolite concentrations are likely to be adjusted to minimize the overall “enzymatic cost” (in terms of space cost).</p>", "<p>Here we study the validity of this hypothesis by focusing on the glycolysis pathway of the yeast, <italic>Saccharomyces cerevisiae</italic>, for which a kinetic model is available. We show that the maximum glycolysis rate determined by the limited solvent capacity is close to the values measured in vivo. Furthermore, the measured concentration of intermediate metabolites and enzyme activities of glycolysis are in agreement with the predicted values necessary to achieve this maximum glycolysis rate. Taken together these results indicate that the limited solvent capacity constraint is relevant for <italic>S. cerevisiae</italic> at physiological conditions. From the modeling point of view, this work demonstrates that a full kinetic model together with the limited solvent capacity constraint can be used to predict not only the metabolite concentrations, but in vivo enzyme activities as well.</p>" ]

[ "<title>Methods</title>", "<title>Kinetic Model of Glycolysis</title>", "<p>We use the <italic>S. cerevisiae</italic> glycolysis model reported in ##REF##11744196##[7]## (see ##SUPPL##0##Protocol S1## for details). The only modification is the extension of the phsophofructokinase (<italic>pfk</italic>) kinetic model from an irreversible to a reversible model.</p>", "<title>Catalytic Constants, Cell Density, Specific Volume</title>", "<p>The catalytic constants were obtained from experimental estimates for <italic>Saccharomyces carlsbergensis</italic>\n##REF##6115423##[16]##, except for glycerol 3-phosphate dehydrogenase that was obtained from an estimate for <italic>Eidolon helvum</italic>\n##REF##11752250##[17]##. For the cell density we use an estimate reported for <italic>E. coli</italic>, <italic>ρ</italic> = 0.34 g/ml ##REF##1748995##[18]##. The specific volume was estimated for several proteins using the molar volumes and masses reported in ##REF##6572909##[6]##, resulting in average of 0.73 ml/g and standard deviation of 0.02 ml/g. See ##SUPPL##0##Protocol S1## for details.</p>", "<title>Optimal Metabolite Concentrations</title>", "<p>The optimal metabolite concentrations are obtained maximizing Equation 4 with respect to the free metabolite concentrations. In the case of ##FIG##1##Figure 2B–2D##, all metabolite concentrations are fixed to their experimental values, except for the metabolite indicated by the X-axis. In the case of ##FIG##2##Figure 3A and 3B##, all intermediate metabolite concentrations are optimized, keeping fixed the concentration of external glucose and cofactors (ATP, ADP, AMP, NADH, NAD). In both cases the reaction rates relative to the glycolysis rate (<italic>r_i</italic>) were taken as input parameters, using the values reported in ##REF##11744196##[7]##. The maximization was performed using simulated annealing ##UREF##6##[19]##.</p>", "<title>Parameter Sensitivity</title>", "<p>To analyze the sensitivity of our predictions to the model parameters we have generated random sets of kinetic parameters, assuming a 10% variation of the fixed metabolite concentrations and all kinetic constants except for the catalytic activities. For the latter we assumed a larger variation of 50%, because they were estimated from a different organism. For each set of parameters we make predictions for the metabolite concentrations and enzyme activities. ##FIG##2##Figure 3## reports the mean values and standard deviations.</p>" ]

[ "<title>Results</title>", "<title>Limited Solvent Capacity Constraint</title>", "<p>The cell's cytoplasm is characterized by a high concentration of macromolecules ##REF##11590012##[1]##,##REF##16825427##[2]## resulting in a limited solvent capacity for the allocation of metabolic enzymes. More precisely, given that enzyme molecules have a finite molar volume <italic>v_i</italic> only a finite number of them fit in a given cell volume <italic>V</italic>. Indeed, if <italic>n_i</italic> is the number of moles of the <italic>i</italic>\n^th enzyme, thenwhere <italic>V</italic>\n₀ accounts for the volume of other cell components and the free volume necessary for cellular transport as well. Equation 1 can be also rewritten aswhere <italic>ρ_i</italic> = <italic>n_im_i</italic>/<italic>V</italic> is the enzyme density (enzyme mass/volume), <italic>μ_i</italic> is the molar mass <italic>v</italic>\n_spec is the specific volume, and <italic>φ</italic> = <italic>V</italic>\n₀/<italic>V</italic> is the fraction of cell volume occupied by cell components other than the enzymes catalyzing the reactions of the pathway under consideration, including the free volume necessary for diffusion. The specific volume has been assumed to be constant for all enzymes, an approximation that has been shown to be realistic at least for globular proteins ##REF##6572909##[6]##. In this new form we can clearly identify the enzyme density (or mass, given that the volume is constant) as the enzyme associated variable contributing to the solvent capacity constraint. This choice is more appropriate than the enzyme concentration <italic>C_i</italic> = <italic>n_i</italic>/<italic>V</italic> (moles/volume) because the specific volume is approximately constant across enzymes, while the molar volume can exhibit significant variations. For example, according to experimental data for several globular proteins ##REF##6572909##[6]##, the molar volume exhibits a 70% variation while the specific volume is almost constant, with a small 2% variation.</p>", "<p>The solvent capacity constraint (Equations 1 and 2) thus imposes a limit to the amount of catalytic units (i.e., enzymes) that can be allocated in the cell cytoplasm. In the following we show that this in turn leads to a constraint for the maximum metabolic rate. The rate of the <italic>i</italic>\n^th reaction per unit of cell dry weight (mol/time/mass) is given bywhere <italic>A_i</italic> is the specific enzyme activity, <italic>C_i</italic> is the enzyme concentration in molar units, <italic>k_i</italic> is the catalytic constant and <italic>M</italic> is the cell mass. The coefficient <italic>x_i</italic> is determined by the specific kinetic model: it takes values in the range of 0≤<italic>x_i</italic>≤1, and it is a function of metabolite concentrations. For example, if the <italic>i</italic>\n^th reaction is described by Michaelis-Menten kinetics with one substrate then <italic>x_i</italic> = <italic>S_i</italic>/(<italic>K_i</italic>+<italic>S_i</italic>), where <italic>S_i</italic> is the substrate concentration and <italic>K_i</italic> is the equilibrium constant. More generally, <italic>x_i</italic> is a function of the concentration of substrates, products and other metabolites regulating the enzyme activity. The fact that the reaction rates are proportional to the enzyme densities (Equation 3) suggests that the limited solvent capacity constraint (Equation 2) has an impact on the reaction rates as well. Indeed, from Equations 2 and 3 we obtainwhere <italic>R</italic> is the cell metabolic rate (or pathway rate), <italic>r_i</italic> = <italic>R_i</italic>/<italic>R</italic> is the rate of reaction <italic>i</italic> relative to the metabolic rate, andwhere <italic>ρ</italic> = <italic>M</italic>/<italic>V</italic> is the cell density. We refer to <italic>a_i</italic> as the crowding coefficients ##REF##17652176##[4]##,##UREF##0##[5]##, because they quantify the contribution of each reaction rate to molecular crowding. The crowding coefficient of a reaction <italic>i</italic> increases with increasing the enzyme's molar mass <italic>μ_i</italic> and decreases with increasing catalytic activity <italic>k_i</italic>. It is also a function of the metabolite concentrations through <italic>x_i</italic>.</p>", "<title>Hypothetical Three Metabolites Pathway</title>", "<p>To illustrate the impact of the limited solvent capacity constraint, we first analyze a hypothetical example, in which we use the relative reaction rates as input parameters, and the metabolite concentrations are the variables to be optimized. Given the reaction rates and the “optimal” metabolite concentrations, the enzyme activities are determined by Equation 3. Finally, the maximum metabolic rate is computed using Equation 4.</p>", "<p>Consider a metabolic pathway consisting of two reversible reactions converting metabolite M₁ into M₂ (reaction 1) and M₂ into M₃ (reaction 2), catalyzed by enzymes <italic>e</italic>\n₁ and <italic>e</italic>\n₂, respectively (##FIG##0##Figure 1, inset##). The reaction rates per unit of cell mass, <italic>R</italic>\n₁ and <italic>R</italic>\n₂, are modeled by reversible Michaelis-Menten rate equations, using Equation 3 with\nwhere <italic>K</italic>\n_1eq and <italic>K</italic>\n_2eq are the equilibrium constants of reaction 1 and 2, respectively, <italic>K_im</italic> is the Michaelis-Menten constant of metabolite <italic>m</italic> in reaction <italic>i</italic>. From Equations 4 to 7 we finally obtainFor the purpose of illustration, we assume 1−<italic>ϕ</italic> = 0.01, (mmol/h/min)⁻¹ (as suggested by typical values reported in ##UREF##0##[5]##), all Michaelis constants equal to 1 mM, and fixed pathway ends metabolite concentrations [M₁] = 3 mM and [M₂] = 1 mM. Furthermore, mass conservation for M₂ implies that <italic>R</italic>\n₁ = <italic>R</italic>\n₂ = <italic>R</italic> (<italic>r</italic>\n₁ = <italic>r</italic>\n₂ = 1) in the steady state, where <italic>R</italic> is the pathway rate. When reaction 1 is close to equilibrium [M₂]≈[M₁]<italic>K</italic>\n_1eq = 3 mM, the first term in the right hand side becomes very large, resulting in a small pathway rate (##FIG##0##Figure 1##). When reaction 2 is close to equilibrium [M₂]≈[M₃]/<italic>K</italic>\n_2eq = 1 mM, the second term in the right hand side becomes very large, again resulting in a small pathway rate (##FIG##0##Figure 1##). At an intermediate [M₂]^* between these two extremes the pathway rate achieves its maximum. Therefore, given the solvent capacity constraint, there is an optimal metabolite concentration resulting in a maximum pathway rate.</p>", "<title>\n<italic>S. cerevisiae</italic> Glycolysis</title>", "<p>Next, we investigate whether the observation of an optimal metabolite concentration for maximum pathway rate extrapolates to a more realistic scenario. For this purpose we use the glycolysis pathway of the yeast <italic>S. cerevisiae</italic> (##FIG##1##Figure 2A##) as a case study. Glycolysis represents a universal pathway for energy production in all domains of life. In <italic>S. cerevisiae</italic> it has been studied extensively resulting in the description of a rate equation model for each of its reactions ##REF##11744196##[7]##,##REF##10951190##[8]##. In particular, we consider the kinetic model developed in ##REF##11744196##[7]## (see <xref ref-type=\"sec\" rid=\"s4\">Methods</xref>). To compare our predictions with experimentally determined values we consider the glycolysis reaction rates and metabolite concentrations reported in ##REF##11744196##[7]## and the enzyme activities reported in ##REF##10951190##[8]##.</p>", "<p>In analogy with the three metabolites case study (##FIG##0##Figure 1##), first we investigate the dependency of the glycolysis rate <italic>R</italic>, represented by the glucose uptake, on the concentration of a given metabolite. In this case we fix all other metabolite concentrations and all relative reaction rates (reaction rate/glycolysis rate) to their experimentally determined values. By doing so the predicted glycolysis rate is an implicit function of the free metabolite concentration alone, through Equation 4. For example, ##FIG##1##Figure 2B## displays the maximum metabolic rate <italic>R</italic> as a function of the concentration of fructose-6-phosphate (F6P). <italic>R</italic> is predicted to achieve a maximum around a F6P concentration of 0.4 mM, close to its experimentally determined value of 0.5 mM ##REF##11744196##[7]## (red triangle in ##FIG##1##Figure 2B##). Similar conclusions are obtained for D-glyceraldehyde-3-phosphate (GAP) (##FIG##1##Figure 2C##) and glycerone-phosphate (DHAP) (##FIG##1##Figure 2D##). This analysis corroborates that there is an optimal metabolite concentration maximizing <italic>R</italic> and, more importantly, that this concentration is very close to the experimentally determined metabolite concentrations. In all cases the measured metabolite concentrations are within the range of 50% or more of the maximum glycolysis rate.</p>", "<p>To further test the optimal metabolite concentration hypothesis, we perform a global optimization and simultaneously compute the optimal concentrations of the glycolysis intermediate metabolites. In this case we fix the concentrations of external glucose and co-factors and all relative reaction rates to their experimentally determined values. By doing so the predicted glycolysis rate is an implicit function of the glycolysis intermediate metabolite concentrations, through Equation 4. The optimal intermediate metabolite concentrations are those maximizing Equation 4. ##FIG##2##Figure 3A## displays the predicted optimal metabolite concentrations as a function of their experimentally determined values (black symbols), the line representing a perfect match. The agreement is remarkably good given the wide range of metabolite concentrations. For phospho-enol-pyruvate (PEP), the predicted value is very sensitive to the model parameters, as indicated by the wide error bars. For fructose 1,6-biphosphate (FBP) the predicted value is smaller by a factor of five than the experimental value, but it is still within range. Taken together, these results indicate that the measured concentrations of intermediate metabolites in the <italic>S. cerevisiae</italic> glycolysis are close to the predicted optimal values maximizing the glycolysis rate given the limited solvent capacity constraint.</p>", "<p>Using the optimal intermediate metabolite concentrations we can make predictions for the enzyme activities as well. Indeed, from the first equality in Equation 3 it follows thatThe reaction rates relative to the glycolysis rate <italic>r_i</italic> are obtained from experimental data, while <italic>x_i</italic> are obtained after substituting the predicted optimal metabolite concentrations on the reaction's kinetic models. ##FIG##2##Figure 3B## displays the predicted enzyme activities (in units of the glycolysis rate) as a function of the experimentally determined values (black symbols), the line representing a perfect match. In most cases we obtain a relatively good agreement between experimentally measured and predicted values, with the exception of phosphofructokinase (<italic>pfk</italic>), for which the measured enzyme activities are significantly overestimated. Of note, for pyruvate kinase (<italic>pk</italic>) the predictions are significantly affected by the model parameters, as indicated by the wide error bars.</p>", "<p>The preceding analysis does not exclude the possibility that other constraints could result in a good agreement as well. To address this point we consider the more general optimization objective <italic>R</italic> = (1−<italic>ϕ</italic>)/Σ^{<italic>N</italic>}\n_{<italic>i</italic> = 1} (<italic>a_ir_i</italic>)<italic>^H</italic>, parametrized by the exponent <italic>H</italic>. Although this objective is not inspired by a biological intuition, it allows us to explore other possibilities beyond the original case <italic>H</italic> = 1. ##FIG##2##Figure 3## show our predictions for the case <italic>H</italic> = 0.1 (red symbols) and <italic>H</italic> = 10 (blue symbols), representing a milder and a stronger dependency with the crowding coefficients <italic>a_i</italic>, respectively. For <italic>H</italic> = 0.1, 1.0 and 10 the predicted metabolite concentrations are in good agreement with the experimental values. Furthermore, when we allow sub-optimal metabolite concentrations resulting in a glycolysis rate below it s maximum our predictions are also in the range of the experimental values (see ##SUPPL##0##Protocol S1##, Table IV). These results indicate that it is sufficient that the optimization objective is a monotonic decreasing function of the crowding coefficients. When the latter is satisfied the metabolite concentrations are up to a great extent constrained by the kinetic model.</p>", "<p>This is not, however, the case for the enzyme activities. For <italic>H</italic> = 0.1 and the enzymes <italic>pfk</italic>, <italic>tpi</italic> and <italic>pk</italic>, there is a large deviation from the perfect match line. For <italic>H</italic> = 10 and the enzymes <italic>tpi</italic> and <italic>pk</italic>, there is a large deviation from the perfect match line as well. Overall, <italic>H</italic> = 1 gives the better agreement between enzyme activity predictions and their measured values. In addition, it provides a clear biophysical interpretation of the solvent capacity constraint (<italic>H</italic> = 1).</p>", "<p>Finally, we use Equation 4 to compute the maximum glycolysis rate as determined by the limited solvent capacity constraint. The global optimization predicts the glycolysis rate <italic>R</italic> = (1−<italic>ϕ</italic>)×12.5 mmol/min/g dry weight. Taking into account that about 30% ##UREF##1##[9]## of the cell is occupied by cell components excluding water, that proteins account for ∼45% of the dry weight ##UREF##2##[10]##, and that of these glycolytic enzymes account for ∼22% ##UREF##3##[11]## of the protein mass we obtain 1−<italic>ϕ</italic>∼0.03. Therefore, given that glycolysis enzymes occupy only 3% of the cell volume, we obtain <italic>R</italic>∼0.38 mmol/min/g dry weight. This prediction is in very good agreement with the experimentally determined glycolysis rate of <italic>S. cerevisiae</italic>, ranging between 0.1 to 0.4 mmol/min/g dry weight ##REF##10951190##[8]##,##UREF##4##[12]##.</p>" ]

[ "<title>Discussion</title>", "<p>The successful modeling of cell metabolism requires the understanding of the physicochemical constraints that are relevant at physiological growth conditions. In our previous work focusing on <italic>E. coli</italic> we have reported results indicating that the limited solvent capacity is an important constraint on cell metabolism, especially under nutrient-rich growth conditions ##REF##17652176##[4]##,##UREF##0##[5]##. Using a flux balance approach that incorporates this constraint we predicted the maximum growth rate in different carbon sources ##REF##17652176##[4]##, the sequence and mode of substrate uptake and utilization from a complex medium ##REF##17652176##[4]##, and the changes in intracellular flux rates upon varying <italic>E. coli</italic> cells' growth rate ##UREF##0##[5]##. More importantly, these predictions were in good agreement with experimentally determined values.</p>", "<p>Here we have extended the study of the impact of the limited solvent capacity by (i) considering a different organism (<italic>S. cerevisiae</italic>), and (ii) a full kinetic model of glycolysis. Using the full kinetic model of <italic>S. cerevisiae</italic> glycolysis, we have demonstrated that the predicted optimal intermediate metabolite concentrations and enzyme activites are in good agreement with the corresponding experimental values. Discrepancies were only observed in association with two different steps in the glycolysis pathway, namely the reaction catalyzed by <italic>pfk</italic> and the reactions between BPG and PEP. The experimental values measurements from cell extracts ##REF##10951190##[8]## and, except for potential experimental caveats, they represent phyiological conditions. We thus we believe that the larger deviations for these enzymes are determined by inconsistencies in the kinetic model equations and/or kinetic model parameters. Finally, the glycolysis rate achieved at the optimal metabolite concentrations is in the range of the experimentally measured values.</p>", "<p>From the quantitative modeling point of view our results indicate that a full kinetic model together with the solvent capacity constraint can be used to make predictions for the metabolite concentrations and enzyme activities. Thus, we propose the simultaneous optimization of intermediate metabolite concentrations, maximizing the metabolic rate given the solvent capacity, as a method to computationally predict the concentrations of a metabolic pathway's intermediate metabolites and enzyme activities. We have demonstrated the applicability of this method by computing the concentration of <italic>S. cerevisiae</italic> glycolysis intermediate metabolites, resulting in a good agreement with published data.</p>", "<p>The hypothesis that high concentration of macromolecules in the cell's cytoplasm imposes a global constraint on the metabolic capacity of an organism has been studied in the past ##REF##1787736##[13]##,##UREF##5##[14]##,##REF##12423338##[15]##. In most cases ##UREF##5##[14]##,##REF##12423338##[15]## it has been postulated that there is a bound to the total enzyme concentration (moles/volume). Yet, -to our knowledge-, no clear explanation has been provided to support that choice. In contrast, our starting postulate is an undeniable physical constraint, the total cell volume (Equation 1). Under this constraint, the enzyme molar volumes are the primary magnitude quantifying the enzymatic cost. In turn, since the enzyme-specific volumes are approximately constant, we can use the enzyme density (mass/volume) as an alternative measure of enzymatic cost.</p>", "<p>This modeling framework has advantages and disadvantages with respect to more traditional approaches based on dynamical systems modeling. As an advantage, our method does not require as input parameters the enzyme activities but rather make quantitative predictions for them. On the other hand, our method is based on a steady-state approximation. Therefore, in its present form, it cannot be used to understand dynamical processes, such as the observed metabolite concentration oscillations in <italic>S. cerevisiae</italic> cells when growing at high glucose concentrations ##REF##11744196##[7]##.</p>" ]

[]

[ "<p>Analyzed the data: AV. Contributed reagents/materials/analysis tools: AV MAdM ALB ZNO. Wrote the paper: AV ZNO.</p>", "<p>The cell's cytoplasm is crowded by its various molecular components, resulting in a limited solvent capacity for the allocation of new proteins, thus constraining various cellular processes such as metabolism. Here we study the impact of the limited solvent capacity constraint on the metabolic rate, enzyme activities, and metabolite concentrations using a computational model of <italic>Saccharomyces cerevisiae</italic> glycolysis as a case study. We show that given the limited solvent capacity constraint, the optimal enzyme activities and the metabolite concentrations necessary to achieve a maximum rate of glycolysis are in agreement with their experimentally measured values. Furthermore, the predicted maximum glycolytic rate determined by the solvent capacity constraint is close to that measured in vivo. These results indicate that the limited solvent capacity is a relevant constraint acting on <italic>S. cerevisiae</italic> at physiological growth conditions, and that a full kinetic model together with the limited solvent capacity constraint can be used to predict both metabolite concentrations and enzyme activities in vivo.</p>", "<title>Author Summary</title>", "<p>The concentration of enzymes and metabolites is continuously adjusted in order to achieve specific metabolic demands. It is highly likely that during evolution global metabolic regulation has evolved such as to achieve a given metabolic demand with an optimal use of intracellular resources. However, the size of enzymes and intermediate metabolites is dramatically different. Enzymes are macromolecules that occupy a relatively large amount of space within a cell's crowded cytoplasm, while metabolites are much smaller. This implies that metabolite concentrations are likely to be adjusted to minimize the overall “enzymatic cost” (in terms of space cost). In this work, we explore this hypothesis using <italic>Saccharomyces cerevisiae</italic> glycolysis as a case study. Our results indicate that metabolite concentrations attain optimal values, minimizing the intracellular space occupied by metabolic enzymes. And, at these optimal concentrations, glycolysis achieves the maximum rate given the intracellular volume fraction occupied by glycolysis enzymes. Taken together with previous studies for <italic>Escherichia coli</italic>, our results indicate that macromolecular crowding is a general constraint on cell metabolism.</p>" ]

[ "<title>Supporting Information</title>" ]

[]

[ "<fig id=\"pcbi-1000195-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000195.g001</object-id><label>Figure 1</label><caption><title>Hypothetical three metabolite pathway.</title><p>The inset shows a hypothetical three metabolite-containing pathway with two reactions. The main panel displays the pathway rate as a function of the concentration of the intermediate metabolite. Of note, at an intermediate metabolite concentration [M₂]^*, the pathway rate achieves a maximum. The plot was obtained using the kinetic parameters indicated in the text.</p></caption></fig>", "<fig id=\"pcbi-1000195-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000195.g002</object-id><label>Figure 2</label><caption><title>\n<italic>S. cerevisiae</italic> glycolysis.</title><p>(A) Schematic representation of glycolysis in <italic>S. cerevisiae</italic>. Metabolites: GLCx, external glucose; GLC, glucose; G6P, glucose 6-phosphate; F6P, fructose 6-phosphate; FBP, fructose 1,6-bisphosphate; DHAP, glycerone phosphate; GAP, D-glyceraldehyde 3-phosphate; BPG, 1,3-bisphosphoglycerate; and PEP, phospho-enol-pyruvate. Reactions: <italic>hxt</italic>, glucose transport; <italic>hk</italic>, hexokinase; <italic>pgi</italic>, phosphogluco isomerase; <italic>pfk</italic>, phospho-fructokinase; <italic>ald</italic>, fructose 1,6-bisphosphate aldolase; <italic>tpi</italic>, triosephosphate isomerase; <italic>gapdh</italic>, D-glyceraldehyde 3-phosphate dehydrogenase; <italic>lpPEP</italic>, reactions from BGP to PEP; <italic>pk</italic>, pyruvate kinase; and <italic>g3pdh</italic>, glycerol 3-phosphate dehydrogenase. (B,C,D) Predicted glycolysis rate as a function of the concentrations of intermediary metabolites in the <italic>S. cerevisiae</italic> glycolysis pathway (in mM). The experimentally determined metabolite levels (from ##REF##11744196##[7]##) are indicated by the red triangles. The dashed lines indicate the concentration intervals resulting in 50% or more of the maximum rate.</p></caption></fig>", "<fig id=\"pcbi-1000195-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000195.g003</object-id><label>Figure 3</label><caption><title>Correlation between predictions vs. experimental data.</title><p>(A) The predicted metabolite concentrations are plotted as a function of the experimentally determined values (black symbols). The error bars represent the standard deviations, upon generating 100 random sets of kinetic parameters. The solid line corresponds with the coincidence of measured and predicted values, indicating a strong correlation between them. (B) The predicted enzyme activities are plotted as a function of the experimentally determined values, measured in units of the glycolysis rate (black symbols). The error bars represent the standard deviations, upon generating 100 random sets of kinetic parameters. The solid line corresponds with the coincidence of measured and predicted values, indicating a strong correlation between them. In both cases, the red and blue symbols were obtained using the more general optimization objective <italic>R</italic> = (1−<italic>ϕ</italic>)/Σ^{<italic>N</italic>}\n_{<italic>i</italic> = 1} (<italic>a_ir_i</italic>)<italic>^H</italic>, with <italic>H</italic> = 0.1 and 10, respectively.</p></caption></fig>" ]

[]

[ "<disp-formula><label>(1)</label></disp-formula>", "<disp-formula><label>(2)</label></disp-formula>", "<disp-formula><label>(3)</label></disp-formula>", "<disp-formula><label>(4)</label></disp-formula>", "<disp-formula><label>(5)</label></disp-formula>", "<disp-formula><label>(6)</label></disp-formula>", "<disp-formula><label>(7)</label></disp-formula>", "<disp-formula><label>(8)</label></disp-formula>", "<inline-formula></inline-formula>", "<disp-formula><label>(9)</label></disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pcbi.1000195.s001\"><label>Protocol S1</label><caption><p>Details on the rate equation model used, the utilized model parameters, and the glycolysis rate and optimal metabolite concentrations.</p><p>(0.10 MB PDF)</p></caption></supplementary-material>" ]

[ "<fn-group><fn fn-type=\"COI-statement\"><p>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p>Research at the Institute of Advanced Study was supported by the Leon Levy Foundation and the Simons Foundation. Research at Northeastern University and at the University of Pittsburgh was supported by National Institutes of Health Grant U01AI070499.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pcbi.1000195.e001.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000195.e002.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000195.e003.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000195.e004.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000195.e005.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000195.e006.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000195.e007.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000195.e008.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pcbi.1000195.e009.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pcbi.1000195.g001\"/>", "<graphic xlink:href=\"pcbi.1000195.g002\"/>", "<graphic xlink:href=\"pcbi.1000195.g003\"/>", "<graphic xlink:href=\"pcbi.1000195.e010.jpg\" mimetype=\"image\" position=\"float\"/>" ]

[ "<media xlink:href=\"pcbi.1000195.s001.pdf\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>\n<italic>Capnocytophaga canimorsus</italic> (formerly Centers for Disease Control group DF-2) has been rarely but regularly isolated from dog or cat bite infections since its discovery in 1976 ##REF##1266816##[1]##,##REF##2915017##[2]##. <italic>C. canimorsus</italic> is a thin Gram-negative rod, found in the normal oral flora of dogs and cats. Clinical infections by <italic>C. canimorsus</italic> generally appear as fulminant septicemia, peripheral gangrene or meningitis ##REF##8816132##[3]##,##REF##12539089##[4]##. Splenectomy, alcohol abuse and immunosuppression have been associated with a number of cases, but more than 40% of the patients have no obvious risk factor ##REF##8905316##[5]##. <italic>Capnocytophaga</italic> belongs to the family of <italic>Flavobacteriaceae</italic> in the phylum of <italic>Bacteroidetes,</italic> which is taxonomically remote from Proteobacteria. The family of <italic>Bacteroidaceae</italic> contains many commensals of the mammalian intestinal system such as <italic>Bacteroides thetaiotaomicron</italic> and <italic>Bacteroides fragilis</italic>\n##REF##17993536##[6]##. The family of <italic>Flavobacteriaceae</italic> includes a variety of environmental and marine bacteria, among which <italic>Flavobacterium johnsoniae</italic> is a common soil and freshwater bacterium studied for its gliding motility ##REF##15170404##[7]##. There are only a few examples of pathogenic bacteria belonging to this family. These are <italic>Flavobacterium psychrophilum,</italic> the causative agent of cold water disease in salmonid fish ##REF##17592475##[8]##, <italic>Ornithobacterium rhinotracheale,</italic> a bacterial pathogen known for causing respiratory disease in poultry ##REF##16177359##[9]##, and <italic>Riemerella anatipestifer</italic> which causes “duckling disease” in waterfowl and turkeys ##REF##8240957##[10]##,##REF##10702488##[11]##. The genus <italic>Capnocytophaga</italic> includes seven species found in normal human oral flora and <italic>C. canimorsus</italic> found in the normal flora of dogs and cats. More than 160 cases of <italic>C. canimorsus</italic> infections have been reported so far ##REF##16455937##[12]## but very few studies have addressed the molecular mechanisms of <italic>C. canimorsus</italic> pathogenesis. Recently, we showed that <italic>C. canimorsus 5 (Cc5</italic>) resists phagocytosis by macrophage cell line ##REF##17205476##[13]##,##UREF##0##[14]##. We also showed that although <italic>C. canimorsus</italic> does not affect the viability of murine or human macrophages, it does not elicit proinflammatory cytokines and it even blocks the proinflammatory response to the LPS from enterobacteria ##REF##17205476##[13]##. In the course of such experiments, <italic>Cc5</italic> exhibited robust growth, although it is usually considered fastidious for growth. In the present study, we show that a surface-localized sialidase plays a key role in initiating an extensive deglycosylation process of host cell glycan structures and that this feeding mechanism serves as a basis for growth and persistence of <italic>C. canimorsus in vivo</italic>.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Bacterial strains and growth conditions</title>", "<p>\n<italic>C. canimorsus 5</italic> was routinely grown on Heart Infusion Agar (HIA; Difco) supplemented with 5% sheep blood (Oxoid) for 2 days at 37°C in presence of 5% CO₂. Bacteria were harvested by gently scraping colonies off the agar surface, washed and resuspended in PBS. <italic>C. canimorsus</italic> was also grown in Heart Infusion Broth (Difco) supplemented with 10% (v/v) fetal bovine serum (FBS; Invitrogen) for approximately 24 h without shaking in an 37°C incubator with 5% CO₂. Selective agents were added at the following concentrations: erythromycin, 10 µg/ml; cefoxitin, 10 µg/ml; gentamicin, 20 µg/ml; ampicillin, 100 µg/ml.</p>", "<title>Cell Culture and Infection</title>", "<p>Murine monocyte-macrophage J774A.1 cells (ATCC TIB-67) were cultured in RPMI 1640 (Invitrogen) supplemented with 10% (v/v) FBS (Invitrogen), 2 mM L-glutamine and 1 mM sodium pyruvate. Human epithelial HeLa cells (<named-content content-type=\"gene\">ATCC CCL-2</named-content>) and canine epithelial MDCK kidney cells (<named-content content-type=\"gene\">ATCC CCL-34</named-content>) were grown in DMEM (Invitrogen) with 10% (v/v) FBS. Cells were seeded in medium without antibiotics at a density of 10⁵/cm² and cultured at 37°C in humidified atmosphere containing 5% CO₂. Unless otherwise indicated, infection was performed after 15 h at a moi of 20 representing 2×10⁶ bacteria per ml in each well at 37°C.</p>", "<p>Monosaccharides and disaccharides (Sigma Aldrich) were added to 0.1% (w/v) final concentration. Neu5Ac and CMP- Neu5Ac were added to 0.01% final concentration.</p>", "<p>\n<italic>Cc5</italic> was pretreated with 1mM Neu5Ac2en at 37°C for 30 min. Subsequently, infection of J774.1 was carried out in presence of 1 mM Neu5Ac2en during 24 h.</p>", "<title>Arbitrarily Primed PCR</title>", "<p>Primers specific to the ends of the transposon and primers of random sequence that may anneal to chromosomal DNA sequences in close proximity to the transposon insertions were used in two rounds of PCR before sequencing. The first round of amplification was carried out in 50 µl containing 100 ng of genomic DNA, 1.5 mM MgCl₂, 200 µM of primers <named-content content-type=\"gene\">5′ CAGAATTCTGTTGCATTTGCAAGTTG 3′</named-content> complementary to Tn<italic>4351</italic> and <named-content content-type=\"gene\">5′ggccacgcgtcgactagtacNNNNNNNNNNacgcc3′</named-content>, 2.5 U of DNA polymerase (DyNAzymeII, Finnzymes), 200 µM of each dNTP, in 10 mM Tris HCl (pH 8.3) for 6 cycles (94°C for 1 min, 30°C for 1 min, 72°C for 2 min) and 30 cycles (94°C for 1 min, 45°C for 1 min, 72°C for 2 min) and final 10 min at 72°C. 10 µl of PCR product containing random fragments was used as template in a second round of 30 cycles of amplification (94°C for 30 sec, 45°C for 30 sec, 72°C for 1 min) using primers <named-content content-type=\"gene\">5′ CAGAATTCTGTTGCATTTGCAAGTTG 3′</named-content> and <named-content content-type=\"gene\">5′ GGCCACGCGTCGACTAGTAC 3′</named-content>, from the 5′ of the random primer. PCR products were purified using NucleoSpin® from Machery Nagel. 20- 50 ng of random sized products were sequenced using an ABI sequencer. The Tn integration site was further confirmed by using primers on chromosomal DNA by sequencing towards the Tn integration site. Primers used were <named-content content-type=\"gene\">5′ AATTGTTGTAACGATTGTCG 3′</named-content> or <named-content content-type=\"gene\">5′ GCGAAGCGTTATCCCAAAGC 3′</named-content> complementary to the <italic>siaC</italic> sequence in a sequencing reaction containing 2 µg genomic DNA of <italic>siaC</italic>, betaine 0.25 M and BigDye Terminator Ready Reaction (PE Biosystems) with an initial denaturation step for 5 min and subsequent 99 cycles (95°C for 30 sec, 50°C for 20 sec, 60°C for 4 min).</p>", "<title>RNA isolation and reverse transcription (RT) PCR</title>", "<p>\n<italic>Cc5</italic> were grown for 2 days on HIA blood plates. RNA was isolated from 5×10⁸ bacteria by a hot phenol/chloroform extraction method followed by DNase I (Amersham Pharmacia) treatment (0.5 U/µg RNA) for 2 h at 37°C. RNA was further cleaned by using a RNeasy kit (Quiagen) and stored at −80°C until use. An additional DNase I digest was introduced with 0.25 U/µg RNA for 15 min at 37°C and stopped by addition of final 2.5 mM EDTA and heat inactivation at 75°C for 10 min. Subsequent reverse transcription was performed with 50 U Superscript II/µg RNA in RT buffer (Invitrogen), 10 mM DTT and 50 µM specific primer (5129: <named-content content-type=\"gene\">5′ GGGTAATCCGCACTTGTCGGG3′</named-content> or 5132: <named-content content-type=\"gene\">5′ GTTTAGTTCTTGATAAATTCC 3′</named-content>) for 60 min at 42°C and stopped at 70°C for 10 min. 10% of cDNA preparation or of a preparation made without addition of reverse transcriptase was subjected to PCR using following primer combinations: 4130 (<named-content content-type=\"gene\">5′ GGGTAACAACAAAAACCACTG 3′</named-content>)+5129; 4132 (<named-content content-type=\"gene\">5′ TATAAGAATAATTGGTGGGC 3′</named-content>)+5129; 4130+5132. 100 ng of genomic DNA from <italic>Cc5</italic> was used as a positive control of the PCR reactions.</p>", "<title>Construction of complementation and expression plasmids</title>", "<p>Full length <italic>siaC</italic> was amplified with <named-content content-type=\"gene\">5′ CATA<underline>CCATGG</underline>GAAATCGAATTTTTTATCTT 3′</named-content> and <named-content content-type=\"gene\">5′ GT<underline>TCTAGA</underline>GAGTTCTTGATAAATTCCTCAACTG 3′</named-content> primers and cloned into the <italic>E. coli- C. canimorsus</italic> shuttle vector pMM47.A ##UREF##1##[16]## with <italic>Nco</italic>I and <italic>Xba</italic>I, leading to the insertion of a glycine at position 2 and a C- terminal histidine 6× tag in plasmid pMM52 (<italic>siaC</italic>\n_FL). Forward primer <named-content content-type=\"gene\">5′ AAAG<underline>CCATGG</underline>GAAACGTAATCGGCGGAGGCG 3′</named-content> was used with the same reverse primer to construct pMM50 (<italic>siaC</italic>\n_Δ1–21), deleting the first 63 bp of <italic>siaC</italic>, but still including methionine and glycine at position 1 and 2, respectively, and using a C-terminal His 6× tag. The catalytic mutation in <italic>siaC</italic> of was introduced by site directed mutagenesis with an inverse PCR on pMM52, using primers <named-content content-type=\"gene\">5′ GAAGGATTTGGGTGTTCGTGTATGTCG 3′</named-content> and <named-content content-type=\"gene\">5′ CGACATACACGAACACCCAAATCCTTC 3′</named-content> leading to pMM59 (<italic>siaC</italic>\n_Y488C). Plasmids derived from pMM47.A contained the <italic>cfxA</italic> gene originating from <italic>Bacteroides sp.</italic> and could be selected in <italic>C. canimorsus</italic> with 10 µg/ml cefoxitin ##UREF##1##[16]##. The beta-lactamase also present on pMM47.A was used as a selection marker in <italic>E. coli</italic>.</p>", "<p>The cDNAs encoding SiaC_Δ1–21 (pHS2) were subsequently amplified using <named-content content-type=\"gene\">5′ GGAATTCCATATGAACGTAATCGGCGGAGGC 3′</named-content> plus <named-content content-type=\"gene\">5′ CGCGGATCCCTAGTTCTTGATAAATTCCTC 3′</named-content> and cloned into the expression vector pET15b(+) (Novagen). Plasmid pHS3 encoding SiaC_{Δ1–21,Y488C} was constructed by site directed mutagenesis on template pHS2 using the same primers as described for pMM59. All constructs were sequenced with an ABI sequencer. The sequence of SiaC was deposited at GenBank (accession number: EU329392).</p>", "<title>Purification of recombinant SiaC and immunoblotting</title>", "<p>Expression of <italic>siaC</italic> constructs in <italic>E. coli</italic> BL21(DE3) was induced with 0.5 mM isopropyl-β-D-1-thiogalactopyranoside at A₆₀₀ = 0.5 for 3 h. Proteins were purified by affinity chromatography using chelating Sepharose (Pharmacia) charged with NiSO₄ according to the manufacturer's instructions. Samples were analyzed by SDS-PAGE by the system of Laemmli, and immunoblotted. Polyclonal serum from rabbit was generated against recombinant SiaC_Δ1–21. The antigen was injected at days 0, 14, 28, and 56 with a final bleeding at day 80 (Laboratoire d'Hormonologie, Marloie, Belgium).</p>", "<title>MUAN hydrolysis</title>", "<p>10⁷ bacteria were incubated with 0.006% 2′-(4-Methylumbelliferyl)-α-D-N-acetylneuraminic acid (MUAN) in 0.25 M sodium acetate pH 7.5 at 37°C for 3 min. Reactions were stopped with 50 mM Na₂CO₃ pH 9.6 and fluorescence was determined at 445 nm with a Wallac Victor² 1420 Multilabel counter (Perkin Elmer).</p>", "<title>Outer Membrane Preparation</title>", "<p>Bacterial cells resuspended in PBS containing DNase and RNase (10 µg/ml), were sonicated on ice. Unbroken cells were removed at 3000× g for 15 min, and total membranes were collected at 20 000× g for 30 min at 4°C. The membranes were suspended in PBS and sarcosyl (N-Lauroylsarcosine sodium salt, Sigma) was added to a final concentration of 1% (v/v). After incubation on ice for 1 h, membranes were collected at 20 000× <italic>g</italic> for 30 min and resuspended in electrophoresis sample buffer and analyzed by SDS-PAGE by the system of Laemmli.</p>", "<title>Immunofluorescence of bacteria</title>", "<p>10⁷ bacteria were incubated on poly-D-lysine (BD) coated glass slides for 1 h at 37°C and subsequently fixed with 3% paraformaldehyde for 15 min. Anti- SiaC polyclonal serum (1∶500) and a FITC conjugated secondary antibody (Goat Anti- Rabbit IgG, Southern Biotech) was used at 1 µg/ml and fluorescence was measured with a Leica DMIRE2 microscope. Pictures were taken with a digital camera (Hamamatsu Photonics) and analyzed with OpenLab software (version 3.1.2) and Adobe Photoshop CS3 (version 10.0.1).</p>", "<title>Lectin Staining</title>", "<p>10⁵ J774.1 macrophages or HeLa epithelial cells were seeded on poly-D-lysine coated slides. Infection was carried out with 4×10⁷ bacteria for 2 h. Uninfected cells were alternatively treated with purified recombinant SiaC at 100 ng/ml. Cells were fixed with 3% paraformaldehyde for 15 min. Biotinylated lectins SNA and PNA (Vector Laboratories) were incubated with cells at 2 µg/ml and 2.5 µg/ml, respectively, for 1 h. After washing with PBS, cells were treated with 1 µg/ml fluorescein conjugated streptavidin (Vector Laboratories) and fluorescence was determined on mounted slides (Vectashield, Vector Laboratories).</p>", "<title>Mice and tissue cage infection model</title>", "<p>12 week-old male C57BL/6 mice were maintained under pathogen-free conditions in the Animal Facility of the Department of Research, University Hospital Basel. Animal experiments were performed in accordance with the guidelines of the Swiss veterinary law. Teflon tissue cages were implanted subcutaneously in the back of anesthetized mice as previously described ##REF##12870123##[19]##. The cages consisted of closed Teflon cylinders (10 mm diameter, 30 mm length, internal volume 1.84 ml) with 130 regularly spaced 0.2 mm holes. 2 weeks after surgery, 200 µl of bacterial suspension was injected percutaneously into the cage. Prior to infection, sterility of the tissue cage was verified. Tissue cage fluid (TCF) was sampled at day 2, 5, 9, 14 and 27 and examined for leukocytes and bacterial viable counts. Leukocytes from TCF were quantified with a Coulter counter (Coulter Electronics) and differentiated by Diff-Quick (Medion Diagnostics) Wright staining of cytospins and examined under light microscopy. The percentage of viable leukocytes was assessed by trypan blue exclusion.</p>", "<p>The survival of <italic>siaC</italic> bacteria in the competition experiment was compared directly with wt <italic>Cc5</italic> in individual animals giving a 1∶1 ratio of wt to mutant bacteria. The number of mutant (erythromycin resistant) and wt bacteria recovered from the TCF of animals was established by plating to media with and without erythromycin. The competitive index was calculated as the (number of mutant/wild-type bacteria recovered from animals)/(number of mutant/wild-type bacteria in the inoculum).</p>", "<title>Statistical analysis</title>", "<p>For growth experiments, means and standard deviation (s.d.) were calculated and statistical significance was evaluated by using a two- tailed, unpaired Student's <italic>t</italic> test. Differences were determined to be significant when p&lt;0.05. For <italic>in vivo</italic> experiments, individual mouse values are shown including the median value of each group. Mann Whitney test with the <italic>post hoc</italic> Bonferroni correction was used for comparison between <italic>Cc5</italic> and <italic>siaC</italic> cfu numbers during infection.</p>" ]

[ "<title>Results</title>", "<title>Growth of <italic>C. canimorsus 5</italic> requires direct contact with cells</title>", "<p>When inoculated at a multiplicity of infection (moi) of 20 to J774.1 murine macrophages cultured in complete RPMI (cRPMI), which includes 10% fetal bovine serum (FBS), <italic>Cc5</italic> multiplied about 100-fold during the 24 h of infection (##FIG##0##Fig. 1A##). This observation could be repeated with non-phagocytic human epithelial HeLa cells and with canine epithelial MDCK kidney cells (##FIG##0##Fig. 1B##). Surprisingly, growth was abolished when J774.1 macrophages were omitted and moreover, <italic>Cc5</italic> was unable to grow in a transwell system, indicating that direct contact is required for bacterial growth (##FIG##0##Fig. 1B##). This implies that <italic>Cc5</italic> may take advantage of some nutrient that is present on the host cell surface. Notably, <italic>Cc5</italic> did not adhere tightly to cells and was not internalized (data not shown). We generated a transposon (Tn) mutant library using Tn<italic>4351</italic> from <italic>B. fragilis</italic>\n##REF##9335266##[15]##,##UREF##1##[16]## and isolated a clone that was unable to grow in the presence of J774.1 cells, but grew normally on blood agar plates. Wild type (wt) and mutant bacteria grew equally well in serum enriched heart infusion medium (##FIG##0##Fig. 1C##). Impaired growth of this Tn mutant was not due to an increased phagocytic uptake by J774.1 since addition of cytochalasin D had little effect on bacterial growth (##FIG##0##Fig. 1B##).</p>", "<title>Surface-localized sialidase is required for the growth of <italic>Cc5</italic> in contact with cells</title>", "<p>The transposon inserted at codon 77 within a gene encoding a protein with similarity to bacterial sialidase, glycosylhydrolase that cleaves terminal sialic acid from glycoconjugates (##FIG##1##Fig. 2A##). The mutated gene, designated <italic>siaC,</italic> was found to be located downstream of genes encoding a predicted transcriptional regulator and a putative N-acyl-glucosamine epimerase (accession number: EU329392). The first gene downstream was found to start 148 bp further from the <italic>siaC</italic> stop codon (##FIG##1##Fig. 2B##). To exclude any polar effects of the Tn integration, we tested whether the downstream gene was transcriptionally linked to <italic>siaC</italic>. Total RNA was isolated from wt <italic>Cc5,</italic> reverse transcribed using two different primers annealing either at the end of <italic>siaC</italic> (5132) or at the end of the downstream gene (5129) and the cDNA was amplified by PCR. As shown in ##FIG##1##Fig 2C##, even though transcripts were present for both genes separately, no transcript spanned <italic>siaC</italic> and the downstream gene. This result indicates that <italic>siaC</italic> is not transcriptionally linked to the downstream gene.</p>", "<p>While intact <italic>Cc5</italic> bacteria cleaved 2′-(4-Methylumbelliferyl)-α-D-N-acetylneuraminic acid (MUAN), the Tn mutant could not, indicating that the mutated gene does indeed encode a sialidase (##FIG##2##Fig. 3A##). We engineered an expression shuttle vector by taking advantage of a cryptic plasmid isolated from another strain of <italic>C. canimorsus</italic> and the promoter of an insertion sequence from <italic>B. fragilis</italic>\n##UREF##1##[16]##. We constructed plasmids encoding full length (FL) SiaC, a variant deprived of the 21 N-terminal residues, predicted to be a signal peptide (Δ1–21), and a catalytic mutant (Y488C). Sialidase activity (##FIG##2##Fig. 3A##) and growth in the presence of J774.1 cells (##FIG##2##Fig. 3B##) was restored by introducing <italic>in trans siaC</italic>\n_FL, but not <italic>siaC</italic>\n_Δ1–21. Sialidase activity was not restored to wt levels by <italic>siaC</italic>\n_Y488C, but it was still significant (##FIG##2##Fig. 3A##), suggesting that this residual activity might account for elevated growth in comparison to the <italic>siaC</italic> mutant (##FIG##2##Fig. 3B##). Using a sarcosyl extraction method, SiaC_FL and SiaC_Y488C were found to be associated with the outer membrane (##FIG##2##Fig. 3C##), whereas SiaC_Δ1–21 was only detected in total cells (##FIG##2##Fig. 3B##). Indirect immunofluorescence using polyclonal anti-SiaC serum on paraformaldehyde fixed but unpermeabilized bacteria confirmed that SiaC is exposed on the bacterial surface unless the signal peptide is removed (##FIG##2##Fig. 3D##). Although it is surface exposed, no SiaC could be detected in the supernatant of infected J774.1 cultures, indicating that it is tightly associated with the outer membrane (##FIG##2##Fig. 3C##). Hence, surface-localized sialidase is required for growth of <italic>Cc5</italic> at the expense of mammalian cells.</p>", "<title>Growth is sustained by N-acetyl glucosamine (GlcNAc) and N-acetyl galactosamine (GalNAc) but not by sialic acids</title>", "<p>Since sialidases cleave terminal sialic acid from glycoconjugates, we first tested whether the addition of sialic acids could restore growth of <italic>siaC</italic>. Addition of neither sialic acid (Neu5Ac, N-Acetyl-2,3-dehydro-2-deoxyneuraminic acid) nor its activated form (CMP-Neu5Ac, Cytidine-5′-monophospho-N-acetylneuraminic acid) restored growth of <italic>siaC</italic> in presence of J774.1. In contrast, growth could be restored by the addition of purified recombinant SiaC or neuraminidase/sialidase NanH from <italic>Clostridium perfringens</italic> to the culture medium, but not by the addition of the catalytically inactive SiaC_Y488C (##FIG##3##Fig. 4A##). This suggested that removal of terminal sialic acids from glycoconjugates is required to make other carbohydrates accessible. Indeed, N-acetyl glucosamine (GlcNAc) and N-acetyl galactosamine (GalNAc), common carbohydrate moieties of glycoconjugates, allowed growth of <italic>siaC</italic> in the presence of macrophages (##FIG##3##Fig. 4B##). Notably, addition of glucose (Glc), galactose (Gal), mannose (Man) or sialyl-lactose (N-acetylneuraminosyl-D-lactose) could not restore growth of <italic>siaC</italic> bacteria (##FIG##3##Fig. 4C##). As galactose (Gal) is a common sugar preceding GlcNAc in glycan molecules, we next tested addition of N-acetyl lactosamine (LacNAc), a disaccharide consisting of β-D-Gal β(1→4) GlcNAc. LacNAc also restored the growth defect of <italic>siaC</italic> indicating the presence of an active β-galactosidase releasing monosaccharides Gal and GlcNAc in wt and <italic>siaC Cc5</italic> (##FIG##3##Fig. 4B##).</p>", "<title>Sialidase desialylates macrophage and epithelial cell surfaces</title>", "<p>J774.1 macrophages were incubated with either wt or <italic>siaC</italic> bacteria and thereafter analyzed for lectin binding to investigate desialyation process on the macrophage cell surface. We used <italic>Sambucus nigra</italic> agglutinin (SNA), which recognizes terminal sialic acids (2- 6 or 2- 3) linked to Gal or to GalNAc, and peanut agglutinin (PNA), a lectin specific for Gal (β 1–3) GalNAc, a disaccharide often forming the core unit of O-linked glycoconjugates (##FIG##4##Fig. 5A##). As shown in ##FIG##4##Fig. 5B##, wt bacteria greatly reduced the amount of sialic acids (SNA panel) and Gal (β 1–3) GalNAc (PNA panel) at the cell surface, while <italic>siaC</italic> bacteria had no effect on glycans masked by sialic acids. When cells were treated simultaneously with purified SiaC and <italic>siaC</italic> bacteria, neither sialic acid nor Gal (β 1–3) GalNAc were detected, indicating that <italic>siaC</italic> bacteria are still proficient in extensive deglycosylation of exposed glycans chains. The same deglycosylation of cell surfaces was observed when HeLa epithelial cells were used (##FIG##4##Fig. 5C##).</p>", "<title>Sialidase inhibitor N-Acetyl-2,3-dehydro-2-deoxyneuraminic acid (Neu5Ac2en) can inhibit growth of <italic>Cc5</italic>\n</title>", "<p>As bacterial and viral sialidases can share common ASP boxes that interact with sialic acid, we postulated that common sialidase inhibitor might have sufficient specificity for the active site in SiaC to inhibit growth of <italic>Cc5</italic> wt bacteria in presence of cells. We tested the anhydro sialate derivative Neu5Ac2en, which is known to inhibit many viral and bacterial neuraminidases ##REF##1797392##[17]##,##REF##1769974##[18]##. Approximately 150 cfu/ml of wt <italic>Cc5</italic> were inoculated to a culture of J774.1 macrophages in the presence of 1mM Neu5Ac2en and growth was monitored after 2, 6, 10 and 24 h (##FIG##5##Fig. 6A##). Between 2 and 24 h post infection, counts of wt <italic>Cc5</italic> were significantly reduced to values close to the <italic>siaC</italic> mutant (##FIG##5##Fig. 6B##). These data indicate that Neu5Ac2en has affinity for the active site of SiaC and restricts the growth of <italic>Cc5</italic> in the presence of J774.1 cells.</p>", "<title>\n<italic>Cc5</italic> but not <italic>siaC</italic> is able to persist in murine tissue cages</title>", "<p>To test whether sialidase could play a role during <italic>C. canimorsus</italic> systemic infection, we selected a murine tissue cage infection model ##REF##12870123##[19]##. Around 10⁷ cfu of wt or <italic>siaC Cc5</italic> bacteria were injected directly into Teflon cages, which had been subcutaneously implanted in C57BL/6 mice. Colony forming units (cfu) counts of wt decreased on day 2 and 5. However, on day 9 they increased by 1 to 3 logs in 4 out of 5 mice, and were able to persist in 3 of 5 mice after 27 days post infection with more than 10⁷ bacteria per ml fluid. The <italic>siaC</italic> bacteria were undetectable after the second day in 5 out of 5 infected mice (##FIG##6##Fig. 7A##). After infection, the total number of leukocytes in tissue cage fluid (1.8×10⁴ +/− 1.3×10⁴ leukocytes/µl, mean +/− standard deviation, s.d.) did not significantly increase and was not related to the bacterial load, suggesting that <italic>Cc5</italic> infection did not lead to strong leukocyte recruitment. This was in agreement with the suppression of inflammation, which we previously reported ##REF##17205476##[13]##. In mixed infections, the competitive index of <italic>siaC</italic> bacteria was 9.7×10⁻⁴, 5.8×10⁻⁷ and 4.7×10⁻⁷ on day 5, 9 and 14, respectively. As observed during infection with wt <italic>Cc5</italic> alone, 3 mice out of 5 that were infected by both strains developed a persistent infection (##FIG##6##Fig. 7B##).</p>", "<p>The fluid from uninfected control cages was collected and the leukocytes and liquid were tested separately for their capacity to sustain growth of <italic>Cc5.</italic> Interestingly, wt <italic>Cc5</italic> did not grow in the presence of the cell-free liquid (data not shown) but they grew in presence of leukocytes whereas <italic>siaC</italic> bacteria did not (##FIG##6##Fig. 7C##). Both strains grew equally well in heart infusion broth supplemented with 10% FBS, indicating a similar fitness <italic>in vitro</italic> (##FIG##0##Fig. 1C##). Mixed cultures in heart infusion broth supplemented with 10% FBS showed comparable growth of wt and <italic>siaC</italic> bacteria. Both strains reached 10⁶ cfu/ml after 24 h (##FIG##6##Fig. 7D##).</p>", "<p>Our data from mixed infection in mice suggest that there is no cross-feeding of nutrients between wt and <italic>siaC Cc5</italic> (##FIG##6##Fig. 7B##). We thus tested whether there would be cross-feeding between wt and <italic>siaC</italic> bacteria when inoculated to J774.1 cultures. When wt and <italic>siaC</italic> were inoculated together at 1∶1 ratio to J774.1 cells, wt <italic>Cc5</italic> reached 10⁸ cfu/ml while <italic>siaC</italic> bacteria only reached 3×10⁶ cfu/ml 24 h post infection (##FIG##6##Fig. 7E##).</p>", "<p>Taken together, these results demonstrate that SiaC plays an essential role in allowing persistence of wt <italic>Cc5</italic> in this tissue cage model and that clearance of <italic>siaC</italic> bacteria is not due to a growth defect per se but to an altered interaction of the mutant with the host. Since sialidase is surface-exposed, one could consider the possibility that it alters the susceptibility to complement. Hence, we checked the susceptibility of wt and <italic>siaC Cc5</italic> to mouse complement and found no difference (data not shown). It is also very unlikely that <italic>siaC</italic> bacteria have an increased sensitivity to killing by mouse leukocytes. Indeed, we tested phagocytosis and killing by human polymorphonuclear leukocytes and found no difference between wt and <italic>siaC</italic> bacteria (Manuscript in preparation). Hence, we conclude that the role of sialidase in infected mice is essentially nutritional.</p>" ]

[ "<title>Discussion</title>", "<p>Sialic acids are a family of nine carbon acid sugars among which Neu5Ac is one of the most widespread variants. Sialic acids are predominantly found at the terminal position of cell-surface and secreted eukaryotic glycan structures and are involved in many physiological processes including binding to microbes and down-regulation of innate immunity ##REF##11421344##[20]##–##REF##17632542##[22]##. Therefore it is not surprising that sialic acids play a role in a variety of host microbe interactions. Several pathogens have evolved ways to expose sialic acid on their surface and hence to escape complement killing and opsonization by mimicry. Sialic acids are incorporated into capsules by <italic>E. coli</italic> K1 ##UREF##2##[23]##, Group B <italic>Streptococci</italic>\n##REF##2554337##[24]##, Serogroups B, C, W135 and Y <italic>Neisseria meningitidis</italic>\n##REF##163259##[25]##. The lipooligosaccharide of <italic>Neisseria gonorrhoeae, Neisseria meningitidis</italic> and <italic>Haemophilus influenzae</italic> are also sialylated ##REF##8330904##[26]##. In this case, a bacterial sialyltransferase uses CMP-Neu5Ac from the host as a substrate ##REF##8330904##[26]##. Sialic acids can also be synthesized from lactate by <italic>Neisseria</italic> itself, demonstrating a close link between metabolism and evasion of innate immune defenses ##REF##15897277##[27]##.</p>", "<p>Besides molecular mimicry, many microbes can utilize sialic acids as a source of carbon and nitrogen like <italic>E. coli</italic> K1, <italic>H. influenzae</italic> or <italic>C. perfringens</italic>\n##REF##15007099##[28]##. Their metabolism comprises a permease for uptake and a neuraminiate lyase for conversion to N-acetyl mannosamine, which is either degraded or used in sialic acid biosynthesis. A number of commensal and pathogenic bacteria are also endowed with a sialidase, a glycosylhydrolase that cleaves sialic acid from sialo-glycoconjugates. Bacterial sialidases have been thought since a long time to contribute to virulence in bacteria that colonize mucosal surfaces such as <italic>Vibrio cholerae, Streptococcus pneumoniae,</italic> group B <italic>streptococci</italic>, <italic>C. perfringens</italic> and <italic>B. fragilis</italic> but the exact role of sialidase on virulence remains controversial ##REF##1472757##[29]##. Recently, it was shown that a sialidase is involved in the formation of <italic>Pseudomonas aeruginosa</italic> biofilms and hence contributes to colonization of the lungs during the initial stages of infection in cystic fibrosis patients ##REF##16862214##[30]##. In <italic>S. pneumoniae,</italic> a sialidase initiates an extensive deglycosylation of different host proteins, including IgA1 and human secretory component ##REF##16420364##[31]##. Furthermore, the sequential action of exoglycosidases sustains growth of <italic>S. pneumoniae</italic> on human α-1 acid glycoprotein, though growth is not as robust as on sucrose and lactose. Although the genetic analysis suggests that sugars from the glycan chain would sustain growth, this has not been shown directly ##UREF##3##[32]##.</p>", "<p>In the present study, which is among the very first on the pathogenesis mechanisms of <italic>C. canimorsus,</italic> we demonstrate that a sialidase allows <italic>C. canimorsus</italic> to feed on glycan chains from glycoproteins. The role of sialidase is not to supply sialic acid since growth of the sialidase-deficient mutant could not be restored by adding sialic acid to the culture medium. Thus, we have a situation similar to that of <italic>S. pneumoniae</italic>: the role of sialidase is to provide access to masked sugars of surface-exposed glycoproteins. Growth of the sialidase-deficient mutant could be restored by amino sugars like GalNAc, GlcNAc and LacNAc but not by glucose, galactose, mannose or sialyl-lactose, indicating that the nutritional requirements of <italic>C. canimorsus</italic> are very different from those of <italic>S. pneumoniae.</italic>\n</p>", "<p>Our study thus confirms the importance of a sialidase to initiate a deglycosylation process for bacterial metabolism. Moreover, in comparison with <italic>S. pneumoniae, C. canimorsus</italic> uses sialidase to feed on glycoproteins exposed at the surface of epithelial cells or even of macrophages, in spite of the fact that they do not adhere to these cells. The observation of extracellular bacteria specifically feeding on the surface of epithelial cells is not unprecedented. It has been described for <italic>B. thetaiotaomicron</italic>, a major commensal from the intestine, which feeds on fucosylated intestinal cells. Colonization by <italic>B. thetaiotaomicron</italic> even triggers the appearance of fucosyltransferase and fucosylated glycan expression ##REF##8703071##[33]##. Recent studies showed that host acquired fucose is incorporated by <italic>B. fragilis</italic> into capsular polysaccharide or glycoproteins, which in turn provides a survival advantage in the mammalian intestinal ecosystem ##REF##15774760##[34]##. As for <italic>C. canimorsus</italic>, it is likely that the capacity to feed on HeLa cells reflects the adaptation to feed on buccal epithelial cells.</p>", "<p>Sialidase, which is pivotal in this feeding process, is surface localized and this surface localization is a prerequisite for unmasking glycan structures. It is not common to find enzymes anchored into the outer membrane, facing the outside of Gram-negative bacteria but there are examples like pullulanase a 116-kDa isoamylase of <italic>Klebsiella oxytoca</italic>\n##REF##2181241##[35]##. Not surprisingly, SiaC is endowed with an N-terminal signal sequence, which turned out to be critical for its targeting. Sialidase thus crosses the cytoplasmic membrane via the Sec pathway but we have at present no explanation on how it crosses the outer membrane and remains anchored. It is probably not by a <italic>C. canimorsus</italic> specific mechanism since sialidase appeared to be also surface-exposed when expressed in <italic>E. coli</italic> (unpublished data). Sialidase could be a lipoprotein, like pullulanase. Alternatively, sialidase could be a surface-anchored auto-transporter protein like the <italic>Y. enterocolitica</italic> YadA ##REF##16675268##[36]##. However, the fact that the C-terminus of sialidase is not involved in the surface localization (unpublished data) argues against this hypothesis. Work in progress tries to address the question of how sialidase is anchored in the outer membrane.</p>", "<p>Unlike what is observed with pullulanase, our data indicate that extremely little sialidase is released from <italic>C. canimorsus.</italic> This observation is in perfect agreement with the fact that <italic>C. canimorsus</italic> needs to be in direct contact with cells to feed on them. It also makes sense in the context of the mouth commensal microflora. Indeed, the oral cavity is occupied by some 500 different bacterial strains ##REF##10588742##[37]##,##REF##11371542##[38]##, creating a fierce competition for nutrition. The fact that <italic>C. canimorsus</italic> does not release this enzyme suggests that <italic>C. canimorsus</italic> maximizes the benefit of sialidase by not sharing this fitness factor with competing bacteria. In agreement with this hypothesis, there is no cross-feeding when wt and <italic>siaC Cc5</italic> bacteria are inoculated together in the presence of macrophages. This implies that wt <italic>C. canimorsus</italic> must be extremely efficient in capturing the aminosugars that it extracts from the surface of cells and we hypothesize that <italic>C. canimorsus</italic> has dedicated high affinity transporters for these in its outer membrane.</p>", "<p>Extracellular <italic>C. canimorsus</italic> replicated very efficiently not only when they were in direct contact with HeLa cells but also with J774.1 macrophages. Thus, <italic>C. canimorsus</italic> not only resists phagocytosis by cultured macrophages ##REF##17205476##[13]##,##UREF##0##[14]##, but they even take advantage of macrophages whose normal function is to engulf and kill microbes. To our knowledge, this is the very first report of a pathogen that can feed on phagocytic cells. This observation suggests that sialidase could contribute to virulence. We used a mouse tissue cage model in which the readout is bacterial persistence and we observed a dramatic difference in persistence between wt and sialidase-deficient <italic>C. canimorsus</italic>. Even more, we gained evidence that <italic>in vivo, C. canimorsus</italic> also feeds on phagocytes. These observations confirm our hypothesis that sialidase contributes to virulence, at least in the mouse model. It seems reasonable to extrapolate that it also plays a critical role during human infections. We would however be reluctant to call sialidase a virulence factor since it most probably evolved as a fitness factor for commensalism in the dog's mouth. Nevertheless, the mouse experiment shows that it may become a persistence factor if <italic>C. canimorsus</italic> is introduced in the tissues from another host. Our study thus shows once again the link between metabolism and virulence, as already well documented in studies on <italic>Salmonella</italic>\n##REF##16541065##[39]##, <italic>Listeria</italic>\n##REF##11572936##[40]## and <italic>Neisseria</italic>\n##REF##15897277##[27]##. However, unlike what was seen with <italic>Salmonella</italic>, there seem to be no or very little redundancy in the <italic>in vivo</italic> metabolism of <italic>C. canimorsus</italic> since the loss of sialidase had dramatic consequences on growth. It is interesting to observe that nutrition <italic>in vivo</italic> may be quite specific in spite of a very rich nutritional environment. Indeed, only GlcNAc and GalNAc could rescue growth while glucose had no effect and galactose was even deleterious. This difference could result from the fact that unlike <italic>Salmonella</italic>, <italic>C. canimorsus</italic> is a commensal highly adapted to its niche and only exceptionally a pathogen. Specialization is probably the hallmark of a bacterium that is primarily a commensal and only rarely a pathogen. Finally, <italic>C. canimorsus</italic> represents one more example illustrating that the distinction between commensals and pathogens is illusive. Commensalism and pathogenesis are two faces of the same coin.</p>", "<p>Influenza neuraminidases have been successfully targeted with chemotherapeutic inhibitors for prophylaxis and treatment ##REF##18049471##[41]##. Given the wide prevalence and important role of sialidases in microbial infections, inhibition of bacterial sialidases could also provide a mechanism to prevent bacterial spreading during infections. Here, we observed a significant inhibition of the growth of <italic>C. canimorsus</italic> in the presence of macrophages by Neu5Ac2en. These preliminary data indicate that microbial sialidases could indeed serve as an attractive drug target to prevent bacterial dissemination.</p>" ]

[]

[ "<p>Conceived and designed the experiments: MM HS RL GRC. Performed the experiments: MM HS CP. Analyzed the data: MM HS RL GRC. Wrote the paper: MM HS RL GRC.</p>", "<p>\n<italic>Capnocytophaga canimorsus</italic>, a commensal bacterium of the canine oral flora, has been repeatedly isolated since 1976 from severe human infections transmitted by dog bites. Here, we show that <italic>C. canimorsus</italic> exhibits robust growth when it is in direct contact with mammalian cells, including phagocytes. This property was found to be dependent on a surface-exposed sialidase allowing <italic>C. canimorsus</italic> to utilize internal aminosugars of glycan chains from host cell glycoproteins. Although sialidase probably evolved to sustain commensalism, by releasing carbohydrates from mucosal surfaces, it also contributed to bacterial persistence in a murine infection model: the wild type, but not the sialidase-deficient mutant, grew and persisted, both when infected singly or in competition. This study reveals an example of pathogenic bacteria feeding on mammalian cells, including phagocytes by deglycosylation of host glycans, and it illustrates how the adaptation of a commensal to its ecological niche in the host, here the dog's oral cavity, contributes to being a potential pathogen.</p>", "<title>Author Summary</title>", "<p>\n<italic>Capnocytophaga canimorsus</italic> is a commensal bacterium of dogs/cats oral flora, which causes rare but severe infections in humans that have been bitten or simply licked by a dog/cat. Fulminant septicemia and peripheral gangrene are most common symptoms. Although splenectomy has been identified as a predisposing factor, some 40% of the patients have no immunosuppression history. <italic>C. canimorsus</italic> belongs to the phylum <italic>Bacteroidetes,</italic> which includes many commensals of the human gut flora but few pathogens. <italic>C. canimorsus</italic> has been shown previously to be immunosuppressive and to resist phagocytosis by macrophages. Here, we show that this bacterium feeds on surface-exposed glycoproteins from cultured mammalian cells. This property, which was found to depend on a bacterial surface-exposed sialidase, suggests that in its natural niche—the dog's oral cavity—<italic>C. canimorsus</italic> may feed on the dog's mucosal cells. Moreover, we found that <italic>C. canimorsus</italic> also feeds on phagocytes and that sialidase contributes to persistence and virulence in a mouse infection model. Thus, by adapting to its ecological niche, <italic>C. canimorsus</italic> also developed the potential to persist within the tissues of an infected host. This observation nicely illustrates how commensalism and pathogenesis are two faces of the same coin.</p>" ]

[]

[ "<p>We thank Z Rajacic, N Jann and M Schmaler for assistance during the <italic>in vivo</italic> studies.</p>" ]

[ "<fig id=\"ppat-1000164-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000164.g001</object-id><label>Figure 1</label><caption><title>Growth of <italic>C. canimorsus 5</italic> is dependent on cell contact.</title><p>(A) Viable counts of 2×10⁶\n<italic>Cc5</italic> after 24 h in presence of J774.1 macrophages in RPMI supplemented with 10% FBS (moi = 20) (black) or in RPMI with FBS without cells (grey) and in a transwell system preventing physical contact between bacteria and macrophages in RPMI with FBS (white). (B) Viable counts of <italic>Cc5</italic> and Tn mutant after 24 h culture with macrophages in RPMI and FBS (black), with macrophages in RPMI and FBS in addition of cytochalasin D (grey), with HeLa cells (light grey) and MDCK cells in DMEM and FBS (white). The grey dotted line represents the bacterial number inoculated. The difference is statistically significant between <italic>Cc5</italic> and Tn mutant (2-tailed unpaired Student's t test p&lt;0.05) in 3 or more experiments. (C) Growth curve of wt <italic>Cc5</italic> (triangles) and Tn mutant (squares) in heart infusion broth (HIB) supplemented with 10% FBS, represented as the mean of 3 or more experiments with the error bars showing the s.d.</p></caption></fig>", "<fig id=\"ppat-1000164-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000164.g002</object-id><label>Figure 2</label><caption><title>Identification of the Tn integration site and analysis of mRNA present in wt <italic>C. canimorsus</italic> 5.</title><p>(A) Amino acid sequence of the <italic>C. canimorsus</italic> sialidase showing the signal peptide (italics) and the BNR/asp repeats (Ser/Thr-X-Asp-X-Gly-X-Thr-Trp/Phe) of bacterial sialidases (boxed). Domain predictions were analyzed by InterProScan ##REF##15980438##[42]##. The residues conserved in sialidases at the C-terminus are underlined and the tyrosine 488 is bold ##REF##2562507##[43]##. The Tn<italic>4351</italic> integration site in SiaC at amino acid 77 is indicated, boxed in grey and bold. (B) Genetic locus of the sialidase gene (<italic>siaC</italic>) including its upstream genes, <italic>gntR</italic>-like gene (CAPCA_MM1) and putative N-acyl-glucosamine epimerase encoding gene (CAPCA_MM2); and downstream coding sequence (CAPCA_MM3). (C) Reverse transcription performed on total RNA with specific primers (5129 or 5132) followed by PCR to identify transcripts present in wt <italic>Cc5</italic> (cDNA). PCR reactions were also performed using genomic DNA (gDNA) as template instead of cDNA as a positive control. As a control, reverse transcription was performed without reverse transcriptase in a parallel assay and used as template for the subsequent PCR reaction (-RT).</p></caption></fig>", "<fig id=\"ppat-1000164-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000164.g003</object-id><label>Figure 3</label><caption><title>Surface localized sialidase is required for growth.</title><p>(A) Sialidase activity of intact bacteria, measured with the substrate MUAN as the mean of triplicate measurements and s.d. of a representative experiment. (B) Viable counts after challenge with 2×10⁶\n<italic>Cc5</italic> (black), <italic>siaC</italic> (light grey) or <italic>siaC</italic> complemented with plasmids containing <italic>siaC</italic>\n_FL, <italic>siaC</italic>\n_Δ1–21 and <italic>siaC</italic>\n_Y488C after 24 h in presence of J774.1 with the grey dotted line indicating the bacterial number inoculated. Sialidase was detected by immunoblotting with a polyclonal antibody against SiaC in total cells (TC). (C) Outer membrane protein fractions (OMP), cell free supernatants (SN) of the J774.1 cultures shown in (B) including as control TC of <italic>Cc5</italic> were analyzed by immunoblotting for the presence of SiaC. (D) Surface localization of SiaC was tested by immunofluorescence on paraformaldehyde fixed but not permeabilized bacteria using anti-SiaC followed by anti- rabbit IgG conjugated to FITC.</p></caption></fig>", "<fig id=\"ppat-1000164-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000164.g004</object-id><label>Figure 4</label><caption><title>Aminosugars but not sialic acids sustain growth of <italic>C. canimorsus.</italic>\n</title><p>Viable counts after challenge with 2×10⁶ wt <italic>Cc5</italic> (black) or <italic>siaC</italic> (grey) grown for 24 h with J774.1 in cRPMI (control) or in the same condition with the addition of Neu5Ac, Neu5Ac- CMP, 12.5 ng/ml enzyme SiaC_FL, SiaC_Y488C or NanH from <italic>C. perfringens</italic> (A) or with the addition of GalNAc, GlcNAc or LacNAc (B) or with the addition of mannose, galactose, glucose or sialyl-lactose (C). Mean values from 3 or more experiments and s.d. are shown including statistical difference between wt <italic>Cc5</italic> and <italic>siaC</italic> with * p&lt;0.05, ** p&lt;0.01 and *** p&lt;0.001 for each pair of columns (2-tailed unpaired Student's t test). The grey dotted line indicates the bacterial number inoculated.</p></caption></fig>", "<fig id=\"ppat-1000164-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000164.g005</object-id><label>Figure 5</label><caption><title>\n<italic>C. canimorsus</italic> desialylates macrophage and epithelial cell surfaces.</title><p>(A) The targets of the lectins used in this study are schematically represented (adapted from ##REF##17460663##[44]##). Surface carbohydrates of J774.1 macrophages (B) or HeLa epithelial cells (C) were analyzed by lectin binding after 2 h of infection with 4×10⁷ wt (<italic>Cc5</italic>) or <italic>siaC</italic> bacteria. Cells were fixed with paraformaldehyde and incubated for 1 h with lectin SNA, which recognizes terminal sialic acids (2- 6 or 2- 3) linked to Gal or to GalNAc or PNA that binds to the disaccharide Gal 1–3 GalNAc only after removal of terminal sialic acids. SiaC was added to cells alone or with <italic>siaC</italic> bacteria at 100 ng/ml. Biotinylated lectins were visualized by FITC conjugated streptavidin.</p></caption></fig>", "<fig id=\"ppat-1000164-g006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000164.g006</object-id><label>Figure 6</label><caption><title>The sialidase inhibitor Neu5Ac2en decreases growth of wt <italic>C. canimorsus 5</italic> in presence of macrophages.</title><p>(A) Viable counts of approximately 150 bacteria grown in cRPMI in the presence of J774.1 cells for 2, 6, 10 and 24 h: <italic>siaC</italic> bacteria (light grey); wt <italic>Cc5</italic> bacteria with 1mM Neu5Ac2en (grey, dotted line); wt <italic>Cc5</italic> in the absence of inhibitor (black). Mean values from 4 experiments and s.d. are shown including statistical difference between <italic>Cc5</italic> and <italic>siaC</italic> in grey or between <italic>Cc5</italic> and <italic>Cc5</italic> treated with Neu5Ac2en in black with * p&lt;0.05, ** p&lt;0.01 and *** p&lt;0.001 (2-tailed unpaired Student's t test). (B) Data from viable counts (mean) shown in (A) is represented as the fold difference compared to wt <italic>Cc5.</italic> The statistical difference is depicted from (a) with * p&lt;0.05, ** p&lt;0.01 and *** p&lt;0.001.</p></caption></fig>", "<fig id=\"ppat-1000164-g007\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.ppat.1000164.g007</object-id><label>Figure 7</label><caption><title>The sialidase mutant is hypo-virulent in a tissue cage mouse infection model.</title><p>Tissue cages were implanted in C57BL/6 mice and infected with 10⁷\n<italic>Cc5</italic> wt and <italic>siaC</italic> bacteria (n = 5) singly (A) or in competition (B). Bacteria were counted in tissue cage fluid (TCF) during 27 days (<italic>Cc5 = </italic>black circles; <italic>siaC</italic> = open circles). Individual values are shown; horizontal lines indicate the median value of each group. The black dotted line is the detection limit of 20 bacteria per ml TCF. (A) Cfu numbers between groups were significantly different on days 2, 5 and 9 with p&lt;0.01 and on days 14 and 27 with p&lt;0.05 (Mann Whitney test). (B) 10⁷ cfu <italic>Cc5</italic> and erythromycin resistant <italic>siaC</italic> were inoculated at a 1∶1 ratio. Bacterial numbers were analyzed for 27 days (n = 5). Viable counts between wt and <italic>siaC</italic> were significantly different on day 2, 5 and 9 with p&lt;0.01 and on day 14 with p&lt;0.05. (C) <italic>Ex vivo</italic> isolated leukocytes were resuspended in serum free RPMI and inoculated at a moi of 20 (2×10⁶ bacteria) or 0.2 (2×10⁴ bacteria) indicated with grey dotted lines and bacterial viable count was monitored after 24 h. Values represent the mean using TCF cells from 3 uninfected mice. TCF leukocytes consist of 68% +/− 4.8% polymorphonuclear neutrophils (PMNs), 18% +/− 3.2% monocytes and 9.1% +/− 3.7% macrophages. Wt and <italic>siaC</italic> numbers were significantly different with p&lt;0.05 (*) and p&lt;0.001 (**) using 2-tailed unpaired student's t test. (D) <italic>In vitro</italic>, <italic>Cc5</italic> and <italic>siaC</italic> were tested in heart infusion broth with FBS inoculated at a 1∶1 ratio with approximately 100 bacteria total and bacterial growth was monitored for 2, 6, 10 and 24 h. (E) Viable counts after challenge with 2×10⁶ (grey dotted line) <italic>Cc5</italic> (black) or <italic>siaC</italic> (grey) grown for 24 h with J774.1 in cRPMI singly (control) or at a 1∶1 ratio (cross-feeding).</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<fn-group><fn fn-type=\"COI-statement\"><p>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p>This work was funded by Swiss National Science Foundation grant Nr 32-65393.01.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"ppat.1000164.g001\"/>", "<graphic xlink:href=\"ppat.1000164.g002\"/>", "<graphic xlink:href=\"ppat.1000164.g003\"/>", "<graphic xlink:href=\"ppat.1000164.g004\"/>", "<graphic xlink:href=\"ppat.1000164.g005\"/>", "<graphic xlink:href=\"ppat.1000164.g006\"/>", "<graphic xlink:href=\"ppat.1000164.g007\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Phase-vanishing reactions, introduced by Ryu [##REF##12405811##1##], Curran [##REF##12405811##1##–##REF##15012090##2##] and Verkade [##REF##14535710##3##], are triphasic reactions, which involve a reagent, a substrate and a liquid perfluoroalkane. A more recent improvement of phase-vanishing brominations is addition of water as the fourth phase to act as an acid scavenger [##UREF##0##4##]. The perfluoroalkane does not dissolve the reactants and is used as a phase screen to separate them. In this procedure, the halogen reagent is of a higher density while the substrate is of a lower density than the perfluoroalkane phase screen. As the reagent in the lower layer diffuses through the perfluoroalkane layer, it reaches the top layer and reacts with it. Since the rate of diffusion is relatively low, the reaction proceeds at a moderate rate, instead of a vigorous, often violent, reaction that would occur if the two reactants were mixed without a phase screen. In the course of the reaction, the reagent disappears (“vanishes”) and the product is mechanically separated from perfluoroalkane, which can be reused.</p>", "<p>Halolactonization and halocyclizations are important reactions in organic synthesis and structure determination [##UREF##1##5##–##UREF##2##6##]. Traditionally, a halolactonization is done in a mixture of aqueous solvent and an organic co-solvent in the presence of a base, such as sodium bicarbonate [##UREF##3##7##–##UREF##4##8##]. Reagents commonly used in halolactonization are I₂/NaHCO₃ [##UREF##5##9##], Br₂ [##UREF##6##10##], IBr [##UREF##7##11##]and ICl [##REF##17979295##12##]. We investigated phase-vanishing reaction conditions as a more efficient and an environmentally friendly alternative to traditional halolactonizations. Reactions were done on neat reagents, which avoided use of any solvent other then a phase screen – perfluorohexane (FC-72), simplified the work up and improved the yields.</p>" ]

[]

[ "<title>Results and Discussion</title>", "<p>Bromolactonization worked very well and was easy to monitor as it was a true “phase-vanishing” reaction. The bromine layer disappeared at the end of the reaction and the products were obtained in good to excellent yields. However, sometimes dibromo derivatives, resulting from addition of bromine to a double bond, were by-products. The reaction mechanism of halolactonization in basic aqueous medium is believed to involve formation of a halonium ion followed by attack of an oxygen nucleophile [##UREF##8##13##].</p>", "<p>Reaction times ranged from 20–60 minutes when Br₂ or ICl were used to several days when I₂ was used. Reaction rates can be controlled by the amount of FC-72 and the rate of stirring – a larger amount (greater depth) of FC-72 resulted in a slower reaction, while increased rate of stirring resulted in a faster reaction. If a very slow rate is desired, the reaction can be done without any stirring. Very fast stirring is not recommended as it may lead to direct mixing of the top and bottom phases and a violent reaction. When a reaction is done with stirring, it is recommended that a vial be clamped and not just placed on the top of a magnetic stirrer. Sometimes dense, solid products form, which encase the stirring bar. A result is that, if not clamped, the entire vial may begin to rotate and it may flip over. Usually, solubility of the reaction products in FC-72 is negligible and it can be reused. However, one should always check the fluorous phase for presence of the product. Finally, due to a low boiling point of FC-72, a considerable amount of it may evaporate in the course of longer reactions unless the reaction vessel is capped. Capping a vial is not always feasible as the reaction byproduct is a gas. In our hands, on ~5 g scale, there were no problems when reaction was done in a capped 20 mL vial.</p>", "<p>In the case of bromination of 4-pentenoic acid (<bold>1</bold>), under phase-vanishing conditions there was competing formation of 4,5-dibromopentenoic acid (<bold>3</bold>). Formation of the dibromo derivative as a minor product in the course of bromolactonization of 4-pentenoic acid has been reported along with the observation that it readily cyclizes to give the corresponding bromolactone [##UREF##9##14##–##UREF##10##15##]. For comparison, the reaction was done in dichloromethane and directly on neat reagents under solvent-free conditions (SFC) (<bold><italic>caution: an extremely violent reaction!</italic></bold>). In dichloromethane as a solvent, bromolactone <bold>2</bold> was produced in a slight excess. Under SFC, the reaction was surprisingly clean in that it gave only a mixture of the bromolactone <bold>2</bold> and dibromoacid <bold>3</bold>. The ratios of the two varied between different runs with the dibromide <bold>3</bold> predominating. Outcome of the reaction under phase-vanishing conditions on neat reactants closely resembled an SFC reaction (##FIG##0##Scheme 1##, see ##SUPPL##0##Supporting Information File 1## and ##SUPPL##1##Supporting Information File 2## for full experimental data). 4,5-Dibromopentanoic acid was identified based on its mass spectrum and was not isolated. Instead, the crude reaction mixture was treated with aqueous sodium bicarbonate and the resulting bromolactone <bold>2</bold> was isolated. Thus, unlike other substrates in this study, 4-pentenoic acid required basic reaction conditions for successful bromolactonization. As bromolactonization of methyl esters <bold>17</bold> and <bold>20</bold> worked very well (<italic>vide infra</italic>), bromolactonization of methyl 4-pentenoate was attempted. The major product was methyl 4,5-dibromopentanoate (~80%) and the bromolactone was only a minor product (~20%).</p>", "<p>Interestingly, when the reaction was done in ethyl acetate, or ethyl acetate was used in a work up, the major isolated product was ethyl 4,5-dibromopentanoate (<bold>4</bold>) (##FIG##1##Scheme 2##). Apparently, 4,5-dibromopentanoic acid underwent hydrogen bromide-catalyzed transesterification with ethyl acetate. In a control experiment, 4-pentenoic acid was transesterified by dissolving it in ethyl acetate and adding a catalytic amount of HBr in acetic acid. The best results, and the best yields of bromolactone, were obtained under phase-vanishing conditions when, upon consumption of bromine, FC-72 was removed and the residue was treated with aqueous NaHCO₃, followed by extraction with ethyl acetate. It should be pointed out that this was the only bromolactonization where we encountered problems. Bromolactonization of other compounds (<bold>7</bold>, <bold>14</bold>, <bold>17</bold> and <bold>20</bold>) worked well.</p>", "<p>While reaction of 4-pentenoic acid with bromine gave the corresponding bromolactone in a high yield, a reaction with iodine did not give good results (##FIG##2##Scheme 3##). The reaction was slow and the product was apparently unstable. Thus, the corresponding iodolactone <bold>5</bold>, along with a number of unidentified byproducts, was observed in GC-MS analysis of the reaction mixture. Isolation of the reaction product was not attempted. When iodine monochloride was used in place of iodine, the reaction was fast and the corresponding iodolactone <bold>5</bold> was obtained cleanly and in a high yield. The only byproduct (&lt;2% according to GC-MS), which was not isolated and was not characterized, had a mass that corresponded to chlorolactone <bold>6</bold>. When iodolactonization of 4-pentenoic acid was done in dichloromethane, the same byproduct formed in a larger amount (8–10%).</p>", "<p>Treatment of 5-norbornene-2-carboxylic acid (<bold>7</bold>) (3:1 mixture of <italic>endo</italic> and <italic>exo</italic> isomers) gave the expected bromolactone <bold>8</bold> along with four byproducts (##FIG##3##Scheme 4##). GC analysis showed the four byproducts as two pairs of closely spaced peaks (retention times of 8.1 and 8.3 min for one pair and 12.5 and 12.6 min for the other pair). Furthermore, GC-MS analysis indicated that the byproducts were the corresponding dibromo compounds. They were tentatively assigned structures <bold>9</bold>–<bold>12</bold>. It was expected that one pair of the dibromo compounds (<bold>11</bold> and <bold>12</bold>) would form as <italic>exo</italic>-5-norbornene-2-carboxylic acid cannot cyclize and, indeed, a pair of dibromo compounds formed in the amount that approximately corresponded to the amount of the starting <italic>exo</italic>-5-norbornene carboxylic acid. Another pair of dibromo compounds (assumed to have structures <bold>9</bold> and <bold>10</bold>) formed in a combined yield of ~10% (according to GC-MS analysis). Thus, formation of dibromo products decreased compared to 4-pentenoic acid. Interestingly, no dibromo derivatives at all were observed in bromolactonizations of diacid <bold>14</bold> and diesters <bold>17</bold> and <bold>20</bold> (##TAB##0##Table 1##, entries 3, 6 and 8).</p>", "<p>Iodine monochloride also worked very well on the same substrates (##TAB##0##Table 1##, entries 2, 4, 7 and 9). However, the reaction was more difficult to monitor. As commercially available iodine monochloride contains an excess of iodine, one has to use an excess of the reagent and, at the end of the reaction, there was some unreacted iodine left over. Thus, one cannot tell when the reaction is over and reaction times were more difficult to determine. Reaction times provided in the ##TAB##0##Table 1## are the times when the reaction was definitively completed. The actual reaction times were probably shorter. As in the case of 4-pentenoic acid, iodolactonization of compounds <bold>7</bold>, <bold>14</bold> and <bold>17</bold> was accompanied by formation of a small amount of a byproduct (&lt;2% according to GC-MS), which were not isolated. Mass spectra of the byproducts were consistent with the corresponding chlorolactones. Under the phase vanishing conditions, the amounts of chlorolactones were relatively small. In dichloromethane the amounts increased to as much as 10% (GC-MS). Interestingly, in both bromolactonization and iodolactonization, diester <bold>17</bold> gave cleaner products and better yields compared to the corresponding diacid <bold>14</bold>. <italic>trans</italic>-Diester <bold>20</bold> gave the best yields in both bromo- and iodolactonization reactions. Neither dibromo compounds (in a reaction with bromine) nor chlorolactone (in a reaction with iodine monochloride) were observed and the corresponding crude products gave single peaks when analyzed by GC and were &gt;95% pure by ¹H NMR.</p>", "<p>It has been reported that iodolatonization of the diacid <bold>14</bold> with iodine failed, while iodolactonization of the corresponding disodium salt gave a mixture of γ- and δ-lactones [##UREF##11##16##]. However, no experimental details were given. Under phase-vanishing conditions, treatment of <bold>14</bold> with iodine gave the corresponding γ-lactone <bold>16</bold> in a moderate yield. The yield was inferior compared to iodolactonization with iodine monochloride. On the other hand, iodine is a safer reagent and one may consider using it under certain circumstances (e.g., a large scale reaction, or need for an inexpensive reagent rather than a high yield).</p>", "<p>Halolactonization reaction worked well both on neat liquid and solid substrates. Although the diesters <bold>17</bold> and <bold>20</bold> are solids, their reactions proceeded through formation of melts [##REF##11535074##17##] and gave the corresponding halolactones in high yields (##TAB##0##Table 1##, entries 6–9). However, stirring was necessary as the respective products were also solids and sometimes they solidified at the interface of FC-72 and ester layers, preventing further reaction. Good stirring prevented formation of a solid layer. Alternatively, the solid layer was occasionally stirred with a small glass rod (a sealed capillary melting point tube) to break the clumps. Solid acid <bold>14</bold> reacted without an apparent formation of a melt and that may explain why the yields were somewhat lower compared to the corresponding diester <bold>17</bold>.</p>" ]

[ "<title>Results and Discussion</title>", "<p>Bromolactonization worked very well and was easy to monitor as it was a true “phase-vanishing” reaction. The bromine layer disappeared at the end of the reaction and the products were obtained in good to excellent yields. However, sometimes dibromo derivatives, resulting from addition of bromine to a double bond, were by-products. The reaction mechanism of halolactonization in basic aqueous medium is believed to involve formation of a halonium ion followed by attack of an oxygen nucleophile [##UREF##8##13##].</p>", "<p>Reaction times ranged from 20–60 minutes when Br₂ or ICl were used to several days when I₂ was used. Reaction rates can be controlled by the amount of FC-72 and the rate of stirring – a larger amount (greater depth) of FC-72 resulted in a slower reaction, while increased rate of stirring resulted in a faster reaction. If a very slow rate is desired, the reaction can be done without any stirring. Very fast stirring is not recommended as it may lead to direct mixing of the top and bottom phases and a violent reaction. When a reaction is done with stirring, it is recommended that a vial be clamped and not just placed on the top of a magnetic stirrer. Sometimes dense, solid products form, which encase the stirring bar. A result is that, if not clamped, the entire vial may begin to rotate and it may flip over. Usually, solubility of the reaction products in FC-72 is negligible and it can be reused. However, one should always check the fluorous phase for presence of the product. Finally, due to a low boiling point of FC-72, a considerable amount of it may evaporate in the course of longer reactions unless the reaction vessel is capped. Capping a vial is not always feasible as the reaction byproduct is a gas. In our hands, on ~5 g scale, there were no problems when reaction was done in a capped 20 mL vial.</p>", "<p>In the case of bromination of 4-pentenoic acid (<bold>1</bold>), under phase-vanishing conditions there was competing formation of 4,5-dibromopentenoic acid (<bold>3</bold>). Formation of the dibromo derivative as a minor product in the course of bromolactonization of 4-pentenoic acid has been reported along with the observation that it readily cyclizes to give the corresponding bromolactone [##UREF##9##14##–##UREF##10##15##]. For comparison, the reaction was done in dichloromethane and directly on neat reagents under solvent-free conditions (SFC) (<bold><italic>caution: an extremely violent reaction!</italic></bold>). In dichloromethane as a solvent, bromolactone <bold>2</bold> was produced in a slight excess. Under SFC, the reaction was surprisingly clean in that it gave only a mixture of the bromolactone <bold>2</bold> and dibromoacid <bold>3</bold>. The ratios of the two varied between different runs with the dibromide <bold>3</bold> predominating. Outcome of the reaction under phase-vanishing conditions on neat reactants closely resembled an SFC reaction (##FIG##0##Scheme 1##, see ##SUPPL##0##Supporting Information File 1## and ##SUPPL##1##Supporting Information File 2## for full experimental data). 4,5-Dibromopentanoic acid was identified based on its mass spectrum and was not isolated. Instead, the crude reaction mixture was treated with aqueous sodium bicarbonate and the resulting bromolactone <bold>2</bold> was isolated. Thus, unlike other substrates in this study, 4-pentenoic acid required basic reaction conditions for successful bromolactonization. As bromolactonization of methyl esters <bold>17</bold> and <bold>20</bold> worked very well (<italic>vide infra</italic>), bromolactonization of methyl 4-pentenoate was attempted. The major product was methyl 4,5-dibromopentanoate (~80%) and the bromolactone was only a minor product (~20%).</p>", "<p>Interestingly, when the reaction was done in ethyl acetate, or ethyl acetate was used in a work up, the major isolated product was ethyl 4,5-dibromopentanoate (<bold>4</bold>) (##FIG##1##Scheme 2##). Apparently, 4,5-dibromopentanoic acid underwent hydrogen bromide-catalyzed transesterification with ethyl acetate. In a control experiment, 4-pentenoic acid was transesterified by dissolving it in ethyl acetate and adding a catalytic amount of HBr in acetic acid. The best results, and the best yields of bromolactone, were obtained under phase-vanishing conditions when, upon consumption of bromine, FC-72 was removed and the residue was treated with aqueous NaHCO₃, followed by extraction with ethyl acetate. It should be pointed out that this was the only bromolactonization where we encountered problems. Bromolactonization of other compounds (<bold>7</bold>, <bold>14</bold>, <bold>17</bold> and <bold>20</bold>) worked well.</p>", "<p>While reaction of 4-pentenoic acid with bromine gave the corresponding bromolactone in a high yield, a reaction with iodine did not give good results (##FIG##2##Scheme 3##). The reaction was slow and the product was apparently unstable. Thus, the corresponding iodolactone <bold>5</bold>, along with a number of unidentified byproducts, was observed in GC-MS analysis of the reaction mixture. Isolation of the reaction product was not attempted. When iodine monochloride was used in place of iodine, the reaction was fast and the corresponding iodolactone <bold>5</bold> was obtained cleanly and in a high yield. The only byproduct (&lt;2% according to GC-MS), which was not isolated and was not characterized, had a mass that corresponded to chlorolactone <bold>6</bold>. When iodolactonization of 4-pentenoic acid was done in dichloromethane, the same byproduct formed in a larger amount (8–10%).</p>", "<p>Treatment of 5-norbornene-2-carboxylic acid (<bold>7</bold>) (3:1 mixture of <italic>endo</italic> and <italic>exo</italic> isomers) gave the expected bromolactone <bold>8</bold> along with four byproducts (##FIG##3##Scheme 4##). GC analysis showed the four byproducts as two pairs of closely spaced peaks (retention times of 8.1 and 8.3 min for one pair and 12.5 and 12.6 min for the other pair). Furthermore, GC-MS analysis indicated that the byproducts were the corresponding dibromo compounds. They were tentatively assigned structures <bold>9</bold>–<bold>12</bold>. It was expected that one pair of the dibromo compounds (<bold>11</bold> and <bold>12</bold>) would form as <italic>exo</italic>-5-norbornene-2-carboxylic acid cannot cyclize and, indeed, a pair of dibromo compounds formed in the amount that approximately corresponded to the amount of the starting <italic>exo</italic>-5-norbornene carboxylic acid. Another pair of dibromo compounds (assumed to have structures <bold>9</bold> and <bold>10</bold>) formed in a combined yield of ~10% (according to GC-MS analysis). Thus, formation of dibromo products decreased compared to 4-pentenoic acid. Interestingly, no dibromo derivatives at all were observed in bromolactonizations of diacid <bold>14</bold> and diesters <bold>17</bold> and <bold>20</bold> (##TAB##0##Table 1##, entries 3, 6 and 8).</p>", "<p>Iodine monochloride also worked very well on the same substrates (##TAB##0##Table 1##, entries 2, 4, 7 and 9). However, the reaction was more difficult to monitor. As commercially available iodine monochloride contains an excess of iodine, one has to use an excess of the reagent and, at the end of the reaction, there was some unreacted iodine left over. Thus, one cannot tell when the reaction is over and reaction times were more difficult to determine. Reaction times provided in the ##TAB##0##Table 1## are the times when the reaction was definitively completed. The actual reaction times were probably shorter. As in the case of 4-pentenoic acid, iodolactonization of compounds <bold>7</bold>, <bold>14</bold> and <bold>17</bold> was accompanied by formation of a small amount of a byproduct (&lt;2% according to GC-MS), which were not isolated. Mass spectra of the byproducts were consistent with the corresponding chlorolactones. Under the phase vanishing conditions, the amounts of chlorolactones were relatively small. In dichloromethane the amounts increased to as much as 10% (GC-MS). Interestingly, in both bromolactonization and iodolactonization, diester <bold>17</bold> gave cleaner products and better yields compared to the corresponding diacid <bold>14</bold>. <italic>trans</italic>-Diester <bold>20</bold> gave the best yields in both bromo- and iodolactonization reactions. Neither dibromo compounds (in a reaction with bromine) nor chlorolactone (in a reaction with iodine monochloride) were observed and the corresponding crude products gave single peaks when analyzed by GC and were &gt;95% pure by ¹H NMR.</p>", "<p>It has been reported that iodolatonization of the diacid <bold>14</bold> with iodine failed, while iodolactonization of the corresponding disodium salt gave a mixture of γ- and δ-lactones [##UREF##11##16##]. However, no experimental details were given. Under phase-vanishing conditions, treatment of <bold>14</bold> with iodine gave the corresponding γ-lactone <bold>16</bold> in a moderate yield. The yield was inferior compared to iodolactonization with iodine monochloride. On the other hand, iodine is a safer reagent and one may consider using it under certain circumstances (e.g., a large scale reaction, or need for an inexpensive reagent rather than a high yield).</p>", "<p>Halolactonization reaction worked well both on neat liquid and solid substrates. Although the diesters <bold>17</bold> and <bold>20</bold> are solids, their reactions proceeded through formation of melts [##REF##11535074##17##] and gave the corresponding halolactones in high yields (##TAB##0##Table 1##, entries 6–9). However, stirring was necessary as the respective products were also solids and sometimes they solidified at the interface of FC-72 and ester layers, preventing further reaction. Good stirring prevented formation of a solid layer. Alternatively, the solid layer was occasionally stirred with a small glass rod (a sealed capillary melting point tube) to break the clumps. Solid acid <bold>14</bold> reacted without an apparent formation of a melt and that may explain why the yields were somewhat lower compared to the corresponding diester <bold>17</bold>.</p>" ]

[ "<title>Conclusion</title>", "<p>In conclusion, a halolactonization of both solid and liquid neat reactants under phase-vanishing conditions is a simple and efficient approach to various halolactones. While a number of methods to prepare those compounds in high yields have been reported, the methodology described avoids use of any solvent other than the phase screen, the reaction was done under neutral reaction conditions, and work-up consisted of mechanical separation of the product from the phase screen. An exception was bromolactonizaiton of 4-pentenoic acid, which required a basic workup. The products were isolated in high yields and require little or no purification. The best electrophiles were Br₂ and ICl while I₂ gave somewhat inferior results. The procedure worked particularly well on methyl esters of the γ,δ-alkenoic acids.</p>" ]

[ "<p>Phase-vanishing reactions are triphasic reactions which involve a reagent, a liquid perfluoroalkane as a phase screen and a substrate. The perfluoroalkane does not dissolve any of the reactants and is used to separate them. Halolactonization of neat substrates under phase-vanishing conditions avoids use of both solvents and basic reaction conditions. Both γ,δ-alkenoic acids as well as the corresponding methyl esters are suitable substrates for phase-vanishing halolactonizations. The reaction works well both on solid and liquid substrates and the products are obtained in good to excellent yields, particularly in the case of rigid bicyclic systems. Bromine (Br₂) and iodine monochloride (ICl) are suitable electrophiles for bromolactonization and iodolactonization, respectively. Although in some cases iodine gave satisfactory yields of the corresponding iodolactone, it is generally inferior to iodine monochloride.</p>" ]

[ "<title>Experimental</title>", "<p>All of the commercially available reagents (bromine, iodine, iodine monochloride, FC-72, 4-pentenoic acid, dicyclopentadiene, acrylic acid, maleic anhydride, dimethyl maleate and dimethyl fumarate) were used as supplied without further purification. Compounds <bold>7</bold>, <bold>17</bold> and <bold>20</bold> were prepared in Diels-Alder reactions between cyclopentadiene and the corresponding dienophiles. Diacid <bold>14</bold> was prepared in a reaction between cyclopentadiene and maleic anhydride followed by aqueous hydrolysis. Those compounds gave satisfactory GC-MS and ¹H NMR spectra.</p>", "<p>Reactions were done in glass vials of various sizes (4–20 mL). Halogen (2.2–30.0 mmol) was added first, followed by of FC-72 (0.5–5 mL) and finally 2.0–25.0 mmol of an alkene. After stirring for 20 min – 3 days, the product was isolated by separating it from the fluorous phase either by filtration (solid products) or by removing the fluorous phase with a pipette.</p>", "<title>Supporting Information</title>" ]

[ "<p>We thank Salvatore Lepore (Department of Chemistry, Florida Atlantic University) for use of his facilities and helpful discussions. Partial financial support in a form of Florida Atlantic University's New Project Development Award (#331006) is gratefully acknowledged.</p>" ]

[ "<fig id=\"C1\" position=\"float\"><label>Scheme 1</label><caption><p>Bromolactonization of 4-pentenoic acid.</p></caption></fig>", "<fig id=\"C2\" position=\"float\"><label>Scheme 2</label><caption><p>Bromolactonization of 4-pentenoic acid in ethyl acetate.</p></caption></fig>", "<fig id=\"C3\" position=\"float\"><label>Scheme 3</label><caption><p>Iodolactonization of 4-pentenoic acid.</p></caption></fig>", "<fig id=\"C4\" position=\"float\"><label>Scheme 4</label><caption><p>Bromolactonization of 5-norbornene-2-carboxylic acid.</p></caption></fig>" ]

[ "<table-wrap id=\"T1\" position=\"float\"><label>Table 1</label><caption><p>Phase-vanishing halolactonization.</p></caption><table frame=\"hsides\" rules=\"groups\"><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Entry</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Starting Material</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Reagent (equiv)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Product (isolated yield, %)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Reaction time</td></tr><tr><td align=\"left\" colspan=\"5\" valign=\"middle\" rowspan=\"1\"><hr/></td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>7</bold> (3:1 <italic>endo</italic>/<italic>exo</italic>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Br₂ (1.05)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>8</bold> (48%, 64% with respect to <italic>endo</italic>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1 h</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>7</bold> (3:1 <italic>endo</italic>/<italic>exo</italic>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">ICl (1.15)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>13</bold> (65%, 87% with respect to <italic>endo</italic>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1 h</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>14</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Br₂ (1.05)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>15</bold> (72%)^a</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1 h</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>14</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">ICl (1.15)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>16</bold> (75%)^a</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1 h</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>14</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">I₂ (1.20)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>16</bold> (54%)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">3 days</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>17</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Br₂ (1.05)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>18</bold> (83%)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1 h</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">7</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>17</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">ICl (1.15)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>19</bold> (86%)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1 h</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">8</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>20</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Br₂ (1.05)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>21</bold> (92%)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">30 min</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>20</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">ICl (1.15)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>22</bold> (94%)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1 h</td></tr></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"SD1\"><label>File 1</label><caption><p>Experimental part</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"SD2\"><label>File 2</label><caption><p>GC-MS and NMR spectra</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"TFN1\"><p>^aIsolated as the corresponding methyl esters.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-g002\"/>", "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-g003\"/>", "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-g004\"/>", "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-g005\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i001.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i002.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i003.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i004.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i005.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i006.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i007.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i008.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i009.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i010.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i011.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-29-i012.jpg\"/>" ]

[ "<media mime-subtype=\"msword\" mimetype=\"application\" xlink:href=\"Beilstein_J_Org_Chem-04-29-s001.doc\" xlink:type=\"simple\" id=\"d39e707\" position=\"anchor\"/>", "<media mime-subtype=\"pdf\" mimetype=\"application\" xlink:href=\"Beilstein_J_Org_Chem-04-29-s002.pdf\" xlink:type=\"simple\" id=\"d39e714\" position=\"anchor\"/>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Our ongoing interest in constructing combinatorial libraries of highly glycosylated beta-peptides that can mimic specific oligosaccharide-protein interactions prompted us to further search for efficient routes toward glycosylated amino acid building blocks derived from asparaginic acid in which the glycon is bound to C-1 of the asparaginic acid through variable spacers (##FIG##0##Figure 1##). Previously, we have prepared a series of glycosylated asparaginic acid building blocks containing as spacers either simple alkyl chains [##UREF##0##1##], or amino alcohols [##UREF##1##2##–##UREF##2##3##]. Such building blocks have been shown to be well suited for combinatorial solid phase or spot synthesis of libraries of highly glycosylated peptides, some members of which were indeed shown to behave like oligosaccharide mimics capable to specifically bind lectins [##UREF##0##1##,##REF##17487187##4##].</p>", "<p>In order to increase the structural diversity of the aforementioned building blocks, we contemplated using as the spacer entity 1,2,3-triazoles which are known to be easily generated through a copper-catalyzed 1,3-dipolar cycloaddition of an organic azide and an alkynyl derivative (Click Reaction) [##UREF##3##5##–##REF##11975567##7##]. For review articles on copper-catalyzed Click Reactions see references [##UREF##5##8##–##REF##17721589##11##]. Recently, we applied this approach to a series of 1,2,3-triazole containing per-<italic>O</italic>-acetyl-glycosides which were prepared by copper-catalyzed 1,3-dipolar cycloaddition either between fully acetylated propargyl 1-thio-glycosides and <italic>t</italic>-butyl (<italic>S</italic>)-4-azido-3-fluorenylmethyloxycarbamido-butyrate or between Fmoc-L-Asp(O<italic>^t</italic>Bu)-propargyl amide and 2,3,4,6-tetra-<italic>O</italic>-acetyl-glycosyl azides and ethyl 2,3,4-tri-<italic>O</italic>-acetyl-6-azido-6-deoxy-1-thio-glycosides, respectively [##UREF##7##12##]. In order to increase the structural diversity of glycosyl amino acid building blocks containing 1,2,3-triazole spacers even more, we next looked at the possibility to use glycosides bearing both, azido and alkynyl groups in copper-catalyzed 1,3-cycloadditions. The results are presented here.</p>" ]

[]

[ "<title>Results and Discussion</title>", "<p>First, 2-propynyl 6-azido-6-deoxy-2,3,4-tri-<italic>O</italic>-acetyl-β-D-glucopyranoside (<bold>4a</bold>) was prepared by the following sequence. 2-Propynyl 2,3,4,6-tetra-<italic>O</italic>-acetyl-β-D-glucopyranoside (<bold>1a</bold>) [##UREF##8##13##] was Zemplén-deacetylated with a catalytic amount of sodium methanolate in methanol. Next, thus obtained crude 2-propynyl β-D-glucopyranoside was regioselectively tosylated at position 6 [##UREF##9##14##] followed by chromatographic purification to afford 6-<italic>O</italic>-<italic>p</italic>-tolylsulfonyl-glucoside <bold>2a</bold> in 66% yield. Acetylation of the latter with acetic anhydride in pyridine gave crude tri-<italic>O</italic>-acetyl-6-<italic>O</italic>-<italic>p</italic>-tolylsulfonyl-glucoside <bold>3a</bold> which was sufficiently pure for the next step. Treatment of <bold>3a</bold> with NaN₃ in DMF finally afforded 6-azido-6-deoxy-glucoside <bold>4a</bold> in 38% yield. When glucoside <bold>4a</bold> was reacted with asparaginic propargyl amide derivative <bold>5</bold> [##UREF##7##12##] in the presence of (EtO)₃PCuI as catalyst and with or without microwave irradiation [##REF##12762694##15##], the induced 1,3-dipolar cycloaddition between the alkynyl and azide moieties (Click Reaction) afforded compound <bold>6</bold> in variable medium yields of approximately 60%. The yield depended on the reaction conditions under which the cycloaddition was carried out. Several byproducts were formed during this cycloaddition reaction which, however, could not be separated and characterized. The amount of these byproducts increased at higher reaction temperatures or upon irradiation with microwave. It was anticipated that the byproducts which lowered the yield of compound <bold>6</bold> might be decomposition products of the starting material <bold>4a</bold>. Therefore, the more stable benzoylated glucoside <bold>3a'</bold> was prepared from <bold>2a</bold>, and converted into the azide <bold>4a'</bold> in 89% and 84% yield, respectively. Treatment of <bold>4a'</bold> with <bold>5</bold> under Cu(I)-catalysis, however, only resulted in a complex mixture of reaction products from which no uniform product could be isolated. Therefore, it was concluded that <bold>4a</bold> and <bold>4a'</bold> may have reacted with themselves resulting in products of oligomerization. Indeed, when <bold>4a</bold> was treated with a catalytic amount of (EtO)₃PCuI , TLC (ethyl acetate/<italic>n</italic>-hexane 1:1) revealed the formation of one faster moving product along with a complex mixture of slower moving products with mobility similar to those previously observed. Careful inspection of the products revealed that dimerisation of <bold>4a</bold> occurred, affording the dimeric glycoside <bold>7a</bold> beside products of oligomerization (##FIG##1##Scheme 1##). The reaction proceeded significantly slower than the coupling of <bold>4a</bold> and <bold>5</bold>. A faster reaction occurred upon irradiation with microwave, which also gave a higher yield (54%) of <bold>7a</bold>. Benzoylated glycoside <bold>4a'</bold> did not give any product of dimerization though. Only oligomers <bold>8</bold> were observed in this case (for details see ##SUPPL##0##Supporting Information File 1##).</p>", "<p>At first, it was unclear whether <bold>7a</bold> was formed by an intramolecular cyclization or a dimerization of <bold>4a</bold> since its concentration-dependent ESI-MS and MALDI-TOF-MS spectra both showed peaks corresponding to the molecular mass of <bold>4a</bold> and <bold>7a</bold>, respectively. However, the dimeric structure of compound <bold>7a</bold> was finally unambiguously assigned by NMR spectroscopy and field desorption (FD) mass spectrometry. The NMR spectra of <bold>7a</bold> showed no conformative anomalies of the pyranose ring what would have been expected if <bold>7a</bold> would have been the product of intramolecular 1,3-dipolar cycloaddition of the azido group and the 2-propynyl aglycon in the starting material <bold>4a</bold>.</p>", "<p>The oligomerization of glycosides containing both, an azido and an alkynyl group upon copper-catalyzed Click-Reaction had been observed previously in two instances. Gin and coworker recently found that 2,3,6-tri-<italic>O</italic>-benzyl-4-<italic>O</italic>-(2-propynyl)-α-D-mannopyranosyl azide affords a cyclic trimer upon 1,3-dipolar cycloaddition of its azido moiety to its propynyl moiety while the corresponding α-1,4-linked manno-disaccharide afforded a cyclic dimer similar to compound <bold>7a</bold> [##REF##16178563##16##]. Jarosz et al. also recently reported about the copper catalyzed reaction of 6-azido-1',2,3,3',4,4'-hexa-<italic>O</italic>-benzyl-6-deoxy-6'-propargyl-sucrose to afford either a product of intramolecular cyclization or a dimeric product, depending on the reaction conditions [##UREF##10##17##]. Likewise, Vasella reported the thermal intramolecular 1,3-dipolar cycloaddition of protected 2-azidoethyl 4⁵-<italic>O</italic>-(2-propynyl)-malto-hexaoside, giving the corresponding isomeric macrocyclic derivatives [##UREF##11##18##]. In the light of Gin's and Jarosz's results and our own unexpected finding that <bold>4a</bold> can form cyclic dimers upon copper-catalyzed Click-Reaction, we investigated several other 2-propynyl and 3-butynyl 6-azido-6-deoxy-glycosides <bold>4</bold> in order to probe their ability to form similar cyclic dimers <bold>7</bold>.</p>", "<p>First, an alternative route to 2-propynyl 6-azido-6-deoxy-glucoside <bold>4a</bold> was attempted (##FIG##2##Scheme 2##). Compound <bold>1a</bold> was deacetylated and treated with <italic>N</italic>-bromosuccinimide and triphenylphosphine in DMF according to Hanessian's procedure [##UREF##12##19##] followed by reacetylation of the OH-groups with acetic anhydride in pyridine to afford 2-propynyl 6-bromo-6-deoxy-2,3,4-tri-<italic>O</italic>-acetyl-β-D-glucopyranoside (<bold>3a''</bold>) in 60% yield. Next, the latter was stirred with NaN₃ in DMF (48 h, 65 °C) to afford <bold>4a</bold> in 44% yield. The preparation of compound <bold>4a</bold>\n<italic>via</italic> the corresponding tosylate <bold>3a</bold> was somewhat more convenient than the synthesis <italic>via</italic> the 6-bromo-6-deoxy counterpart <bold>3a''</bold> and resulted in a similar overall yield. Therefore, all other 6-azido-6-deoxy-glycosides <bold>4</bold> were prepared via the corresponding tosylates <bold>3</bold> as described above. ##FIG##2##Scheme 2## summarizes the yields for the preparation of the tosylates <bold>3</bold> and 6-azido-6-deoxy-glycosides <bold>4</bold>. Starting materials <bold>1</bold> were prepared following known procedures for <bold>1a</bold> [##UREF##8##13##,##UREF##13##20##], <bold>1b</bold> [##UREF##14##21##], <bold>1d</bold> [##UREF##8##13##], <bold>1f</bold> [##UREF##13##20##,##UREF##15##22##] and <bold>1g</bold> [##UREF##14##21##]. 2-Propynyl 2,3,4,6-tetra-<italic>O</italic>-acetyl-α-D-glycopyranosides <bold>1c</bold> and <bold>1e</bold> have not been described previously. They were prepared from D-glucose and D-galactose in 20% and 22% yield, respectively via classical Fischer-Glycosylation in 2-propynol as the solvent under acidic conditions followed by acetylation of the intermediate glycosides and chromatographic separation of the anomeric acetates.</p>", "<p>Next, glycosides <bold>4a</bold>–<bold>g</bold> were submitted to dimerization by 1,3-dipolar cycloaddition reaction. As the catalyst, 10 mol% (EtO)₃PCuI was applied and used along with three equivalents diisopropyl ethylamine in toluene [##REF##12762694##15##]. Microwave irradiation [##UREF##16##23##] reduced the reaction time significantly but also resulted in decomposition of the starting material in some cases. ##TAB##0##Table 1## summarizes the results for the dimerization of <bold>4a</bold>–<bold>g</bold> to <bold>7a</bold>–<bold>g</bold>.</p>", "<p>Yields of the copper-catalyzed dimerizations were low to medium (14–54%) and depended on the sugar moiety, the anomeric configuration and the ring size which was formed during the Click-Reaction. In general, no other cyclization product could be isolated from the reaction mixtures although significant amounts of byproducts were formed. These byproducts were slower moving compounds on TLC (ethyl acetate/<italic>n</italic>-hexane 1:1) and appeared to be products of oligomerization of the starting material <bold>4</bold>. In the case of benzoylated glycoside <bold>4a'</bold> where no cyclic dimer could be isolated from the complex reaction mixture, FAB MS of the purified mixture indeed revealed the presence of linear dimers, trimers and tetramers.</p>", "<p>α-Galactoside <bold>4e</bold> did not give any isolable dimer <bold>7e</bold> at all (<italic>cf.</italic>\n##TAB##0##Table 1##, entry 5). Similarly, α-glucoside <bold>4c</bold> resulted in a lower yield of the corresponding dimer compared to β-glucoside <bold>4a</bold> (<italic>cf.</italic>\n##TAB##0##Table 1##, entries 1 and 3). This may be attributed to a significant ring-strain in the α-linked dimers. For example, the ¹H NMR of compound <bold>7c</bold> showed an unusually small coupling constant between H-1 and H-2 (&lt;1.0 Hz) and H-2 and H-3 (3.1 Hz) which is indicative that the sugar moieties in <bold>7c</bold> are no longer in a chair conformation (see Table 1 in the ##SUPPL##0##Supporting Information File 1##). No such effects were observed in the manno series though (<italic>cf.</italic>\n##TAB##0##Table 1##, entries 6 and 7). Here, the corresponding dimers <bold>7f</bold> and <bold>7g</bold> showed regular coupling constants in their NMR spectra.</p>", "<p>The effect of microwave irradiation on the outcome of the dimerization is somewhat confusing. In general, microwave irradiation resulted in a faster reaction, <italic>i.e.</italic> faster disappearance of the starting material (<italic>cf.</italic>\n##TAB##0##Table 1##, entry 1). Similar accelerations of Click-Reactions upon microwave irradiation had been observed previously as well [##REF##12762694##15##]. However, the higher temperature associated with the microwave irradiation also resulted in a more pronounced decomposition of the starting material, and thus resulted in a lower yield of the dimers (<italic>cf.</italic>\n##TAB##0##Table 1##, entries 1, 3 and 4) while heating of the reaction mixture alone resulted in complex product mixtures from which no dimerization products could be isolated. In the case of compounds <bold>4b</bold> and <bold>4e</bold>–<bold>g</bold>, no reaction occurred at room temperature (<italic>cf.</italic>\n##TAB##0##Table 1##, entries 2 and 5–7).</p>" ]

[ "<title>Results and Discussion</title>", "<p>First, 2-propynyl 6-azido-6-deoxy-2,3,4-tri-<italic>O</italic>-acetyl-β-D-glucopyranoside (<bold>4a</bold>) was prepared by the following sequence. 2-Propynyl 2,3,4,6-tetra-<italic>O</italic>-acetyl-β-D-glucopyranoside (<bold>1a</bold>) [##UREF##8##13##] was Zemplén-deacetylated with a catalytic amount of sodium methanolate in methanol. Next, thus obtained crude 2-propynyl β-D-glucopyranoside was regioselectively tosylated at position 6 [##UREF##9##14##] followed by chromatographic purification to afford 6-<italic>O</italic>-<italic>p</italic>-tolylsulfonyl-glucoside <bold>2a</bold> in 66% yield. Acetylation of the latter with acetic anhydride in pyridine gave crude tri-<italic>O</italic>-acetyl-6-<italic>O</italic>-<italic>p</italic>-tolylsulfonyl-glucoside <bold>3a</bold> which was sufficiently pure for the next step. Treatment of <bold>3a</bold> with NaN₃ in DMF finally afforded 6-azido-6-deoxy-glucoside <bold>4a</bold> in 38% yield. When glucoside <bold>4a</bold> was reacted with asparaginic propargyl amide derivative <bold>5</bold> [##UREF##7##12##] in the presence of (EtO)₃PCuI as catalyst and with or without microwave irradiation [##REF##12762694##15##], the induced 1,3-dipolar cycloaddition between the alkynyl and azide moieties (Click Reaction) afforded compound <bold>6</bold> in variable medium yields of approximately 60%. The yield depended on the reaction conditions under which the cycloaddition was carried out. Several byproducts were formed during this cycloaddition reaction which, however, could not be separated and characterized. The amount of these byproducts increased at higher reaction temperatures or upon irradiation with microwave. It was anticipated that the byproducts which lowered the yield of compound <bold>6</bold> might be decomposition products of the starting material <bold>4a</bold>. Therefore, the more stable benzoylated glucoside <bold>3a'</bold> was prepared from <bold>2a</bold>, and converted into the azide <bold>4a'</bold> in 89% and 84% yield, respectively. Treatment of <bold>4a'</bold> with <bold>5</bold> under Cu(I)-catalysis, however, only resulted in a complex mixture of reaction products from which no uniform product could be isolated. Therefore, it was concluded that <bold>4a</bold> and <bold>4a'</bold> may have reacted with themselves resulting in products of oligomerization. Indeed, when <bold>4a</bold> was treated with a catalytic amount of (EtO)₃PCuI , TLC (ethyl acetate/<italic>n</italic>-hexane 1:1) revealed the formation of one faster moving product along with a complex mixture of slower moving products with mobility similar to those previously observed. Careful inspection of the products revealed that dimerisation of <bold>4a</bold> occurred, affording the dimeric glycoside <bold>7a</bold> beside products of oligomerization (##FIG##1##Scheme 1##). The reaction proceeded significantly slower than the coupling of <bold>4a</bold> and <bold>5</bold>. A faster reaction occurred upon irradiation with microwave, which also gave a higher yield (54%) of <bold>7a</bold>. Benzoylated glycoside <bold>4a'</bold> did not give any product of dimerization though. Only oligomers <bold>8</bold> were observed in this case (for details see ##SUPPL##0##Supporting Information File 1##).</p>", "<p>At first, it was unclear whether <bold>7a</bold> was formed by an intramolecular cyclization or a dimerization of <bold>4a</bold> since its concentration-dependent ESI-MS and MALDI-TOF-MS spectra both showed peaks corresponding to the molecular mass of <bold>4a</bold> and <bold>7a</bold>, respectively. However, the dimeric structure of compound <bold>7a</bold> was finally unambiguously assigned by NMR spectroscopy and field desorption (FD) mass spectrometry. The NMR spectra of <bold>7a</bold> showed no conformative anomalies of the pyranose ring what would have been expected if <bold>7a</bold> would have been the product of intramolecular 1,3-dipolar cycloaddition of the azido group and the 2-propynyl aglycon in the starting material <bold>4a</bold>.</p>", "<p>The oligomerization of glycosides containing both, an azido and an alkynyl group upon copper-catalyzed Click-Reaction had been observed previously in two instances. Gin and coworker recently found that 2,3,6-tri-<italic>O</italic>-benzyl-4-<italic>O</italic>-(2-propynyl)-α-D-mannopyranosyl azide affords a cyclic trimer upon 1,3-dipolar cycloaddition of its azido moiety to its propynyl moiety while the corresponding α-1,4-linked manno-disaccharide afforded a cyclic dimer similar to compound <bold>7a</bold> [##REF##16178563##16##]. Jarosz et al. also recently reported about the copper catalyzed reaction of 6-azido-1',2,3,3',4,4'-hexa-<italic>O</italic>-benzyl-6-deoxy-6'-propargyl-sucrose to afford either a product of intramolecular cyclization or a dimeric product, depending on the reaction conditions [##UREF##10##17##]. Likewise, Vasella reported the thermal intramolecular 1,3-dipolar cycloaddition of protected 2-azidoethyl 4⁵-<italic>O</italic>-(2-propynyl)-malto-hexaoside, giving the corresponding isomeric macrocyclic derivatives [##UREF##11##18##]. In the light of Gin's and Jarosz's results and our own unexpected finding that <bold>4a</bold> can form cyclic dimers upon copper-catalyzed Click-Reaction, we investigated several other 2-propynyl and 3-butynyl 6-azido-6-deoxy-glycosides <bold>4</bold> in order to probe their ability to form similar cyclic dimers <bold>7</bold>.</p>", "<p>First, an alternative route to 2-propynyl 6-azido-6-deoxy-glucoside <bold>4a</bold> was attempted (##FIG##2##Scheme 2##). Compound <bold>1a</bold> was deacetylated and treated with <italic>N</italic>-bromosuccinimide and triphenylphosphine in DMF according to Hanessian's procedure [##UREF##12##19##] followed by reacetylation of the OH-groups with acetic anhydride in pyridine to afford 2-propynyl 6-bromo-6-deoxy-2,3,4-tri-<italic>O</italic>-acetyl-β-D-glucopyranoside (<bold>3a''</bold>) in 60% yield. Next, the latter was stirred with NaN₃ in DMF (48 h, 65 °C) to afford <bold>4a</bold> in 44% yield. The preparation of compound <bold>4a</bold>\n<italic>via</italic> the corresponding tosylate <bold>3a</bold> was somewhat more convenient than the synthesis <italic>via</italic> the 6-bromo-6-deoxy counterpart <bold>3a''</bold> and resulted in a similar overall yield. Therefore, all other 6-azido-6-deoxy-glycosides <bold>4</bold> were prepared via the corresponding tosylates <bold>3</bold> as described above. ##FIG##2##Scheme 2## summarizes the yields for the preparation of the tosylates <bold>3</bold> and 6-azido-6-deoxy-glycosides <bold>4</bold>. Starting materials <bold>1</bold> were prepared following known procedures for <bold>1a</bold> [##UREF##8##13##,##UREF##13##20##], <bold>1b</bold> [##UREF##14##21##], <bold>1d</bold> [##UREF##8##13##], <bold>1f</bold> [##UREF##13##20##,##UREF##15##22##] and <bold>1g</bold> [##UREF##14##21##]. 2-Propynyl 2,3,4,6-tetra-<italic>O</italic>-acetyl-α-D-glycopyranosides <bold>1c</bold> and <bold>1e</bold> have not been described previously. They were prepared from D-glucose and D-galactose in 20% and 22% yield, respectively via classical Fischer-Glycosylation in 2-propynol as the solvent under acidic conditions followed by acetylation of the intermediate glycosides and chromatographic separation of the anomeric acetates.</p>", "<p>Next, glycosides <bold>4a</bold>–<bold>g</bold> were submitted to dimerization by 1,3-dipolar cycloaddition reaction. As the catalyst, 10 mol% (EtO)₃PCuI was applied and used along with three equivalents diisopropyl ethylamine in toluene [##REF##12762694##15##]. Microwave irradiation [##UREF##16##23##] reduced the reaction time significantly but also resulted in decomposition of the starting material in some cases. ##TAB##0##Table 1## summarizes the results for the dimerization of <bold>4a</bold>–<bold>g</bold> to <bold>7a</bold>–<bold>g</bold>.</p>", "<p>Yields of the copper-catalyzed dimerizations were low to medium (14–54%) and depended on the sugar moiety, the anomeric configuration and the ring size which was formed during the Click-Reaction. In general, no other cyclization product could be isolated from the reaction mixtures although significant amounts of byproducts were formed. These byproducts were slower moving compounds on TLC (ethyl acetate/<italic>n</italic>-hexane 1:1) and appeared to be products of oligomerization of the starting material <bold>4</bold>. In the case of benzoylated glycoside <bold>4a'</bold> where no cyclic dimer could be isolated from the complex reaction mixture, FAB MS of the purified mixture indeed revealed the presence of linear dimers, trimers and tetramers.</p>", "<p>α-Galactoside <bold>4e</bold> did not give any isolable dimer <bold>7e</bold> at all (<italic>cf.</italic>\n##TAB##0##Table 1##, entry 5). Similarly, α-glucoside <bold>4c</bold> resulted in a lower yield of the corresponding dimer compared to β-glucoside <bold>4a</bold> (<italic>cf.</italic>\n##TAB##0##Table 1##, entries 1 and 3). This may be attributed to a significant ring-strain in the α-linked dimers. For example, the ¹H NMR of compound <bold>7c</bold> showed an unusually small coupling constant between H-1 and H-2 (&lt;1.0 Hz) and H-2 and H-3 (3.1 Hz) which is indicative that the sugar moieties in <bold>7c</bold> are no longer in a chair conformation (see Table 1 in the ##SUPPL##0##Supporting Information File 1##). No such effects were observed in the manno series though (<italic>cf.</italic>\n##TAB##0##Table 1##, entries 6 and 7). Here, the corresponding dimers <bold>7f</bold> and <bold>7g</bold> showed regular coupling constants in their NMR spectra.</p>", "<p>The effect of microwave irradiation on the outcome of the dimerization is somewhat confusing. In general, microwave irradiation resulted in a faster reaction, <italic>i.e.</italic> faster disappearance of the starting material (<italic>cf.</italic>\n##TAB##0##Table 1##, entry 1). Similar accelerations of Click-Reactions upon microwave irradiation had been observed previously as well [##REF##12762694##15##]. However, the higher temperature associated with the microwave irradiation also resulted in a more pronounced decomposition of the starting material, and thus resulted in a lower yield of the dimers (<italic>cf.</italic>\n##TAB##0##Table 1##, entries 1, 3 and 4) while heating of the reaction mixture alone resulted in complex product mixtures from which no dimerization products could be isolated. In the case of compounds <bold>4b</bold> and <bold>4e</bold>–<bold>g</bold>, no reaction occurred at room temperature (<italic>cf.</italic>\n##TAB##0##Table 1##, entries 2 and 5–7).</p>" ]

[ "<title>Conclusion</title>", "<p>We describe for the first time the copper-catalyzed dimerization of simple acetylated 2-propynyl and 3-butynyl 6-azido-6-deoxy-glycosides in the gluco, galacto and manno series leading to macrocyclic rings containing two sugar moieties and two 1,2,3-triazole moieties. For instance, such compounds may function as novel ligands for the preparation of metal complexes [##UREF##17##24##]. Further examples for cyclizations of other azido-alkynyl-glycosides are under investigation.</p>" ]

[ "<p>Copper-catalyzed, thermal or microwave promoted 1,3-dipolar cycloaddition (Click Reaction) of 2-propynyl and 3-butynyl 2,3,4-tri-<italic>O</italic>-acetyl-6-azido-6-deoxy-glycopyranosides in the D-gluco, D-galacto and D-manno series afford the corresponding dimeric cycloaddition products.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>This work was financially supported by the Fonds der Chemischen Industrie. We thank the groups of Professor Albert and Professor Zeller for measuring the NMR and MS spectra.</p>" ]

[ "<fig id=\"F1\" position=\"float\"><label>Figure 1</label><caption><p>Schematic representation of glycosylated building blocks for the combinatorial synthesis of glycopeptides.</p></caption></fig>", "<fig id=\"C1\" position=\"float\"><label>Scheme 1</label><caption><p>Synthesis and reaction of compounds <bold>4a</bold> and <bold>4a</bold>'.</p></caption></fig>", "<fig id=\"C2\" position=\"float\"><label>Scheme 2</label><caption><p>Preparation of compounds <bold>4a</bold>–<bold>g</bold>.</p></caption></fig>" ]

[ "<table-wrap id=\"T1\" position=\"float\"><label>Table 1</label><caption><p>Dimerization of Glycosides <bold>4a</bold>–<bold>g</bold> under Cu-Catalysis.</p></caption><table frame=\"hsides\" rules=\"groups\"><tr><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Entry</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Glycoside <bold>4</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Product <bold>7</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Conditions</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Yield</td></tr><tr><td align=\"left\" colspan=\"5\" valign=\"middle\" rowspan=\"1\"><hr/></td></tr><tr><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>4a</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">\n<bold>7a</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">12 h rt<break/>1 h 80 °C, 20 W MW</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">54%<break/>20%</td></tr><tr><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>4b</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">\n<bold>7b</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">12 h rt<break/>1 h 80 °C, 20 W MW</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">-<break/>32%</td></tr><tr><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>4c</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">\n<bold>7c</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">12 h rt<break/>1 h 80 °C, 20 W MW</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">14%<break/>-</td></tr><tr><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>4d</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">\n<bold>7d</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">12 h rt<break/>1 h 80 °C, 20 W MW</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">28%<break/>-</td></tr><tr><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>4e</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">\n<bold>7e</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">12 h rt<break/>1 h 80 °C, 20 W MW</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">-<break/>-</td></tr><tr><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>4f</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">\n<bold>7f</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">12 h rt<break/>1 h 80 °C, 20 W MW</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">-<break/>30%</td></tr><tr><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">7</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><break/><bold>4g</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">\n<bold>7g</bold></td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">12 h rt<break/>1 h 80 °C, 20 W MW</td><td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">-<break/>53%</td></tr></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"SD1\"><label>File 1</label><caption><p>Experimental Data</p></caption></supplementary-material>" ]

[]

[ "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-g002\"/>", "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-g003\"/>", "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-g004\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i001.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i002.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i003.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i004.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i005.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i006.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i007.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i008.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i009.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i010.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i011.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i012.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i013.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-30-i014.jpg\"/>" ]

[ "<media mime-subtype=\"msword\" mimetype=\"application\" xlink:href=\"Beilstein_J_Org_Chem-04-30-s001.doc\" xlink:type=\"simple\" id=\"d39e878\" position=\"anchor\"/>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>The cycloadditions of anthrones <bold>1</bold> and <italic>N</italic>-substituted maleimides <bold>2</bold> are prominent examples of asymmetric catalysis exerted by chiral Brønsted bases. Moderate to excellent stereoselectivities of products <bold>3</bold> have been reported using pyrrolidines <bold>4</bold> [##UREF##0##1##–##UREF##1##2##], cyclic guanidine <bold>5</bold> [##REF##17044689##3##], or cinchona alkaloids <bold>6</bold> [##UREF##2##4##] as catalysts. Recently, we could promote this type of cycloaddition by metal-free bisoxazolines <bold>7</bold> in up to 70% <italic>ee</italic>, in spite of their limited Brønsted-basicity [##UREF##3##5##] (##FIG##0##Scheme 1##).</p>", "<p>Our study was motivated by the structural similarity of bisoxazolines <bold>7</bold> and bisamidines <bold>8</bold>. Bisamidines <bold>8</bold>, readily accessible from malonodinitrile in two steps, prefer the conjugated tautomeric form (enamine-imine) in the monoprotonated state, which is characterised by an almost planar structure [##REF##17585816##6##] (##FIG##1##Scheme 2##).</p>", "<p>The aqueous p<italic>K</italic>_a of <bold>8</bold>·H⁺ is approximately 11, sufficient to allow deprotonation of anthrones <bold>1</bold> (p<italic>K</italic>_a around 10, [##UREF##4##7##–##UREF##5##8##]) by bisamidines to a significant extent. Here we report on the use of neutral bisamidines <bold>8</bold> as asymmetric Brønsted base catalysts in the cycloaddition of anthrones <bold>1</bold> and maleimides <bold>2</bold>.</p>" ]

[]

[ "<title>Results and Discussion</title>", "<p>Analogous to the synthesis of compound <bold>8a</bold> [##REF##17585816##6##], the other bisamidines were prepared as hydrochlorides in 60–79% yield from the corresponding chiral diamines <bold>9</bold> and bisimidate <bold>10</bold> in refluxing ethanol. Simple extraction in the presence of Na₂CO₃ afforded the neutral bases <bold>8b</bold>–<bold>c</bold> and <italic>ent-</italic><bold>8d</bold> in almost quantitative yield. The <italic>S</italic>,<italic>S</italic> configurated diamines <bold>9b</bold> and <bold>9c</bold> were prepared from L-(+)-tartaric acid (<italic>R,R</italic>) via the vicinal diazide using Saalfrank's procedure [##UREF##6##9##]. <bold>9d</bold> was purchased as the dihydrochloride salt and then deprotonated by aqueous sodium hydroxide. As an “artefact” of the sequence rule, the <italic>S</italic>,<italic>S</italic> configurated diamine <bold>9d</bold> leads to bisamidine <italic>ent</italic>-<bold>8d</bold> (##FIG##2##Scheme 3##).</p>", "<p>The anthrones <bold>1b</bold> (R¹: H; R²: Cl) and <bold>1c</bold> (R¹: Cl; R²: H) resulted from regioselective reductions of 1,8-dichloroanthraquinone [##UREF##7##10##–##REF##11667122##11##]. Aliphatic side chains of compounds <bold>2</bold> could be introduced by a Mitsunobu alkylation of maleimide [##UREF##8##12##]. Alternatively, substituted maleimides were prepared by reaction of maleic anhydride with the corresponding amines followed by ring closure [##UREF##9##13##–##UREF##10##14##].</p>", "<p>Cycloaddition kinetics of <bold>1a</bold> and <bold>2a</bold> was examined first by ¹H NMR in CD₂Cl₂ at room temperature. In the absence of catalyst, no product could be observed after 4 days. 5 mol% of the bisamidinium salt <bold>8a</bold>·H⁺ with tetrakis(3,5-bis(trifluoromethyl)phenyl)borate (TFPB^-) as weakly coordinating anion resulted in 7% yield of <bold>3a</bold> after 4 h. In contrast, only 1 mol% of the free Brønsted base <bold>8a</bold> led to a high rate increase in the first 30 min. After 90 min no further conversion was observed indicating product inhibition (##FIG##3##Figure 1##). Accordingly, the reaction runs best in the base-catalyzed mode. Compared to the bisoxazolines <bold>7</bold>, bisamidines <bold>8</bold> as stronger Brønsted bases induced much higher rates in all subsequent experiments.</p>", "<p>In the next series of experiments, bisamidines <bold>8a</bold>–<bold>c</bold> and <italic>ent</italic>-<bold>8d</bold> were compared as catalysts of the cycloaddition forming <bold>3a</bold> from <italic>N</italic>-phenylmaleimide (<bold>2a</bold>) and anthrone (<bold>1a</bold>). Using 0.25 equiv of catalyst at room temperature, isolated yields between 71% and 86% were obtained after 30 min. The best enantioselectivity, albeit low, was induced by amidine <bold>8c</bold> (24% <italic>ee</italic>). As expected, in the presence of catalyst <italic>ent</italic>-<bold>8d</bold> product <italic>ent</italic>-<bold>3a</bold> was formed preferentially (##TAB##0##Table 1##).</p>", "<p>In a solvent screening using 10 mol% of TBDPS-protected bisamidine <bold>8c</bold>, best results were obtained in dichloromethane (84% yield; 30% <italic>ee</italic>). Even higher yields were accessible in aromatic solvents, however, at the price of reduced stereoselectivity (##TAB##1##Table 2##).</p>", "<p>Lowering the reaction temperature from 23 to −20 °C (<bold>8c</bold>, dichloromethane) retarded the cycloaddition but did not change enantioselectivities. After extended reaction times, excellent yields were still observed. Up to 39% <italic>ee</italic> was finally obtained at −70 °C. However, such conditions resulted in lower yields, even with increased catalyst loads and further extended reaction times. Best results, 96% yield and 36% <italic>ee</italic> with only 10 mol% of catalyst, were found at −40 °C (##TAB##2##Table 3##).</p>", "<p>Having identified suitable experimental conditions, we explored the scope of the bisamidine-catalyzed Diels-Alder reaction. The results are summarized in ##TAB##3##Table 4##. Both electron-donating and electron-withdrawing substituents were tolerated and furnished products in good to excellent yields and with moderate values of <italic>ee</italic>. A remarkable increase in enantioselectivity was observed using maleimide <bold>2i</bold>. The steric hindrance imposed by the large 2,6-diisopropylphenyl moiety of <bold>2i</bold> resulted in 76% <italic>ee</italic> at −70 °C but also lowered reaction rates.</p>", "<p>Only 13% yield could be obtained under such conditions. Yields rose to 65% at room temperature (51% <italic>ee</italic>; entries 11 and 12). With other sterically hindered dienophiles such as <italic>N</italic>-<italic>tert</italic>-butylmaleimide (<bold>2c</bold>), the level of <italic>ee</italic> remained low (entry 4). The halogen-substituted anthrones <bold>1b</bold>–<bold>c</bold> did not react with <bold>2i</bold> at −70 °C. At room temperature, however, <bold>1b</bold> and <bold>2i</bold> were efficiently transformed into <bold>3m</bold> by catalyst <bold>8a</bold> with 76% yield and 54% <italic>ee</italic>. A single recrystallisation step afforded an almost enantiopure product (96% <italic>ee</italic>). The <italic>R</italic>,<italic>R</italic> configuration of compound <bold>3m</bold> was determined by anomalous X-ray diffraction using a single crystal of <bold>3m</bold> with 96% <italic>ee</italic> (##FIG##4##Figure 2##).</p>", "<p>A mechanistic rationalisation is proposed in ##FIG##5##Scheme 4##. The catalyst deprotonates the anthrone in the initial step. This assumption is supported by the p<italic>K</italic>_a values of compounds <bold>2a</bold> (10, [##UREF##4##7##–##UREF##5##8##]) and <bold>8</bold>·H⁺ (~11, [##REF##17585816##6##]). Furthermore, the appearance of the yellow color of enolates (<bold>1</bold>·H⁺) shows significant proton transfer when bisamidine <bold>8a</bold> is added to anthrones <bold>1a</bold>, <bold>1b</bold>, or <bold>1c</bold>. A chiral contact ion pair <bold>A</bold> is formed and controls the stereochemical course of the Diels-Alder reaction with maleimides. In the last step, the catalyst-product-complex <bold>B</bold> dissociates and regenerates the unprotonated bisamidine.</p>" ]

[ "<title>Results and Discussion</title>", "<p>Analogous to the synthesis of compound <bold>8a</bold> [##REF##17585816##6##], the other bisamidines were prepared as hydrochlorides in 60–79% yield from the corresponding chiral diamines <bold>9</bold> and bisimidate <bold>10</bold> in refluxing ethanol. Simple extraction in the presence of Na₂CO₃ afforded the neutral bases <bold>8b</bold>–<bold>c</bold> and <italic>ent-</italic><bold>8d</bold> in almost quantitative yield. The <italic>S</italic>,<italic>S</italic> configurated diamines <bold>9b</bold> and <bold>9c</bold> were prepared from L-(+)-tartaric acid (<italic>R,R</italic>) via the vicinal diazide using Saalfrank's procedure [##UREF##6##9##]. <bold>9d</bold> was purchased as the dihydrochloride salt and then deprotonated by aqueous sodium hydroxide. As an “artefact” of the sequence rule, the <italic>S</italic>,<italic>S</italic> configurated diamine <bold>9d</bold> leads to bisamidine <italic>ent</italic>-<bold>8d</bold> (##FIG##2##Scheme 3##).</p>", "<p>The anthrones <bold>1b</bold> (R¹: H; R²: Cl) and <bold>1c</bold> (R¹: Cl; R²: H) resulted from regioselective reductions of 1,8-dichloroanthraquinone [##UREF##7##10##–##REF##11667122##11##]. Aliphatic side chains of compounds <bold>2</bold> could be introduced by a Mitsunobu alkylation of maleimide [##UREF##8##12##]. Alternatively, substituted maleimides were prepared by reaction of maleic anhydride with the corresponding amines followed by ring closure [##UREF##9##13##–##UREF##10##14##].</p>", "<p>Cycloaddition kinetics of <bold>1a</bold> and <bold>2a</bold> was examined first by ¹H NMR in CD₂Cl₂ at room temperature. In the absence of catalyst, no product could be observed after 4 days. 5 mol% of the bisamidinium salt <bold>8a</bold>·H⁺ with tetrakis(3,5-bis(trifluoromethyl)phenyl)borate (TFPB^-) as weakly coordinating anion resulted in 7% yield of <bold>3a</bold> after 4 h. In contrast, only 1 mol% of the free Brønsted base <bold>8a</bold> led to a high rate increase in the first 30 min. After 90 min no further conversion was observed indicating product inhibition (##FIG##3##Figure 1##). Accordingly, the reaction runs best in the base-catalyzed mode. Compared to the bisoxazolines <bold>7</bold>, bisamidines <bold>8</bold> as stronger Brønsted bases induced much higher rates in all subsequent experiments.</p>", "<p>In the next series of experiments, bisamidines <bold>8a</bold>–<bold>c</bold> and <italic>ent</italic>-<bold>8d</bold> were compared as catalysts of the cycloaddition forming <bold>3a</bold> from <italic>N</italic>-phenylmaleimide (<bold>2a</bold>) and anthrone (<bold>1a</bold>). Using 0.25 equiv of catalyst at room temperature, isolated yields between 71% and 86% were obtained after 30 min. The best enantioselectivity, albeit low, was induced by amidine <bold>8c</bold> (24% <italic>ee</italic>). As expected, in the presence of catalyst <italic>ent</italic>-<bold>8d</bold> product <italic>ent</italic>-<bold>3a</bold> was formed preferentially (##TAB##0##Table 1##).</p>", "<p>In a solvent screening using 10 mol% of TBDPS-protected bisamidine <bold>8c</bold>, best results were obtained in dichloromethane (84% yield; 30% <italic>ee</italic>). Even higher yields were accessible in aromatic solvents, however, at the price of reduced stereoselectivity (##TAB##1##Table 2##).</p>", "<p>Lowering the reaction temperature from 23 to −20 °C (<bold>8c</bold>, dichloromethane) retarded the cycloaddition but did not change enantioselectivities. After extended reaction times, excellent yields were still observed. Up to 39% <italic>ee</italic> was finally obtained at −70 °C. However, such conditions resulted in lower yields, even with increased catalyst loads and further extended reaction times. Best results, 96% yield and 36% <italic>ee</italic> with only 10 mol% of catalyst, were found at −40 °C (##TAB##2##Table 3##).</p>", "<p>Having identified suitable experimental conditions, we explored the scope of the bisamidine-catalyzed Diels-Alder reaction. The results are summarized in ##TAB##3##Table 4##. Both electron-donating and electron-withdrawing substituents were tolerated and furnished products in good to excellent yields and with moderate values of <italic>ee</italic>. A remarkable increase in enantioselectivity was observed using maleimide <bold>2i</bold>. The steric hindrance imposed by the large 2,6-diisopropylphenyl moiety of <bold>2i</bold> resulted in 76% <italic>ee</italic> at −70 °C but also lowered reaction rates.</p>", "<p>Only 13% yield could be obtained under such conditions. Yields rose to 65% at room temperature (51% <italic>ee</italic>; entries 11 and 12). With other sterically hindered dienophiles such as <italic>N</italic>-<italic>tert</italic>-butylmaleimide (<bold>2c</bold>), the level of <italic>ee</italic> remained low (entry 4). The halogen-substituted anthrones <bold>1b</bold>–<bold>c</bold> did not react with <bold>2i</bold> at −70 °C. At room temperature, however, <bold>1b</bold> and <bold>2i</bold> were efficiently transformed into <bold>3m</bold> by catalyst <bold>8a</bold> with 76% yield and 54% <italic>ee</italic>. A single recrystallisation step afforded an almost enantiopure product (96% <italic>ee</italic>). The <italic>R</italic>,<italic>R</italic> configuration of compound <bold>3m</bold> was determined by anomalous X-ray diffraction using a single crystal of <bold>3m</bold> with 96% <italic>ee</italic> (##FIG##4##Figure 2##).</p>", "<p>A mechanistic rationalisation is proposed in ##FIG##5##Scheme 4##. The catalyst deprotonates the anthrone in the initial step. This assumption is supported by the p<italic>K</italic>_a values of compounds <bold>2a</bold> (10, [##UREF##4##7##–##UREF##5##8##]) and <bold>8</bold>·H⁺ (~11, [##REF##17585816##6##]). Furthermore, the appearance of the yellow color of enolates (<bold>1</bold>·H⁺) shows significant proton transfer when bisamidine <bold>8a</bold> is added to anthrones <bold>1a</bold>, <bold>1b</bold>, or <bold>1c</bold>. A chiral contact ion pair <bold>A</bold> is formed and controls the stereochemical course of the Diels-Alder reaction with maleimides. In the last step, the catalyst-product-complex <bold>B</bold> dissociates and regenerates the unprotonated bisamidine.</p>" ]

[ "<title>Conclusion</title>", "<p><italic>C</italic>₂-symmetric bisamidines were shown to be potent chiral Brønsted base catalysts for the Diels-Alder reaction of <italic>N</italic>-substituted maleimides and anthrones. Compared to bisoxazolines <bold>7</bold>, much shorter reaction times under comparable conditions were sufficient with the more basic bisamidine catalysts <bold>8</bold> (~50-fold [##UREF##3##5##]). The higher intrinsic reactivity of the bisamidines allowed to run the reactions at lower temperatures. In both groups of catalysts, the phenyl substituted species induced the lowest enantioselectivities. Bisamidine <bold>8a</bold> performed better than the corresponding bisoxazoline. Increasing the size of substituents in catalysts <bold>8b</bold>–<bold>d</bold> also improved stereoselectivities, but not to high levels. This may be due to the flexible nature of the substituents present in bisamidines <bold>8b</bold> and <bold>8c</bold>. It is instructive, therefore, to compare with the bisoxazolines <bold>7</bold>. By far the best enantioselectivities were observed in this series with the <italic>t</italic>-Bu derivative (47% <italic>ee</italic> versus 3% for the phenyl analogue in the reaction of <bold>1a</bold> and <bold>2a</bold>). Keeping in mind that even the less selective bisamidine <bold>8a</bold> could induce up to 76% <italic>ee</italic> in favorable cases, replacing the phenyl moieties of <bold>8a</bold> by <italic>t</italic>-Bu is an attractive option for future studies on bisamidine-mediated organocatalytic transformations.</p>" ]

[ "<p><italic>C</italic>₂-symmetric bisamidines <bold>8</bold> have been tested as chiral Brønsted bases in the Diels-Alder reaction of anthrones and <italic>N</italic>-substituted maleimides. High yields of cycloadducts and significant asymmetric inductions up to 76% <italic>ee</italic> are accessible. The proposed mechanism involves proton transfer between anthrone and bisamidine, association of the resulting ions and finally a cycloaddition step stereoselectively controlled by the chiral ion pair.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>Financial support provided by the DFG (Priority Program “Organocatalysis” SPP1179) and by the FAZIT STIFTUNG is gratefully acknowledged.</p>" ]

[ "<fig id=\"C1\" position=\"float\"><label>Scheme 1</label><caption><p>Diels-Alder reaction of anthrones <bold>1</bold> and maleimides <bold>2</bold> catalyzed by chiral Brønsted bases <bold>4</bold>–<bold>8</bold>.</p></caption></fig>", "<fig id=\"C2\" position=\"float\"><label>Scheme 2</label><caption><p>Protonation states and tautomerism of <italic>C</italic>₂-symmetric bisamidine <bold>8a</bold> [##REF##17585816##6##].</p></caption></fig>", "<fig id=\"C3\" position=\"float\"><label>Scheme 3</label><caption><p>Synthesis of <italic>C</italic>₂-symmetric bisamidines <bold>8b</bold>–<bold>c</bold> and <italic>ent</italic>-<bold>8d</bold>.</p></caption></fig>", "<fig id=\"F1\" position=\"float\"><label>Figure 1</label><caption><p>Kinetic measurements of <bold>1a</bold> with <bold>2a</bold> catalyzed by 5 mol% of <bold>8a</bold>·H⁺·TFPB^- (black line) and 1 mol% <bold>8a</bold> (free base; red line).</p></caption></fig>", "<fig id=\"F2\" position=\"float\"><label>Figure 2</label><caption><p>Molecular structure of <bold>3m</bold> (C: black; N: blue; O: red; Cl: green; hydrogen atoms are omitted for the sake of clarity).</p></caption></fig>", "<fig id=\"C4\" position=\"float\"><label>Scheme 4</label><caption><p>Proposed mechanism of the Diels-Alder reaction.</p></caption></fig>" ]

[ "<table-wrap id=\"T1\" position=\"float\"><label>Table 1</label><caption><p>First evaluation step of chiral bisamidine catalysts.</p></caption><table frame=\"hsides\" rules=\"groups\"><tr><td align=\"center\" colspan=\"4\" valign=\"middle\" rowspan=\"1\"></td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">entry^a</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">catalyst</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">yield [%]^b</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><italic>ee</italic> [%]^c</td></tr><tr><td align=\"left\" colspan=\"4\" valign=\"middle\" rowspan=\"1\"><hr/></td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>8a</bold></td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">86</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">11</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>8b</bold></td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">78</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">17</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>8c</bold></td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">85</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">24</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><italic>ent</italic>-<bold>8d</bold></td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">71</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">−17^d</td></tr></table></table-wrap>", "<table-wrap id=\"T2\" position=\"float\"><label>Table 2</label><caption><p>Influence of the solvent on the bisamidine catalyzed Diels-Alder reaction.</p></caption><table frame=\"hsides\" rules=\"groups\"><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">entry^a</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">solvent</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">yield [%]^b</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><italic>ee</italic> [%]^c</td></tr><tr><td align=\"left\" colspan=\"4\" valign=\"middle\" rowspan=\"1\"><hr/></td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">dichloromethane</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">84</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">30</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">chloroform</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">86</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">18</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">benzene</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">98</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">21</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">toluene</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">99</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">16</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">α,α,α-trifluorotoluene</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">99</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">13</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">dibutyl ether</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">89</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">11</td></tr></table></table-wrap>", "<table-wrap id=\"T3\" position=\"float\"><label>Table 3</label><caption><p>Influence of temperature on the Diels-Alder reaction.</p></caption><table frame=\"hsides\" rules=\"groups\"><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">entry^a</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">reaction temperature [°C]</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">reaction time [h]</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">yield [%]^b</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><italic>ee</italic> [%]^c</td></tr><tr><td align=\"left\" colspan=\"5\" valign=\"middle\" rowspan=\"1\"><hr/></td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">23</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">84</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">30</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">0</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">24</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">96</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">29</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">−20</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">24</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">98</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">31</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">−40</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">48</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">96</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">36</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">−70</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">96</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">71</td><td align=\"right\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">39</td></tr></table></table-wrap>", "<table-wrap id=\"T4\" position=\"float\"><label>Table 4</label><caption><p>Scope of the Diels-Alder-reaction.</p></caption><table frame=\"hsides\" rules=\"groups\"><tr><td align=\"center\" colspan=\"7\" valign=\"middle\" rowspan=\"1\"></td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">entry^a</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1</bold> [R¹, R²]</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">R³</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">condition^b</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">yield [%]^c</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><italic>ee</italic> [%]^d</td></tr><tr><td align=\"left\" colspan=\"7\" valign=\"middle\" rowspan=\"1\"><hr/></td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1a</bold> [H, H,]</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Ph (<bold>2a</bold>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">A</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">96</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">36</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1b</bold> [H, Cl]</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>2a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">A</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3b</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">95</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">41</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">iPr (<bold>2b</bold>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">B</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3c</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">74</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">26</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><italic>t</italic>-Bu (<bold>2c</bold>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">B</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3d</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">45</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">30</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Cy (<bold>2d</bold>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">B</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3e</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">83</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">42</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1c</bold> [Cl, H]</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>2d</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">B</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3f</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">90</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">19</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">7</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">Bn (<bold>2e</bold>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">A</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3g</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">95</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">20</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">8</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">CHPh₂ (<bold>2f</bold>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">A</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3h</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">85</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">26</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">4-Br-(C₆H₄)- (<bold>2g</bold>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">B</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3i</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">70</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">13</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">10</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">4-MeO-(C₆H₄)- (<bold>2h</bold>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">A</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3j</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">82</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">32</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">11</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">2,6-iPr₂-(C₆H₃)- (<bold>2i</bold>)</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">B</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3k</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">13</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>76</bold></td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">12</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1a</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>2i</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3k</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">65</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">51</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">13</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1c</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>2i</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3l</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">77</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">34</td></tr><tr><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">14</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>1b</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>2i</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">C</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\"><bold>3m</bold></td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">76</td><td align=\"left\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">54 (<bold>96</bold>)^e</td></tr></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"SD1\"><label>File 1</label><caption><p>Supporting information features characterisation data and copies of ¹H- and ¹³C-NMR spectra of anthrones <bold>1</bold>, maleimides <bold>2</bold>, Diels-Alder adducts <bold>3</bold>, bisamidine hydrochlorides <bold>8b</bold>–<bold>d</bold>·H⁺·Cl^-, neutral bisamidines <bold>8b</bold>–<bold>d</bold> and diamines <bold>9b</bold>–<bold>c</bold>, plus copies of chromatograms obtained with chiral columns.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"SD2\"><label>File 2</label><caption><p>X-Ray data of compound <bold>3k</bold></p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"SD3\"><label>File 3</label><caption><p>X-Ray data of compound <bold>3m</bold></p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"TFN1\"><p>^aAll reactions were carried out using 0.1 mmol maleimide <bold>2a</bold>, 1.1 equiv anthrone (<bold>1a</bold>) and 0.25 equiv of catalyst in 1 mL abs. dichloromethane at room temperature for 30 minutes. ^bIsolated yield after column chromatography. ^cThe enantiomeric excess was determined by HPLC using a Chiralpak IA column. ^dA negative <italic>ee</italic> stands for an excess of <italic>ent</italic>-<bold>3a</bold>.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"TFN2\"><p>^aAll reactions were carried out using 0.1 mmol maleimide <bold>2a</bold>, 1.1 equiv anthrone (<bold>1a</bold>) and 0.1 equiv of <bold>8c</bold> in 1 mL abs. solvent at room temperature for 60 minutes. ^bIsolated yield after column chromatography. ^cThe enantiomeric excess was determined by HPLC using a Chiralpak IA column.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"TFN3\"><p>^aAll reactions were carried out using 0.1 mmol maleimide <bold>2a</bold>, 1.1 equiv anthrone (<bold>1a</bold>) and 0.1 (entry 1–4) or 0.25 equiv (entry 5) of <bold>8c</bold> in 1 mL abs. dichloromethane. ^bIsolated yield after column chromatography. ^cEnantiomeric excess was determined by HPLC using Chiralpak IA column.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"TFN4\"><p>^aAll reactions were carried out using 0.1 mmol maleimide, 1.1 equiv anthrone in 1 mL abs. CH₂Cl₂. ^bA = 10 mol% <bold>8c</bold>, −40 °C, 48 h; B = 25 mol% <bold>8a</bold>, −70 °C, 96 h; C = 25 mol% <bold>8a</bold>, r.t., 3 h. ^cIsolated yield after column chromatography. ^dThe enantiomeric excess was determined by HPLC using a Chiralpak IA column. ^eRecrystallized from 2-propanol/<italic>n</italic>-hexane.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-28-g004\"/>", "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-28-g005\"/>", "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-28-g006\"/>", "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-28-g002\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-28-i001.jpg\"/>", "<inline-graphic xlink:href=\"Beilstein_J_Org_Chem-04-28-i002.jpg\"/>", "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-28-g003\"/>", "<graphic xlink:href=\"Beilstein_J_Org_Chem-04-28-g007\"/>" ]

[ "<media mime-subtype=\"msword\" mimetype=\"application\" xlink:href=\"Beilstein_J_Org_Chem-04-28-s001.doc\" xlink:type=\"simple\" id=\"d39e1561\" position=\"anchor\"/>", "<media mime-subtype=\"x-cif\" mimetype=\"chemical\" xlink:href=\"Beilstein_J_Org_Chem-04-28-s002.cif\" xlink:type=\"simple\" id=\"d39e1570\" position=\"anchor\"/>", "<media mime-subtype=\"x-cif\" mimetype=\"chemical\" xlink:href=\"Beilstein_J_Org_Chem-04-28-s003.cif\" xlink:type=\"simple\" id=\"d39e1579\" position=\"anchor\"/>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>The Mediterranean diet, representing the dietary pattern usually consumed among the populations bordering the Mediterranean sea, has been widely reported to be a model of healthy eating for its contribution to a favourable health status and a better quality of life.##UREF##0##1##\n##REF##16532897##2## Since the first data from the seven countries study,##UREF##1##3## several studies in different populations have established a beneficial role for the main components of the Mediterranean diet on the occurrence of cardiovascular diseases and chronic degenerative diseases.##REF##16532897##2##\n##UREF##2##4## However, research interest in this field over the past years has been focused on estimating adherence to the whole Mediterranean diet rather than analysing the individual components of the dietary pattern in relation to the health status of the population.##REF##11790957##5## This because the analyses of single nutrients ignore important interactions between components of a diet and, more importantly, because people do not eat isolated nutrients. Hence, dietary scores estimating adherence to a Mediterranean diet, devised a priori on the basis of the characteristic components of the traditional diet of the Mediterranean area, have been found to be associated with a reduction of overall mortality and mortality from cardiovascular diseases and cancer.##REF##16512961##6## The aim of this study was to do a systematic review with meta-analysis of all the available prospective cohort studies that have assessed the association between adherence to a Mediterranean diet and adverse outcomes, in order to establish the role of adherence to a Mediterranean diet in primary prevention.</p>" ]

[ "<title>Methods</title>", "<title>Data sources</title>", "<p>We focused on prospective studies investigating the association between adherence to a Mediterranean diet and health outcomes. We searched PubMed, Embase, Web of Science, and the Cochrane Central Register of Controlled Trials databases up to 30 June 2008, using a search strategy that included both truncated free text and exploded MeSH terms. MeSH headings included “Mediterranean”, “diet”, “dietary pattern”, “disease”, “health”, “cardiovascular disease”, “cerebrovascular disease”, “coronary heart disease”, “degenerative diseases”, “cancer”, “neoplasm”, “prospective”, “follow-up”, or “cohort”, and their variants. The search strategy had no language restrictions. We also consulted references from the extracted articles and reviews to complete the data bank. When multiple articles for a single study were present, we used the latest publication and supplemented it, if necessary, with data from the most complete or updated publication. We assessed the relevance of studies by using a hierarchical approach based on title, abstract, and the full manuscript.</p>", "<title>Study selection</title>", "<p>We identified studies that prospectively evaluated the association of an a priori score used for assessing adherence to a Mediterranean diet and adverse clinical outcomes. We excluded the studies if they had a cross sectional or case-control design, if they analysed adherence to a non-specific dietary pattern or to a recommended dietary guideline and not to a Mediterranean diet, if they evaluated a cohort of patients with a previous clinical event (that is, secondary prevention), if they did not adjust for potential confounders, and if they did not report an adequate statistical analysis.</p>", "<p>Figure 1 shows the process of study selection. Our initial search yielded 62 reports, of which we excluded 20 on the basis of the title or abstract. Of the remaining 42 articles, we excluded 26 for the following reasons: a non-specific dietary pattern, instead of a Mediterranean diet, was evaluated (n=3); cross sectional or case-control design was used (n=18); and the study population was in secondary prevention (n=5). We excluded four additional articles because they represented duplicate studies, so we included only the latest or the more complete paper in the final analysis. Finally, 12 articles fulfilled our inclusion criteria.^w1-w12</p>", "<title>Data extraction</title>", "<p>We extracted the following baseline characteristics from the original reports by using a standardised data extraction form and included them in the meta-analysis: lead author, year of publication, cohort name, country of origin of the cohort, sample size of the cohort and number of outcomes, duration of follow-up, age at entry, sex, outcome, components of the score for adherence to a Mediterranean diet, and variables that entered into the multivariable model as potential confounders (table 1). Two investigators (FS and FC) collected the data, and disagreements were solved by consensus and by the opinion of a third author (AC), if necessary. Outcomes of interest were overall mortality, mortality from cardiovascular diseases, incidence of or mortality from cancer, as well as occurrence of Parkinson’s disease and Alzheimer’s disease.</p>", "<p>We assessed the quality of the studies according to the number of participants, the duration of follow-up, and adjustment for potential confounders. We considered studies with a high number of participants; long duration of follow-up; and adjustment for confounders including demographic, anthropometric, and traditional risk factors to be of high quality.</p>", "<title>Definition of adherence to Mediterranean diet</title>", "<p>Adherence to a Mediterranean diet was defined through scores that estimated the conformity of the dietary pattern of the studied population with the traditional Mediterranean dietary pattern. Values of zero or one were assigned to each dietary component by using as cut offs the overall sex specific medians among the study participants. Specifically, people whose consumption of components considered to be part of a Mediterranean diet (vegetables, fruits, legumes, cereals, fish, and a moderate intake of red wine during meals) was above the median consumption of the population were assigned a value of one, whereas a value of zero was given to those with consumptions below the median. By contrast, people whose consumption of components presumed not to form part of a Mediterranean diet (red and processed meats, dairy products) was above the median consumption of the population had a value of zero assigned, and the others had a value of one. However, some differences among the studies existed, especially in relation to the food category of vegetables (grouped with potatoes in one study^w5), meat and meat products (grouped with poultry in some studies^{w4 w6}), and nuts and seeds (grouped with fruits in some studies,^{w4 w6 w7 w12} with legumes in one study,^w5 and considered a group by themselves in some others^{w8 w10 w11}), as well as milk and dairy products (not present in some studies^{w8 w10 w11}) and fish (present only in more recent studies^w4-w12). Thus, the total adherence scores (estimated as the sum of the above indicated scores of zero and one) varied from a minimum of 0 points indicating low adherence to a maximum of 7-9 points reflecting high adherence to a Mediterranean diet.</p>", "<title>Statistical analysis</title>", "<p>We used RevMan, version 4.2 for Windows by the Cochrane Collaboration to analyse data. We used the results of the original studies from multivariable models with the most complete adjustment for potential confounders; table 1 shows the confounding variables included in this analysis. We used a random effects model that accounts for interstudy variation and provides a more conservative effect than a fixed model. We calculated random summary relative risks with 95% confidence intervals by using an inverse variance method. </p>", "<p>We grouped the studies according to the different clinical outcomes (mortality from all causes, mortality from cardiovascular diseases, incidence of or mortality from cancer, and incidence of Parkinson’s disease and Alzheimer’s disease). We assessed the potential sources of heterogeneity by using the standard χ² test. In addition, we used the I² statistic to investigate heterogeneity by examining the extent of inconsistency across the study results. To examine the potential source of heterogeneity across studies evaluating overall mortality, we did sensitivity analyses according to some characteristics of the studies—sex (male, female), country of origin (European countries, United States, other countries), follow-up time (below or above the median follow-up time of the studies: 8 years), and the quality of the studies (low, high). To assess the presence of publication bias, we computed the “failsafe N” for each of the main outcomes; this value is an estimate of the number of studies with null results that would need to be added to the meta-analysis to reduce the overall observed significant result to non-significance.</p>" ]

[ "<title>Results</title>", "<title>Characteristics of study cohorts</title>", "<p>Sample sizes varied between 161 and 214 284, with a follow-up time ranging from 3.7 to 18 years. Outcomes of interest were overall mortality, cardiovascular mortality, incidence of or mortality from neoplastic disease, and incidence of Parkinson’s disease and Alzheimer’s disease. Only six out of 12 studies were done in Mediterranean populations.^{w1 w3-w6 w12} The remaining cohorts comprised US populations,^w8-w11 northern Europeans,^w5-w7 and a cohort of Europeans living in Australia.^w3 The total number of subjects in the included studies was 1 574 299.</p>", "<title>Main outcomes</title>", "<p>According to the different clinical outcomes, overall mortality was evaluated in eight cohorts (nine studies) for a total of 514 816 subjects and 33 576 deaths, cardiovascular mortality in three cohorts (four studies) including a total of 404 491 subjects and 3876 fatal events, cancer incidence/mortality in five cohorts (six studies) comprising 521 366 subjects and 10 929 events, and incidence of Parkinson’s disease and Alzheimer’s disease in two cohorts (three studies) for a total of 133 626 subjects and 783 cases.</p>", "<p>Figure 2 shows the cumulative analysis for studies that analysed overall mortality as the primary clinical outcome. Using a random effects model, we found that a two point increase in score for adherence to a Mediterranean diet was significantly associated with a reduced risk of mortality from any cause (relative risk 0.91, 95% confidence interval 0.89 to 0.94; P&lt;0.0001). Significant heterogeneity was present among the studies (I₂=48.8%; P=0.05). However, after exclusion of the paper by Trichopoulou et al 2003 that analysed the same cohort as Trichopoulou et al 2005,^{w4 w6} the significant association with overall mortality remained (relative risk 0.92, 0.91 to 0.94; P&lt;0.0001), showing no significant heterogeneity (I₂=18.3%; P=0.3).</p>", "<p>Similarly figure 3 shows that a greater adherence to a Mediterranean diet significantly reduced the risk of mortality from cardiovascular diseases (relative risk 0.91, 0.87 to 0.95; P&lt;0.0001) with non-significant heterogeneity (I₂=32.6%; P=0.2). Furthermore, greater adherence to a Mediterranean diet significantly reduced the occurrence of and mortality from neoplasm (relative risk 0.94, 0.92 to 0.96; P&lt;0.0001) (I₂=0%; P=0.5) (fig 4). Finally, the overall analysis showed a significant reduction in incidence of Parkinson’s disease and Alzheimer’s disease associated with a higher score of adherence to a Mediterranean diet (relative risk 0.87, 0.80 to 0.96; P=0.004), with no heterogeneity among the studies (I₂=0%; P=0.5) (fig 5).</p>", "<title>Sensitivity analyses</title>", "<p>Studies included in this meta-analysis varied in some characteristics. Because such heterogeneity of studies is likely to produce heterogeneity of effect sizes across studies, we did some sensitivity analyses. Table 2 shows the different subgroup analyses on studies evaluating overall mortality as clinical outcome. These analyses showed no significant influence of any variable (country of origin of the study, sex, follow-up time, quality of the studies) on the overall results of the meta-analysis.</p>", "<title>Publication bias</title>", "<p>To assess the presence of publication bias, we computed the failsafe N for each of the main outcomes. Each failsafe N (580 for studies evaluating overall mortality as main outcome, 68 for studies with cardiovascular mortality as main outcome, 72 for incidence of or mortality from cancer, and 43 for incidence of degenerative diseases) far exceeded Rosenthal’s recommendation (failsafe N&gt;5k+10, where k is the number of studies included in the analysis) for a robust effect of the overall analysis.</p>" ]

[ "<title>Discussion</title>", "<p>This meta-analysis shows, in an overall analysis comprising more than 1.5 million healthy subjects and 40 000 fatal and non-fatal events, that greater adherence to a Mediterranean diet is significantly associated with a reduced risk of overall mortality, cardiovascular mortality, cancer incidence and mortality, and incidence of Parkinson’s disease and Alzheimer’s disease. The cumulative analysis of 12 cohort studies shows that a two point increase in the score for adherence to a Mediterranean diet determines a 9% reduction in overall mortality, a 9% reduction in mortality from cardiovascular diseases, a 6% reduction in incidence of or mortality from neoplasm, and a 13% reduction in incidence of Parkinson’s disease and Alzheimer’s disease. To the best of our knowledge, this is the first report that has systematically assessed, through meta-analysis, the possible association between adherence to a Mediterranean diet, mortality, and the occurrence of chronic diseases in the general population.</p>", "<title>Diet and disease</title>", "<p>The effect of diet on human health has been amply reported in many epidemiological, population based, and randomised clinical trials, providing evidence that a dietary pattern rich in some beneficial food groups such as fruit, vegetables, whole grains, and fish can reduce the incidence of cardiovascular and neoplastic diseases.##REF##16512956##7## However, until now, the vast majority of studies followed the approach of assessing single nutrients or food groups in relation to the occurrence of disease.##UREF##2##4##\n##REF##3990713##8##\n##REF##18076281##9## This approach seems to have several conceptual and methodological limitations, because food components of diet present synergistic and antagonist interactions and because people eat a complex of nutrients.##REF##11790957##5## Therefore, over the past few years, researchers have shifted their attention from the evaluation of single nutrients to the analysis of dietary pattern as a whole.##REF##16512961##6##\n^w1-w10 As a result, an increasing number of studies have been done by summing foods considered to be important for health to provide an overall measure of dietary quality—that is, a quality diet score.##REF##16512961##6##</p>", "<p>In this context, a prominent position has been occupied by studies evaluating adherence to a Mediterranean diet, because of its well known and evidence based beneficial effects on human health. Indeed, since the early 1970s many investigators have reported the beneficial role of the Mediterranean diet, as originally reported by Keys in the pioneering seven countries study.##UREF##1##3## A diet rich in fruits, vegetables, legumes, and cereals, with olive oil as the only source of fat, moderate consumption of red wine especially during meals, and low consumption of red meat has been shown to be beneficial for all cause and cardiovascular mortality, lipid metabolism, blood pressure, and several different disease states such as endothelial dysfunction and overweight.##REF##16512956##7##</p>", "<title>Practical implications</title>", "<p>In this study we aimed to systematically analyse all the prospective cohort studies that evaluated the effect of a computational score estimating adherence to a Mediterranean diet on health status. From the overall analysis of 11 cohort studies, of which eight assessed the risk of overall mortality, four assessed cardiovascular mortality, six assessed incidence of or mortality from neoplasm, and three assessed incidence of Parkinson’s disease and Alzheimer’s disease, we report a significant reduction in risk of all the main clinical outcomes with an increasing score for adherence to a Mediterranean diet. This observation seems to show that a score based on a theoretically defined Mediterranean diet is an effective preventive tool for measuring the risk of mortality and morbidity in the general population.</p>", "<p>A Mediterranean diet has been shown to have a beneficial effect on the occurrence of diseases in industrialised and non-industrialised countries. All the major scientific associations, in fact, strongly encourage people to consume a Mediterranean-like dietary pattern to reduce their risk of disease.##UREF##3##10##\n##REF##16785338##11##\n##UREF##4##12## Unfortunately, despite this worldwide promotion of the Mediterranean diet, a progressive shift to a non-Mediterranean dietary pattern, even in countries bordering the Mediterranean sea, has progressively developed.##REF##15741987##13## It thus seems urgent to identify an effective preventive strategy to decrease the risk burden related to dietary habits in the general population; the use of such a tool could be important in increasing the implementation of dietary guidelines.</p>", "<title>Limitations</title>", "<p>Some limitations of this study can be identified. The Mediterranean diet is not a homogeneous pattern of eating, and heterogeneity on the score items exists. How to group some food categories such as legumes, nuts, and milk and dairy products; the real importance of different types of meat; and the establishment of the moderate amount of alcohol intake are still matters of dispute among researchers and can differ among the selected studies. None the less, the key characteristics of a Mediterranean diet were present in all the studies, and the overall analysis seemed not to be significantly influenced by these differences. In addition, the use of a score for estimating a dietary pattern is limited by subjectivity, conditioned by the available data and the main objectives of the study, and so possibly determining a great variability in the interpretation of the results.</p>", "<p>Finally, a further limitation exists in the different adjustment for potential confounders seen among the included studies. This difference could have determined a residual confounding within the studies, especially for the non-Mediterranean cohorts. However, the sensitivity analysis according to the quality of the studies, which also included the presence or not of adjustment factors, showed no significant influence of residual confounding on the overall findings of our meta-analysis.</p>", "<title>Conclusions</title>", "<p>This meta-analysis shows that adherence to a Mediterranean diet can significantly decrease the risk of overall mortality, mortality from cardiovascular diseases, incidence of or mortality from cancer, and incidence of Parkinson’s disease and Alzheimer’s disease. These results seem to be clinically relevant in terms of public health, particularly for reducing the risk of premature death in the general population, and are strictly concordant with current guidelines and recommendations from all the major scientific associations that strongly encourage a Mediterranean-like dietary pattern for primary and secondary prevention of major chronic diseases.</p>" ]

[ "<title>Conclusions</title>", "<p>This meta-analysis shows that adherence to a Mediterranean diet can significantly decrease the risk of overall mortality, mortality from cardiovascular diseases, incidence of or mortality from cancer, and incidence of Parkinson’s disease and Alzheimer’s disease. These results seem to be clinically relevant in terms of public health, particularly for reducing the risk of premature death in the general population, and are strictly concordant with current guidelines and recommendations from all the major scientific associations that strongly encourage a Mediterranean-like dietary pattern for primary and secondary prevention of major chronic diseases.</p>" ]

[ "<p><bold>Objective</bold> To systematically review all the prospective cohort studies that have analysed the relation between adherence to a Mediterranean diet, mortality, and incidence of chronic diseases in a primary prevention setting. </p>", "<p><bold>Design</bold> Meta-analysis of prospective cohort studies.</p>", "<p><bold>Data sources</bold> English and non-English publications in PubMed, Embase, Web of Science, and the Cochrane Central Register of Controlled Trials from 1966 to 30 June 2008. </p>", "<p><bold>Studies reviewed</bold> Studies that analysed prospectively the association between adherence to a Mediterranean diet, mortality, and incidence of diseases; 12 studies, with a total of 1 574 299 subjects followed for a time ranging from three to 18 years were included.</p>", "<p><bold>Results</bold> The cumulative analysis among eight cohorts (514 816 subjects and 33 576 deaths) evaluating overall mortality in relation to adherence to a Mediterranean diet showed that a two point increase in the adherence score was significantly associated with a reduced risk of mortality (pooled relative risk 0.91, 95% confidence interval 0.89 to 0.94). Likewise, the analyses showed a beneficial role for greater adherence to a Mediterranean diet on cardiovascular mortality (pooled relative risk 0.91, 0.87 to 0.95), incidence of or mortality from cancer (0.94, 0.92 to 0.96), and incidence of Parkinson’s disease and Alzheimer’s disease (0.87, 0.80 to 0.96).</p>", "<p><bold>Conclusions</bold> Greater adherence to a Mediterranean diet is associated with a significant improvement in health status, as seen by a significant reduction in overall mortality (9%), mortality from cardiovascular diseases (9%), incidence of or mortality from cancer (6%), and incidence of Parkinson’s disease and Alzheimer’s disease (13%). These results seem to be clinically relevant for public health, in particular for encouraging a Mediterranean-like dietary pattern for primary prevention of major chronic diseases.</p>" ]

[]

[ "<p><bold>Cite this as:</bold>\n<italic>BMJ</italic> 2008;337:a1344</p>" ]

[ "<fig id=\"fig1\" position=\"float\"><caption><p><bold>Fig 1</bold> Process of study selection</p></caption></fig>", "<fig id=\"fig2\" position=\"float\"><caption><p><bold>Fig 2</bold> Risk of all cause mortality associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect size</p></caption></fig>", "<fig id=\"fig3\" position=\"float\"><caption><p><bold>Fig 3</bold> Risk of mortality from cardiovascular diseases associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect size</p></caption></fig>", "<fig id=\"fig4\" position=\"float\"><caption><p><bold>Fig 4</bold> Risk of occurrence of or mortality from cancer associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect size</p></caption></fig>", "<fig id=\"fig5\" position=\"float\"><caption><p><bold>Fig 5</bold> Risk of Parkinson’s disease and Alzheimer’s disease associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect size</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\" position=\"float\"><label>Table 1</label><caption><p> Study characteristics</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"bottom\">Author, year</th><th colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"bottom\">Country</th><th colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"bottom\">No of outcomes/No in cohort</th><th colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"bottom\">Outcome</th><th colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"bottom\">Follow-up (years)</th><th colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"bottom\">Age at entry (years)</th><th colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"bottom\">Sex</th><th colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"bottom\">Components of score</th><th colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"bottom\">Adjustment</th></tr></thead><tbody><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Trichopoulou et al, 1995^w1</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Greece</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">53/182</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Overall mortality</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">4-5</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">&gt;70 (mean 75.4)</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">M/F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High cereals; 3. High fruits; 4. High vegetables; 5. High MUFA:SFA ratio; 6. Moderate alcohol; 7. Low meat and meat products; 8. Low milk and dairy products</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, sex, smoking habit, total diet score</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Kouris-Blazos et al, 1999^w2</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Australia</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">36/330</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Overall mortality</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">4-6</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">≥70</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">M/F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High cereals<break/>3. High fruits; 4. High vegetables; 5. High MUFA:SFA ratio; 6. Moderate alcohol; 7. Low meat and meat products; 8. Low milk and dairy products</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, sex, smoking habit, ethnic origin</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Lasheras et al, 2000^w3</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Spain</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">96/161</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Overall mortality</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">9.5</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">65-80</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">M/F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High cereals; 3. High fruits; 4. High vegetables; 5. High MUFA:SFA ratio; 6. Moderate alcohol; 7. Low meat and meat products; 8. Low milk and dairy products</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, sex, total diet score, albumin, dieting in response to chronic conditions, BMI, self assessment of health, physical activity</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Trichopoulou et al, 2003^w4 (EPIC)</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Greece</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">275/22 043; 54/22 04</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Overall mortality; CHD mortality</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">3.7</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">20-86</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">M/F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High cereals; 3. High fruits/nuts; 4. High vegetables; 5. High fish; 6. High MUFA:SFA ratio; 7. Moderate alcohol; 8. Low meat and poultry; 9. Low dairy products</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, sex, smoking habit, years of education, BMI, waist to hip ratio, energy expenditure score, energy intake, consumption of potatoes and eggs</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Knoops et al, 2004^w5 (HALE project: SENECA and FINE)</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Belgium, Denmark, Finland, France, Greece, Hungary, Italy, Netherlands, Portugal, Spain, Switzerland</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">935/2339; 122/2152; 371/2152; 233/2152</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Overall mortality; CHD mortality; CVD mortality; cancer mortality</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">10</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">70-90</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">M/F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes/nuts/seeds; 2. High cereals; 3. High fruits; 4. High vegetables/potatoes; 5. High fish; 6. High MUFA:SFA ratio; 7. Low meat and meat products; 8. Low dairy products</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, sex, smoking habit, physical activity, BMI, dietary habits, alcohol, years of education, study population</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Trichopoulou et al, 2005^w6 (EPIC-elderly)</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Denmark, France, Germany, Greece, Italy, Netherlands, Norway, Spain, Sweden, UK</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">3810/67 228 (after exclusion of Greek patients)</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Overall mortality</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">7.4</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">&gt;60</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">M/F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High cereals; 3. High fruits/nuts; 4. High vegetables; 5. High fish; 6. High MUFA+PUFA:SFA ratio; 7. Moderate alcohol; 8. Low meat and poultry; 9. Low dairy products</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age; sex; smoking habit; diabetes; education; BMI; waist to hip ratio; physical activity; energy; consumption of potatoes, eggs, sugar, and confectionery</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Lagiou et al, 2006^w7</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Sweden</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">572/42 237; 280/42 237</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Overall mortality; cancer mortality</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">12</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">30-49</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High cereals; 3. High fruits/nuts; 4. High vegetables; 5. High fish; 6. High MUFA:SFA ratio; 7. Moderate alcohol; 8. Low meat and meat products; 9. Low dairy products</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age; height; BMI; smoking habit; physical activity; education; energy; consumption of potatoes, eggs, PUFA, sweets, and non-alcoholic beverages</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Fung et al, 2006^w8 (nurses’ health study)</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">USA</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">3580/71 058</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Breast cancer</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">18</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">30-55</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High wholegrain products; 3. High fruits; 4. High nuts; 5. High vegetables; 6. High fish; 7. High MUFA:SFA ratio; 8. Moderate alcohol; 9. Low red and processed meats</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, smoking habit, BMI, multivitamins, energy, physical activity, family history of breast cancer and of benign breast disease, menopause, HRT, weight change since age 18 years</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Scarmeas et al, 2006^w9</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">USA</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">85/2258</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Alzheimer’s disease</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">4</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">Mean 77.2</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">M/F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High cereals; 3. High fruits; 4. High vegetables; 5. High fish; 6. High MUFA:SFA ratio; 7. Moderate alcohol; 8. Low meat; 9. Low dairy products</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, sex, cohort, ethnicity, smoking habit, BMI, education, energy, ApoE genotype, comorbidity index</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Gao et al, 2007^w10 (health professionals and nurses’ health studies)—men</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">USA</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">508/49 692</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Parkinson’s disease</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">16</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">40-75</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">M</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High wholegrain products; 3. High fruits; 4. High nuts; 5. High vegetables; 6. High fish; 7. High MUFA:SFA ratio; 8. Moderate alcohol; 9. Low red and processed meat</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, smoking habit, BMI, use of NSAIDs, energy, caffeine intake</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Gao et al, 2007^w10 (health professionals and nurses’ health studies)—women</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">USA</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">190/81 676</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Parkinson’s disease</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">16</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">40-75</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High wholegrain products; 3. High fruits; 4. High nuts; 5. High vegetables; 6. High fish; 7. High MUFA:SFA ratio; 8. Moderate alcohol; 9. Low red and processed meat</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, smoking habit, BMI, use of NSAIDs, energy, caffeine intake</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Mitrou et al, 2007^w11 (National Institutes of Health diet and health study)—men</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">USA</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">18 126/214 284; 2425/214 284; 3717/214 284</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Overall mortality; CVD deaths; cancer deaths</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">10</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">50-71</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">M</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High wholegrain products; 3. High fruits; 4. High nuts; 5. High vegetables; 6. High fish; 7. High MUFA:SFA ratio; 8. Moderate alcohol; 9. Low red and processed meat</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, race, smoking habit, energy, BMI, education, physical activity</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Mitrou et al, 2007^w11 (National Institutes of Health diet and health study)—women</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">USA</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">9673/166 012; 1026/166 012; 2268/166 012</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Overall mortality; CVD deaths; cancer deaths</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">10</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">50-71</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High wholegrain products; 3. High fruits; 4. High nuts; 5. High vegetables; 6. High fish; 7. High MUFA:SFA ratio; 8. Moderate alcohol; 9. Low red and processed meat</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age, race, smoking habit, energy, BMI, education, physical activity, HRT</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Benetou et al, 2008^w12 (EPIC)</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Greece</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">851/25 623</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Incident cancers (excluding non-melanoma skin cancers)</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">7.9</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">20-86</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">M/F</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">1. High legumes; 2. High cereals; 3. High fruits/nuts; 4. High vegetables; 5. High fish; 6. High MUFA:SFA ratio; 7. Moderate alcohol; 8. Low meat and poultry; 9. Low dairy products</td><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age; sex; smoking habit; years of education; BMI; height; physical activity; total energy; consumption of potatoes, eggs, confectionery, and non-alcoholic beverages</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\" position=\"float\"><label>Table 2</label><caption><p> Sensitivity analysis</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"bottom\">Variable (No of studies)</th><th colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"bottom\">Relative risk (95% CI)</th></tr></thead><tbody><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Sex:</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\"/></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Male (7)</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">0.89 (0.85 to 0.94)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Female (8)</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">0.90 (0.86 to 0.94)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Country of origin:</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\"/></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Europe (6)</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">0.87 (0.81 to 0.94)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> United States/Australia (3)</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">0.93 (0.91 to 0.94)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Follow-up time:</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\"/></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> &lt;8 years (4)</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">0.82 (0.69 to 0.97)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> &gt;8 years (5)</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">0.92 (0.91 to 0.94)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Study quality:</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\"/></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Low (3)</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">0.89 (0.83 to 0.99)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> High (6)</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">0.92 (0.90 to 0.94)</td></tr></tbody></table></table-wrap>" ]

[]

[ "<boxed-text position=\"float\" content-type=\"style3\"><sec><title>What is already known on this topic</title><list list-type=\"simple\"><list-item><p>The Mediterranean diet is a well known model of diet for primary and secondary prevention of major chronic diseases</p></list-item><list-item><p>An adherence score can be used to assess the adherence of a specific population to the rules of a traditional Mediterranean diet</p></list-item></list></sec><sec><title>What this study adds</title><list list-type=\"simple\"><list-item><p>Greater adherence to a Mediterranean diet confers a significant protection for overall mortality, as well as cardiovascular disease mortality and incidence of cancer and degenerative diseases</p></list-item><list-item><p>The adherence score based on a theoretically defined Mediterranean diet could be an effective preventive tool for reducing the risk of mortality and morbidity in the general population</p></list-item></list></sec></boxed-text>" ]

[]

[]

[]

[]

[ "<table-wrap-foot><p>ApoE=apolipoprotein E; BMI=body mass index; CHD=coronary heart disease; CVD=cardiovascular disease; HRT=hormone replacement therapy; MUFA=mono-unsaturated fatty acids; NSAIDs=non-steroidal anti-inflammatory drugs; PUFA=polyunsaturated fatty acids; SFA=saturated fatty acids.</p></table-wrap-foot>", "<fn-group><fn fn-type=\"participating-researchers\"><p>Contributors: FS and AC contributed to conception and design. All authors were involved in analysis and interpretation of the data. FS and FC drafted the manuscript, which was critically revised for important intellectual content by RA, GFG, and AC. All authors approved the final version. FS provided statistical expertise and is the guarantor.</p></fn><fn fn-type=\"financial-disclosure\"><p>Funding: None.</p></fn><fn fn-type=\"conflict\"><p>Competing interests: None declared.</p></fn><fn><p>Ethical approval: Not needed.</p></fn><fn><p>Provenance and peer review: Not commissioned; externally peer reviewed.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"soff557223.f1\"/>", "<graphic xlink:href=\"soff557223.f2\"/>", "<graphic xlink:href=\"soff557223.f3\"/>", "<graphic xlink:href=\"soff557223.f4\"/>", "<graphic xlink:href=\"soff557223.f5\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Food advertising of non-nutritious products targets children throughout the world, not only in developed and wealthy countries.##REF##15829531##1##\n##REF##9612782##2##\n##REF##16775233##3## Children in Lao Popular Democratic Republic (Laos) are targets of food advertising that contributes to poor nutrition, overweight, and underweight. Markets and school playgrounds are full of attractively labelled, non-nutritious foods such as crisps, fizzy drinks, and sweets.</p>", "<p>Of particular concern is the breaching of the international code of marketing of breast milk substitutes, which has been repeatedly advocated in recent years in developing ##REF##12531842##4##\n##REF##18450233##5## and industrialised countries.##REF##16925828##6## The code is intended as a minimum requirement by all governments and aims to protect infant health by preventing inappropriate marketing of breast milk substitutes. “The code applies to the marketing of breast-milk substitutes . . . including bottle-fed complementary foods, when marketed or otherwise represented to be suitable for use as a partial or total replacement of breast milk.”##REF##7279477##7## Product labels are often the only source of information available to consumers regarding the content and uses of a product, and public interest groups have documented multiple instances of formula companies providing misleading information on infant formula labels.##UREF##0##8##</p>", "<p>In Laos a popular coffee creamer is marketed with an illustration of a mother bear holding a baby bear in the breastfeeding position (figure). The first listed ingredient is sugar and it contains 3.6% of calories as protein and 27.3% as fat. For infant formulas the recommended content is 7.2-12.0% protein and 40-54% fat.##REF##16254515##9## The coffee creamer is sold in many rural roadside shops. The company uses the same Bear Brand logo on its canned sterilised cows’ milk product and on infant formula products for infants from 6 months. A warning on the can states “This product is not to be used as breast milk substitute” in English, Thai, and Lao. There is also an illustration of a feeding bottle with a cross through it.</p>", "<p>We have encountered infants and children admitted to hospital with protein-calorie malnutrition who were fed this product exclusively. We conducted two surveys to determine whether or not the practice of feeding this coffee creamer to infants is widespread in Laos. We also explored the impact of the cartoon logo on people’s perception of the product’s appropriateness for infants.</p>" ]

[ "<title>Methods</title>", "<title>Study site</title>", "<p>Laos is one of the poorest countries in southeast Asia, 135th out of 177 countries evaluated according to the human development index.##REF##16501731##10##\n##UREF##1##11## It is a multiethnic and multilingual country with more than 45 languages spoken. The average official literacy rate in Laos (73%) does not reflect the disparity between the rate in urban areas (89%) and that in rural areas (54%), which can be assumed to be overestimated in remote areas.##UREF##2##12## The documented high prevalence of stunting (41%) and wasting (15%) among children (0-5 years) has not improved over recent decades.##REF##8820613##13##\n##UREF##3##14## Stunting rates reach up to 65% in children aged 12-23 months and in some ethnic groups.##REF##17445054##15## Up to 74% of children aged over 5 in northern rural areas are stunted.##REF##15915154##16## In Laos, 95% of the mothers breast feed their newborn children but the rate of exclusive breast feeding is below 28%.##UREF##4##17## Food taboos during the postpartum period and inappropriate early supplementary feeding in infants are common nationwide.##UREF##5##18##\n##REF##17704036##19## Inadequate breast feeding and weaning practices contribute to high rates of malnutrition and infant and child mortality.##UREF##6##20##\n##REF##16925871##21## Micronutrient deficiency, beri-beri (clinical vitamin B-1 deficiency), and a high incidence of children with bladder stones have been associated with a diet with low diversity and consumption of glutinous rice.##REF##15906634##22##\n##REF##14567833##23##</p>", "<title>Study procedure</title>", "<p>In Laos in 2006, 26 paediatricians had completed paediatric residency training and were working in eight of the 17 provinces. One author (LMS) interviewed these paediatricians to collect information on parents’ use of the Bear Brand coffee creamer as a substitute for breast milk.</p>", "<p>In 2007 we conducted a sample survey in Laos. Of 18 provinces, we chose five as representative of the geographical strata of south (Attapeu, Savannakhet provinces), central (Vientiane Capital and Vientiane provinces), and northern Laos (Luang Nam Tha province). In one district we applied a two stage random sample procedure. At the first stage we randomly selected a village and at the second stage, households. Random numbers were used to select the villages from the list of villages per district and households from the list of households in the villages. We randomly chose one adult per household to be interviewed in the Lao language from among all present household members aged over 18. We randomly selected and enrolled 84 villages and an average of 14 households with one adult from each. A semistructured questionnaire examined knowledge and use of the coffee creamer with the Bear Brand logo.</p>", "<p>Participants gave their informed consent to participate in the survey.</p>", "<title>Sample size estimation, definitions, and analysis</title>", "<p>We calculated that we needed a sample size of 983 people, based on an estimated use of coffee creamer of 20%, a 4% precision with α=0.05, and 90% power, and we added 10% more people to account for incomplete data, resulting in a sample of 1080.</p>", "<p>Data were double entered with Epidata (<ext-link ext-link-type=\"uri\" xlink:href=\"www.epidata.dk\">www.epidata.dk</ext-link>, Odense, Denmark) and Stata, version 8 (Stata Cooperation, College Station, TX). We performed analyses only in people who recognised the coffee creamer can. We used χ² and Fisher’s exact tests for categorical variables and Student’s <italic>t</italic> test and analysis of variance (<italic>F</italic> test) for normally distributed continuous data. We considered P&lt;0.05 as significant.</p>" ]

[ "<title>Results</title>", "<p>Of the 26 paediatricians interviewed, 13 reported that parents “often” feed the Bear Brand coffee creamer to infants as a substitute for breast milk. Eleven reported that parents “sometimes” feed the product to infants. Paediatricians in Vientiane reported that mothers use the product when they return to work. In the countryside poor families use the product as a breast milk substitute when the mother is ill or dies.</p>", "<p>Paediatricians have encountered infants and children admitted to hospital with protein calorie malnutrition who had been fed this product exclusively and often reported similar stories (see box).</p>", "<p>Table 1 provides the main characteristics of the population surveyed. Of 1098 respondents, 570 (51%) lived in remote rural area, 364 (33%) lived in a semirural area (&lt;30 km from an urban centre), and 164 (15%) lived in an urban area (Vientiane or Savannakhet). Of the adults interviewed, 1031/1098 (94%) recognised the can; 994/1031 (96%) believed that it contained milk; and only 21/1031 (2%) identified the contents correctly as coffee creamer. In total, 191 (19%) reported giving the coffee creamer with the Bear Brand logo to infants at a mean age of 4.7 months (95% confidence interval 4.1 to 5.3). The main reasons given were that they thought it complemented breast feeding (40%), was good for infants’ growth (19%), was a substitute for breast milk (17%), and was cheap (11%).</p>", "<p>Of 1031 people, 824 (80%) said they had not read the text warning on the can. Tables 2 and 3 describe the respondents’ understanding of the Bear Brand logo and the feeding bottle with a cross through it.</p>" ]

[ "<title>Discussion</title>", "<p>Bear Brand coffee creamer is a recognised and well distributed product in diverse parts of Laos. Nearly half of surveyed adults believe that the cartoon logo on the can means that the product is “good for infants” or “a replacement for breast milk.” Nearly a fifth of parents had given the product to their young infants. Paediatricians confirmed that parents use this product as a substitute for breast feeding.</p>", "<p>Nearly half of respondents did not notice the written warning on the label or the picture of a baby’s bottle with a cross through it. Some (12%) did not understand the meaning of the cross through the bottle. The cartoon bear and her cub seem to provide the most salient misinformation, and the warnings on the label are inadequate to ensure safe and appropriate use of this product. The data suggest that the image of the Bear Brand misleads parents, who believe the coffee creamer is a suitable food for babies.</p>", "<p>Protection of breast feeding by limiting the active promotion of the use of breast milk substitutes by formula companies is the central goal of the international code of marketing of breast milk substitutes.##REF##12531842##4##\n##REF##12531818##24##\n##REF##9552947##25## The sugar based coffee creamer, according to its label, is “not . . . a breast-milk substitute.” The logo on the label of the bear holding a cub in the breastfeeding position, however, conveys quite the opposite message, making this a somewhat different type of code violation.</p>", "<title>Strengths and limitations</title>", "<p>Our study was limited by quota samples in four diverse Lao provinces. We interviewed the families at home during the day and therefore interviewed more women. The included villages had to have road access, which excluded more remote villages and limited ethnic diversity. As our sample was more literate than the national average (90% <italic>v</italic> 70%), we might have underestimated the level of misconceptions and the nationwide use of the coffee creamer as a substitute for breast milk because secondary education was associated with a lower risk of giving the coffee creamer to infants (unpublished data).</p>", "<p>Formula companies have been shown to provide misleading information on infant formula labels.##UREF##0##8## We have shown further misconceptions associated with the use of the Bear Brand logo for a coffee creamer. Because of its ease of misinterpretation, this logo should not be permitted on products that are not infant formula.</p>", "<p>Our research was limited to Laos, so the relevance to other southeast Asian countries is not known. Future research is required to determine the distribution of this logo in developing countries throughout the world.</p>", "<title>Conclusion</title>", "<p>The Bear Brand logo’s non-verbal message implies that the product contained is intended for infants. The powerful visual message is not mitigated by the addition of warning text or by the confusing symbol of the feeding bottle with a cross through it. The sale of coffee creamer with this logo places the health of infants and children at risk in a developing nation that already has extreme levels of malnutrition. </p>" ]

[ "<title>Conclusion</title>", "<p>The Bear Brand logo’s non-verbal message implies that the product contained is intended for infants. The powerful visual message is not mitigated by the addition of warning text or by the confusing symbol of the feeding bottle with a cross through it. The sale of coffee creamer with this logo places the health of infants and children at risk in a developing nation that already has extreme levels of malnutrition. </p>" ]

[ "<p><bold>Objective</bold> To investigate the use of Bear Brand coffee creamer as a food for infants and the impact on consumers of the logo of a cartoon baby bear held by its mother in the breastfeeding position.</p>", "<p><bold>Design</bold> Interviews with paediatricians throughout the country and a national survey of potential consumers regarding their perceptions and use of the Bear Brand coffee creamer.</p>", "<p><bold>Setting</bold> 84 randomised villages in south, central, and northern Laos.</p>", "<p><bold>Participants</bold> 26 Lao paediatricians and 1098 adults in households in a cluster sampling.</p>", "<p><bold>Results</bold> Of the 26 paediatricians, 24 said that parents “often” or “sometimes” fed this product to infants as a substitute for breast milk. In the capital city, paediatricians said that mothers used the product when they returned to work. In the countryside, they reported that poor families used it when the mother was ill or died. Of 1098 adults surveyed, 96% believed that the can contains milk; 46% believed the Bear Brand logo indicates that the product is formulated for feeding to infants or to replace breast milk; 80% had not read the written warning on the can; and over 18% reported giving the product to their infant at a mean age of 4.7 months (95% confidence interval 4.1 to 5.3).</p>", "<p><bold>Conclusion</bold> The Bear Brand coffee creamer is used as a breast milk substitute in Laos. The cartoon logo influences people’s perception of the product that belies the written warning “This product is not to be used as a breast milk substitute.” Use of this logo on coffee creamer is misleading to the local population and places the health of infants at risk.</p>" ]

[]

[ "<p><bold>Cite this as:</bold>\n<italic>BMJ</italic> 2008;337:a1379</p>" ]

[ "<fig id=\"fig1\" position=\"float\"><caption><p>Label from Bear Brand coffee creamer</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\" position=\"float\"><label>Table 1</label><caption><p> Characteristics of 1098 adult respondents to survey about use of Bear Brand coffee creamer in Laos</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"bottom\"/><th colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"bottom\">No (%) or mean (95% CI)</th></tr></thead><tbody><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Women</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">612 (63.6)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Age (years)*</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">42.0 (41.2 to 42.7)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Illiterate</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">110 (10.0)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Secondary education or above</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">559 (50.0)</td></tr><tr><td colspan=\"2\" rowspan=\"1\" align=\"left\" valign=\"top\">Ethnic group:</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Lao Loum</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">1010 (91.9)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Lao Theung</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">70 (6.4)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Lao Soum</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">18 (1.6)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Family size*</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">5.9 (5.7 to 6.0)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Households with no children</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">132 (12.0)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Households with TV or radio</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">847 (83.6)</td></tr><tr><td colspan=\"2\" rowspan=\"1\" align=\"left\" valign=\"top\">Occupation:</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Farmer</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">388 (35.1)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Shopkeeper or craftsman</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">260 (23.6)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> No job/housewife/retired</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">219 (20.0)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Civil servant</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">129 (11.7)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Worker</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">33 (3.0)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\"> Other</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">69 (6.3)</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\" position=\"float\"><label>Table 2</label><caption><p> Responses of 1023 adults when asked “What does this logo mean?” (respondents shown Bear Brand logo, see figure)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"bottom\">Response</th><th colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"bottom\">No (%)</th></tr></thead><tbody><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Product is good for infants</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">402 (39.3)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Product is good for children</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">115 (11.2)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Advertising</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">110 (10.7)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Product is replacement for breast milk</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">66 (6.5)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Product is for adults</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">13 (1.3)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Product is made from animals</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">6 (0.6)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Bears love their children</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">4 (0.4)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Product is for everyone</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">3 (0.3)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Not sure</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">242 (23.7)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Other response</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">2 (0.2)</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl3\" position=\"float\"><label>Table 3</label><caption><p> Responses of 1018 adults when asked “What does this picture mean” (respondents shown picture of feeding bottle with cross through it, see figure)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"bottom\">Response</th><th colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"bottom\">No (%)</th></tr></thead><tbody><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Never noticed it/don’t know</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">557 (49.7)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Product not for infants</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">319 (31.3)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Product should not be given by bottle</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">98 (9.6)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Product is dangerous</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">16 (1.6)</td></tr><tr><td colspan=\"1\" rowspan=\"1\" align=\"left\" valign=\"top\">Product is good for babies</td><td colspan=\"1\" rowspan=\"1\" align=\"center\" valign=\"top\">10 (1.0)</td></tr></tbody></table></table-wrap>" ]

[]

[ "<boxed-text position=\"float\" content-type=\"style4\"><sec><title>Examples of cases</title><p>In the remote countryside of Luang Nam Tha Province an 18 month old boy who weighed 6 kg was admitted to hospital with evidence of kwashiorkor. His grandparents had fed him Bear Brand coffee creamer since the age of 2 months, when his mother had died. The family was of the ethnic group Lenten and did not read the Lao language. They believed the can contained cows’ milk. The child had never grown well and had never moved on to solid foods, but his condition improved rapidly with nutritional support and family education in the hospital.</p><p>In the capital city, Vientiane, a 3 month old infant with the oedema and skin changes of kwashiorkor died in hospital from complications of severe malnutrition and pneumonia. Her mother had been concerned that the infant was vomiting with breast feeding shortly after birth. She stopped breast feeding and bought the coffee creamer with the Bear Brand logo. The mother stated that she read the label and understood the creamer was not for use instead of breast milk, but noted that the can had a mother bear with a baby bear on its label, and that people in her village consider this an acceptable infant food.</p></sec></boxed-text>", "<boxed-text position=\"float\" content-type=\"style3\"><sec><title>What is already known on this topic</title><list list-type=\"simple\"><list-item><p>Exclusive breast feeding protects infant health and is recommended for the first 6 months of life</p></list-item><list-item><p>The international code on marketing of breast milk substitutes aims to protect infant health by preventing inappropriate promotion of breast milk substitutes and is considered the minimum requirement to be adopted by all governments</p></list-item><list-item><p>Violations have been reported and addressed</p></list-item></list></sec><sec><title>What this study adds</title><list list-type=\"simple\"><list-item><p>A cartoon bear holding a cub in the breastfeeding position pictured on cans of coffee creamer implies that the product is suitable for young infants, despite a written warning to the contrary</p></list-item><list-item><p>This product is widely available in the country of Laos and is commonly given to young infants</p></list-item><list-item><p>Use of this logo on coffee creamer is misleading to the local population and places the health of infants at risk</p></list-item></list></sec></boxed-text>" ]

[]

[]

[]

[]

[ "<table-wrap-foot><p>* Mean (95% confidence interval).</p></table-wrap-foot>", "<fn-group><fn><p>We thank the P8 Masters’ students of the Institut de la Francophonie pour la Médecine Tropicale (Agence Universitaire de la Francophonie), J P René, and Bryan Watt; the Lao paediatricians and the staff of Health Frontier, Vientiane; and the families and health authorities of Vientiane Municipality, Savannakhet, Hinheup, Atapeu, and Luang Nam Tha.</p></fn><fn fn-type=\"participating-researchers\"><p>Contributors: HB was responsible for the overall coordination of the study and contributed to the study design, enrolment, field supervision, and data analysis; he is also guarantor. TA and VL enrolled and followed up the children and collected and analysed the data. MA and LMS were responsible for the first survey and contributed to the interpretation of the data. All authors contributed to the writing of the paper.</p></fn><fn fn-type=\"financial-disclosure\"><p>Funding: Agence Universitaire de la Francophonie (AUF).</p></fn><fn fn-type=\"conflict\"><p>Competing interests: None declared.</p></fn><fn><p>Ethical approval: National ethical review board of Laos.</p></fn><fn><p>Provenance and peer review: Not commissioned; externally peer reviewed.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"barh579086.f1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The majority of DNA microarrays in use today are created from single genomes that do not reflect the genetic diversity of a group of heterogeneous entities. Mixed-DNA microarrays offer an alternative for \"capturing\" genetic diversity and can be used for classification purposes such as identifying pathogens or determining genetic relationships for epidemiology studies [##REF##17209070##1##, ####REF##12554104##2##, ##REF##12574259##3##, ##REF##16405681##4####16405681##4##]. DNA from one or more reference strains or plasmids is shotgun-cloned, and a mixed-genome or mixed-plasmid microarray is generated from randomly selected, PCR-amplified clone inserts [##REF##12554104##2##,##REF##12574259##3##]. Unlike most fingerprinting tools, the mixed-array format permits identification of informative probes that can be retrieved from the clone library for sequencing [##REF##16517630##5##]. However, redundant sequences and limited representation of diversity can limit the application of these tools [##REF##12574259##3##,##REF##16405681##4##]. Fortunately, a growing public database of genomes offers a new opportunity to incorporate non-redundant and diverse sequences into a mixed-microarray format. These arrays can be used to quickly assess the distribution of genetic diversity across multiple species and niches.</p>", "<p>This work focuses on the optimal design of classification arrays. By optimal we mean minimizing the complexity and cost of an array by using as few probes as possible while still rendering sufficient information to discriminate between strains or groups of organisms and to avoid bias; the goal is to remove irrelevant probes (probes that contain no useful information) and reduce the number of redundant probes (probes that contain the same information) in such a way that the chosen probes will allow us to perform the desired classification task accurately. Selection of an optimal set of probes is a key factor in designing a successful mixed microarray to suit a particular need. The effects of probe length and the number of probes per gene have been discussed in [##REF##15243142##6##]. A method for finding unique and valid oligonucleotides or probes was proposed in [##REF##16049019##7##], which tries to identify probes for a gene such that there is no similar occurrence in other locations of a genome. A tool for choosing optimal DNA oligos is reported in [##REF##12824409##8##], which identifies oligo sequences that occur in members of the target group but not in the non-target group. However, these methods are used for genome-wide probe selection and are not intended to identify minimum probe sets for classification problems.</p>", "<p>A number of methods have been introduced for designing optimal probe sets. Pre-filtering methods [##REF##12603017##9##] use clustering of all probes to find similar probe groups. Similar probes are discarded; the remaining probes are ranked, and top-ranked probes are kept for further analysis. A similar method [##UREF##0##10##] uses K-means to cluster all genes, and the means of different gene clusters are used as prototype genes. The limitation of these methods is that the number of clusters must be specified. A hybrid approach [##REF##15585531##11##] ranks the probes first and selects a set of top-ranked probes. Hierarchical clustering is then used on these probes to generate a dendrogram. The optimal probes are selected by collapsing dense clusters. In this manner a small set of probes is identified that has a similar prediction accuracy to one that uses more probes.</p>", "<p>The methods described above identify optimal probes using training data when the structure of the data is given. Such information, however, is usually unavailable for microarray data sets. A tool is still needed to help design mixed microarrays when prior knowledge of a microarray data set is unavailable. The focus of this paper is a software program, PLASMID, used for selecting an optimal set of probe sequences without a <italic>priori </italic>knowledge that will enable correct classification of groups of plasmids or bacteria. Data used to identify probe candidates can be either existing microarray data (or similar hybridization data) or sequence data from a public database such as GenBank. The latter are converted to \"probe\" sequences, and virtual hybridization is used to generate data for probe selection [##REF##17209070##1##]. To demonstrate the generality of PLASMID, we include an example whereby the program can also be applied to develop a minimum probe set to distinguish between two classes of leukemia using data from an expression array.</p>" ]

[ "<title>Methods</title>", "<title>Finding meaningful clusters in hybridization data</title>", "<p>Finding meaningful clusters of samples (e.g., plasmids) from a given set of hybridization or sequence data is the starting point for the design of an optimal microarray; our tool provides several clustering options. Clustering methods can be divided into two general groups: distance-based methods and model-based methods. Distance-based methods are either non-hierarchical or hierarchical, and each method has its particular strengths and weaknesses. Currently our tool includes the K-means non-hierarchical clustering algorithm and hierarchical clustering by means of Unweighted Pair Group Method with Arithmetic mean (UPGMA), neighbor joining, or Ward's minimum variance method, all of which are widely used in microarray data analysis [##UREF##1##12##]. Two distance metrics have been implemented, Euclidean distance and Pearson's correlation coefficient, from which users can choose. The distance-based methods listed above are standard clustering techniques. In addition to these, we have also implemented the model-based clustering method described below.</p>", "<title>Model-based genetic clustering</title>", "<p>Distance-based methods are simple to use, and the clustering results are easy to explain. However, it is hard to obtain information about the number of clusters, the confidence level of the clustering results, and so on, from these methods. To avoid some of these issues, model-based clustering methods can be used as an alternative. Model-based clustering methods assume that the data can be clustered according to a set of underlying distributions. These underlying distributions can be modeled, and finding a suitable model can be construed as an optimization problem. We assume that <italic>M </italic>is the underlying model for a data set represented by a matrix X where each row of the matrix represents the data for a given sample (e.g., plasmid). The best clustering result is represented by partition <italic>P </italic>of X. A measure is used to determine which <italic>P </italic>is most likely for X. In our tool the measure is the likelihood of all possible partitions <italic>P</italic>. A number of different optimization methods can be used to find the solution for <italic>P</italic>. In our tool, we have chosen to use a genetic algorithm because of its simplicity and efficiency in addition to its ability to find the optimal solution. Usually model-based clustering methods are based on the Expectation-Maximization (EM) method. However, EM algorithms tend to break down for microarray data because an inversion of the covariance matrix must be performed. In genetic algorithms, a search method is used to circumvent the need for this computation, thereby making genetic model-based methods more stable.</p>", "<p>To find the best partition <italic>P </italic>we want to maximize the posterior probability <italic>f</italic>(<italic>P</italic>|X). According to Bayes' theorem, where <italic>f</italic>(<italic>P</italic>) is the prior probability. Recasting Bayes' theorem in terms of the likelihood ℒ(X|<italic>P</italic>) gives <italic>f</italic>(<italic>P</italic>|X) ∝ ℒ(X|<italic>P</italic>) <italic>f </italic>(<italic>P</italic>)–that is, the posterior probability is proportional to the product of the likelihood and prior probability. Now if we assume a uniform distribution for <italic>P</italic>, then <italic>f</italic>(<italic>P</italic>) is constant and maximizing the posterior probability <italic>f</italic>(<italic>P</italic>|X) is equivalent to maximizing the likelihood ℒ(X|<italic>P</italic>).</p>", "<p>If we assume the rows of the matrix X in each cluster of the partition are independent and identically distributed, we can compute the likelihood of a partition. For this work, we assume the rows in each cluster are normally distributed with mean <italic>μ</italic>_{<italic>i </italic>}and variance , and we assume a normal distribution for all <italic>μ</italic>_{<italic>i </italic>}and an inverse-Γ distribution for all . This leads to:</p>", "<p></p>", "<p>where <italic>k </italic>is the index of clusters, <italic>j </italic>is the index of probes, <italic>n</italic>_{<italic>k </italic>}is the number of samples in the <italic>k</italic>th cluster, <italic>k</italic>_{<italic>i </italic>}is the index of samples in the <italic>k</italic>th cluster, and <italic>μ</italic>₀and are the overall mean and variance of all the data [##REF##16766561##13##].</p>", "<p>Using this as a measure, the genetic algorithm is used to find the partition that maximizes the likelihood. The steps of the genetic algorithm are summarized as follows:</p>", "<p>1. Generate <italic>N </italic>random partitions. Each partition is represented by a vector [1 2 1 ⋯] where each term is the index of a cluster.</p>", "<p>2. Prior knowledge of pairs of samples highly unlikely to be in the same cluster can be incorporated into the partition likelihood by creating a text file with each pair of samples, together with a small weighting factor, on one line. The weighting factor must be smaller than 1, but how much smaller has to be determined empirically based on the end result. A weighting factor of zero indicates that the pair cannot be in the same cluster.</p>", "<p>3. Compute the likelihood ℒ for all partitions.</p>", "<p>4. Repeat the following steps until the the maximum iterations (<italic>Max</italic>) has been reached or the difference between the likelihood of two successive iterations is less than <italic>ε</italic>, where <italic>Max </italic>and <italic>ε </italic>are given.</p>", "<p>(a) Select the two partitions with the highest scores.</p>", "<p>(b) Do crossover and mutation on these two partitions to generate new partitions. Crossover is accomplished by randomly selecting sections of equal length from each partition and exchanging them. Mutation is performed following crossover and is accomplished by randomly selecting one term in each of the partitions and changing it to a different value.</p>", "<p>(c) Compute the likelihood ℒ for these two new partitions (offspring).</p>", "<p>(d) Replace the two lowest-ranked partitions with the offspring.</p>", "<p>Other measures can be used including Bayesian Information Criteria and minimum description length. These measures will be included in future versions of PLASMID.</p>", "<title>Probe ranking for classification</title>", "<p>In a DNA microarray data set there are usually many more probes than the number of samples (e.g., plasmids) to be classified, and often some probes either convey no useful information or convey the same information. Thus, in the design of an optimal probe set for sample classification, one objective is to identify and remove irrelevant and redundant probes. In this section, we describe our method for removing irrelevant probes; in the next section redundancy reduction is described.</p>", "<p>Irrelevant probes are removed using probe ranking on the clusters of samples obtained in the previous step. There are two basic approaches to probe ranking: filter techniques and wrapper techniques. Because of their simplicity, filter procedures are used most commonly for DNA microarrays. The filter procedure ranks each probe using a metric based on its classification relevance. Top-ranked probes are then selected to perform classification. Numerous filter metrics are described in the literature [##REF##12912841##14##]: probabilistic and distance metrics, dependence measures, scores based on information theory, etc. In our tool, filter metrics are determined using two different statistical tests, the ANOVA-<italic>F </italic>and Brown-Forsythe tests. Other tests considered were the Welch, adjusted Welch, Cochran, and Kruskal-Wallis test statistics [##REF##16046818##15##].</p>", "<p>The test statistic is used as a metric to evaluate the discriminating power of a probe. Higher values represent more discriminating probes. For some applications, clusters may include an insufficient number of samples for meaningful statistical analysis. Such cases can be handled by generating random samples that differ only slightly from the original samples. These samples can be included in the statistical analysis and then discarded without compromising the probe ranking procedure. The purpose of adding these samples is for computational convenience only; they do not add more information.</p>", "<p>The end result of the probe ranking function is a list of all probes ranked by their classification relevance. At this point, the user can either stop and use some chosen number of the top-ranked probes for the array probe set or continue with probe reduction and stepwise discriminant analysis to remove redundant probes and assign weights to the probes.</p>", "<title>Stepwise discriminant analysis</title>", "<p>Probe ranking is used to remove irrelevant probes that convey little or no information. Nevertheless, while the top-ranked probes are informative, at least some of them are likely to convey redundant information. The next task is to remove this unnecessary redundancy. K-means clustering is usually used to cluster samples (e.g., plasmids) as described in an earlier section, but here we use it in a novel way to cluster probes. A set of top-ranked probes is clustered into <italic>κ </italic>groups where the value of <italic>κ </italic>is evaluated empirically to maximize classification accuracy; probes in the same group are highly correlated with each other but uncorrelated or loosely correlated with probes in other groups. The probe closest to the center of a group is chosen to be representative of that group, and the <italic>κ </italic>representative probes are used with stepwise discriminant analysis (SDA) [##UREF##2##16##] which identifies the optimal probe set from the <italic>κ </italic>probes. At each step of the SDA, an statistic is computed for each probe; this value is used to determine whether including the probe or excluding the probe from will significantly improve sample differentiation. The SDA process starts with an empty probe set , and an iterative process of adding a probe to or removing a probe from continues until no probes can be added or removed. is used for the probes in , and is used for the probes not in . The probe in with the smallest value of less than a chosen threshold value, usually 1.0, is removed; the probe not in with the largest value of greater than the threshold value is added to . The formulas used to compute are:</p>", "<p> values:</p>", "<p></p>", "<p></p>", "<p>Wilks' Λ:</p>", "<p></p>", "<p>Within-group covariance matrix:</p>", "<p></p>", "<p>Among-group covariance matrix:</p>", "<p></p>", "<p>where <italic>q </italic>is the number of clusters, <italic>n</italic>_{<italic>m </italic>}is the number of samples in the cluster <italic>m</italic>, <italic>x</italic>_{<italic>mki </italic>}is the value of the <italic>i</italic>th probe for the <italic>k</italic>th sample in the <italic>m</italic>th cluster, <italic>n </italic>is the total number of samples, <italic>r </italic>is the number of probes currently included in , |<italic>p </italic>denotes a new group of probes which is obtained by adding the probe <italic>p </italic>to , and \\<italic>p </italic>denotes a new group of probes which is obtained by removing the probe <italic>p </italic>from .</p>", "<p>At the conclusion of SDA, the optimal probe set is determined based on the prediction accuracy of the selected probes. Because there are typically a small number of samples associated with microarray data, prediction accuracy is computed using the leave-one-out (LOO) cross validation method [##REF##15585531##11##,##REF##16046818##15##]. The set of probes associated with the highest LOO predication accuracy is written to a file together with its associated weights. It is important to note that when SDA is used to obtain the final probe set, the weights associated with the probes must be used for classification of new empirical data obtained using the probes. The probes should not be treated with equal weight.</p>", "<title>Probe selection for a classification microarray</title>", "<p>In summary, the steps in our design of an optimal probe set are:</p>", "<p>1. Cluster the samples (e.g., plasmids) using microarray or sequence data and select clusters of interest using a hierarchical, non-hierarchical, and/or model-based method. <italic>A priori </italic>clustering is also permitted.</p>", "<p>2. Use the probe ranking procedure with the sample clusters to rank the probes for relevance.</p>", "<p>3. Repeat K-means clustering of probes for probe reduction until satisfied:</p>", "<p>(a) Select <italic>j </italic>top-ranked probes.</p>", "<p>(b) Repeat for <italic>κ </italic>in a chosen range:</p>", "<p>i. Cluster the <italic>j </italic>top-ranked probes into <italic>κ </italic>clusters.</p>", "<p>ii. Choose <italic>κ </italic>representative probes, one from each cluster.</p>", "<p>iii. Use SDA to find a set of probes from the <italic>κ </italic>representative probes and compute the LOO prediction accuracy.</p>", "<p>4. Save the set of probes associated with the highest LOO prediction accuracy together with its weights. After constructing the optimized microarray, a set of independent control samples should be hybridized to empirically assess the accuracy of the microarray results.</p>", "<p>A flowchart of the process is shown in Fig. ##FIG##0##1##. It should be pointed out that the optimal number of probes computed by this process does not take into account the effects of noise and other random experimental effects. The sample-to-feature (SFR) ratio gives the minimum number of probes that should be used to create a microarray. The rule of thumb is given by [##REF##12912828##17##]:</p>", "<p></p>", "<p>In this paper we refer to features as probes. The SFR should be used in conjunction with the results to choose the optimal probe set.</p>" ]

[ "<title>Results and Discussion</title>", "<p>In this section we present results obtained using PLASMID to analyze a mixed-plasmid microarray data set [##REF##16405681##4##] and a simulated mixed-genome microarray data set [##REF##17209070##1##]. We also present results for publically-available leukemia expression array data [##REF##10521349##18##]. For this latter data set, clusters (i.e., types of leukemia) are pre-assigned so only probe ranking, reduction of probe redundancy, and stepwise discriminant analysis (SDA) are used to determine the optimal probe set. PLASMID's performance in probe selection is evaluated using the leave-one-out (LOO) approach for which one sample is excluded and the remaining samples are used to obtain the discriminant functions. Each sample is, in turn, excluded and a corresponding set of discriminant functions is used to classify it. The prediction accuracy, the percentage of times the withheld samples are correctly classified, is used as the performance metric.</p>", "<title>Mixed-plasmid microarray data</title>", "<p>A mixed-plasmid microarray has been used to compare the genetic composition of plasmids [##REF##16405681##4##]. The microarray consists of 576 probes composed of randomly selected fragments of plasmid DNA, and the data were obtained from hybridization experiments with 43 plasmids. The data are composed of hybridization signal intensities for each microarray probe [see Additional file ##SUPPL##0##1##].</p>", "<p>First we used the Ward's minimim variance hierarchical clustering algorithm to create a dendrogram. To test the two-class problem, we divided the dendrogram into two clusters. One cluster consisted of 15 plasmids which, with one exception (the <italic>peSSuTet </italic>plasmid), have the <italic>bla</italic>_CMY-2antibiotic resistance gene; the other cluster consisted of 28 plasmids. We then used the probe ranking function, choosing the ANOVA-<italic>F </italic>test statistic, and generated a scatter plot (Fig. ##FIG##1##2##). The scatter plot shows that the majority of the probes have statistical values close to zero and, thus, that ANOVA-<italic>F </italic>test statistics can be used to distinguish between informative (<italic>F </italic>&gt; 0) and uninformative (<italic>F </italic>≈ 0) probes. This result also serves to highlight the need for optimization algorithms, as the majority of probes provide limited discrimination.</p>", "<p>For the two-cluster case, we chose 1, 20, and 200 top-ranked probes for comparison. Using reduction of probe redundancy and SDA, we found that a single probe (5-E3, a transposase gene associated with the <italic>bla</italic>_CMY-2element [##REF##16569893##19##]) correctly classified all but two of the plasmids [##REF##16405681##4##]. Interestingly, in the original study one of these two plasmids (<italic>pe1171sT</italic>) was classified with plasmids that harbor the <italic>bla</italic>_CMY-2gene even though it does not carry this gene. Analysis with PLASMID separated <italic>pe1171sT </italic>from the <italic>bla</italic>_{CMY -2}plasmids. In addition, a different plasmid (<italic>pe7594T</italic>) that harbors the <italic>bla</italic>_CMY-2gene was classified with other <italic>bla</italic>_CMY-2positive plasmids. Thus, analysis using PLASMID more accurately reflects the phenotypic properties of the plasmids included in the study. The one exception was the <italic>peSSuT </italic>plasmid that was consistently classified with <italic>bla</italic>_CMY-2plasmids while not harboring this gene [##REF##12824409##8##].</p>", "<p>Next we divided the original dendrogram into five plasmid clusters and ranked probes as before. As expected, the number of probe clusters <italic>κ </italic>specified for the reduction of probe redundancy affects the prediction accuracy (Table ##TAB##0##1##). Small values of <italic>κ </italic>certainly reduce redundancy, but they also reduce specificity. The optimal set of probes is identified using SDA with the LOO method to determine the highest prediction accuracy. In this case, the smallest number of probes from the top-most ranked probes with the highest prediction accuracy is 10. Thus, PLASMID analysis reduced the original data set of 576 probes to 10 probes that are needed to accurately classify plasmids into one of five groups. Non-hierarchical clustering followed by probe ranking, probe reduction, and SDA gave similar results (data not shown).</p>", "<p>In addition to hierarchical and non-hiearchical clustering methods, we can obtain classification results using our model-based method, which is based on a genetic algorithm. The genetic algorithm predicted that the most likely number of plasmid clusters is five (Table ##TAB##1##2##). Comparison of Tables ##TAB##0##1## and ##TAB##1##2## shows that prediction accuracies depend on the initial clustering method used. For this case, the prediction accuracies for the model-based clustering method are larger for a given number of probe clusters than those obtained via the hierarchical method. Furthermore, the variance in prediction accuracies is lower as a function of the number of top-ranked probes when clusters are initially assigned using the model-based method. For other data sets, however, another clustering model might give the best results.</p>", "<p>Based on the sample-to-feature ratio (SFR), at least 9 probes (features) are required for classifying 43 plasmids (samples). Tables ##TAB##0##1## and ##TAB##1##2## show several choices for 10 probes with equivalent performance. When additional information is available, it should be used to assist with the choice of a final set.</p>", "<title>Virtual Streptococcus mixed-genome microarray data</title>", "<p>A virtual <italic>Streptococcus </italic>mixed-genome microarray was constructed by Wan <italic>et al</italic>. [##REF##17209070##1##]. To create the equally-represented, 4000-probe virtual array, 800 gene segments each 600-bp long were randomly selected from genomes of fifteen strains of five bacterial species–that is, each species was represented by 800 different probes. Virtual hybridization was accomplished using BLAST scores as proxies for array probe intensities [see Additional file ##SUPPL##1##2##], and PLASMID was used to analyze the data. In the initial analysis one bacterial species was excluded from the study because it was represented by only a single strain (<italic>S. mutans </italic>UA159). Because we knew a <italic>priori </italic>that the samples belonged to four different species, the goal was to find an optimal set of probes to classify these four. ANOVA-<italic>F </italic>tests were used to rank the 4000 probes, and LOO analysis was performed on different numbers of the highest ranked probes. In fact, we found the LOO prediction accuracy to be 100% for differentiating the four different species using only the single top-ranked probe. On examination we found that the hybridization values (BLAST scores) for this probe for strains from different groups were well separated (<italic>i.e</italic>., different from each other), while the hybridization values for strains from the same group were very similar. While it appears that successful classification can be achieved with a single probe when classification relies on differences in hybridization signal, given inherent sources of variation in microarray hybridization, it would be prudent to include additional probes to increase classification confidence for empirical data. For example, the minimum recommended probe set in this case would be 3 according to the SFR.</p>", "<p>In the second analysis, our model-based clustering method identified two clusters, one with the two <italic>S. pneumoniae </italic>strains and the other with the remaining 13 strains. After probe ranking, reduction of probe redundancy, and SDA, a single probe could be used to differentiate these two groups. We also used non-hierarchical clustering of the samples followed by probe ranking, probe reduction, and SDA. When the number of clusters was chosen to be <italic>k </italic>= 2, the result was identical to the result obtained using our model-based cluster method. When the number of clusters was chosen to be <italic>k </italic>= 3 or <italic>k </italic>= 4, the two <italic>S. pneumoniae </italic>genomes were placed into different groups. A dendrogram constructed using the neighbor joining method shows a clear distinction between the two <italic>S. pneumoniae </italic>samples and the remaining bacteria (Fig. ##FIG##2##3##). When these two samples are excluded, PLASMID groups the remaining thirteen samples correctly into four species clusters. The results shown in Table ##TAB##2##3## are obtained using non-hierarchical clustering, probe ranking, probe reduction, and SDA. As this table illustrates, only 2 probes are needed to obtain 100% prediction accuracy by species. These 2 probes are from the genomes of <italic>S. pneumoniae </italic>TIGR4 and either <italic>S. pyogenes </italic>M1 GAS or <italic>S. pyogenes </italic>MGAS5005. Based on the SFR rule of thumb, at least 3 probes are needed. Several choices exist that suffice for this condition.</p>", "<p>For virtual microarrays, BLAST scores are used to obtain hybridization intensities, and the accuracy of the scores will affect the choice of an optimal probe set. While error could be modeled from real data, the best measure of reliability will be obtained using actual hybridization experiments.</p>", "<title>Public ALL/AML leukemia data</title>", "<p>The ALL/AML leukemia data set, obtained from expression arrays, has been widely used in the literature. It consists of two classes of leukemia, acute lymphoblastic leukemia (ALL) and acute myeloblastic leukemia (AML), and there are 72 samples (47 ALL and 25 AML) and 7129 probes. Table ##TAB##3##4## shows prediction accuracy results after probe ranking, probe redundancy reduction, and SDA have been performed. When the top 50 probes were selected, the highest accuracy was achieved when probes were clustered into 10 groups. A set of 10 probes was identifed with a prediction accuracy of 97.22%. Using additional probes does not lead to improvement. According to the SFR rule of thumb, at least 20 probes should be used in the actual microarray design; several choices of 20 probes exist and all produce robust prediction results (Table ##TAB##3##4##).</p>" ]

[ "<title>Results and Discussion</title>", "<p>In this section we present results obtained using PLASMID to analyze a mixed-plasmid microarray data set [##REF##16405681##4##] and a simulated mixed-genome microarray data set [##REF##17209070##1##]. We also present results for publically-available leukemia expression array data [##REF##10521349##18##]. For this latter data set, clusters (i.e., types of leukemia) are pre-assigned so only probe ranking, reduction of probe redundancy, and stepwise discriminant analysis (SDA) are used to determine the optimal probe set. PLASMID's performance in probe selection is evaluated using the leave-one-out (LOO) approach for which one sample is excluded and the remaining samples are used to obtain the discriminant functions. Each sample is, in turn, excluded and a corresponding set of discriminant functions is used to classify it. The prediction accuracy, the percentage of times the withheld samples are correctly classified, is used as the performance metric.</p>", "<title>Mixed-plasmid microarray data</title>", "<p>A mixed-plasmid microarray has been used to compare the genetic composition of plasmids [##REF##16405681##4##]. The microarray consists of 576 probes composed of randomly selected fragments of plasmid DNA, and the data were obtained from hybridization experiments with 43 plasmids. The data are composed of hybridization signal intensities for each microarray probe [see Additional file ##SUPPL##0##1##].</p>", "<p>First we used the Ward's minimim variance hierarchical clustering algorithm to create a dendrogram. To test the two-class problem, we divided the dendrogram into two clusters. One cluster consisted of 15 plasmids which, with one exception (the <italic>peSSuTet </italic>plasmid), have the <italic>bla</italic>_CMY-2antibiotic resistance gene; the other cluster consisted of 28 plasmids. We then used the probe ranking function, choosing the ANOVA-<italic>F </italic>test statistic, and generated a scatter plot (Fig. ##FIG##1##2##). The scatter plot shows that the majority of the probes have statistical values close to zero and, thus, that ANOVA-<italic>F </italic>test statistics can be used to distinguish between informative (<italic>F </italic>&gt; 0) and uninformative (<italic>F </italic>≈ 0) probes. This result also serves to highlight the need for optimization algorithms, as the majority of probes provide limited discrimination.</p>", "<p>For the two-cluster case, we chose 1, 20, and 200 top-ranked probes for comparison. Using reduction of probe redundancy and SDA, we found that a single probe (5-E3, a transposase gene associated with the <italic>bla</italic>_CMY-2element [##REF##16569893##19##]) correctly classified all but two of the plasmids [##REF##16405681##4##]. Interestingly, in the original study one of these two plasmids (<italic>pe1171sT</italic>) was classified with plasmids that harbor the <italic>bla</italic>_CMY-2gene even though it does not carry this gene. Analysis with PLASMID separated <italic>pe1171sT </italic>from the <italic>bla</italic>_{CMY -2}plasmids. In addition, a different plasmid (<italic>pe7594T</italic>) that harbors the <italic>bla</italic>_CMY-2gene was classified with other <italic>bla</italic>_CMY-2positive plasmids. Thus, analysis using PLASMID more accurately reflects the phenotypic properties of the plasmids included in the study. The one exception was the <italic>peSSuT </italic>plasmid that was consistently classified with <italic>bla</italic>_CMY-2plasmids while not harboring this gene [##REF##12824409##8##].</p>", "<p>Next we divided the original dendrogram into five plasmid clusters and ranked probes as before. As expected, the number of probe clusters <italic>κ </italic>specified for the reduction of probe redundancy affects the prediction accuracy (Table ##TAB##0##1##). Small values of <italic>κ </italic>certainly reduce redundancy, but they also reduce specificity. The optimal set of probes is identified using SDA with the LOO method to determine the highest prediction accuracy. In this case, the smallest number of probes from the top-most ranked probes with the highest prediction accuracy is 10. Thus, PLASMID analysis reduced the original data set of 576 probes to 10 probes that are needed to accurately classify plasmids into one of five groups. Non-hierarchical clustering followed by probe ranking, probe reduction, and SDA gave similar results (data not shown).</p>", "<p>In addition to hierarchical and non-hiearchical clustering methods, we can obtain classification results using our model-based method, which is based on a genetic algorithm. The genetic algorithm predicted that the most likely number of plasmid clusters is five (Table ##TAB##1##2##). Comparison of Tables ##TAB##0##1## and ##TAB##1##2## shows that prediction accuracies depend on the initial clustering method used. For this case, the prediction accuracies for the model-based clustering method are larger for a given number of probe clusters than those obtained via the hierarchical method. Furthermore, the variance in prediction accuracies is lower as a function of the number of top-ranked probes when clusters are initially assigned using the model-based method. For other data sets, however, another clustering model might give the best results.</p>", "<p>Based on the sample-to-feature ratio (SFR), at least 9 probes (features) are required for classifying 43 plasmids (samples). Tables ##TAB##0##1## and ##TAB##1##2## show several choices for 10 probes with equivalent performance. When additional information is available, it should be used to assist with the choice of a final set.</p>", "<title>Virtual Streptococcus mixed-genome microarray data</title>", "<p>A virtual <italic>Streptococcus </italic>mixed-genome microarray was constructed by Wan <italic>et al</italic>. [##REF##17209070##1##]. To create the equally-represented, 4000-probe virtual array, 800 gene segments each 600-bp long were randomly selected from genomes of fifteen strains of five bacterial species–that is, each species was represented by 800 different probes. Virtual hybridization was accomplished using BLAST scores as proxies for array probe intensities [see Additional file ##SUPPL##1##2##], and PLASMID was used to analyze the data. In the initial analysis one bacterial species was excluded from the study because it was represented by only a single strain (<italic>S. mutans </italic>UA159). Because we knew a <italic>priori </italic>that the samples belonged to four different species, the goal was to find an optimal set of probes to classify these four. ANOVA-<italic>F </italic>tests were used to rank the 4000 probes, and LOO analysis was performed on different numbers of the highest ranked probes. In fact, we found the LOO prediction accuracy to be 100% for differentiating the four different species using only the single top-ranked probe. On examination we found that the hybridization values (BLAST scores) for this probe for strains from different groups were well separated (<italic>i.e</italic>., different from each other), while the hybridization values for strains from the same group were very similar. While it appears that successful classification can be achieved with a single probe when classification relies on differences in hybridization signal, given inherent sources of variation in microarray hybridization, it would be prudent to include additional probes to increase classification confidence for empirical data. For example, the minimum recommended probe set in this case would be 3 according to the SFR.</p>", "<p>In the second analysis, our model-based clustering method identified two clusters, one with the two <italic>S. pneumoniae </italic>strains and the other with the remaining 13 strains. After probe ranking, reduction of probe redundancy, and SDA, a single probe could be used to differentiate these two groups. We also used non-hierarchical clustering of the samples followed by probe ranking, probe reduction, and SDA. When the number of clusters was chosen to be <italic>k </italic>= 2, the result was identical to the result obtained using our model-based cluster method. When the number of clusters was chosen to be <italic>k </italic>= 3 or <italic>k </italic>= 4, the two <italic>S. pneumoniae </italic>genomes were placed into different groups. A dendrogram constructed using the neighbor joining method shows a clear distinction between the two <italic>S. pneumoniae </italic>samples and the remaining bacteria (Fig. ##FIG##2##3##). When these two samples are excluded, PLASMID groups the remaining thirteen samples correctly into four species clusters. The results shown in Table ##TAB##2##3## are obtained using non-hierarchical clustering, probe ranking, probe reduction, and SDA. As this table illustrates, only 2 probes are needed to obtain 100% prediction accuracy by species. These 2 probes are from the genomes of <italic>S. pneumoniae </italic>TIGR4 and either <italic>S. pyogenes </italic>M1 GAS or <italic>S. pyogenes </italic>MGAS5005. Based on the SFR rule of thumb, at least 3 probes are needed. Several choices exist that suffice for this condition.</p>", "<p>For virtual microarrays, BLAST scores are used to obtain hybridization intensities, and the accuracy of the scores will affect the choice of an optimal probe set. While error could be modeled from real data, the best measure of reliability will be obtained using actual hybridization experiments.</p>", "<title>Public ALL/AML leukemia data</title>", "<p>The ALL/AML leukemia data set, obtained from expression arrays, has been widely used in the literature. It consists of two classes of leukemia, acute lymphoblastic leukemia (ALL) and acute myeloblastic leukemia (AML), and there are 72 samples (47 ALL and 25 AML) and 7129 probes. Table ##TAB##3##4## shows prediction accuracy results after probe ranking, probe redundancy reduction, and SDA have been performed. When the top 50 probes were selected, the highest accuracy was achieved when probes were clustered into 10 groups. A set of 10 probes was identifed with a prediction accuracy of 97.22%. Using additional probes does not lead to improvement. According to the SFR rule of thumb, at least 20 probes should be used in the actual microarray design; several choices of 20 probes exist and all produce robust prediction results (Table ##TAB##3##4##).</p>" ]

[ "<title>Conclusion</title>", "<p>In this paper we describe a new software tool, PLASMID, for selecting an optimal set of probes for the design of a classification microarray. The tool provides the user with several clustering methods, a probe ranking method, probe redundancy reduction, and probe selection using stepwise discriminant analysis. Images can be saved in several different formats, and weights generated using SDA can be stored for use in analysis of experimental data. In addition, PLASMID can be used to construct virtual microarrays with genomes from public databases; these can then be used to determine an optimal probe set for use in actual microarray experiments. The software package has been applied to data from a mixed-plasmid microarray, a virtual mixed-genome microarray, and an expression microarray. Robust results have been obtained for all three sets of data.</p>", "<p>Although many methods are available for determining a set of probes for a given microarray data set, these methods require the classification information to be known in advance. PLASMID was designed to be used prior to implementation of a microarray when no such information is available, although the program can also be used when clusters are known a <italic>priori</italic>.</p>", "<p>PLASMID can be obtained by following the link from <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.vetmed.wsu.edu/research_vmp/MicroArrayLab/\"/>.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Classification microarrays are used for purposes such as identifying strains of bacteria and determining genetic relationships to understand the epidemiology of an infectious disease. For these cases, mixed microarrays, which are composed of DNA from more than one organism, are more effective than conventional microarrays composed of DNA from a single organism. Selection of probes is a key factor in designing successful mixed microarrays because redundant sequences are inefficient and limited representation of diversity can restrict application of the microarray. We have developed a Java-based software tool, called PLASMID, for use in selecting the minimum set of probe sequences needed to classify different groups of plasmids or bacteria.</p>", "<title>Results</title>", "<p>The software program was successfully applied to several different sets of data. The utility of PLASMID was illustrated using existing mixed-plasmid microarray data as well as data from a virtual mixed-genome microarray constructed from different strains of <italic>Streptococcus</italic>. Moreover, use of data from expression microarray experiments demonstrated the generality of PLASMID.</p>", "<title>Conclusion</title>", "<p>In this paper we describe a new software tool for selecting a set of probes for a classification microarray. While the tool was developed for the design of mixed microarrays–and mixed-plasmid microarrays in particular–it can also be used to design expression arrays. The user can choose from several clustering methods (including hierarchical, non-hierarchical, and a model-based genetic algorithm), several probe ranking methods, and several different display methods. A novel approach is used for probe redundancy reduction, and probe selection is accomplished via stepwise discriminant analysis. Data can be entered in different formats (including Excel and comma-delimited text), and dendrogram, heat map, and scatter plot images can be saved in several different formats (including jpeg and tiff). Weights generated using stepwise discriminant analysis can be stored for analysis of subsequent experimental data. Additionally, PLASMID can be used to construct virtual microarrays with genomes from public databases, which can then be used to identify an optimal set of probes.</p>" ]

[ "<title>System Overview and Implementation</title>", "<p>Our software tool PLASMID is implemented as a Java application. The NetBeans platform was chosen for development because addition of new functions is easily implemented. Also, many of the tasks common to desktop applications are provided by NetBeans. These include user interface management (e.g., menus and toolbars), user settings management, storage management (saving and loading any kind of data), window management, and wizard framework (supporting step-by-step dialogs). Each function is implemented as a NetBeans module and can be installed or removed easily without affecting existing functions. Java is a platform-independent programming language, so although PLASMID has been developed using the Windows operating system, it will be relatively easy to adapt it to other operating systems. We intend to extend PLASMID to both the Linux and Mac OS X operating systems. In addition to Java, PLASMID uses code written using the C++ programming language. C++ is needed for computationally intensive tasks that require greater speed and efficiency. The use of two different programming languages is transparent to the user.</p>", "<p>PLASMID provides an integrated environment for designing an optimal classification microarray. As such, PLASMID v0.91 includes the following services:</p>", "<p>1. Loading and management of different kinds of input data, including plasmid sequence data, hybridization data, virtual hybridization data, and probe sequences. Data may be in tab-delimited or comma-delimited text format or in Microsoft Excel spreadsheet format.</p>", "<p>2. Different methods for processing hybridization data. The tool provides several data preprocessing methods, including normalization and noise filtering. It also provides hierarchical, non-hierarchical, and model-based methods for clustering samples; two different statistical tests for ranking probes; use of K-means clustering for reduction of probe redundancy; and stepwise discriminant analysis with assignment of weights to probes.</p>", "<p>3. Design of mixed arrays using existing hybridization data or virtual hybridization data. An optimal set of probes is identified, and weights associated with each probe are stored for analysis of experimental results.</p>", "<p>4. Construction of virtual microarrays to obtain virtual hybridization data using genomes from the National Center for Biotechnology Information (NCBI) database. Genomes for probes can be chosen by accession number or by gene sequence.</p>", "<p>5. Visualization of microarray data and data processing results, including dendrograms, heat maps, and scatter plots. Plots can be saved in different image formats.</p>", "<p>6. Automatic probe design after the user has specified the parameters. A step-by-step wizard guides the user through the various steps.</p>", "<p>Experimental data obtained from microarrays designed using PLASMID can be used as input data and analyzed using the weighted classification function obtained in 3.</p>", "<title>Availability and requirements</title>", "<p>• Project name: PLASMID</p>", "<p>• Project home page: <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.vetmed.wsu.edu/research_vmp/MicroArrayLab/\"/></p>", "<p>• Operating system: Windows but to be ported to Linux and Mac OS X</p>", "<p>• Programming languages: Java and C++ (with gcc compiler)</p>", "<p>• Other requirements: Java Runtime Environment</p>", "<p>• License: Free to academic and nonprofit organizations</p>", "<title>Authors' contributions</title>", "<p>DM and SLB developed PLASMID, DM was responsible for programming PLASMID, and DRC provided the necessary microbiology expertise. All authors have read and approved the final manuscript.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>This project has been funded by the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Services, under Contract No. NO-1-AI-30055; by the Agricultural Animal Health Program, College of Veterinary Medicine, Washington State University, Pullman, WA; and by the Carl M. Hansen Foundation.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Flowchart of PLASMID</bold>. Flowchart of the probe selection process using PLASMID.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Scatter plot of ANOVA-<italic>F </italic>test statistics for the mixed-plasmid microarray probes</bold>. The scatter plot shows that the majority of the probes have statistical values close to zero and, thus, that ANOVA-<italic>F </italic>test statistics can be used to distinguish between informative (<italic>F </italic>&gt; 0) and uninformative (<italic>F </italic>≈ 0) probes.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Dendrogram for <italic>Streptococcus </italic>MGM data</bold>. The dendrogram constructed using the neighbor joining method shows a clear distinction between the two <italic>S. pneumoniae </italic>samples and the remaining bacteria.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Classification accuracy of mixed-plasmid data using hierachical clustering with five sample (plasmid) clusters. PA is the prediction accuracy.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"12\">Number of clusters of probes, <italic>κ</italic></td></tr><tr><td/><td colspan=\"12\"><hr/></td></tr><tr><td/><td align=\"center\" colspan=\"2\">2</td><td align=\"center\" colspan=\"2\">5</td><td align=\"center\" colspan=\"2\">10</td><td align=\"center\" colspan=\"2\">20</td><td align=\"center\" colspan=\"2\">30</td><td align=\"center\" colspan=\"2\">40</td></tr><tr><td/><td colspan=\"12\"><hr/></td></tr><tr><td align=\"center\">Number of top-ranked probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td></tr></thead><tbody><tr><td align=\"center\">100</td><td align=\"center\">72.09</td><td align=\"center\">2</td><td align=\"center\">72.09</td><td align=\"center\">5</td><td align=\"center\">72.09</td><td align=\"center\">10</td><td align=\"center\">69.77</td><td align=\"center\">19</td><td align=\"center\">69.77</td><td align=\"center\">29</td><td align=\"center\">69.77</td><td align=\"center\">32</td></tr><tr><td align=\"center\">150</td><td align=\"center\">86.05</td><td align=\"center\">2</td><td align=\"center\">93.02</td><td align=\"center\">5</td><td align=\"center\">95.35</td><td align=\"center\">10</td><td align=\"center\">95.35</td><td align=\"center\">20</td><td align=\"center\">95.35</td><td align=\"center\">29</td><td align=\"center\">95.35</td><td align=\"center\">36</td></tr><tr><td align=\"center\">200</td><td align=\"center\">74.42</td><td align=\"center\">2</td><td align=\"center\">90.70</td><td align=\"center\">5</td><td align=\"center\">93.02</td><td align=\"center\">10</td><td align=\"center\">93.02</td><td align=\"center\">20</td><td align=\"center\">93.02</td><td align=\"center\">30</td><td align=\"center\">95.35</td><td align=\"center\">35</td></tr><tr><td align=\"center\">250</td><td align=\"center\">76.74</td><td align=\"center\">2</td><td align=\"center\">95.35</td><td align=\"center\">5</td><td align=\"center\">95.35</td><td align=\"center\">10</td><td align=\"center\">95.35</td><td align=\"center\">20</td><td align=\"center\">95.35</td><td align=\"center\">30</td><td align=\"center\">90.70</td><td align=\"center\">34</td></tr><tr><td align=\"center\">300</td><td align=\"center\">46.51</td><td align=\"center\">2</td><td align=\"center\">88.37</td><td align=\"center\">5</td><td align=\"center\">93.02</td><td align=\"center\">10</td><td align=\"center\">93.02</td><td align=\"center\">20</td><td align=\"center\">93.02</td><td align=\"center\">30</td><td align=\"center\">90.70</td><td align=\"center\">35</td></tr><tr><td align=\"center\">350</td><td align=\"center\">76.74</td><td align=\"center\">2</td><td align=\"center\">93.02</td><td align=\"center\">5</td><td align=\"center\">93.02</td><td align=\"center\">10</td><td align=\"center\">95.35</td><td align=\"center\">20</td><td align=\"center\">95.35</td><td align=\"center\">30</td><td align=\"center\">95.35</td><td align=\"center\">33</td></tr><tr><td align=\"center\">400</td><td align=\"center\">69.77</td><td align=\"center\">2</td><td align=\"center\">93.02</td><td align=\"center\">5</td><td align=\"center\">90.70</td><td align=\"center\">10</td><td align=\"center\">93.02</td><td align=\"center\">20</td><td align=\"center\">90.70</td><td align=\"center\">30</td><td align=\"center\">93.02</td><td align=\"center\">35</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Classification accuracy of mixed-plasmid data with model-based clustering. PA is the prediction accuracy.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"12\">Number of clusters of probes, <italic>κ</italic></td></tr><tr><td/><td colspan=\"12\"><hr/></td></tr><tr><td/><td align=\"center\" colspan=\"2\">2</td><td align=\"center\" colspan=\"2\">5</td><td align=\"center\" colspan=\"2\">10</td><td align=\"center\" colspan=\"2\">20</td><td align=\"center\" colspan=\"2\">30</td><td align=\"center\" colspan=\"2\">40</td></tr><tr><td/><td colspan=\"12\"><hr/></td></tr><tr><td align=\"center\">Number of top-ranked probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td></tr></thead><tbody><tr><td align=\"center\">100</td><td align=\"center\">83.72</td><td align=\"center\">2</td><td align=\"center\">95.35</td><td align=\"center\">5</td><td align=\"center\">95.35</td><td align=\"center\">10</td><td align=\"center\">95.35</td><td align=\"center\">19</td><td align=\"center\">95.35</td><td align=\"center\">36</td><td align=\"center\">95.35</td><td align=\"center\">33</td></tr><tr><td align=\"center\">150</td><td align=\"center\">53.49</td><td align=\"center\">2</td><td align=\"center\">90.70</td><td align=\"center\">5</td><td align=\"center\">93.02</td><td align=\"center\">10</td><td align=\"center\">93.02</td><td align=\"center\">20</td><td align=\"center\">93.02</td><td align=\"center\">36</td><td align=\"center\">93.02</td><td align=\"center\">28</td></tr><tr><td align=\"center\">200</td><td align=\"center\">79.07</td><td align=\"center\">2</td><td align=\"center\">93.02</td><td align=\"center\">5</td><td align=\"center\">93.02</td><td align=\"center\">10</td><td align=\"center\">93.02</td><td align=\"center\">20</td><td align=\"center\">93.02</td><td align=\"center\">36</td><td align=\"center\">93.02</td><td align=\"center\">35</td></tr><tr><td align=\"center\">250</td><td align=\"center\">76.74</td><td align=\"center\">2</td><td align=\"center\">95.35</td><td align=\"center\">5</td><td align=\"center\">95.35</td><td align=\"center\">10</td><td align=\"center\">95.35</td><td align=\"center\">20</td><td align=\"center\">95.35</td><td align=\"center\">35</td><td align=\"center\">93.02</td><td align=\"center\">32</td></tr><tr><td align=\"center\">300</td><td align=\"center\">69.77</td><td align=\"center\">2</td><td align=\"center\">93.02</td><td align=\"center\">5</td><td align=\"center\">93.02</td><td align=\"center\">10</td><td align=\"center\">95.35</td><td align=\"center\">20</td><td align=\"center\">95.35</td><td align=\"center\">34</td><td align=\"center\">95.35</td><td align=\"center\">35</td></tr><tr><td align=\"center\">350</td><td align=\"center\">67.44</td><td align=\"center\">2</td><td align=\"center\">93.02</td><td align=\"center\">5</td><td align=\"center\">93.02</td><td align=\"center\">10</td><td align=\"center\">93.02</td><td align=\"center\">20</td><td align=\"center\">93.02</td><td align=\"center\">35</td><td align=\"center\">93.02</td><td align=\"center\">35</td></tr><tr><td align=\"center\">400</td><td align=\"center\">69.77</td><td align=\"center\">2</td><td align=\"center\">93.02</td><td align=\"center\">5</td><td align=\"center\">93.02</td><td align=\"center\">10</td><td align=\"center\">93.02</td><td align=\"center\">20</td><td align=\"center\">93.02</td><td align=\"center\">37</td><td align=\"center\">95.35</td><td align=\"center\">35</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Classification accuracy using mixed-genome array data with non-hierarchical clustering for four sample (bacterial species) clusters. PA is the prediction accuracy.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"12\">Number of clusters of probes, <italic>κ</italic></td></tr><tr><td/><td colspan=\"12\"><hr/></td></tr><tr><td/><td align=\"center\" colspan=\"2\">2</td><td align=\"center\" colspan=\"2\">5</td><td align=\"center\" colspan=\"2\">10</td><td align=\"center\" colspan=\"2\">20</td><td align=\"center\" colspan=\"2\">30</td><td align=\"center\" colspan=\"2\">40</td></tr><tr><td/><td colspan=\"12\"><hr/></td></tr><tr><td align=\"center\">Number of top-ranked probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td></tr></thead><tbody><tr><td align=\"center\">50</td><td align=\"center\">100</td><td align=\"center\">2</td><td align=\"center\">100</td><td align=\"center\">5</td><td align=\"center\">100</td><td align=\"center\">7</td><td align=\"center\">100</td><td align=\"center\">7</td><td align=\"center\">100</td><td align=\"center\">1</td><td align=\"center\">75</td><td align=\"center\">1</td></tr><tr><td align=\"center\">100</td><td align=\"center\">100</td><td align=\"center\">2</td><td align=\"center\">100</td><td align=\"center\">5</td><td align=\"center\">100</td><td align=\"center\">7</td><td align=\"center\">100</td><td align=\"center\">7</td><td align=\"center\">100</td><td align=\"center\">7</td><td align=\"center\">100</td><td align=\"center\">1</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Classification accuracy using ALL/AML leukemia data. PA is the prediction accuracy.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"12\">Number of clusters of probes, <italic>κ</italic></td></tr><tr><td/><td colspan=\"12\"><hr/></td></tr><tr><td/><td align=\"center\" colspan=\"2\">2</td><td align=\"center\" colspan=\"2\">5</td><td align=\"center\" colspan=\"2\">10</td><td align=\"center\" colspan=\"2\">20</td><td align=\"center\" colspan=\"2\">30</td><td align=\"center\" colspan=\"2\">40</td></tr><tr><td/><td colspan=\"12\"><hr/></td></tr><tr><td align=\"center\">Number of top-ranked probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td><td align=\"center\">PA (%)</td><td align=\"center\">No. of probes</td></tr></thead><tbody><tr><td align=\"center\">50</td><td align=\"center\">94.44</td><td align=\"center\">2</td><td align=\"center\">94.44</td><td align=\"center\">5</td><td align=\"center\">97.22</td><td align=\"center\">10</td><td align=\"center\">97.22</td><td align=\"center\">20</td><td align=\"center\">97.22</td><td align=\"center\">30</td><td align=\"center\">97.22</td><td align=\"center\">39</td></tr><tr><td align=\"center\">100</td><td align=\"center\">88.89</td><td align=\"center\">2</td><td align=\"center\">95.83</td><td align=\"center\">5</td><td align=\"center\">94.44</td><td align=\"center\">10</td><td align=\"center\">97.22</td><td align=\"center\">20</td><td align=\"center\">97.22</td><td align=\"center\">30</td><td align=\"center\">97.22</td><td align=\"center\">40</td></tr><tr><td align=\"center\">150</td><td align=\"center\">83.33</td><td align=\"center\">2</td><td align=\"center\">95.83</td><td align=\"center\">5</td><td align=\"center\">97.22</td><td align=\"center\">10</td><td align=\"center\">97.22</td><td align=\"center\">20</td><td align=\"center\">97.22</td><td align=\"center\">30</td><td align=\"center\">97.22</td><td align=\"center\">39</td></tr><tr><td align=\"center\">200</td><td align=\"center\">79.17</td><td align=\"center\">2</td><td align=\"center\">80.56</td><td align=\"center\">5</td><td align=\"center\">97.22</td><td align=\"center\">10</td><td align=\"center\">97.22</td><td align=\"center\">20</td><td align=\"center\">97.22</td><td align=\"center\">30</td><td align=\"center\">97.22</td><td align=\"center\">39</td></tr><tr><td align=\"center\">250</td><td align=\"center\">79.17</td><td align=\"center\">2</td><td align=\"center\">79.17</td><td align=\"center\">5</td><td align=\"center\">97.22</td><td align=\"center\">10</td><td align=\"center\">97.22</td><td align=\"center\">20</td><td align=\"center\">97.22</td><td align=\"center\">30</td><td align=\"center\">97.22</td><td align=\"center\">39</td></tr></tbody></table></table-wrap>" ]

[ "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1471-2105-9-328-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>f</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>P</mml:mi><mml:mo>|</mml:mo><mml:mtext>X</mml:mtext><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>f</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mtext>X</mml:mtext><mml:mo>|</mml:mo><mml:mi>P</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>f</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>P</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>f</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mtext>X</mml:mtext><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M2\" name=\"1471-2105-9-328-i3\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>σ</mml:mi><mml:mi>i</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M3\" name=\"1471-2105-9-328-i3\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>σ</mml:mi><mml:mi>i</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM1\"><label>(1)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M4\" name=\"1471-2105-9-328-i4\" overflow=\"scroll\">\n <mml:semantics>\n <mml:mrow>\n <mml:mi>ℒ</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mtext>X</mml:mtext>\n <mml:mo>|</mml:mo>\n <mml:mi>P</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n <mml:mo>=</mml:mo>\n <mml:mstyle displaystyle=\"true\">\n <mml:munder>\n <mml:mo>∏</mml:mo>\n <mml:mi>k</mml:mi>\n </mml:munder>\n <mml:mrow>\n <mml:mstyle displaystyle=\"true\">\n <mml:munder>\n <mml:mo>∏</mml:mo>\n <mml:mi>j</mml:mi>\n </mml:munder>\n <mml:mrow>\n <mml:mfrac>\n <mml:mrow>\n <mml:mn>2</mml:mn>\n <mml:msubsup>\n <mml:mi>σ</mml:mi>\n <mml:mn>0</mml:mn>\n <mml:mn>2</mml:mn>\n </mml:msubsup>\n </mml:mrow>\n <mml:mrow>\n <mml:mi>Γ</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mn>1</mml:mn>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n </mml:mfrac>\n <mml:mfrac>\n <mml:mrow>\n <mml:msup>\n <mml:mrow>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mn>2</mml:mn>\n <mml:mi>π</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n <mml:mrow>\n <mml:mo>−</mml:mo>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:msub>\n <mml:mi>n</mml:mi>\n <mml:mi>k</mml:mi>\n </mml:msub>\n <mml:mo>/</mml:mo>\n <mml:mn>2</mml:mn>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n </mml:msup>\n </mml:mrow>\n <mml:mrow>\n <mml:msqrt>\n <mml:mrow>\n <mml:msub>\n <mml:mi>n</mml:mi>\n <mml:mi>k</mml:mi>\n </mml:msub>\n <mml:mo>+</mml:mo>\n <mml:mn>1</mml:mn>\n </mml:mrow>\n </mml:msqrt>\n </mml:mrow>\n </mml:mfrac>\n <mml:mfrac>\n <mml:mrow>\n <mml:mi>Γ</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:msub>\n <mml:mi>n</mml:mi>\n <mml:mi>k</mml:mi>\n </mml:msub>\n <mml:mo>/</mml:mo>\n <mml:mn>2</mml:mn>\n <mml:mo>+</mml:mo>\n <mml:mn>1</mml:mn>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n <mml:mrow>\n <mml:msup>\n <mml:mrow>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mn>2</mml:mn>\n <mml:msubsup>\n <mml:mi>σ</mml:mi>\n <mml:mn>0</mml:mn>\n <mml:mn>2</mml:mn>\n </mml:msubsup>\n <mml:mo>+</mml:mo>\n <mml:mn>0.5</mml:mn>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mstyle displaystyle=\"true\">\n <mml:msub>\n <mml:mo>∑</mml:mo>\n <mml:mi>i</mml:mi>\n </mml:msub>\n <mml:mrow>\n <mml:msubsup>\n <mml:mi>x</mml:mi>\n <mml:mrow>\n <mml:msub>\n <mml:mi>k</mml:mi>\n <mml:mi>i</mml:mi>\n </mml:msub>\n <mml:mi>j</mml:mi>\n </mml:mrow>\n <mml:mn>2</mml:mn>\n </mml:msubsup>\n <mml:mo>+</mml:mo>\n <mml:msubsup>\n <mml:mi>μ</mml:mi>\n <mml:mn>0</mml:mn>\n <mml:mn>2</mml:mn>\n </mml:msubsup>\n </mml:mrow>\n </mml:mstyle>\n <mml:mo>−</mml:mo>\n <mml:mfrac>\n <mml:mrow>\n <mml:mstyle displaystyle=\"true\">\n <mml:msub>\n <mml:mo>∑</mml:mo>\n <mml:mi>i</mml:mi>\n </mml:msub>\n <mml:mrow>\n <mml:msup>\n <mml:mrow>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:msub>\n <mml:mi>x</mml:mi>\n <mml:mrow>\n <mml:msub>\n <mml:mi>k</mml:mi>\n <mml:mi>i</mml:mi>\n </mml:msub>\n <mml:mi>j</mml:mi>\n </mml:mrow>\n </mml:msub>\n <mml:mo>+</mml:mo>\n <mml:msub>\n <mml:mi>μ</mml:mi>\n <mml:mn>0</mml:mn>\n </mml:msub>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n <mml:mn>2</mml:mn>\n </mml:msup>\n </mml:mrow>\n </mml:mstyle>\n </mml:mrow>\n <mml:mrow>\n <mml:msub>\n <mml:mi>n</mml:mi>\n <mml:mi>k</mml:mi>\n </mml:msub>\n <mml:mo>+</mml:mo>\n <mml:mn>1</mml:mn>\n </mml:mrow>\n </mml:mfrac>\n <mml:mo stretchy=\"false\">)</mml:mo>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n <mml:mrow>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:msub>\n <mml:mi>n</mml:mi>\n <mml:mi>k</mml:mi>\n </mml:msub>\n <mml:mo>/</mml:mo>\n <mml:mn>2</mml:mn>\n <mml:mo>+</mml:mo>\n <mml:mn>1</mml:mn>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n </mml:msup>\n </mml:mrow>\n </mml:mfrac>\n </mml:mrow>\n </mml:mstyle>\n </mml:mrow>\n </mml:mstyle>\n </mml:mrow>\n \n </mml:semantics>\n </mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M5\" name=\"1471-2105-9-328-i5\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>σ</mml:mi><mml:mn>0</mml:mn><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M6\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M7\" name=\"1471-2105-9-328-i7\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">F</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M8\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M9\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M10\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M11\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M12\" name=\"1471-2105-9-328-i8\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi mathvariant=\"script\">F</mml:mi><mml:mrow><mml:mi>r</mml:mi><mml:mi>e</mml:mi><mml:mi>m</mml:mi><mml:mi>o</mml:mi><mml:mi>v</mml:mi><mml:mi>e</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M13\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M14\" name=\"1471-2105-9-328-i9\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi mathvariant=\"script\">F</mml:mi><mml:mrow><mml:mi>e</mml:mi><mml:mi>n</mml:mi><mml:mi>t</mml:mi><mml:mi>e</mml:mi><mml:mi>r</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M15\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M16\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M17\" name=\"1471-2105-9-328-i8\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi mathvariant=\"script\">F</mml:mi><mml:mrow><mml:mi>r</mml:mi><mml:mi>e</mml:mi><mml:mi>m</mml:mi><mml:mi>o</mml:mi><mml:mi>v</mml:mi><mml:mi>e</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M18\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M19\" name=\"1471-2105-9-328-i9\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi mathvariant=\"script\">F</mml:mi><mml:mrow><mml:mi>e</mml:mi><mml:mi>n</mml:mi><mml:mi>t</mml:mi><mml:mi>e</mml:mi><mml:mi>r</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M20\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M21\" name=\"1471-2105-9-328-i7\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">F</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M22\" name=\"1471-2105-9-328-i7\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">F</mml:mi></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM2\"><label>(2)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M23\" name=\"1471-2105-9-328-i10\" overflow=\"scroll\">\n <mml:semantics>\n <mml:mrow>\n <mml:msub>\n <mml:mi mathvariant=\"script\">F</mml:mi>\n <mml:mrow>\n <mml:mi>r</mml:mi>\n <mml:mi>e</mml:mi>\n <mml:mi>m</mml:mi>\n <mml:mi>o</mml:mi>\n <mml:mi>v</mml:mi>\n <mml:mi>e</mml:mi>\n </mml:mrow>\n </mml:msub>\n <mml:mo>=</mml:mo>\n <mml:mfrac>\n <mml:mrow>\n <mml:mi>n</mml:mi>\n <mml:mo>−</mml:mo>\n <mml:mi>r</mml:mi>\n <mml:mo>−</mml:mo>\n <mml:mi>q</mml:mi>\n <mml:mo>+</mml:mo>\n <mml:mn>1</mml:mn>\n </mml:mrow>\n <mml:mrow>\n <mml:mi>q</mml:mi>\n <mml:mo>−</mml:mo>\n <mml:mn>1</mml:mn>\n </mml:mrow>\n </mml:mfrac>\n <mml:mfrac>\n <mml:mrow>\n <mml:mi>Λ</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi mathvariant=\"script\">G</mml:mi>\n <mml:mo>\\</mml:mo>\n <mml:mi>p</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n <mml:mo>−</mml:mo>\n <mml:mi>Λ</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi mathvariant=\"script\">G</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n <mml:mrow>\n <mml:mi>Λ</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi mathvariant=\"script\">G</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n </mml:mfrac>\n </mml:mrow>\n \n </mml:semantics>\n </mml:math></disp-formula>", "<disp-formula id=\"bmcM3\"><label>(3)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M24\" name=\"1471-2105-9-328-i11\" overflow=\"scroll\">\n <mml:semantics>\n <mml:mrow>\n <mml:msub>\n <mml:mi mathvariant=\"script\">F</mml:mi>\n <mml:mrow>\n <mml:mi>e</mml:mi>\n <mml:mi>n</mml:mi>\n <mml:mi>t</mml:mi>\n <mml:mi>e</mml:mi>\n <mml:mi>r</mml:mi>\n </mml:mrow>\n </mml:msub>\n <mml:mo>=</mml:mo>\n <mml:mfrac>\n <mml:mrow>\n <mml:mi>n</mml:mi>\n <mml:mo>−</mml:mo>\n <mml:mi>r</mml:mi>\n <mml:mo>−</mml:mo>\n <mml:mi>q</mml:mi>\n </mml:mrow>\n <mml:mrow>\n <mml:mi>q</mml:mi>\n <mml:mo>−</mml:mo>\n <mml:mn>1</mml:mn>\n </mml:mrow>\n </mml:mfrac>\n <mml:mfrac>\n <mml:mrow>\n <mml:mi>Λ</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi mathvariant=\"script\">G</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n <mml:mo>−</mml:mo>\n <mml:mi>Λ</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi mathvariant=\"script\">G</mml:mi>\n <mml:mo>|</mml:mo>\n <mml:mi>p</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n <mml:mrow>\n <mml:mi>Λ</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi mathvariant=\"script\">G</mml:mi>\n <mml:mo>|</mml:mo>\n <mml:mi>p</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n </mml:mfrac>\n </mml:mrow>\n \n </mml:semantics>\n </mml:math></disp-formula>", "<disp-formula id=\"bmcM4\"><label>(4)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M25\" name=\"1471-2105-9-328-i12\" overflow=\"scroll\">\n <mml:semantics>\n <mml:mrow>\n <mml:mi>Λ</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi mathvariant=\"script\">G</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n <mml:mo>=</mml:mo>\n <mml:mfrac>\n <mml:mrow>\n <mml:mi>det</mml:mi>\n <mml:mo>⁡</mml:mo>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi>W</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n <mml:mrow>\n <mml:mi>det</mml:mi>\n <mml:mo>⁡</mml:mo>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi>T</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n </mml:mfrac>\n </mml:mrow>\n \n </mml:semantics>\n </mml:math></disp-formula>", "<disp-formula id=\"bmcM5\"><label>(5)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M26\" name=\"1471-2105-9-328-i13\" overflow=\"scroll\">\n <mml:semantics>\n <mml:mrow>\n <mml:mi>W</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi mathvariant=\"script\">G</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n <mml:mo>=</mml:mo>\n <mml:mstyle displaystyle=\"true\">\n <mml:munderover>\n <mml:mo>∑</mml:mo>\n <mml:mrow>\n <mml:mi>m</mml:mi>\n <mml:mo>=</mml:mo>\n <mml:mn>1</mml:mn>\n </mml:mrow>\n <mml:mi>q</mml:mi>\n </mml:munderover>\n <mml:mrow>\n <mml:mstyle displaystyle=\"true\">\n <mml:munderover>\n <mml:mo>∑</mml:mo>\n <mml:mrow>\n <mml:mi>m</mml:mi>\n <mml:mo>=</mml:mo>\n <mml:mn>1</mml:mn>\n </mml:mrow>\n <mml:mrow>\n <mml:msub>\n <mml:mi>n</mml:mi>\n <mml:mi>m</mml:mi>\n </mml:msub>\n </mml:mrow>\n </mml:munderover>\n <mml:mrow>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:msub>\n <mml:mi>x</mml:mi>\n <mml:mrow>\n <mml:mi>m</mml:mi>\n <mml:mi>k</mml:mi>\n <mml:mi>i</mml:mi>\n </mml:mrow>\n </mml:msub>\n <mml:mo>−</mml:mo>\n <mml:msub>\n <mml:mi>x</mml:mi>\n <mml:mrow>\n <mml:mi>m</mml:mi>\n <mml:mi>i</mml:mi>\n <mml:mo>.</mml:mo>\n </mml:mrow>\n </mml:msub>\n <mml:mo stretchy=\"false\">)</mml:mo>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:msub>\n <mml:mi>x</mml:mi>\n <mml:mrow>\n <mml:mi>m</mml:mi>\n <mml:mi>k</mml:mi>\n <mml:mi>j</mml:mi>\n </mml:mrow>\n </mml:msub>\n <mml:mo>−</mml:mo>\n <mml:msub>\n <mml:mi>x</mml:mi>\n <mml:mrow>\n <mml:mi>m</mml:mi>\n <mml:mi>j</mml:mi>\n <mml:mo>.</mml:mo>\n </mml:mrow>\n </mml:msub>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n </mml:mstyle>\n </mml:mrow>\n </mml:mstyle>\n </mml:mrow>\n \n </mml:semantics>\n </mml:math></disp-formula>", "<disp-formula id=\"bmcM6\"><label>(6)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M27\" name=\"1471-2105-9-328-i14\" overflow=\"scroll\">\n <mml:semantics>\n <mml:mrow>\n <mml:mi>T</mml:mi>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:mi mathvariant=\"script\">G</mml:mi>\n <mml:mo stretchy=\"false\">)</mml:mo>\n <mml:mo>=</mml:mo>\n <mml:mstyle displaystyle=\"true\">\n <mml:munderover>\n <mml:mo>∑</mml:mo>\n <mml:mrow>\n <mml:mi>m</mml:mi>\n <mml:mo>=</mml:mo>\n <mml:mn>1</mml:mn>\n </mml:mrow>\n <mml:mi>q</mml:mi>\n </mml:munderover>\n <mml:mrow>\n <mml:mstyle displaystyle=\"true\">\n <mml:munderover>\n <mml:mo>∑</mml:mo>\n <mml:mrow>\n <mml:mi>m</mml:mi>\n <mml:mo>=</mml:mo>\n <mml:mn>1</mml:mn>\n </mml:mrow>\n <mml:mrow>\n <mml:msub>\n <mml:mi>n</mml:mi>\n <mml:mi>m</mml:mi>\n </mml:msub>\n </mml:mrow>\n </mml:munderover>\n <mml:mrow>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:msub>\n <mml:mi>x</mml:mi>\n <mml:mrow>\n <mml:mi>m</mml:mi>\n <mml:mi>k</mml:mi>\n <mml:mi>i</mml:mi>\n </mml:mrow>\n </mml:msub>\n <mml:mo>−</mml:mo>\n <mml:msub>\n <mml:mi>x</mml:mi>\n <mml:mrow>\n <mml:mi>i</mml:mi>\n <mml:mn>..</mml:mn>\n </mml:mrow>\n </mml:msub>\n <mml:mo stretchy=\"false\">)</mml:mo>\n <mml:mo stretchy=\"false\">(</mml:mo>\n <mml:msub>\n <mml:mi>x</mml:mi>\n <mml:mrow>\n <mml:mi>m</mml:mi>\n <mml:mi>k</mml:mi>\n <mml:mi>j</mml:mi>\n </mml:mrow>\n </mml:msub>\n <mml:mo>−</mml:mo>\n <mml:msub>\n <mml:mi>x</mml:mi>\n <mml:mrow>\n <mml:mi>j</mml:mi>\n <mml:mn>..</mml:mn>\n </mml:mrow>\n </mml:msub>\n <mml:mo stretchy=\"false\">)</mml:mo>\n </mml:mrow>\n </mml:mstyle>\n </mml:mrow>\n </mml:mstyle>\n </mml:mrow>\n \n </mml:semantics>\n </mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M28\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M29\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M30\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M31\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M32\" name=\"1471-2105-9-328-i6\" overflow=\"scroll\"><mml:semantics><mml:mi mathvariant=\"script\">G</mml:mi></mml:semantics></mml:math></inline-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M33\" name=\"1471-2105-9-328-i15\" overflow=\"scroll\">\n <mml:semantics>\n <mml:mrow>\n <mml:mi>S</mml:mi>\n <mml:mi>F</mml:mi>\n <mml:mi>R</mml:mi>\n <mml:mo>=</mml:mo>\n <mml:mfrac>\n <mml:mrow>\n <mml:mi>n</mml:mi>\n <mml:mi>u</mml:mi>\n <mml:mi>m</mml:mi>\n <mml:mi>b</mml:mi>\n <mml:mi>e</mml:mi>\n <mml:mi>r</mml:mi>\n <mml:mtext> </mml:mtext>\n <mml:mi>o</mml:mi>\n <mml:mi>f</mml:mi>\n <mml:mtext> </mml:mtext>\n <mml:mi>s</mml:mi>\n <mml:mi>a</mml:mi>\n <mml:mi>m</mml:mi>\n <mml:mi>p</mml:mi>\n <mml:mi>l</mml:mi>\n <mml:mi>e</mml:mi>\n <mml:mi>s</mml:mi>\n </mml:mrow>\n <mml:mrow>\n <mml:mi>n</mml:mi>\n <mml:mi>u</mml:mi>\n <mml:mi>m</mml:mi>\n <mml:mi>b</mml:mi>\n <mml:mi>e</mml:mi>\n <mml:mi>r</mml:mi>\n <mml:mtext> </mml:mtext>\n <mml:mi>o</mml:mi>\n <mml:mi>f</mml:mi>\n <mml:mtext> </mml:mtext>\n <mml:mi>f</mml:mi>\n <mml:mi>e</mml:mi>\n <mml:mi>a</mml:mi>\n <mml:mi>t</mml:mi>\n <mml:mi>u</mml:mi>\n <mml:mi>r</mml:mi>\n <mml:mi>e</mml:mi>\n <mml:mi>s</mml:mi>\n </mml:mrow>\n </mml:mfrac>\n <mml:mo>≤</mml:mo>\n <mml:mfrac>\n <mml:mn>1</mml:mn>\n <mml:mn>5</mml:mn>\n </mml:mfrac>\n <mml:mo>.</mml:mo>\n </mml:mrow>\n \n </mml:semantics>\n </mml:math></disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p><bold>MPM</bold>. This file contains the information for the mixed-plasmid microarray in the form of a 44-row × 577-column matrix. The first row gives the 576 probe names, and the first column gives the 43 plasmid names. The remaining rows and columns contain the hybridization intensity data.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional file 2</title><p><bold>MGM</bold>. This file contains the information for the mixed-genome virtual microarray in the form of a 15-row × 4001-column matrix where the first column gives the 15 names of the streptococcus strains. The probes were not given specific names. All remaining columns contain the virtual hybridization intensity data.</p></caption></supplementary-material>" ]

[]

[ "<graphic xlink:href=\"1471-2105-9-328-1\"/>", "<graphic xlink:href=\"1471-2105-9-328-2\"/>", "<graphic xlink:href=\"1471-2105-9-328-3\"/>" ]

[ "<media xlink:href=\"1471-2105-9-328-S1.txt\" mimetype=\"text\" mime-subtype=\"plain\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2105-9-328-S2.txt\" mimetype=\"text\" mime-subtype=\"plain\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The prevalence of gestational diabetes mellitus (GDM) is increasing all over the world [##REF##15735191##1##,##REF##15339790##2##]. In Australia the recent prevalence estimates for GDM ranged from 5.2% to 8.8% [##REF##12832303##3##]. The risks for both mothers with GDM and their infants are well-documented. For the infants, these include an increased risk of macrosomia, birth injuries such as shoulder dystocia, bone fracture and nerve palsies, hypoglycaemia, and hyperbilirubinaemia [##REF##4069477##4##, ####REF##15846171##5##, ##REF##8166187##6##, ##REF##729447##7####729447##7##]. Women with GDM are at increased risk of developing pre-eclampsia and have an increased chance of need for induction of labour and caesarean section. Gestational diabetes is also a strong risk factor for later development of type 2 diabetes [##REF##12351492##8##].</p>", "<p>Although the risks associated with GDM are well recognised, the impact on maternal and infant health outcomes is less clear for borderline gestational diabetes mellitus (BGDM), which is characterised by values of glucose tolerance intermediate between normal and gestational diabetes. A recent 10 year audit examining the influence of different levels of glucose tolerance on pregnancy complications, [##REF##17627686##9##] revealed a significantly increased risk of pre-eclampsia, caesarean section, neonatal hypoglycaemia and hyperbilirubinaemia for women with BGDM compared with women with normal glucose tolerance. The results are consistent with other literature reports, which identified an increasing risk of adverse maternal and infant outcomes with increasing plasma glucose values [##REF##7631672##10##, ####REF##9704225##11##, ##REF##15547538##12####15547538##12##].</p>", "<p>It is estimated that 6.6% of pregnant women or approximately 16,500 women have BGDM each year in Australia [##REF##17627686##9##]. Given the uncertainty surrounding BGDM, we assessed data from participants in the Australian Collaborative Trial of Supplements with antioxidants Vitamin C and Vitamin E to pregnant women for the prevention of pre-eclampsia (ACTS) [##REF##16641396##13##] to compare the maternal demographic, pregnancy and infant health outcomes of women who had BGDM (screened positive for GDM on oral glucose challenge test (OGCT) but their subsequent oral glucose tolerance test (OGTT) was normal) with women who screened negative on OGCT for GDM.</p>" ]

[ "<title>Methods</title>", "<p>The study population included women participating in the ACTS trial [##REF##16641396##13##], a multi-centre randomised placebo controlled trial of antioxidant (vitamins C and E) supplements for the prevention of perinatal complications, who had an OGCT as screening for gestational diabetes. The methods and results of this trial have been reported previously [##REF##16641396##13##]. Briefly, eligible women were: nulliparous, with a singleton pregnancy between 14 and 22 weeks of gestation with a normal blood pressure at the time of recruitment and who gave informed consent. Women with any of the following were ineligible: known multiple pregnancy, known lethal fetal anomaly, known thrombophilia, chronic renal failure, antihypertensive therapy or contraindication to vitamin C or E therapy including haemochromatosis or anticoagulant therapy. Randomisation was performed through a central telephone randomisation service. Women assigned to the vitamin group were provided a daily dose of 1000 mg vitamin C and 400 IU vitamin E until birth, and women in the control group were provided a matching placebo. An OGTT was offered between 24–30 weeks gestation, for those women who screened positive on OGCT test. The study protocol was approved by the research and ethics committees at the nine collaboration hospitals around Australia.</p>", "<p>We compared demographic, obstetric and neonatal outcomes between women with BGDM and those who screened normal on OGCT. As the ACTS found no significant differences between the antioxidant and placebo groups for the risk of pre-eclampsia, intrauterine growth restriction or other serious outcomes for the infant, the analyses include the combined populations of women who received either antioxidant or placebo supplements.</p>", "<title>Data collection</title>", "<p>Pregnancy outcome data including OGCT and OGTT results were collected prospectively from women's medical records. Sociodemographic variables were collected either from women's medical records or self-completed questionnaires at trial entry and included: maternal age, ethnicity, body mass index (BMI), social-economic status as measured by socio-economic index for area (SEIFA) score [##UREF##0##14##], maternal education, smoking status, blood pressure at trial entry, and family history of pre-eclampsia. Complete outcome data were available for all 1877 women randomised.</p>", "<title>Outcome variables</title>", "<p>BGDM was defined as a positive OGCT (blood glucose ≥7.8 mmol/L 1 hour after a 50 g glucose load) and normal 75 g OGTT (fasting blood glucose &lt;5.5 mmol/L and 2 hour blood glucose &lt;7.8 mmol/L). Pregnancy outcomes assessed included: maternal adverse outcomes (a composite outcome defined as any of the following until six weeks postpartum: death, pulmonary oedema, eclampsia, stroke, thrombocytopenia, renal insufficiency, respiratory arrest, placental abruption, abnormal liver function, preterm prelabour rupture of membranes, major postpartum haemorrhage, postpartum pyrexia, pneumonia, deep-vein thrombosis, or pulmonary embolus requiring anticoagulant therapy) [##REF##16641396##13##]; pregnancy induced hypertension (PIH); pre-eclampsia (defined as systolic blood pressure ≥140 mmHg or diastolic blood pressure [Korokoff V] ≥90 mmHg on at least two occasions four or more hours apart, or both arising after 20 weeks' gestation and one or more of the following: proteinuria, renal insufficiency, liver disease, neurological problems, haematologic disturbances, or fetal growth restriction) [##REF##10925900##15##]; antenatal hospitalisation; preterm prelabour rupture of the membranes; induction of labour; mode of birth; postnatal complications such as postpartum haemorrhage and infection; and length of hospital stay.</p>", "<p>Neonatal outcomes included a composite outcome of death or infant adverse outcome defined as: stillbirth or death of a liveborn infant before hospital discharge, birthweight &lt;3^rdcentile for gestational age, severe respiratory distress syndrome, chronic lung disease, intraventricular haemorrhage grade 3 or 4, cystic periventricular leukomalacia, retinopathy of prematurity grade 3 or 4, necrotizing enterocolitis, 5 minute Apgar score &lt;4, seizures before 24 hours of age or requiring 2 or more drugs to control, hypotonia for ≥2 hours, stupor, decreased response to pain or coma, tube feeding for ≥4 days, care in the neonatal intensive care unit (NICU) &gt;4 days, or use of ventilation for ≥24 hours [##REF##16641396##13##]; gestational age at birth; preterm birth (&lt;37 weeks); 5 minute Apgar score &lt;7, infant body size at birth (weight, length and head circumference), small and large-for-gestational age (defined as a birth weight below the 10^thpercentile or above 90^thpercentile for gestation according to fetal sex on standardized birth-weight charts, respectively), macrosomia (defined as birthweight ≥4.5 kg), admission to NICU or neonatal nursery, respiratory distress syndrome, mechanical ventilation, antibiotics use after birth, encephalopathy (Sarnat 2 or 3 score) and length of hospital stay.</p>", "<title>Statistical analysis</title>", "<p>Statistical analysis was carried out using SAS software, version 9.1. Dichotomous variables were analysed using log-binomial regression and presented as relative risks, with 95% confidence intervals; and continuous variables, if normally distributed, were analysed using Student's t-test and presented as mean differences, with 95% confidence intervals; non-parametric tests were used for skewed data. Analyses were then adjusted for maternal age and BMI given the strong association of these factors with GDM. A p value of 0.05 or less was considered to indicate statistical significance.</p>" ]

[ "<title>Results</title>", "<p>Of the 1877 women enrolled in the ACTS trial, 1804 (96%) did not have a fetal loss and underwent screening using a 50 g oral glucose challenge test for gestational diabetes. Of the women screened 1596 (88%) had a normal OGCT screening result, 68 (4%) had an abnormal OGTT and 140 (8%) had BGDM (screened positive on OGCT, normal OGTT).</p>", "<p>Overall, women with BGDM and women with a normal OGCT had similar characteristics at entry to the study including ethnicity, socio-economic status and educational attainment (Table ##TAB##0##1##). Compared with women with a normal OGCT, women with BGDM were older (mean difference 1.3 years, [95%CI 0.3, 2.2], p = 0.01), less likely to have a normal BMI (RR 0.82, [95%CI 0.67, 0.99], p = 0.04) and almost twice as likely to be obese (RR 1.92, [95%CI 1.41, 2.62], p &lt; 0.0001) (Table ##TAB##0##1##). There was no statistically significant difference found between these groups in the number of women who smoked or in their mean systolic or diastolic blood pressure at study entry.</p>", "<p>In unadjusted analyses women with BGDM were more likely to experience a maternal adverse outcome (RR 1.59, [95%CI 1.00, 2.52], p = 0.05) and to develop pregnancy induced hypertension (RR 1.51, [95%CI 1.03, 2.20], p = 0.03) compared with women with a normal OGCT. These differences were not seen when adjustment was made for maternal age and BMI. There was no significant difference in the rate of pre-eclampsia between the two comparison groups (Table ##TAB##1##2##).</p>", "<p>The rate of induction of labour was similar and the overall caesarean section rate did not differ between groups. In the unadjusted analyses significantly more women with BGDM gave birth by caesarean section for fetal distress (RR 1.63, [95%CI 1.10, 2.41], p = 0.01) compared with women with a normal OGCT although this was not significant in the adjusted analyses. The length of postnatal hospital stay was significantly longer (mean difference 0.4 days, [95%CI 0.1, 0.7], p = 0.01) for women with BGDM compared to women with normal OGCT in the unadjusted analyses but not when adjusted for maternal age and BMI (Table ##TAB##1##2##).</p>", "<p>Overall there was no difference in the risk of death or infant adverse outcome between the two groups (Table ##TAB##2##3##). In unadjusted analyses infants born to women with BGDM were at increased risk of being born preterm (RR 1.68, [95%CI 1.00, 2.80], p = 0.05) and were also significantly more likely to be macrosomic (birthweight ≥4.5 kg) (RR 2.53, [95%CI 1.06, 6.03], p = 0.04) compared with infants born to women with a normal OGCT. When adjusted for maternal age and BMI the association with an earlier gestational age at birth and the risk of being macrosomic remained for infants born to women with BGDM compared with infants born to women with a normal OGCT (Table ##TAB##2##3##).</p>", "<p>The hospital stay was significantly longer (unadjusted p = 0.001, adjusted p = 0.01) for infants born to BGDM mothers compared with infants born to mothers with a normal OGCT (Table ##TAB##2##3##). Infants born to BGDM mothers were more than twice as likely to be admitted to NICU (unadjusted p = 0.03, adjusted p = 0.04) and more likely to be admitted to the neonatal nursery (unadjusted p = 0.002, adjusted p = 0.01). Antibiotic use less than 48 hours after birth was significantly greater among infants born to BGDM mothers (unadjusted and adjusted p = 0.01) and more infants born to the BGDM women had Sarnat stage 2 or 3 encephalopathy (unadjusted and adjusted p = 0.03) compared with infants born to women with a normal OGCT (Table ##TAB##2##3##).</p>" ]

[ "<title>Discussion</title>", "<p>In this cohort of primiparous women in Australia, 8% were found to have BGDM. In this study, associations with BGDM were identified for maternal obesity and increasing maternal age, similar to those identified for gestational diabetes in other literature [##REF##12020334##16##, ####REF##3996763##17##, ##REF##11284643##18####11284643##18##].</p>", "<p>In our study, women with BGDM had a higher risk of adverse health outcomes overall, and were more likely to develop pregnancy induced hypertension, require a caesarean section for fetal distress and have a longer postnatal hospital stay. However, we did not detect a statistically significant increase in the risk of pre-eclampsia or caesarean section overall among women with BGDM, which has been reported by previous studies [##REF##17627686##9##, ####REF##7631672##10##, ##REF##9704225##11####9704225##11##]. Increasing maternal age and BMI are strongly associated with adverse maternal health outcomes. When these factors were adjusted for, no differences were seen for health outcomes between women with BGDM and normal women.</p>", "<p>We identified an increased risk of preterm birth amongst BGDM mothers. The reason for this is not readily apparent, given that there is no difference in the rate of induction of labour between the two groups. Infants of BGDM mothers were more likely to require a NICU and/or nursery admission and longer hospital stays. This may be explained by the higher rate of pregnancy induced hypertension, caesarean section for fetal distress, preterm birth and encephalopathy (Sarnat stage 2 or 3) in this group. Infants born to BGDM mothers in both unadjusted analyses and when adjusted for maternal age and maternal BMI were also at higher risk of macrosomia, which is consistent with previous studies [##REF##7631672##10##,##REF##9704225##11##].</p>", "<p>Our study has identified increased risks of maternal adverse health outcomes overall and a range of infant adverse health outcomes associated with BGDM. In Australia, there are over 250,000 births annually [##UREF##1##19##]. Our data suggest that a substantial number of Australian pregnant women, over 20,000 each year, will have BGDM and therefore maternal and infant adverse health outcomes that are directly or indirectly attributable to BGDM. Evidence from the Australian Carbohydrate Study in Pregnant Women (ACHOIS) trial [##REF##15951574##20##] confirmed that untreated mild GDM is associated with relatively rare but nonetheless significant adverse perinatal outcomes. The trial demonstrated that the risk of these outcomes can be reduced with standard treatment consisting of individual dietary and lifestyle advice during pregnancy. There is, however, insufficient evidence regarding the benefits and harms of similar intervention for women with BGDM, with only one small clinical trial identifying a significantly reduced risk of large-for-gestational age infants with dietary advice and regular blood glucose monitoring for women with borderline glucose intolerance [##REF##16241919##21##]. Data from our analysis highlight the need for well-designed large randomised clinical trials to investigate the benefits and harms of such treatment for women with BGDM.</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>The impact of borderline gestational diabetes mellitus (BGDM), defined as a positive oral glucose challenge test (OGCT) and normal oral glucose tolerance test (OGTT), on maternal and infant health is unclear. We assessed maternal and infant health outcomes in women with BGDM and compared these to women who had a normal OGCT screen for gestational diabetes.</p>", "<title>Methods</title>", "<p>We compared demographic, obstetric and neonatal outcomes between women participating in the Australian Collaborative Trial of Supplements with antioxidants Vitamin C and Vitamin E to pregnant women for the prevention of pre-eclampsia (ACTS) who had BGDM and who screened negative on OGCT.</p>", "<title>Results</title>", "<p>Women who had BGDM were older (mean difference 1.3 years, [95% confidence interval (CI) 0.3, 2.2], p = 0.01) and more likely to be obese (27.1% vs 14.1%, relative risk (RR) 1.92, [95% CI 1.41, 2.62], p &lt; 0.0001) than women who screened negative on OGCT. The risk of adverse maternal outcome overall was higher (12.9% vs 8.1%, RR 1.59, [95% CI 1.00, 2.52], p = 0.05) in women with BGDM compared with women with a normal OGCT. Women with BGDM were more likely to develop pregnancy induced hypertension (17.9% vs 11.8%, RR 1.51, [95% CI 1.03, 2.20], p = 0.03), have a caesarean for fetal distress (17.1% vs 10.5%, RR 1.63, [95% CI 1.10, 2.41], p = 0.01), and require a longer postnatal hospital stay (mean difference 0.4 day, [95% CI 0.1, 0.7], p = 0.01) than those with a normal glucose tolerance.</p>", "<p>Infants born to BGDM mothers were more likely to be born preterm (10.7% vs 6.4%, RR 1.68, [95% CI 1.00, 2.80], p = 0.05), have macrosomia (birthweight ≥4.5 kg) (4.3% vs 1.7%, RR 2.53, [95% CI 1.06, 6.03], p = 0.04), be admitted to the neonatal intensive care unit (NICU) (6.5% vs 3.0%, RR 2.18, [95% CI 1.09, 4.36], p = 0.03) or the neonatal nursery (40.3% vs 28.4%, RR 1.42, [95% CI 1.14, 1.76], p = 0.002), and have a longer hospital stay (p = 0.001). More infants in the BGDM group had Sarnat stage 2 or 3 neonatal encephalopathy (12.9% vs 7.8%, RR 1.65, [95% CI 1.04, 2.63], p = 0.03).</p>", "<title>Conclusion</title>", "<p>Women with BGDM and their infants had an increased risk of adverse health outcomes compared with women with a negative OGCT. Intervention strategies to reduce the risks for these women and their infants need evaluation.</p>", "<title>Trial registration</title>", "<p>Current Controlled Trials ISRCTN00416244</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>All authors contributed to the study design, interpretation of the data and preparation of the drafts of the manuscript. In addition CAC and ARR coordinated the study and the collection of data. KJW performed the data analyses. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2393/8/31/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We are indebted to the women and their children who participated in the ACTS study, the ACTS Study Group [##REF##16641396##13##], Helena Oakey for statistical support and Melissa Ewens for administration in support for this paper.</p>", "<p>The ACTS trial was funded by grants from the National Health and Medical Research Council, Australia and the Channel 7 Research Foundation, South Australia.</p>", "<p>Ethics</p>", "<p>Children, Youth and Women's Health Service (CYWHS) Human Research and Ethics Committee (HREC)</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Demographics of women with borderline GDM compared with women with a normal OGCT</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Characteristics</bold><break/></td><td align=\"left\"><bold>Borderline GDM </bold><break/>n = 140 (%)</td><td align=\"left\"><bold>Normal OGCT </bold><break/>n = 1596 (%)</td><td align=\"left\"><bold>Relative risk </bold><break/>[95% CI]</td><td align=\"left\"><bold>p value</bold><break/></td></tr></thead><tbody><tr><td align=\"left\"><bold>Age</bold>^{<bold>a </bold>}(years)</td><td align=\"left\">27.5 ± 5.4</td><td align=\"left\">26.3 ± 5.8</td><td align=\"left\">1.3 [0.3, 2.2]</td><td align=\"left\">0.01</td></tr><tr><td align=\"left\"><bold>Race</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Caucasian</td><td align=\"left\">129 (92.1)</td><td align=\"left\">1517 (95.1)</td><td align=\"left\">0.97 [0.92, 1.02]</td><td align=\"left\">0.22</td></tr><tr><td align=\"left\"> Asian</td><td align=\"left\">5 (3.6)</td><td align=\"left\">47 (2.9)</td><td align=\"left\">1.21 [0.49, 3.00]</td><td align=\"left\">0.68</td></tr><tr><td align=\"left\"> Other</td><td align=\"left\">6 (4.3)</td><td align=\"left\">32 (2.0)</td><td align=\"left\">2.14 [0.91, 5.02]</td><td align=\"left\">0.08</td></tr><tr><td align=\"left\"><bold>BMI</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Underweight (&lt;18.5)</td><td align=\"left\">3 (2.3)</td><td align=\"left\">59 (4.0)</td><td align=\"left\">0.59 [0.19, 1.80]</td><td align=\"left\">0.34</td></tr><tr><td align=\"left\"> Normal (18.5 – &lt;25)</td><td align=\"left\">59 (45.7)</td><td align=\"left\">818 (55.9)</td><td align=\"left\">0.82 [0.67, 0.99]</td><td align=\"left\">0.04</td></tr><tr><td align=\"left\"> Overweight (25 – &lt;30)</td><td align=\"left\">32 (24.8)</td><td align=\"left\">380 (26.0)</td><td align=\"left\">0.96 [0.70, 1.31]</td><td align=\"left\">0.78</td></tr><tr><td align=\"left\"> Obese (≥30)</td><td align=\"left\">35 (27.1)</td><td align=\"left\">207 (14.1)</td><td align=\"left\">1.92 [1.41, 2.62]</td><td align=\"left\">&lt;0.0001</td></tr><tr><td align=\"left\"><bold>SEIFA</bold>^{<bold>b</bold>}</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Low</td><td align=\"left\">40 (28.6)</td><td align=\"left\">429 (26.9)</td><td align=\"left\">1.06 [0.81, 1.40]</td><td align=\"left\">0.66</td></tr><tr><td align=\"left\"> Low-Mid</td><td align=\"left\">25 (17.9)</td><td align=\"left\">288 (17.9)</td><td align=\"left\">1.00 [0.69, 1.44]</td><td align=\"left\">0.99</td></tr><tr><td align=\"left\"> Mid-High</td><td align=\"left\">37 (26.4)</td><td align=\"left\">391 (24.5)</td><td align=\"left\">1.08 [0.81, 1.44]</td><td align=\"left\">0.61</td></tr><tr><td align=\"left\"> High</td><td align=\"left\">38 (27.1)</td><td align=\"left\">490 (30.7)</td><td align=\"left\">0.88 [0.67, 1.17]</td><td align=\"left\">0.39</td></tr><tr><td align=\"left\"><bold>Education</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Secondary or lower</td><td align=\"left\">54 (39.7)</td><td align=\"left\">704 (45.0)</td><td align=\"left\">0.88 [0.71, 1.09]</td><td align=\"left\">0.25</td></tr><tr><td align=\"left\"> TAFE or equivalent</td><td align=\"left\">38 (27.9)</td><td align=\"left\">361 (23.1)</td><td align=\"left\">1.21 [0.91, 1.61]</td><td align=\"left\">0.19</td></tr><tr><td align=\"left\"> University</td><td align=\"left\">44 (32.4)</td><td align=\"left\">498 (31.9)</td><td align=\"left\">1.02 [0.79, 1.31]</td><td align=\"left\">0.91</td></tr><tr><td align=\"left\"><bold>Smoking</bold></td><td align=\"left\">37 (26.4)</td><td align=\"left\">340 (21.3)</td><td align=\"left\">1.24 [0.93, 1.66]</td><td align=\"left\">0.15</td></tr><tr><td align=\"left\"><bold>BP at trial entry</bold>^{<bold>a </bold>}(mmHg)</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Systolic BP</td><td align=\"left\">110.7 ± 11.2</td><td align=\"left\">110.1 ± 10.5</td><td align=\"left\">0.6 [-1.3, 2.4]</td><td align=\"left\">0.55</td></tr><tr><td align=\"left\"> Diastolic BP</td><td align=\"left\">66.4 ± 9.0</td><td align=\"left\">65.3 ± 8.0</td><td align=\"left\">1.1 [-0.3, 2.5]</td><td align=\"left\">0.12</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Clinical outcomes among women with borderline GDM compared with women with a normal OGCT</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Outcome</bold><break/></td><td align=\"left\"><bold>Borderline GDM </bold><break/>n = 140 (%)</td><td align=\"left\"><bold>Normal OGCT </bold><break/>n = 1596 (%)</td><td align=\"left\"><bold>Unadjusted relative risk </bold><break/>[95% CI]</td><td align=\"left\"><bold>p value</bold><break/></td><td align=\"left\"><bold>Adjusted relative risk </bold><break/>[95% CI]</td><td align=\"left\"><bold>p value</bold><break/></td></tr></thead><tbody><tr><td align=\"left\"><bold>Maternal adverse outcome</bold></td><td align=\"left\">18 (12.9)</td><td align=\"left\">129 (8.1)</td><td align=\"left\">1.59 [1.00, 2.52]</td><td align=\"left\">0.05</td><td align=\"left\">1.47 [0.92, 2.34]</td><td align=\"left\">0.11</td></tr><tr><td align=\"left\"><bold>Pregnancy induced hypertension</bold></td><td align=\"left\">25 (17.9)</td><td align=\"left\">189 (11.8)</td><td align=\"left\">1.51 [1.03, 2.20]</td><td align=\"left\">0.03</td><td align=\"left\">1.31 [0.90, 1.90]</td><td align=\"left\">0.16</td></tr><tr><td align=\"left\"><bold>Pre-eclampsia</bold></td><td align=\"left\">9 (6.4)</td><td align=\"left\">86 (5.4)</td><td align=\"left\">1.19 [0.61, 2.32]</td><td align=\"left\">0.60</td><td align=\"left\">1.08 [0.56. 2.10]</td><td align=\"left\">0.82</td></tr><tr><td align=\"left\"><bold>Antenatal hospitalisation</bold></td><td align=\"left\">29 (20.7)</td><td align=\"left\">287 (18.0)</td><td align=\"left\">1.15 [0.82, 1.62]</td><td align=\"left\">0.42</td><td align=\"left\">1.17 [0.83, 1.65]</td><td align=\"left\">0.36</td></tr><tr><td align=\"left\"><bold>PPROM</bold></td><td align=\"left\">6 (4.3)</td><td align=\"left\">41 (2.6)</td><td align=\"left\">1.67 [0.72, 3.86]</td><td align=\"left\">0.23</td><td align=\"left\">1.54 [0.66, 3.57]</td><td align=\"left\">0.32</td></tr><tr><td align=\"left\"><bold>Induction of labour</bold></td><td align=\"left\">49 (35.0)</td><td align=\"left\">498 (31.2)</td><td align=\"left\">1.12 [0.88, 1.42]</td><td align=\"left\">0.34</td><td align=\"left\">1.06 [0.84. 1.34]</td><td align=\"left\">0.62</td></tr><tr><td align=\"left\"><bold>Vaginal birth</bold></td><td align=\"left\">94 (67.1)</td><td align=\"left\">1184 (74.2)</td><td align=\"left\">0.90 [0.80, 1.02]</td><td align=\"left\">0.10</td><td align=\"left\">0.96 [0.86, 1.08]</td><td align=\"left\">0.48</td></tr><tr><td align=\"left\"> Normal vaginal birth</td><td align=\"left\">70 (50.0)</td><td align=\"left\">865 (54.2)</td><td align=\"left\">0.92 [0.78, 1.10]</td><td align=\"left\">0.36</td><td align=\"left\">1.00 [0.85. 1.17]</td><td align=\"left\">1.00</td></tr><tr><td align=\"left\"> Instrumental vaginal birth</td><td align=\"left\">24 (17.1)</td><td align=\"left\">319 (20.0)</td><td align=\"left\">0.86 [0.59, 1.25]</td><td align=\"left\">0.42</td><td align=\"left\">0.87 [0.60, 1.27]</td><td align=\"left\">0.48</td></tr><tr><td align=\"left\"><bold>Caesarean section</bold></td><td align=\"left\">46 (32.9)</td><td align=\"left\">412 (25.8)</td><td align=\"left\">1.27 [0.99, 1.64]</td><td align=\"left\">0.06</td><td align=\"left\">1.13 [0.89. 1.43]</td><td align=\"left\">0.33</td></tr><tr><td align=\"left\"> Elective</td><td align=\"left\">10 (7.1)</td><td align=\"left\">101 (6.3)</td><td align=\"left\">1.13 [0.60, 2.11]</td><td align=\"left\">0.70</td><td align=\"left\">1.01 [0.54, 1.88]</td><td align=\"left\">0.98</td></tr><tr><td align=\"left\"> Emergency</td><td align=\"left\">36 (25.7)</td><td align=\"left\">311 (19.5)</td><td align=\"left\">1.32 [0.98, 1.78]</td><td align=\"left\">0.07</td><td align=\"left\">1.17 [0.87, 1.56]</td><td align=\"left\">0.30</td></tr><tr><td align=\"left\"><bold>Caesarean section for fetal distress</bold></td><td align=\"left\">24 (17.1)</td><td align=\"left\">168 (10.5)</td><td align=\"left\">1.63 [1.10, 2.41]</td><td align=\"left\">0.01</td><td align=\"left\">1.43 [0.97, 2.11]</td><td align=\"left\">0.07</td></tr><tr><td align=\"left\"><bold>Major postpartum haemorrhage</bold></td><td align=\"left\">4 (2.9)</td><td align=\"left\">42 (2.6)</td><td align=\"left\">1.09 [0.40, 2.98]</td><td align=\"left\">0.87</td><td align=\"left\">0.96 [0.35, 2.66]</td><td align=\"left\">0.94</td></tr><tr><td align=\"left\"><bold>Postpartum pyrexia</bold></td><td align=\"left\">3 (2.1)</td><td align=\"left\">13 (0.8)</td><td align=\"left\">2.63 [0.76, 9.12]</td><td align=\"left\">0.13</td><td align=\"left\">2.33 [0.66, 8.17]</td><td align=\"left\">0.19</td></tr><tr><td align=\"left\"><bold>Maternal length of stay</bold>^a(days)</td><td align=\"left\">3.5 ± 2.0</td><td align=\"left\">3.1 ± 1.7</td><td align=\"left\">0.4 [0.1, 0.7]</td><td align=\"left\">0.01</td><td align=\"left\">0.3 [-0.0. 0.6]</td><td align=\"left\">0.06</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Clinical outcomes among babies born to women with borderline GDM compared with women with a normal OGCT</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Outcome</bold><break/></td><td align=\"left\"><bold>Borderline GDM</bold><break/></td><td align=\"left\"><bold>Normal OGCT</bold><break/></td><td align=\"left\"><bold>Unadjusted relative risk </bold><break/>[95% CI]</td><td align=\"left\"><bold>p value</bold><break/></td><td align=\"left\"><bold>Adjusted relative risk </bold><break/>[95% CI]</td><td align=\"left\"><bold>p value</bold><break/></td></tr></thead><tbody><tr><td align=\"left\"><bold><italic>Births</italic></bold></td><td align=\"left\">n = 140 (%)</td><td align=\"left\">n = 1596 (%)</td><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>Infant death or adverse outcome</bold></td><td align=\"left\">18 (12.9)</td><td align=\"left\">162 (10.2)</td><td align=\"left\">1.27 [0.80, 2.00]</td><td align=\"left\">0.31</td><td align=\"left\">1.25 [0.79. 1.98]</td><td align=\"left\">0.34</td></tr><tr><td align=\"left\"><bold>Stillbirth</bold></td><td align=\"left\">1 (0.7)</td><td align=\"left\">13 (0.8)</td><td align=\"left\">0.88 [0.12, 6.65]</td><td align=\"left\">0.90</td><td align=\"left\">0.79 [0.10, 6.08]</td><td align=\"left\">0.82</td></tr><tr><td align=\"left\"><bold>Neonatal death</bold></td><td align=\"left\">0</td><td align=\"left\">5 (0.3)</td><td align=\"left\">--</td><td align=\"left\">--</td><td align=\"left\">--</td><td align=\"left\">--</td></tr><tr><td align=\"left\"><bold>Perinatal death</bold></td><td align=\"left\">1 (0.7)</td><td align=\"left\">18 (1.1)</td><td align=\"left\">0.63 [0.09, 4.71]</td><td align=\"left\">0.66</td><td align=\"left\">0.56 [0.07, 4.21]</td><td align=\"left\">0.57</td></tr><tr><td align=\"left\"><bold>GA at birth</bold>^a(weeks)</td><td align=\"left\">39.7 (38.5–40.9)</td><td align=\"left\">40.1 (39.0–41.0)</td><td align=\"left\">--</td><td align=\"left\">0.004</td><td align=\"left\">--</td><td align=\"left\">0.003</td></tr><tr><td align=\"left\"><bold>Preterm birth </bold>(GA &lt;37 weeks)</td><td align=\"left\">15 (10.7)</td><td align=\"left\">102 (6.4)</td><td align=\"left\">1.68 [1.00, 2.80]</td><td align=\"left\">0.05</td><td align=\"left\">1.64 [0.97, 2.75]</td><td align=\"left\">0.06</td></tr><tr><td align=\"left\"><bold>Very preterm birth </bold>(GA &lt;34 weeks)</td><td align=\"left\">4 (2.9)</td><td align=\"left\">33 (2.1)</td><td align=\"left\">1.38 [0.50, 3.84]</td><td align=\"left\">0.54</td><td align=\"left\">1.40 [0.50, 3.91]</td><td align=\"left\">0.53</td></tr><tr><td align=\"left\"><bold>Extremely preterm birth </bold>(GA &lt;28 weeks)</td><td align=\"left\">0</td><td align=\"left\">12 (0.8)</td><td align=\"left\">--</td><td align=\"left\">--</td><td align=\"left\">--</td><td align=\"left\">--</td></tr><tr><td align=\"left\"><bold>Apgar 5 minute &lt;7</bold></td><td align=\"left\">3 (2.1)</td><td align=\"left\">33 (2.1)</td><td align=\"left\">1.04 [0.32, 3.33]</td><td align=\"left\">0.95</td><td align=\"left\">1.05 [0.32, 3.40]</td><td align=\"left\">0.94</td></tr><tr><td align=\"left\"><bold>Birthweight</bold>^b(g)</td><td align=\"left\">3375 ± 626.3</td><td align=\"left\">3388 ± 593.4</td><td align=\"left\">-13.0 [-116, 89.5]</td><td align=\"left\">0.80</td><td align=\"left\">-28.8 [-132, 73.9]</td><td align=\"left\">0.58</td></tr><tr><td align=\"left\"><bold>Birth length</bold>^b(cm)</td><td align=\"left\">50.2 ± 2.5</td><td align=\"left\">50.3 ± 3.3</td><td align=\"left\">-0.06 [-0.6, 0.5]</td><td align=\"left\">0.82</td><td align=\"left\">-0.09 [-0.7, 0.5]</td><td align=\"left\">0.75</td></tr><tr><td align=\"left\"><bold>Birth head circumference</bold>^b(cm)</td><td align=\"left\">34.3 ± 1.8</td><td align=\"left\">34.4 ± 1.9</td><td align=\"left\">-0.10 [-0.4, 0.2]</td><td align=\"left\">0.55</td><td align=\"left\">-0.17 [-0.5, 0.2]</td><td align=\"left\">0.34</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\"><bold><italic>Liveborns</italic></bold></td><td align=\"left\">n = 139 (%)</td><td align=\"left\">n = 1583 (%)</td><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>SFGA </bold>(Birthweight &lt;10^thpercentile)</td><td align=\"left\">10 (7.2)</td><td align=\"left\">153 (9.7)</td><td align=\"left\">0.74 [0.40, 1.38]</td><td align=\"left\">0.35</td><td align=\"left\">0.76 [0.41. 1.42]</td><td align=\"left\">0.39</td></tr><tr><td align=\"left\"><bold>LFGA </bold>(Birthweight ≥90^thpercentile)</td><td align=\"left\">19 (13.7)</td><td align=\"left\">153 (9.7)</td><td align=\"left\">1.41 [0.91, 2.20]</td><td align=\"left\">0.13</td><td align=\"left\">1.29 [0.83, 2.00]</td><td align=\"left\">0.27</td></tr><tr><td align=\"left\"><bold>Macrosomia </bold>(Birthweight ≥4.5 kg)</td><td align=\"left\">6 (4.3)</td><td align=\"left\">27 (1.7)</td><td align=\"left\">2.53 [1.06, 6.03]</td><td align=\"left\">0.04</td><td align=\"left\">2.27 [0.97, 5.34]</td><td align=\"left\">0.06</td></tr><tr><td align=\"left\"><bold>Length of stay</bold>^a(days)</td><td align=\"left\">3 (3–5)</td><td align=\"left\">3 (2–4)</td><td align=\"left\">--</td><td align=\"left\">0.001</td><td align=\"left\">--</td><td align=\"left\">0.01</td></tr><tr><td align=\"left\"><bold>Admission to nursery</bold></td><td align=\"left\">56 (40.3)</td><td align=\"left\">450 (28.4)</td><td align=\"left\">1.42 [1.14, 1.76]</td><td align=\"left\">0.002</td><td align=\"left\">1.35 [1.09, 1.68]</td><td align=\"left\">0.01</td></tr><tr><td align=\"left\"><bold>Admission to NICU</bold></td><td align=\"left\">9 (6.5)</td><td align=\"left\">47 (3.0)</td><td align=\"left\">2.18 [1.09, 4.36]</td><td align=\"left\">0.03</td><td align=\"left\">2.05 [1.02, 4.13]</td><td align=\"left\">0.04</td></tr><tr><td align=\"left\"><bold>RDS</bold></td><td align=\"left\">0</td><td align=\"left\">14 (0.9)</td><td align=\"left\">--</td><td align=\"left\">--</td><td align=\"left\">--</td><td align=\"left\">--</td></tr><tr><td align=\"left\"><bold>Mechanical ventilation</bold></td><td align=\"left\">2 (1.4)</td><td align=\"left\">33 (2.1)</td><td align=\"left\">0.69 [0.17, 2.85]</td><td align=\"left\">0.61</td><td align=\"left\">0.65 [0.16, 2.71]</td><td align=\"left\">0.56</td></tr><tr><td align=\"left\"><bold>Antibiotics &lt;48 hours</bold></td><td align=\"left\">14 (10.1)</td><td align=\"left\">78 (4.9)</td><td align=\"left\">2.04 [1.19, 3.51]</td><td align=\"left\">0.01</td><td align=\"left\">2.14 [1.24, 3.68]</td><td align=\"left\">0.01</td></tr><tr><td align=\"left\"><bold>Sarnat stage 2 or 3 encephalopathy</bold></td><td align=\"left\">18 (12.9)</td><td align=\"left\">124 (7.8)</td><td align=\"left\">1.65 [1.04, 2.63]</td><td align=\"left\">0.03</td><td align=\"left\">1.69 [1.06, 2.69]</td><td align=\"left\">0.03</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>^aValues are mean ± standard deviation, and the comparisons are mean difference (95% CI)</p><p>^bLower scores indicate lower socioeconomic status</p><p>BMI, Body mass index; BP, Blood pressure; SEIFA, Socio-economic index for area</p></table-wrap-foot>", "<table-wrap-foot><p>^aValue is mean ± standard deviation, and the comparison is mean difference (95% CI).</p></table-wrap-foot>", "<table-wrap-foot><p>^aValues are median (IR range). ^bValue is mean ± standard deviation, and the comparison is mean difference (95% CI). GA, gestational age; SFGA, small for gestational age; LFGA, large for gestational age; NICU, neonatal intensive care unit; RDS, respiratory distress syndrome</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Crohn's disease is an inflammatory bowel disease that affects the entire gut, but mostly the terminal ileum of the small bowel is involved. Due to the chronic inflammation the affected bowel segment is scarred and may become stenotic. Although medical treatment aims to reduce the inflammation, many patients eventually will have surgery because of obstructive complaints [##REF##16378007##1##].</p>", "<p>Primary medical treatment is still considered the preferred treatment because of the potential morbidity associated with surgery. Furthermore, medical treatment might avert surgery. Medical therapy consists of remission induction by a short course of steroids most often followed by maintenance therapy with immunomodulating agents. Recurrence of disease activity is primarily treated with steroids. Frequent disease exacerbations and steroid dependency are an indication for treatment with infliximab. Infliximab is a chimeric anti-TNF monoclonal antibody against tumor necrosis factor, an important proinflammatory cytokine in Crohn's disease. Treatment with this biological is effective in inducing and maintaining response and remission in patients with moderate to severe Crohn's disease.</p>", "<p>Infliximab therapy once initiated is best continued at 8 weeks intervals, although interval therapy is often used to reduce costs and to avoid the risks of long-term immune suppression. Major drawbacks of medical therapy are long-term use of medication with associated impairment of quality of life, morbidity and high costs. Furthermore, infliximab treatment is an open-ended medical treatment: it is unclear for how long therapy should be continued. Interrupting the treatment is undesirable since it is associated with loss of response due to anti-infliximab antibody formation [##REF##12584368##2##, ####REF##15168370##3##, ##REF##16481629##4####16481629##4##]. It remains unclear in how many patients with recurrent Crohn's disease surgery can eventually be avoided [##REF##16378007##1##] Thus, patients with recurrent Crohn's disease encompass a heterogeneous group of patients some of which will respond to (long-term) medical treatment whereas in others surgery cannot be averted by medical treatment.</p>", "<p>It is well established that an ileocolic resection is an effective and low morbidity operation resulting in a quick relieve of complaints and fast restoration of quality of life. Most frequent complications requiring reoperation are anastomotic dehiscence and intra-abdominal abscess. In several publications analyzing safety of laparoscopic ileocolic resection, the percentage of complications requiring reoperation varied from 0 to 7.6% [##REF##16481628##5##]. After ileocolic resection, medication can be stopped or limited to prophylactic medication when indicated [##REF##18266036##6##]. The length of loss of small bowel is generally limited and averages 20–25 cm in patients who had surgery for obstructive symptoms refractory to medical treatment. Long-term surgical recurrence occurs in 20–25% over an 8–9 years period in patients refractory to medication [##REF##16328612##7##,##REF##16148548##8##].</p>", "<p>Patients are generally young and in the middle of building their socioeconomic life. Disease activity with its associated complaints and long-term therapy have a pronounced effect on quality of life characterized by sick leave and non-attendance of social activities [##REF##1556402##9##,##REF##12742203##10##]. Patients that have a clinical recurrence after medical treatment can be considered as patients having a more severe type of the disease. To date consensus statements offer either treatment with infliximab or surgical resection in limited disease, because no comparative studies on the two alternatives exist. It can be hypothesized that surgery may avoid long-term or ineffective medical treatment improving quality of life and reducing costs. With the implementation of the laparoscopic approach, morbidity and overall costs are further reduced, and body image and cosmesis are maintained [##REF##16481628##5##,##REF##12004208##11##, ####REF##9788857##12##, ##REF##16432345##13##, ##REF##17164582##14##, ##REF##11353959##15####11353959##15##]. For these reasons time has come to compare laparoscopic ileocolic resection with infliximab treatment in terms of quality of life, sick leave and costs.</p>" ]

[ "<title>Methods/design</title>", "<title>Study objectives</title>", "<p>To compare, in a prospective randomized setting, the short-term and medium-term effectiveness and costs of ileocolic resection versus infliximab therapy in patients with recurrent Crohn's disease of the distal ileum. Two research questions can be defined:</p>", "<p><italic>1. </italic>How does infliximab treatment of patients with recurrent Crohn's disease of the distal ileum compare with laparoscopic ileocolic resection in terms of quality of life, hospital stay, morbidity, sick leave and surgical (re)interventions?</p>", "<p><italic>2. </italic>What are the 12 months cumulative total costs of infliximab treatment versus laparoscopic ileocolic resection in patients with recurrent Crohn's disease of the distal ileum?</p>", "<title>Study design</title>", "<p>The LIR!C-study is a randomized multicenter trial with participation of at least five academic and 20 regional hospitals. Patients presenting with recurrent or resistant Crohn's disease of the distal ileum will be counseled and asked for informed consent if the inclusion and exclusion criteria are met. Randomization will take place after informed consent has been obtained. Patients will be randomized to either a laparoscopic ileocolic resection or to treatment with infliximab (see Figure ##FIG##0##1##).</p>", "<p>Randomization is performed by an Internet randomization module prepared by the Department of Clinical Research and Data management. Biased-coin randomization is used and the randomization is stratified for the randomizing centers and the presence or absence of peri-anal fistulas.</p>", "<title>Primary and secondary endpoints</title>", "<p>The primary endpoints of the LIR!C trial are the disease-specific quality of life as measured by the IBDQ [##REF##9817123##16##,##REF##10109801##17##] and the costs per QALY.</p>", "<p>Secondary outcome parameters are general quality of life, as measured by the SF-36 [##REF##2644154##18##] and EQ 5D [##REF##12439115##19##,##REF##15290922##20##] questionnaires; number of days on sick leave; morbidity (due to either surgery or medical treatment), total in and out hospital medical and non-medical costs and body image and cosmesis as measured by the body image questionnaire (BIQ) [##REF##16432345##13##].</p>", "<title>Study population</title>", "<p>The study population consists of patients with recurrent or resistant Crohn's disease of the distal ileum, not responding to immunomodulating therapy.</p>", "<p>Inclusion criteria are: age in between 18 and 80 years, recurrent Crohn's disease of the distal ileum, a completed IBDQ, SF-36 and EQ-5D and BIQ before randomization, informed consent.</p>", "<p>Exclusion criteria are: prior ileocolic resection for Crohn's disease, obstructive Crohn's disease of the distal ileum requiring surgery, diseased small bowel segment longer than 40 cm, abdominal abscesses and abdominal fluid collections, American Society of Anesthesiologists (ASA) III en IV, insufficient understanding of the Dutch language or cognitively inability to complete Dutch questionnaires.</p>", "<title>Participating centers</title>", "<p>Twenty-seven centers will enroll patients. Five of these hospitals are academic hospitals.</p>", "<title>Ethics</title>", "<p>The study is conducted in accordance with the principles of the Declaration of Helsinki and 'good clinical practice' guidelines. The protocol has been approved by the Medical Ethical Committee of the Academic Medical Center in Amsterdam and the local Ethical Committees of the participating centers. Prior to randomization informed consent will be obtained from patients.</p>", "<title>Study Outline</title>", "<title>Recruitment</title>", "<p>Patients will be recruited in the outpatients IBD clinics of the participating medical centers. Patients may not want to participate in the study because they reject the principle of randomization, especially since the randomization of this trial will result in two very divergent treatment strategies. A small pilot study in the AMC IBD clinic learned that most patients would participate as long as careful and clear explanation of the study is offered. In a prior study comparing laparoscopic with open ileocolic resection performed by our institute only 2 out of 62 patient refused participation or randomization [##REF##17164582##14##].</p>", "<title>Infliximab</title>", "<p>Patients randomly allocated to infliximab treatment arm will undergo remission induction consisting of three subsequent infusions at week 0, 2 and 6 in a dose of 5 mg/kg. Infliximab maintenance therapy, consisting of infusions of 5 mg/kg at 8 to 12 weeks intervals, will be given to patients with active disease after an episode of disease activity that was treated with infliximab remission induction. In case of disease recurrence during infliximab treatment intervals will be shortened to 6 weeks and/or the dose level increased to 10 mg/kg. Infusion reactions will be treated with 25 mg prednisolone and 2 mg clemastine intravenously prior to subsequent infusions. Infliximab therapy will be combined with azathioprine immunomodulation in a dose of 2,5 mg/kg daily or 6-mercaptopurine 1,5 mg/kg daily. In the case of intolerance to these immunomodulating agents methotrexate will be given in an intramuscular dose of 15 mg once a week. Infliximab will be given without any co-medication in case of intolerance to abovementioned immunomodulating drugs or in case of contra-indications for the use of these drugs. Patients total blood counts and liver enzymes will be monitored 2 and 4 weeks after initiation of therapy and subsequently at 3 months intervals.</p>", "<title>(Laparoscopic) ileocolic resection</title>", "<p>Patients randomly allocated to surgery receive a short course of steroids to reduce the inflammation, consisting of prednisolone 40 mg oral dose (OD) for one week, 30 mg OD during one week, 20 mg OD for 1 week, followed by a dose of 10 mg. Once steroid therapy has been tapered to a dose of 20 mg/day ileocolic resection can be performed.</p>", "<p>Surgery will be performed under general anesthesia. Patients will receive antibiotics for 24 hours. Ileocolic resection is done preferably laparoscopically. A variety of techniques can be applied performing a laparoscopic ileocolic resection ranging from a facilitated (laparoscopic mobilization of the right colon followed by extracorporeal vascular ligation, bowel transsection and reanastomosing) to a total laparoscopic procedure (all steps are done intracorporeally including anastomosis making). Generally, 3 or 4 trocarts suffice. The minilaparotomy preferably is done as an up and down transumbilical incision or in case of a large specimen as a Pfannenstiehl incision.</p>", "<title>Statistical analysis</title>", "<title>Intention to treat</title>", "<p>The analysis will be performed in accordance with the intention to treat principle.</p>", "<title>Sample size calculation</title>", "<p>The primary outcome of the study is a difference in IBDQ total score between the two randomized groups at one year. In order to assess the sample size for this study, a 0.5 between-group effect size on the IBDQ total score at week 48 was considered to be relevant. A modest effect size of 0.50 is generally considered to be clinically relevant. With a 5% two-sided significance level, 65 patients per study arm will be needed to achieve an 80% power to detect such a difference with a two-sided t-test [##UREF##0##21##]. Additional mixed-models repeated measures analysis of variance will be used to investigate whether there is a different pattern of change over time between the two study arms in the four IBDQ dimensions and the EQ 5D [##REF##12719681##22##].</p>", "<title>Data collection and monitoring</title>", "<p>Patients will be followed for a period of 12 months. Seven times during this follow-up period patients will complete a set of questionnaires (the IBDQ, EQ 5D, SF-36 and BIQ): patients will complete the first set of questionnaires before randomization, the next set at week 2 of their therapy, the third set at week 6 and after that every 3 months (3, 6, 9 and 12 months after start of therapy). The questionnaires will be sent to the patients by post accompanied by a return envelop provided with postage stamps and the address of the hospital. Collection of the questionnaires will be safeguarded by the trial coordinator.</p>", "<p>Additional to the questionnaires, disease activity will be assessed by calculating the CDAI. For this calculation patients will be asked to keep a diary for seven days. In total, patients will fill in 7 diaries. During a visit to the gastro-enterologist or trial-nurse, hematocrit, presence or absence of an abdominal mass and number of complications will be assessed. CDAI can be calculated from these data combined with data from the CDAI-diaries. These visits will coincide with visits to the outpatient department for conventional patient care. At the end of the study period, after 12 months, patients will undergo an endoscopy to measure the extent of inflammation 12 months after therapy. Patients that received an ileocolic resection will be scored using the Rutgeerts endoscopic score.</p>", "<p>Patients will be contacted by telephone every month by a trial nurse to assess complications, additional interventions, re-admissions, duration of hospital and intensive care stay and visits to the outpatient clinic, number of days of sick leave and of social in attendance and to ensure completions of the questionnaires.</p>", "<p>An electronic Case Record Form (CRF) will include general patient's data (sex, age, medical history etc), patient's response to the questionnaires and data concerning type of intervention, complications, mortality, duration of hospital and intensive care stay.</p>", "<p>An independent trial monitor from the IBD trial-department will monitor the study procedure and the data of included patients.</p>", "<title>Data analysis</title>", "<p>As we do not expect a difference in mortality, data on quality of life will be the key outcome measure in the comparison. Differences in quality of life and morbidity will be analyzed using mixed-models analysis of variance for repeated measures, accounting for differences in survival between groups. Mortality will be compared using Kaplan-Meier curves and log-rank statistics.</p>", "<p>To analyze the secondary outcomes (general quality of life, number of days on sick leave, morbidity (due to either surgery or medical treatment) and results of the body image questionnaire) the two groups will be compared using the statistical program SPSS 14.0^®.</p>", "<p>A data and safety monitoring committee will safeguard trial continuation based on safety and effectiveness data. They will perform an interim analysis after 60 included patients have reached a one month follow up.</p>", "<title>Economic evaluation</title>", "<p>The marginal direct medical, non-medical and time cost, costs per QALY and cost-utility ratio will be calculated for the surgical and medical treatment strategies. Cost items will include costs of hospital admissions and readmissions (operation, nursing days, outpatient visits), institutional care (nursing homes, hospice), home care, medication and other health care providers as well as direct non-medical costs (travel expenses). Costs will be calculated by counting resource use in the diaries, questionnaires and additional 3 month interviews and multiplying these with unit prices. Standard unit prices will be used when available, complemented by results from cost calculations where needed.</p>", "<p>The cumulative total costs will be calculated for the 12 month study period. In addition, the cumulative costs for each cost category will be calculated.</p>", "<p>The EQ-5D score profiles will be transposed to health utility values following scoring algorithms based on time trade-off elicitation techniques applied in the general population. Both the UK and the Dutch scoring algorithms will be applied and compared in sensitivity analyses. QALY's are calculated as the product-sum of health utilities and the lengths of the preceding period in-between measurements during follow-up. In the final analysis, a 12 month difference in average QALY's will be calculated.</p>" ]

[]

[ "<title>Discussion</title>", "<p>In the Netherlands infliximab treatment is indicated for patients with Crohn's disease that are either steroid refractory or steroid dependent following treatment with steroids alone or in combination with immunomodulatory drugs such as azathioprine or methotrexate [##REF##16929083##23##]. These guidelines are in concert with the European consensus on infliximab treatment [##REF##16481629##4##]. However, the ECCO also recommends in its consensus statement the need for trials comparing infliximab and surgery stating that 'infliximab should be considered for steroid or immunomodulatory disease or intolerance, although surgical options should be considered and discussed.' The LIR!C-trial aims to diminish this discussion and to provide an evidence-based best treatment strategy.</p>", "<p>Besides infliximab (Remicade^®), new biologic agents have been introduced in the past few years. Adalimumab (Humira^®), a human anti-TNF monoclonal antibody, has possible advantages over infliximab in administration route and costs. In the future, therefore, adalimumab might become the preferred biological treatment. Nevertheless, at this moment adalimumab as therapy for Crohn's disease is relatively young and still proving itself. Four placebo-controlled trials assessing efficacy in Crohn's disease have been conducted so far, analyzing 1400 patients. Conclusions were that adalimumab is more effective than placebo for remission-induction and maintaining remission [##REF##17241859##24##, ####REF##16472588##25##, ##REF##17470824##26##, ##REF##17299059##27####17299059##27##]. No trials comparing adalimumab and infliximab head to head have been published yet. Because of lack of sufficient long-term data we chose to not use adalimumab in this study. Certoluzimab Pegol (Cimzia^®), a polyethylene glycolated anti-TNFα antibody fragment was associated with a modest improvement in response rates but with no significant improvements in remission rates if compared to placebo [##REF##17634458##28##]. As maintenance therapy outcomes on response and remission were better compared to placebo-therapy [##REF##17634459##29##]. However, it has not been registered in Europe yet for the therapy of Crohn's disease and therefore it is not included in this trial either.</p>", "<p>This trial compares medical therapy with (minimal invasive) surgery for Crohn's disease of the distal ileum. The first analysis will provide short- and medium-term results up to one year of follow-up. However, since long-term data of this cohort are especially of importance, we aim to develop a follow-up study to continue follow-up after the first year.</p>", "<p>Considering the drawbacks of infliximab treatment, ileocolic resection can be an equivalent alternative treatment, in spite of a small risk on serious surgical complications. Both strategies have not been compared in a clinical trial so far [##REF##16481629##4##]. Infliximab treatment may be less cost-effective when compared to laparoscopic ileocolic resection and may show less effective when assessed by means of quality of life. Therefore this study aims to answer the question which treatment is to be preferred for recurrent distal ileitis: medical therapy or early surgery.</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>With the availability of infliximab, nowadays recurrent Crohn's disease, defined as disease refractory to immunomodulatory agents that has been treated with steroids, is generally treated with infliximab. Infliximab is an effective but expensive treatment and once started it is unclear when therapy can be discontinued. Surgical resection has been the golden standard in recurrent Crohn's disease. Laparoscopic ileocolic resection proved to be safe and is characterized by a quick symptom reduction.</p>", "<p>The objective of this study is to compare infliximab treatment with laparoscopic ileocolic resection in patients with recurrent Crohn's disease of the distal ileum with respect to quality of life and costs.</p>", "<title>Methods/design</title>", "<p>The study is designed as a multicenter randomized clinical trial including patients with Crohn's disease located in the terminal ileum that require infliximab treatment following recent consensus statements on inflammatory bowel disease treatment: moderate to severe disease activity in patients that fail to respond to steroid therapy or immunomodulatory therapy. Patients will be randomized to receive either infliximab or undergo a laparoscopic ileocolic resection. Primary outcomes are quality of life and costs. Secondary outcomes are hospital stay, early and late morbidity, sick leave and surgical recurrence. In order to detect an effect size of 0.5 on the Inflammatory Bowel Disease Questionnaire at a 5% two sided significance level with a power of 80%, a sample size of 65 patients per treatment group can be calculated. An economic evaluation will be performed by assessing the marginal direct medical, non-medical and time costs and the costs per Quality Adjusted Life Year (QALY) will be calculated. For both treatment strategies a cost-utility ratio will be calculated. Patients will be included from December 2007.</p>", "<title>Discussion</title>", "<p>The LIR!C-trial is a randomized multicenter trial that will provide evidence whether infliximab treatment or surgery is the best treatment for recurrent distal ileitis in Crohn's disease.</p>", "<title>Trial registration</title>", "<p>Nederlands Trial Register NTR1150</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>EJE drafted the manuscript. WAB and PCFS co-authored the writing of the manuscript. PCFS and WAB are the principal investigators. All other authors participated in the design of the study during several meetings and are local investigators at the participating centers. All authors edited the manuscript and read and approved the final manuscript.</p>", "<title>Acknowledgements</title>", "<p>ZonMw, grant number 10788.2201</p>", "<p>All authors are member of the LIR!C-study group.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2482/8/15/prepub\"/></p>" ]

[]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>LIR!C-trial flowchart.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1471-2482-8-15-1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Homocysteine is a key metabolite of the one carbone metabolism, at the cross point between the remethylation pathway that produces methionine and the transulfuration pathway that produces cysteine [##REF##15189131##1##]. Plasma homocysteine (t-Hcys) is influenced by genetic polymorphisms of the key enzymes of the remethylation pathway, <italic>methylenetetrahydrofolate reductase </italic>(<italic>MTHFR C677T</italic>) [##REF##7647779##2##,##REF##9545395##3##], <italic>methionine synthase </italic>(<italic>MTR A2756G</italic>) [##REF##11257268##4##] and <italic>methionine synthase reductase </italic>(<italic>MTRR A66G</italic>) [##REF##9501215##5##]. MTHFR catalyses the synthesis of methyletetrahydrofolate, the methyl donor of homocysteine and MTR and MTRR are the two key enzymes of the synthesis of methionine by remethylation of homocysteine. The <italic>C677T </italic>genetic polymorphism in the <italic>MTHFR </italic>gene is found to be associated with a thermo-labile variant enzyme that shows a reduced activity [##REF##7647779##2##]. A genetic polymorphism in the <italic>transcobalamin </italic>gene (<italic>TCN2 C776G</italic>) is another genetic trait that may influence the vitamin B12 cellular delivery and consequently, homocysteine metabolism [##REF##11159542##6##]. Transcobalamin is a specific plasma transporter of cobalamin (vitamin B12) and facilitates the cellular uptake of the vitamin by receptor-mediated endocytosis [##REF##11159542##6##].</p>", "<p>Hyperhomocysteinemia is recognized as a risk factor for both venous and arterial thrombosis in the general population [##REF##9869152##7##,##REF##7475822##8##]. In this regard, genetic determinants of homocysteine metabolism have received increasing attention over the past decade as potential contributors to the greater risk of thrombosis in inflammatory bowel disease (IBD) subjects [##UREF##0##9##]. Homocysteine has also a crucial role in cellular stress, epigenetic events, inflammatory processes and host-microbial interactions. Hyperhomocysteinemia might therefore influence the clinical history of IBD, including disease severity [##UREF##0##9##]. Mahmud and colleagues were the first to report a higher prevalence of MTHFR <italic>677T </italic>allele in IBD patients, compared to controls [##REF##10446107##10##]. Nevertheless a number of subsequent studies have produced conflicting results, possibly related with complex interactions between environmental and genetic factors [##REF##11569694##11##, ####REF##11467646##12##, ##UREF##1##13##, ##REF##11374681##14##, ##REF##10961731##15##, ##REF##15831933##16##, ##REF##14560147##17####14560147##17##]. Ethnic and geographical variations in the distribution and phenotypic influence of <italic>MTHFR </italic>variants may explain, at least in part, discrepancies between case control studies on the association between <italic>MTHFR </italic>polymorphism and IBD [##REF##16522920##18##,##REF##17220211##19##]. Polymorphisms of other homocysteine metabolism-related enzymes may also contribute to the link between homocysteine metabolism and IBD patients. The influence of <italic>MTHFR, MTR, MTRR </italic>and <italic>TCN2 </italic>polymorphisms on the primary and secondary risk of ulcerative colitis have never been evaluated in a same sample population. Most of the case-control studies have evaluated Caucasian populations from North America and Europe, where the prevalence of Crohn disease is predominant, in contrast with populations from Asia, where UC is the most frequent form of IBD. In contrast to <italic>MTHFR 677T </italic>allele, the G allele frequency of <italic>TCN2 </italic>polymorphism is dramatically higher in Central China, compared with that reported in Caucasian populations [##REF##15831933##16##,##REF##16522920##18##,##REF##17220211##19##].</p>", "<p>The aim of this study was therefore to evaluate the influence of genetic determinants of homocysteine metabolism, <italic>MTHFR</italic>, <italic>MTR</italic>, <italic>MTRR </italic>and <italic>TCN2 </italic>with the primary risk and the clinical manifestations of UC, in a case control study of a sample population from Central China.</p>" ]

[ "<title>Methods</title>", "<title>Study subjects</title>", "<p>Eligible patients were males or females with established ulcerative colitis UC. 168 patients who were recruited from one single department (Department of Gastroenterology and Reseach center of Digestive diseases, Zhongnan Hospital of Wuhan University, Wuhan, province of Hubei, RP China) were prospectively enrolled in a case-control study, during a 2-years period. Information regarding concomitant medications was collected at study entry. Extent of UC was defined according to Montreal Classification [##REF##16698746##20##]. The term \"Pancolitis\" means \"ulcerative colitis that involves the entire colon.\" The UC subjects were compared to 219 healthy volunteers from a blood donor Center of Wuhan, who presented without anemia, abdominal pain, inflammation, diarrhea nor blood in the stools. The controls were extracted from a larger cohort and were matched for age and sex with UC patients. All individuals belonged to Han ethnicity and originated from the Hubei province, in Central China. This study was approved by the local ethic committee, and the subjects included gave informed consent, according to the Helsinki declaration.</p>", "<title>Assays and DNA genotyping procedures</title>", "<p>Fasting venous blood was collected in EDTA-containing tubes, immediately centrifuged, and stored at -20°C until analysis. DNA was isolated from a lymphocyte-enriched fraction of whole blood with NUCLEON BACC3 for extraction of genomic DNA kit (Amersham Pharmacia Biotech, Milan, Italy). The procedures for detecting the <italic>C677T </italic>and <italic>A1298C </italic>polymorphisms of <italic>MTHFR</italic>, as well as the <italic>A2756G MTR </italic>and the <italic>A66G MTRR </italic>polymorphisms, were based on polymerase chain reaction (PCR) amplification, restriction cleavage and separation of the DNA fragments by 15% non denaturant polyacrylamide gel electrophoresis (SDS-PAGE), as previously described [##REF##16522920##18##,##REF##17220211##19##]. Genotyping of the <italic>TCN2 C776G </italic>polymorphism was performed by the amplification-refractory mutation system, as described recently by us [##REF##11159542##6##]. DNA samples corresponding to amplified DNA of the <italic>MTHFR</italic>, <italic>MTR </italic>and <italic>MTRR </italic>genotypes were sequenced and subsequently used as controls in all series of genotype determination. Five % of the samples were re-genotyped to check for genotype calling consistency, with a genotyping success rate higher than 98%.</p>", "<title>Statistical analysis</title>", "<p>The tested <italic>a priori </italic>hypotheses were the association of the genetic determinants of homocysteine metabolism with the primary risk and the clinical characteristics of the disease, respectively. Categorical variables were reported as counts, percentages and 95% confidence interval, and continuous variables as means ± SD. For categorical variables, continuity-corrected chi-square test or exact Fisher test (when variables were stratified in function of clinical characteristics) were used to assess differences. For continuous variables, a Mann-Whitney U-test was employed. The significance, odds ratios (OR) and 95% confidence interval of independent determinants regarding the risk of IBD were determined by logistic regression analysis using a model that included age, sex and the genetic and clinical variables. The residual model of logistic regression multivariate analysis considered only the variables with P-value &lt; 0.10 in the univariate analysis. The minimal size of our sample was estimated at 150 patients, with a study power 1-β = 0.8 and α = 0.05, assuming a 1.5-fold difference in the less frequent alleles between controls and patients. P-value lower than 0.05 indicated statistical significance and Bonferroni correction for multiple testing was used as described previously, when several dependent or independent statistical tests were being performed simultaneously [##REF##16522920##18##]. (Data were prospectively collected and analyzed using the Statview 5 software for Windows (SAS Institute, Berkley, California, USA) and the SPSS 10.0 software for Windows (SPSS, Paris, France). HWE was tested using the chi-square two-tailed calculation of Knud Christensen, population genetics, www.kursus.kvl.dk.</p>" ]

[ "<title>Results</title>", "<title>Clinical characteristics of UC patients</title>", "<p>There were no significant difference in sex ratio between controls and UC patients, with respectively 38% and 42% of females (P = 0.3660); the age was respectively 41 ± 15 and 41 ± 12 years (P = 0.7428). The age onset was 38 ± 15, without difference among gender. The main clinical characteristic of the UC patients are summarized in table ##TAB##0##1##. A quarter presented with an extensive form of colitis and about 57% with lesions limited to rectum and/or sigmoid. None of the cases presented with a reported thromboembolic episode.</p>", "<title>Comparison of the genetic characteristics of UC patients and controls</title>", "<p>There was no linkage desequilibrium among different genotypes. The genotype of <italic>MTHFR C677T, A1298C</italic>, <italic>MTR</italic>, <italic>MTRR</italic>, and <italic>TCN </italic>could be determined in respectively 215, 213, 218, 219, 216 of the 219 DNA samples from controls and in respectively 168, 160, 156, 164, 146 of the 168 DNA samples from UC cases. The genotype distributions of <italic>MTHFR</italic>, <italic>MTR</italic>, <italic>MTRR</italic>, and <italic>TCN </italic>polymorphisms of patients were in Hardy-Weinberg equilibrium (P-values of two tailed chi-square 0.5071, 0.8231, 0.3055, 0.7184, 0.4237, respectively) and were not influenced by gender (data not shown). <italic>MTHFR 677T </italic>and <italic>1298C </italic>alleles were in complete disequlibrium, as previously reported in other populations [##REF##9545395##3##].</p>", "<p>The genotype distributions of <italic>MTHFR</italic>, <italic>MTRR </italic>and <italic>TCN </italic>polymorphisms did not differ between controls and UC patients (table ##TAB##1##2##). The same observation was made for allele frequencies (table ##TAB##1##2##). In contrast, the carriers of <italic>MTR 2756AG/GG </italic>genotypes were more frequent in UC than in the control group, with respective percentages of 26.3 (95% C.I. 20.0–33.7) and 16.5 (12.2–22.0) (P = 0.0212). The same observation was made with the <italic>MTR 2756G </italic>allele frequency, with a difference even more significant (table ##TAB##1##2##). In logistic regression, carriage of <italic>MTR 2756G </italic>allele was an independent predictor of UC, with an Odds ratio estimated at 1.8 (table ##TAB##1##2##).</p>", "<title>Association of genetic polymorphisms with clinical characteristics of UC patients</title>", "<p>We next examined the association of polymorphisms of homocysteine metabolism-related enzymes with the clinical characteristics. None of the polymorphisms influenced the age onset of the disease. In univariate analysis, <italic>MTR</italic>, <italic>MTRR </italic>and <italic>TCN </italic>polymorphisms were not associated with location and extent of UC while <italic>MTHFR 677TT </italic>genotype was associated with pancolitis. The frequency of the <italic>TT </italic>genotype of <italic>MTHFR C677T </italic>polymorphism was 2.7-fold higher in UC individuals with pancolitis than in other UC cases, with respective percentages of 27.3 (95% C.I.16.4–42.0) and 10.5 (95% C.I. 6.3–17.1) (P = 0.0123). The frequency of subjects who presented with either <italic>677TT </italic>or the double heterozygous <italic>677CT/1298AC </italic>genotype was also significantly different between subjects with pancolitis and those with left colitis or proctitis, with respective percentages of 43.2 (95% C.I. 29.6–43.2) and 20.2 (95% C.I. 14.1–28.1) (P = 0.0048).</p>", "<p>Finally, the association of gene polymorphisms of genes of the remethylation pathway with the risk of pancolitis was estimated by logistic regression analysis using a model that included age, sex, medications and genetic polymorphisms. We confirmed that the only significant independent gene predictor of extensive lesions was <italic>MTHFR</italic>. The sex and age adjusted odds ratio of the association between pancolitis and <italic>MTHFR 677 TT </italic>was estimated to 3.82 (95% C.I. 1.3–11.7, P = 0.0179). When the age and sex adjusted analysis was performed by considering the carriers of <italic>677TT </italic>or <italic>C677CT/1298AC </italic>genotypes of <italic>MTHFR </italic>instead of <italic>MTHFR 677TT </italic>only, the same conclusion was reached, with an Odds ratio at 3.3 (95% C.I. 1.4–7.9, P = 0.0084).</p>" ]

[ "<title>Discussion</title>", "<p>Genetic factors possibly associated with ulcerative colitis remain poorly known. This is the first study that evaluated the association of genes of the remethylation pathway of homocysteine, <italic>MTHFR</italic>, <italic>MTR</italic>, <italic>MTRR </italic>and <italic>TCN2</italic>, in a case population restricted to UC, in China. It evidenced a significant association of <italic>MTR </italic>and <italic>MTHFR </italic>variants with the primary risk and the extent of the disease, respectively; <italic>677TT </italic>and <italic>C677CT</italic>/<italic>1298AC </italic>genotypes of <italic>MTHFR </italic>were predictors of extensive colitis, while no association was found with the primary risk of UC. In contrast, <italic>MTR 2756G </italic>allele was associated with an increased risk of UC and had no influence on the extent of the disease. MTR and MTFHR have a complementary role in this pathway. During the remethylation of homocysteine into methionine, a methyl group provided by 5-methyltetrahydrofolate (5-methylTHF) is transferred to homocysteine by MTR. In this remethylation pathway, cobalamin (vitamin B12) is involved as an intermediate carrier of the methyl group while the 5-methylTHF is synthesized by MTHFR. The phenotypic influence of the 2756G allele on the activity of MTR has not been clearly evidenced [##REF##11257268##4##]. In contrast, both <italic>677TT </italic>and <italic>1298AC/677CT </italic>genotypes are known to produce a decreased catalytic enzyme activity of MTHFR [##REF##7647779##2##]. Notably, the <italic>677TT </italic>genotype of <italic>MTHFR </italic>is encoding a thermolabile variant characterized by an alanine to valine substitution at position 222, and a 50% reduction in enzyme activity [##REF##7647779##2##]. As a consequence, <italic>MTHFR C677T </italic>polymorphism produces a decreased cellular level of methyl-THF and a cellular accumulation of homocysteine, particularly in patients with insufficient folate supply [##REF##11282420##21##]. Interestingly, recent experimental findings raise the possibility that homocysteine-induced cellular and vascular stress may contribute to the maintenance of a chronic mucosal inflammatory state in IBD [##UREF##0##9##,##REF##15866230##22##].</p>", "<p>Clinical implications of the <italic>MTHFR C677T </italic>polymorphism include increased risk for several diseases, such as vascular, neurological and for birth defects [##REF##11282420##21##,##REF##14656028##23##]. Its phenotypic influence on homocysteine metabolism is neutral in subjects with sufficient folate supply, as in South Europe [##REF##16522920##18##]. Similarly, the association of <italic>MTHFR </italic>with other diseases than IBD seems to be neutral in South European countries, where the status of populations in folate is better than in North Europe, while some studies have showed association with <italic>MTR </italic>and <italic>MTRR </italic>in these countries [##REF##14656028##23##,##UREF##2##24##]. This could explain some of the discrepant results produced on association studies with IBD in populations with contrasted status in folate. Indeed, two studies performed in Northern Europe reported a significant increased frequency of the <italic>MTHFR TT </italic>genotype [##REF##10446107##10##,##REF##11843038##25##], whereas four studies in Southern Europe (two Italian, one French, one Portuguese) and one Chinese study showed no difference [##REF##11569694##11##,##REF##11374681##14##, ####REF##10961731##15##, ##REF##15831933##16##, ##REF##14560147##17####14560147##17##]. In contrast, we found a higher frequency of <italic>MTRR 66A </italic>allele in patients with Crohn's disease compared with controls, and we observed an influence of this gene variant on the extent of the disease [##REF##17925002##26##]. We also evidenced that vitamin B12 was a nutritional determinant of homocysteinemia, that was under the influence of oxidative stress [##REF##17925002##26##]. Consistently, a recent report showed that folate deficiency was not a predictor of homocysteine level in case of increased oxidative stress, underscoring the implication of other nutritional determinants, such as vitamin B12, under pathological conditions [##REF##17652662##27##]. Taken together, these results are in agreement with our findings indicating that <italic>methionine synthase A2756G </italic>polymorphism, a vitamin B12-dependent enzyme, may predict ulcerative colitis. The data are also difficult to compare among contrasted populations since the frequency of these gene polymorphisms and their influence on homocysteine fluctuate greatly, worldwide [##REF##16522920##18##,##REF##17220211##19##]. The prevalence of the homozygous <italic>T </italic>allele seems to be influenced by folate status, the prevalence being the highest in Mexicans, Chinese and Italians and the lowest in West Africans [##REF##16522920##18##,##REF##17220211##19##]. It would be therefore useful to revisit the studies on IBD and <italic>MTHFR </italic>that have been published in Europe by evaluating also the association with the other gene variants related with the remethylation pathway. Finally, the discrepancies observed in previous studies on association between <italic>MTHFR </italic>and IBD may also correspond in part to differences in the proportion of cases with extensive colitis, as suggested by our observation that MTFHR is a predictor of pancolitis, at least in UC. This observation was in agreement with our previous study of a smaller series of UC patients [##REF##15831933##16##]. In addition, the association with severity of the disease may be related with a deficit in folate and vitamin B12, as previously observed by us in Caucasians with Crohn's disease [##REF##17925002##26##]. Vitamin B12 metabolism influenced Crohn's disease activity by modulating oxidative stress, measured by superoxide dismutase activity [##REF##17925002##26##]. Homocysteine levels were recently correlated with activity, number of flares and duration of the disease [##REF##16677162##28##]. Unfortunately, we could not determine the blood level of these determinants in the present series as no serum or plasma sample was available. Furthermore, we recently found that azathioprine therapy decreased plasma homocysteine level [##REF##17925002##26##], suggesting interactions between azathioprine and homocysteine metabolisms. Genetic variation in the <italic>MTHFR </italic>gene may result in reduced S-adenosylmethionine concentations, leading to enhanced TPMT enzyme degradation and possibly modulating azathioprine efficacy [##REF##16306100##29##].</p>", "<p>Over the past decade, hyperhomocysteinemia and <italic>MTHFR C677T </italic>polymorphism have been regarded solely as potential contributors to the greater risk of thrombosis in IBD [##REF##11374681##14##,##REF##10961731##15##,##REF##9798965##30##, ####REF##11051355##31##, ##REF##17100967##32####17100967##32##]. Our data indicated clearly that it influenced the severity of UC, independently of the risk of thrombosis, as none of the case from our series underwent a thrombosis episode.</p>" ]

[ "<title>Conclusion</title>", "<p><italic>MTR and MTHFR </italic>are two genes of vitamin B12 and folate cellular metabolism associated respectively with the primary risk and the severity of UC, at least in central China. These findings might open new insights into the pathogenesis and assessment of UC, particularly for the potential interest in treating the patients presenting with the <italic>677TT MTHFR </italic>genetic trait and a deficit in folate. However, our results await confirmation in a large cohort of patients with UC.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>The association of genetic polymorphisms related to metabolism of homocysteine with inflammatory bowel disease has been evidenced in Crohn disease and remains an open question in ulcerative colitis. We evaluated the association of the polymorphisms of MTHFR, MTR, MTRR and TCN2 genes with ulcerative colitis in Central China.</p>", "<title>Methods</title>", "<p>168 patients were genotyped for these polymorphisms and compared to 219 matched controls.</p>", "<title>Results</title>", "<p><italic>Methionine synthase </italic>2756G allele frequency was higher in ulcerative colitis than in controls 0.15 (95% C.I. 0.11–0.19) vs 0.09 (95% C.I. 0.07 – 0.12), (P = 0.0137) and predicted ulcerative colitis risk in logistic regression, with an Odds ratio at 1.8 (95% C.I. 1.15–2.84). <italic>Methylenetetrahydrofolate reductase</italic> 677TT genotype was 2.7-fold more prevalent in individuals with pancolitis than in those with left colitis or proctitis, with respective percentages of 27.3 (95% C.I.16.4–42.0) and 10.5 (95% C.I. 6.3–17.1) (P = 0.0123). The carriage of 677TT or 677CT/1298AC genotypes of <italic>methylenetetrahydrofolate reductase</italic> was more frequent in cases with pancolitis than in subjects with left colitis or proctitis (P = 0.0048), with an Odds ratio adjusted by age and sex at 3.3 (95% C.I. 1.4–7.9), P = 0.0084) in logistic regression.</p>", "<title>Conclusion</title>", "<p><italic>Methionine synthase </italic>and <italic>methylenetetrahydrofolate reductase</italic> are genes of vitamin B12 and folate cellular metabolism associated respectively with risk and extent of ulcerative colitis, at least in Central China. This finding may open new insights, particularly for the potential interest in treating patients carrying the 677TT MTHFR genetic trait and a deficit in folate.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>J–LG was the guarantor and supervisor of the study. MC, LP–B, BX and J–LG contributed to conception and design, recruitment of patients, analysis and interpretation of data and drafting of the manuscript. R–MG–R carried out the biological and genetics analyses, interpretation of data, statistical analyses, revising of the manuscript. BX and MC recruited the patients and M–AB and J–PB contributed in interpretating the data. LP–B and J–LG revised the manuscript. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2350/9/78/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>Institutional funding was received from Region Lorraine and Inserm (France) and a grant of international collaboration was supported from China Hubei provincial science &amp; technology department (2007CA003). No private sponsor was involved in the study. Min Chen was recipient of a fellowship grant from Chinese service center for scholarly exchange, PR of China, and from Region Lorraine, France.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Categorical clinical characteristics of the 168 patients with ulcerative colitis.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Characteristic</bold></td><td align=\"center\"><bold>Number of cases</bold></td><td align=\"center\"><bold>Percentage (95% confidence interval)</bold></td></tr></thead><tbody><tr><td align=\"left\">Male sex</td><td align=\"center\">97</td><td align=\"center\">57.7 (50.2-57.7)</td></tr><tr><td align=\"left\">Past smoking</td><td align=\"center\">14</td><td align=\"center\">8.3 (5.1–13.5)</td></tr><tr><td align=\"left\">Current smoker</td><td align=\"center\">12</td><td align=\"center\">7.1 (4.2–12.1)</td></tr><tr><td align=\"left\"><bold>Concomitant medications:</bold></td><td/><td/></tr><tr><td align=\"left\"> Steroids</td><td align=\"center\">50</td><td align=\"center\">29.8 (23.4–29.8)</td></tr><tr><td align=\"left\"> Sulfasalazine</td><td align=\"center\">98</td><td align=\"center\">58.3 (50.8–65.5)</td></tr><tr><td align=\"left\"> 5-aminosalicylate</td><td align=\"center\">69</td><td align=\"center\">41.1 (33.9–48.6)</td></tr><tr><td align=\"left\"> Antibiotics</td><td align=\"center\">15</td><td align=\"center\">8.9 (5.5–14.2)</td></tr><tr><td align=\"left\"><bold>Disease location:</bold></td><td/><td/></tr><tr><td align=\"left\"> Rectum</td><td align=\"center\">78</td><td align=\"center\">46.4 (39.0–54.0)</td></tr><tr><td align=\"left\"> Left colon</td><td align=\"center\">46</td><td align=\"center\">27.4 (21.2–34.6)</td></tr><tr><td align=\"left\"> Pancolitis</td><td align=\"center\">44</td><td align=\"center\">26.2 (20.1–33.3)</td></tr><tr><td align=\"left\"><bold>Personal history of surgery:</bold></td><td/><td/></tr><tr><td align=\"left\"> Colectomy</td><td align=\"center\">2</td><td align=\"center\">1.2 (0.4–4.2)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Genotype frequencies and minor allele frequencies of genetic polymorphisms in ulcerative colitis patients and controls.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Controls</bold><break/>Frequencies (%)</td><td align=\"center\"><bold>UC patients</bold><break/>Frequencies (%)</td><td align=\"center\"><bold>P-value</bold></td><td align=\"center\"><bold>OR (95% CI)*</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold><italic>MTHFR 677</italic></bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold><italic>CC</italic>^#</bold></td><td align=\"center\">40.9</td><td align=\"center\">38.1</td><td/><td/></tr><tr><td align=\"left\"><bold><italic>CT</italic>^#</bold></td><td align=\"center\">43.3</td><td align=\"center\">45.2</td><td align=\"center\">0.6053^a</td><td align=\"center\">1.12 (0.72–1.75)</td></tr><tr><td align=\"left\"><bold><italic>TT</italic></bold></td><td align=\"center\">15.8</td><td align=\"center\">16.7</td><td align=\"center\">0.6821^a</td><td align=\"center\">1.13 (0.62–2.05)</td></tr><tr><td align=\"left\"><bold>Total number of patients analysed</bold></td><td align=\"center\">215</td><td align=\"center\">168</td><td/><td/></tr><tr><td align=\"left\"><bold>Allele <italic>T</italic></bold></td><td align=\"center\">0.37</td><td align=\"center\">0.39</td><td align=\"center\">0.6027</td><td align=\"center\">1.09 (0.81–1.46)</td></tr><tr><td align=\"left\"><bold><italic>MTHFR 1298</italic></bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold><italic>AA</italic>^#</bold></td><td align=\"center\">66.2</td><td align=\"center\">69.4</td><td/><td/></tr><tr><td align=\"left\"><bold><italic>AC</italic></bold></td><td align=\"center\">29.1</td><td align=\"center\">28.1</td><td align=\"center\">0.7277^a</td><td align=\"center\">0.92 (0.58–1.46)</td></tr><tr><td align=\"left\"><bold><italic>CC</italic></bold></td><td align=\"center\">4.7</td><td align=\"center\">2.5</td><td align=\"center\">0.2631^a</td><td align=\"center\">0.51 (0.15–1.66)</td></tr><tr><td align=\"left\"><bold>Total number of patients analysed</bold></td><td align=\"center\">213</td><td align=\"center\">160</td><td/><td/></tr><tr><td align=\"left\"><bold>Allele <italic>C</italic></bold></td><td align=\"center\">0.19</td><td align=\"center\">0.17</td><td align=\"center\">0.3455</td><td align=\"center\">0.86 (0.59–1.26)</td></tr><tr><td align=\"left\"><bold><italic>MTR 2756</italic></bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold><italic>AA</italic>^#</bold></td><td align=\"center\">83.5</td><td align=\"center\">73.7</td><td/><td/></tr><tr><td align=\"left\"><bold><italic>AG</italic></bold></td><td align=\"center\">15.1</td><td align=\"center\">23.1</td><td align=\"center\">0.0422^a</td><td align=\"center\">1.72 (1.02–2.92)</td></tr><tr><td align=\"left\"><bold><italic>GG</italic></bold></td><td align=\"center\">1.4</td><td align=\"center\">3.2</td><td align=\"center\">0.1899^a</td><td align=\"center\">2.64 (0.62–11.25)</td></tr><tr><td align=\"left\"><bold>Total number of patients analysed</bold></td><td align=\"center\">218</td><td align=\"center\">156</td><td/><td/></tr><tr><td align=\"left\"><bold>Allele <italic>G</italic></bold></td><td align=\"center\">0.09</td><td align=\"center\">0.15</td><td align=\"center\">0.0137</td><td align=\"center\">1.81 (1.15–2.84)</td></tr><tr><td align=\"left\"><bold><italic>MTRR 66</italic></bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold><italic>AA</italic>^#</bold></td><td align=\"center\">58.0</td><td align=\"center\">53.7</td><td/><td/></tr><tr><td align=\"left\"><bold><italic>AG</italic></bold></td><td align=\"center\">35.6</td><td align=\"center\">38.4</td><td align=\"center\">0.4839^a</td><td align=\"center\">1.17 (0.76–1.79)</td></tr><tr><td align=\"left\"><bold><italic>GG</italic></bold></td><td align=\"center\">6.4</td><td align=\"center\">7.9</td><td align=\"center\">0.4746^a</td><td align=\"center\">1.34 (0.60–2.99)</td></tr><tr><td align=\"left\"><bold>Total number of patients analysed</bold></td><td align=\"center\">219</td><td align=\"center\">164</td><td/><td/></tr><tr><td align=\"left\"><bold>Allele <italic>G</italic></bold></td><td align=\"center\">0.24</td><td align=\"center\">0.27</td><td align=\"center\">0.2401</td><td align=\"center\">1.18 (0.85–1.64)</td></tr><tr><td align=\"left\"><bold><italic>TCN2 776</italic></bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold><italic>CC</italic>^#</bold></td><td align=\"center\">21.8</td><td align=\"center\">18.5</td><td/><td/></tr><tr><td align=\"left\"><bold><italic>CG</italic></bold></td><td align=\"center\">38.9</td><td align=\"center\">45.2</td><td align=\"center\">0.3358^a</td><td align=\"center\">1.26 (0.79–2.01)</td></tr><tr><td align=\"left\"><bold><italic>GG</italic></bold></td><td align=\"center\">39.3</td><td align=\"center\">36.3</td><td align=\"center\">0.7835^a</td><td align=\"center\">1.09 (0.61–1.95)</td></tr><tr><td align=\"left\"><bold>Total number of patients analysed</bold></td><td align=\"center\">216</td><td align=\"center\">146</td><td/><td/></tr><tr><td align=\"left\"><bold>Allele <italic>G</italic></bold></td><td align=\"center\">0.58</td><td align=\"center\">0.59</td><td align=\"center\">0.9447</td><td align=\"center\">1.0 (0.74–1.35)</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>*(OR: unadjusted odds ratio; 95 CI%: 95% confidence interval)</p><p>Abbreviations :<italic>methylenetetrahydrofolate reductase </italic>(<italic>MTHFR</italic>), <italic>methionine synthase </italic>(<italic>MTR</italic>), <italic>methylenetetrahydrofolate reductase </italic>(<italic>MTRR</italic>) and <italic>transcobalamin </italic>(<italic>TCN2</italic>)</p><p>^#Dominant genotype. ^aComparison with dominant genotype.</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Approximately 85% of the 2.3 million HIV-infected children under 15 years worldwide live in sub-Saharan Africa.[##UREF##0##1##] While antiretroviral therapy (ART) of children in Africa has resulted in dramatically improved survival, clinical, immunological and virological status, less than 15% of children needing ART on the continent currently receive it. [##REF##15352588##2##, ####REF##15353976##3##, ##REF##17099313##4##, ##REF##17407046##5##, ##REF##17367540##6##, ##REF##17117017##7##, ##REF##17077930##8##, ##REF##16988517##9##, ##UREF##1##10##, ##UREF##2##11####2##11##] Excellent adherence is one of the most important factors in determining treatment success and preventing viral resistance, and the need for near-perfect adherence to lifelong therapy from an early age has been identified as a major challenge in the administration of ART to HIV-infected children.[##UREF##1##10##,##REF##10877736##12##, ####REF##10462336##13##, ##REF##17469726##14##, ##UREF##3##15####3##15##] There is concern about the extent to which this is achievable for children in resource-limited settings, particularly in the context of the rapid scale-up of pediatric treatment programs required to address the HIV burden on children in Africa.[##REF##17469726##14##,##REF##16343374##16##]</p>", "<p>Research from rich countries suggests that adherence may be more complex in children compared to adults due to many factors including reliance on caregivers who may themselves be ill or may not be the child's parent, complex dosing regimens, lack of availability of pediatric fixed-dose combinations, poor drug palatability, difficulty with taking tablets/capsules and interference with daily routines. [##REF##10462336##13##,##UREF##4##17##, ####REF##12544410##18##, ##UREF##5##19##, ##UREF##6##20##, ##UREF##7##21##, ##REF##11927734##22##, ##REF##12794557##23####12794557##23##] Adherence estimates of 50 to 75% have been reported, well below the required 90 to 95% to achieve optimal viral suppression.[##REF##10877736##12##,##REF##10462336##13##,##UREF##4##17##, ####REF##12544410##18##, ##UREF##5##19##, ##UREF##6##20##, ##UREF##7##21##, ##REF##11927734##22##, ##REF##12794557##23####12794557##23##]</p>", "<p>While African adult studies show that good adherence to ART is possible despite poor social circumstances, there are limited studies in African children.[##REF##16343374##16##,##REF##15322481##24##, ####REF##12799558##25##, ##REF##16896111##26##, ##REF##17121449##27####17121449##27##] Health service challenges as well as individual factors such as poor socio-economic circumstances, poor literacy and the prohibitive cost of liquid drug formulations necessitating tablet/capsule administration to very young children are additional potential barriers to adherence in African children.[##UREF##1##10##,##UREF##3##15##] In Kampala, Uganda, 72% of children aged 2–18 years had adherence ≥ 95% measured with home-based unannounced pill counts, compared to 89% using 3-day self-reported adherence and 94% using clinic-based pill counts.[##REF##17565809##28##] Muller et al. similarly found discrepant results using different measures of adherence in young children (median age 38 months) in South Africa.[##UREF##8##29##] Using electronic means (Medication Event Monitoring System (MEMS)) to monitor adherence, only 36% of patients achieved ≥ 95% adherence, in comparison to 91% of caregivers reporting excellent adherence on a visual analogue scale.[##UREF##8##29##] The only other published adherence studies of African children include only older children and measure self-reported adherence only, with varying results. In Côte d'Ivoire and Uganda, approximately one third of caregiver-child pairs reported missing doses and in South Africa, Reddi et al. describe 10% of children missing ≥ 3 doses during the previous month.[##REF##17367540##6##,##REF##16280708##30##,##REF##16791525##31##] African studies thus concur with international literature that more objective measures of adherence (e.g. unannounced pill counts and MEMS) tend to be more sensitive to lapses in adherence.[##UREF##7##21##,##REF##11352698##32##] However, such measures are not feasible in resource-limited settings with large-scale programs, and there is a need to determine the utility of simpler measures of adherence such as clinic-based pill-counts and self-report in predicting virological response in the African context. Furthermore, all published African studies have been conducted over short periods (≤ 3 months), mostly in older chidren, and may not reflect longer term adherence patterns in very young children.</p>", "<p>We therefore aimed to measure the level of adherence to ART in infants and young children during the first year of treatment using both medication returned as measured at the clinic and caregiver self-report, to assess the extent to which such measured adherence predicts viral and immunological outcomes and to identify factors associated with good adherence.</p>" ]

[ "<title>Methods</title>", "<title>Study design, setting and population</title>", "<p>This was a prospective cohort study. All HIV infected children commenced on antiretroviral triple therapy between July 2002 and January 2004 (n = 122) as part of the ART program of the Red Cross Children's Hospital, a tertiary care institution in Cape Town, South Africa, were eligible and agreed to participate in the study. The ART program at the hospital began prior to the official national government ART program and was donor funded during this time. Selection criteria for commencement of ART have been described elsewhere.[##REF##15352588##2##] Briefly, clinical and immunological criteria as recommended by the 2001 European treatment guidelines were followed.[##REF##12139662##33##] In addition, the following limited social criteria needed to be met: having an identifiable caregiver to administer medication and attend clinic appointments; resident in Cape Town for at least 3 months; caregiver compliance with last 3 clinic appointments and caregiver willingness to comply with ongoing regular clinic attendance and monitoring. The majority of children were commenced on stavudine (d4T), lamivudine (3TC) and efavirenz (EFV – children &gt;10 kg or &gt;3 years) or ritonavir (RTV – children &lt;10 kg or &lt;3 years) as no other protease inhibitor was readily available in suitable formulation and dosage in South Africa at the time. Children were followed up with monthly clinical visits for the first year. Viral load, CD4 cell count and percentage were determined using standard laboratory methods at commencement of ART and after 1 year of treatment. The definition of undetectable viral load was &lt;400 copies/ml. The study was approved by the University of Cape Town Research Ethics Committee [Ref: 261/2002], and all parents provided written informed consent for their and their children's participation.</p>", "<title>Measurements of adherence and associated factors</title>", "<p>Clinical and socio-demographic characteristics at commencement of ART were determined by the clinician treating each child and recorded on standardized data collection forms. Children were retrospectively re-staged according to medical record information using the WHO 4-stage clinical classification for the purpose of this analysis.[##UREF##3##15##] Weight-for-age, height-for-age and weight-for-height z-scores were calculated with EpiInfo 2000, version 1.0 (Division of Surveillance and Epidemiology, CDC, Atlanta, Georgia).</p>", "<title>Measurement of adherence by medication return (MR)</title>", "<p>At every monthly visit for one year, caregivers were requested to return all empty medicine containers and unused medication. A dedicated program pharmacist measured the amount of unused medication volumetrically for syrups/solutions and by pill count for tablets/capsules. The percentage adherence for each antiretroviral medication was calculated by dividing actual use (determined from returned containers and unused medication) by expected use (determined from the previous month's script).</p>", "<p>A composite measure of annual average percentage adherence by MR was calculated by determining the arithmetic mean of the percentage adherence for each drug at each monthly visit. For caregiver-child pairs who did not return medication at every visit as requested, the annual average percentage adherence was calculated using the number of months for which medication was actually returned as the denominator when determining the arithmetic mean (per protocol analysis). Sensitivity analysis was additionally performed assigning adherence &lt;90% for months in which medication was not returned and recalculating annual average percentage adherence for children with missing medication returns. A small amount of extra medication (in excess of what was prescribed) was issued at each visit so that patients would not be without medication if drugs were spilled or additional doses required due to vomiting or spitting out. For a number of medication returns, therefore, more drug was used than prescribed (i.e. adherence &gt;100%) so adherence was capped at 100% per return when calculating annual average adherence. An uncapped annual average MR adherence was also calculated for some of the analyses.</p>", "<title>Measurement of adherence by questionnaire</title>", "<p>A standardized interview was administered by the treating clinician to each caregiver after the child had completed 3 months of ART. The interview script was based on Pediatric Aids Clinical Trials Group (PACTG) adherence questionnaires modules 1 and 2 and assessed caregiver's ability to accurately describe the ART regimen, recall of missed doses in the past 3 days, difficulties experienced with giving medication and beliefs about ART.[##UREF##9##34##,##UREF##10##35##] Based on reported missed doses, children were classified as not fully adherent (NFA) if ≥ 1 dose was missed in the previous 3 days. Interpreters were used so that interviews were conducted in the language of the caregiver's choice.</p>", "<title>Analysis</title>", "<p>All statistical analysis was done using Stata (version 10) (Stata Corporation, College Station, Texas, USA). The effects of low annual average (capped) MR adherence, average (capped) MR adherence in the month immediately preceding viral load measurement and reported NFA on viral load suppression and immunological response were determined using logistic regression models adjusted for other determinants of outcome. MR adherence was dichotomized as ≥ 90% or &lt;90% as this threshold explained the largest amount of variability in the outcome. Univariate and multivariate analysis of the association between demographic, social and clinical factors as well as experiencing problems with giving medication and annual average MR adherence ≥ 90% were examined using Wilcoxon rank sum (Mann-Whitney), Student's t-test, chi²or Fisher's exact tests and logistic regression models as indicated. Agreement between MR adherence &lt;90% and caregiver reported NFA was calculated using the kappa statistic.</p>", "<p>All multivariate models were built by sequentially adding the next most significant predictor variable from the univariate analysis, and variables with a p-value &lt;0.1 after adjustment for those already included in the model, or that changed the OR for variables included in the model by more than 10%, were retained. Since variables reflecting severity of illness (WHO stage, weight-for-height z-score, CD4 percent and absolute count and log viral load) were highly correlated with one another, only the single most predictive variable, i.e. weight-for-height z-score, was included in the multivariate models for virological and immunological outcome. Similarly, only one measure of socio-economic status (formal housing, access to water and electricity, access to a refrigerator and employment) was used at a time in each model, i.e. access to water and electricity. Only socio-demographic factors and regimen were included in building the multivariate model of MR adherence ≥ 90% as the sample size did not allow for inclusion of clinical factors as well. P-values for all statistical analyses are reported exactly with no particular cut-off used to define significance.[##REF##11159626##36##]</p>", "<p>At the time that the study was designed, few data on paediatric adherence from resource-constrained countries were available on which to base sample size calculations. The change from donor-funded ART to the government ART program necessitated ending of the study as the government program did not fund adherence monitoring. By this stage the accumulated sample size was sufficient to detect a 50% reduction in proportion with viral suppression in those with MR adherence &lt;90% with 90% power, assuming that 75% of children had MR adherence ≥ 90%.</p>" ]

[ "<title>Results</title>", "<p>Medication was returned on at least one occasion for 115/122 (94%) children who commenced ART with the remaining children deceased (n = 6) or lost to follow-up (n = 1) before their first follow-up visit (figure ##FIG##0##1##). After 1 year, 88 children were alive and remained in care on ART. Table ##TAB##0##1## shows the demographic and clinical characteristics and ART drugs prescribed at baseline. Children were young with a median age (IQR) of 37 (16 – 61) months. Although overall socio-economic status was poor with high levels of caregiver unemployment (73%) and informal housing (49%), most caregivers (88%) had at least secondary education and the majority of children (80%) were from households with access to water and electricity.</p>", "<title>Annual average MR adherence</title>", "<p>A total of 91/115 (79%) children achieved annual average MR adherence ≥ 90% with 73% of these having adherence ≥ 95% (figure ##FIG##1##2##). Only 9 (8%) children had average adherence for the full year below 80%. These percentages did not change substantially if adherence &lt;90% was assigned for missing returns in children not returning medication on all possible occasions. The number of children remaining in care at each month and the proportion returning medication as requested is shown in figure ##FIG##2##3##. There was no change in the proportion of children returning medication over time (p = 0.17). Among only those children who remained alive and in care for the entire first year of treatment (excluding those deceased and LFU; n = 88), the proportion with adherence &lt;90% decreased over time with an OR for having low adherence of 0.91 (95% CI: 0.87 – 0.96; p = 0.000) for each additional month (figure ##FIG##3##4##). Annual average MR adherence &lt;90% was more likely among the 34/115 (30%) child-caregiver pairs who failed to return empty medication containers/unused medication at more than one follow-up visit (OR = 4.97; 95% CI:1.92 – 12.87; p = 0.001).</p>", "<title>Relationship between MR adherence and viral/immunological outcomes</title>", "<p>Undetectable viral load was achieved in 62/80 (78%) children with annual average MR adherence ≥ 90% compared to 2/8 (25%) of children with lower adherence (OR = 10.3; 95% CI: 1.92 – 55.7; p = 0.005). In univariate analysis, other factors significantly associated with viral suppression were less wasting as reflected in higher weight-for-height z-score, less severe disease (WHO stage 2 or 3 vs WHO stage 4) and being on a non-ritonavir containing regimen (Table ##TAB##1##2##). Since nutritional status is an important determinant of WHO stage, only weight-for-height z-score and regimen were included in building an adjusted logistic regression model for the relationship between adherence and viral suppression. After adjustment for wasting, regimen was no longer associated with viral load outcome, however adherence ≥ 90% remained an important covariate associated with viral suppression (Adjusted OR = 5.5 [95%CI: 0.8–35.6], p = 0.075). Sensitivity analysis was performed by recalculating average annual adherence for caregiver-child pairs who did not return medication for one or more months by assigning adherence &lt;90% for months in which medication was not returned, and there remained a significant univariate association between MR adherence ≥ 90% and undetectable viral load, and a strong trend towards an association after adjustment for baseline weight-for-height z-score. There was no association between MR adherence in the month or 2 months preceding viral load measurement and viral suppression in either univariate analysis or after adjustment for baseline weight-for-height z-score. Similarly no association was found between the proportion of visits in which caregivers failed to return medication and viral suppression in either univariate or adjusted analyses.</p>", "<p>The median (IQR) changes in CD4 percent and absolute count were 10.1% (5.7% – 15.2%) and 393 cells/μl (113 – 654) respectively. There was no association between annual average MR adherence ≥ 90% and either CD4 percent at 1 year, or change from baseline CD4 percent or absolute count over 1 year in either univariate analysis or models adjusted for other determinants of immunological response.</p>", "<title>Excess MR Adherence</title>", "<p>Using the uncapped annual average MR adherence, 47/115 (40.9%) of children had adherence &gt;100%. This was more common among children taking ritonavir (30/50 [60%]) compared to those on efavirenz (17/65 [26%]; p = 0.000). Adherence &gt;100% was also more common in those 2 years of age or less (28/37 [75.7%]) compared to older children (19/78 [24.4%]; p = 0.000). This age-related difference in excess adherence was maintained after stratifying for ritonavir-based regimen (p = 0.02 and 0.05 for those with ritonavir-based and efavirenz-based regimens respectively).</p>", "<title>Questionnaire responses</title>", "<p>Of 98 children alive and in care after 3 months of ART, 87 (90%) completed adherence questionnaires (figure ##FIG##0##1##). Annual average MR adherence ≥ 90% was more common in those who completed questionnaires (59/87 [68%] vs 4/11 [36%]; p = 0.051).</p>", "<title>Reported missed doses</title>", "<p>NFA (missing ≥ 1 dose in the previous 3 days) was present in 12/87 (13.8%) children, and was more common in those whose MR adherence for that month was &lt;90% (7/22 [31.8%]) vs 5/62 [8.1%]; p = 0.006). Nevertheless, the sensitivity of NFA for poor MR adherence in the preceding month was only 31.8% (95% CI: 10.7% – 53.0%) and agreement between the two measures was only slightly better than that expected by chance (κ = 0.278; p = 0.003). There was no association between NFA and viral response at 1 year (p = 0.965). Although 38/87 (43.7%) of caregivers were unable to describe how to give their child's ART regimen exactly, this was not associated with either NFA or MR adherence &lt;90% (p = 0.88 and p = 0.68 respectively).</p>", "<title>Experience with giving medication</title>", "<p>A notable number of caregivers (33/87 [38.4%]) experienced problems with giving ART medication. Poor palatability of medication was the most common problem (21.8% of caregivers), with 68% of these being attributed to ritonavir. Change in daily routine was a problem for 12.6% of caregivers. Experiencing problems with giving medication did not affect reported (OR = 2.13; 95%CI: 0.59 – 7.65; p = 0.32) or MR adherence (OR = 0.61; 95%CI: 0.22 – 1.65; p = 0.32) in the month in which problems were reported, but was associated with annual average MR adherence &lt;90% (OR = 3.07; 95% CI: 0.91 – 10.38; p = 0.06).</p>", "<p>Most (65/87 [74.7%]) caregivers used at least one method to assist with remembering and giving medication. The most commonly used aids were activities of daily living reminders (35 [40%]) and treatment partners (23 [26%]). While the vast majority of caregivers (81 [93%]) believed that ART medication was improving their child's health, a significant number (24 [28%]) were unsure or believed that their children would not deteriorate if ART was stopped. Caregiver beliefs did not influence adherence by any measure.</p>", "<title>Determinants of annual average MR adherence ≥ 90%</title>", "<p>Socio-demographic variables most strongly associated with annual average MR adherence ≥ 90% on univariate analysis were caregivers having secondary education and household access to water and electricity as well as a refrigerator (tables ##TAB##0##1## and ##TAB##2##3##). Having secondary education was not significantly associated with any of the indicators of socio-economic status. Being on an efavirenz-based regimen was less strongly associated with good adherence. Secondary education of the caregiver and household access to water and electricity were both independent socio-demographic predictors of annual MR adherence ≥ 90%, while taking ritonavir was negatively associated with MR adherence ≥ 90%.</p>" ]

[ "<title>Discussion</title>", "<p>This study extends to infants and young children the finding that good adherence to ART in Africa is achievable with nearly 80% of children obtaining average MR adherence ≥ 90% over the first year of ART. [##REF##16343374##16##,##REF##12799558##25##,##REF##16896111##26##,##REF##17565809##28##] This is at least as good as pediatric adherence in rich countries.[##REF##10462336##13##,##UREF##4##17##, ####REF##12544410##18##, ##UREF##5##19##, ##UREF##6##20##, ##UREF##7##21##, ##REF##11927734##22####11927734##22##] This excellent adherence occurred despite nearly 40% of caregivers experiencing subjective difficulty with administering medication. Secondary education, access to water and electricity and a non-ritonavir based regimen were all independently associated with better adherence. This study demonstrates that although clinic-based medication measures may not be as sensitive as unannounced home-based measures or MEMS monitoring to detect poor adherence, they are still strongly predictive of virologic response. Annual average MR adherence ≥ 90% was associated with a greater than 5 fold increased likelihood of suppressing viral load at the end of the year.</p>", "<p>The excellent adherence seen in our study may, however, not be representative of current pediatric adherence in Africa as ART is scaled up. These were the first children to receive ART at our tertiary care institution and many were well-known to the HIV service as adherent with other medications and clinic visits. In addition, the social criteria used to determine ART eligibility may have further selected those patients more likely to be adherent. Furthermore, treatment was donor-funded in the context of no access to ART through government health services so caregivers may have felt that ART was a privilege. Moreover, regular adherence monitoring may have enhanced adherence. While the halving of the proportion of children with MR adherence &lt;90% during the year supports this, it is a problem inherent to all adherence studies. In addition, the actual estimate of adherence may be inflated by use of clinic-based medication measures. The finding by Muller et al.[##UREF##8##29##] of much lower adherence using MEMS caps in a recent post-national ART roll-out study in a similar setting to ours supports the likelihood of both particularly good adherence in our cohort, and of adherence being over-estimated by our method of measurement. Nevertheless, the strong association with viral suppression in our study indicates at least reasonable accuracy of adherence measurement.</p>", "<p>In contrast, reported NFA showed no relationship to viral outcome and was only 32% sensitive for MR adherence &lt;90%. While the failure to find an association with viral suppression may well be due to the fact that reported adherence was only measured once long before viral load measurement, the tendency for caregivers and patients to over-estimate adherence is well established.[##UREF##3##15##,##REF##12794557##23##,##REF##17565809##28##,##UREF##8##29##,##REF##11352698##32##] In our study, this may have been exacerbated by interviews being conducted by clinicians to whom caregivers might be reluctant to admit to missing doses, particularly in the context of limited access to ART. This, however, reflects clinical practice.[##UREF##3##15##], and any attempts to develop a screening tool for poor adherence in busy roll-out clinics must take into account under-reporting of non-adherence to clinicians. In this study, the additional measuring of adherence by medication return should have, at least in part, mitigated against under-reporting.</p>", "<p>Interestingly we found no association between MR adherence in the month or 2 months immediately prior to viral load measurement, in contrast to the strong association with average MR adherence over the whole year. Other studies have found strong associations between short-term adherence and viral suppression immediately thereafter. [##REF##10877736##12##,##UREF##8##29##] Having a single viral load measurement only after a full year on treatment made it impossible to examine the effect of short-term fluctuations in adherence on viral outcome. The small sample in this study may also have reduced its power to show such an association particularly as a far smaller number of children than expected had poor adherence. In addition, the sample size limited the number of predictor variables that could be used in analytic models. We were only able to include a single indicator of disease severity in the model of viral suppression and only able to include socio-demographic and regimen factors as predictors of MR adherence ≥ 90%.</p>", "<p>It is notable that nearly 40% of caregivers experienced problems with giving ART, and that this adversely affected adherence over the whole year. Poor palatability of medication was the most common problem reported, especially in those taking ritonavir. This problem is unique to infants and very young children taking liquid formulations of drugs, and concurs with international studies.[##REF##12544410##18##,##REF##11927734##22##] A limitation of our study is that drug formulation was not recorded so its effect on adherence could not be examined. However, the finding that a greater proportion of children under 2 years of age (who would all be taking liquid formulations) had adherence &gt;100% suggests that repeat dosing of syrups/solutions is frequently necessary, placing an additional burden on caregivers.</p>", "<p>It is further noteworthy that MR adherence results in the ritonavir group appear paradoxical: taking ritonavir was both negatively associated with (capped) MR adherence ≥ 90%, and positively associated with (uncapped) MR adherence &gt;100%. The excess adherence is explained by frequent need for repeat dosing. However poor adherence in the ritonavir group when measures are capped indicates that there is not always compliance with frequent dosing with an unpalatable drug ultimately impacting negatively on adherence. Indeed the poor adherence in the ritonavir group was attributable to poor adherence to ritonavir alone, with adherence to the other two drugs in the regimen being acceptable (data not shown). Although ritonavir is no longer used as a single third agent in ART regimens, Kaletra^®(lopinavir/ritonavir) is also unpalatable and is recommended in the South African national guidelines as part of the first-line regimen in children under 3 years of age.[##UREF##11##37##] Moreover, drugs other than ritonavir accounted for nearly a third of palatability problems. The need for pediatric-friendly formulations of ART cannot be over-emphasized.</p>", "<p>While some studies have suggested that older children are more likely to be non-adherent, we found no relationship between age and adherence.[##UREF##6##20##,##REF##16280708##30##] However, in our study the median age was young (36 months) with few children approaching adolescence, so its potential negative effect could not be determined. Nevertheless, it should be noted that the age of children commencing ART at our institution has decreased since this initial program, with the majority of children now being less than 2 year old.[##REF##17077930##8##] Our findings may therefore not be applicable to our current patient cohort, and there is a need for further research into adherence in very young infants in Africa.</p>", "<p>While previous research in Africa has found little impact of socio-economic status on adherence, our study suggests that better caregiver education and socio-economic status are both strongly independently associated with better adherence. [##REF##16343374##16##,##REF##12799558##25##,##REF##17121449##27##,##UREF##8##29##] Caregivers with secondary education and those with access to water and electricity are 4.5 times and 2.7 times more likely to have adherence ≥ 90% respectively. The lack of association between caregiver education and any of the indicators of socio-economic status emphasizes that these both impact on adherence and one is not simply a surrogate marker of the other. Compared to adult ART, administering medication to children, particularly infants, is more complex and requires exact measurement of dosages, often to a fraction of a milliliter, and compliance with stringent storage requirements. In addition, unlike adults where dosage remains constant over a long period, dosages for children change frequently due to their rapid growth. Although we did not examine the effect of complex dosages and dosage changes on MR adherence, the positive impact on adherence in children of better education and socio-economic status of their caregivers is not surprising. However, the failure to find an association with caregiver's ability to describe a regimen and adherence suggests that it is the caregiver's overall education that impacts more on treatment adherence than treatment literacy per se.</p>" ]

[ "<title>Conclusion</title>", "<p>This study demonstrates the potential for caregivers of African young children to achieve adherence comparable with that of wealthier countries. The association between MR adherence and viral response attests to the value of a relatively simple low technology tool for measuring adherence i.e. clinic-based medication return for detecting lapses in adherence. Nevertheless, repeated measurements of medication returned, particularly of syrups/solutions, are not feasible in most large program scale-up settings and the poor sensitivity of low reported adherence for low MR adherence highlights the need to develop practical easy-to-use reliable screening tools to detect children in whom more intensive adherence monitoring is indicated. [##REF##16280708##30##] The negative impact of problems experienced with giving ART, unpalatable drugs, poor caregiver education and socio-economic status on adherence in this study underscores the need for more pediatric-friendly drug formulations as well as the importance of supporting caregivers in providing ART to children.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Antiretroviral therapy (ART) dramatically improves outcomes for children in Africa; however excellent adherence is required for treatment success. This study describes the utility of different measures of adherence in detecting lapses in infants and young children in Cape Town, South Africa.</p>", "<title>Methods</title>", "<p>In a prospective cohort of 122 HIV-infected children commenced on ART, adherence was measured monthly during the first year of treatment by medication return (MR) for both syrups and tablets/capsules. A questionnaire was administered to caregivers after 3 months of treatment to assess experience with giving medication and self-reported adherence. Viral and immune response to treatment were assessed at the end of one year and associations with measured adherence determined.</p>", "<title>Results</title>", "<p>Medication was returned for 115/122 (94%) children with median age (IQR) of 37 (16 – 61) months. Ninety-one (79%) children achieved annual average MR adherence ≥ 90%. This was an important covariate associated with viral suppression after adjustment for disease severity (OR = 5.5 [95%CI: 0.8–35.6], p = 0.075), however was not associated with immunological response to ART. By 3 months on ART, 13 (10%) children had deceased and 11 (10%) were lost to follow-up. Questionnaires were completed by 87/98 (90%) of caregivers of those who remained in care. Sensitivity of poor reported adherence (missing ≥ 1 dose in the previous 3 days) for MR adherence &lt;90% was only 31.8% (95% CI: 10.7% – 53.0%). Caregivers of 33/87 (38.4%) children reported difficulties with giving medication, most commonly poor palatability (21.8%). Independent socio-demographic predictors of MR adherence ≥ 90% were secondary education of caregivers (OR = 4.49; 95%CI: 1.10 – 18.24) and access to water and electricity (OR = 2.65; 95%CI: 0.93 – 7.55). Taking ritonavir was negatively associated with MR adherence ≥ 90% (OR = 0.37; 95%CI: 0.13 – 1.02).</p>", "<title>Conclusion</title>", "<p>Excellent adherence to ART is possible in African infants and young children and the relatively simple low technology measure of adherence by MR strongly predicts viral response. Better socio-economic status and more palatable regimens are associated with better adherence.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>MD conceived of the study, participated in its design and co-ordination, performed statistical analysis and drafted the manuscript. BE participated in the design and co-ordination of the study and advised on analysis. AB provided advice on statistical analysis. JN participated in the design and co-ordination of the study. TF carried out all medication measurements. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2431/8/34/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We would like to thank staff members of the Infectious Diseases Clinic and Immunology Laboratory of Red Cross Children's Hospital who provided clinical care to the patients and laboratory services respectively. Donations to fund the program were received from Syfrets Trust Ltd, Merck (Pty) Ltd, Bristol-Myers Squibb Foundation, Durbanville High School and the University of Cape Town. Mary-Ann Davies and Andrew Boulle receive support from the International Epidemiological Databases to Evaluate AIDS in Southern Africa (IeDEASA) collaboration which is funded by the National Institutes for Health (NIH; U01 AI069924-01)</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Profile of study.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Annual average MR adherence (n = 115).</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Number of children in care at the end of each month on ART</bold>. Total number of children in care at the end of each month divided into those returning and not returning medication</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Changes in proportion with MR adherence &lt;90% during the first year on ART</bold>. Only children who remained alive and in care for entire first year of treatment (excludes children deceased or LFU; n = 88).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Social, demographic and clinical characteristics at start of ART according to annual average MR adherence.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Variable</bold></td><td align=\"left\"><bold>All children (n = 122)</bold></td><td align=\"left\"><bold>Adherence ≥ 90% (n = 91) (79%)</bold></td><td align=\"left\"><bold>Adherence&lt;90% (n = 24) (21%)</bold></td><td align=\"left\"><bold>p-value</bold></td></tr></thead><tbody><tr><td align=\"left\">Median age (months) (IQR)</td><td align=\"left\">37 (16 – 61)</td><td align=\"left\">37 (16 – 57)</td><td align=\"left\">49 (20 – 72)</td><td align=\"left\">0.49*</td></tr><tr><td align=\"left\">Gender</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Female (%)</td><td align=\"left\">52 (43)</td><td align=\"left\">39 (43)</td><td align=\"left\">12 (50)</td><td align=\"left\">0.53^†</td></tr><tr><td align=\"left\">WHO stage</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> 4 (%)</td><td align=\"left\">54 (44)</td><td align=\"left\">34 (37)</td><td align=\"left\">14 (58%)</td><td align=\"left\">0.06^†</td></tr><tr><td align=\"left\">Mean weight-for-age z-score (sd)</td><td align=\"left\">-2.02 (1.54)</td><td align=\"left\">-1.82 (1.55)</td><td align=\"left\">-2.41 (1.37)</td><td align=\"left\">0.09‡</td></tr><tr><td align=\"left\">Mean height-for-age z-score (sd)</td><td align=\"left\">-2.71 (1.36)</td><td align=\"left\">-2.66 (1.36)</td><td align=\"left\">-2.74 (1.33)</td><td align=\"left\">0.78‡</td></tr><tr><td align=\"left\">Mean weight-for-height z-score (sd)</td><td align=\"left\">-0.56 (1.44)</td><td align=\"left\">-0.31 (1.34)</td><td align=\"left\">-1.08 (1.42)</td><td align=\"left\">0.02‡</td></tr><tr><td align=\"left\">Median CD4 percent (IQR)</td><td align=\"left\">11.1 (6.9 – 15.0)</td><td align=\"left\">11.4 (8.0 – 15.1)</td><td align=\"left\">10.1 (4.2 – 14.3)</td><td align=\"left\">0.18*</td></tr><tr><td align=\"left\">Median CD4 count (IQR)</td><td align=\"left\">556 (242 – 908)</td><td align=\"left\">592 (273 – 938)</td><td align=\"left\">518 (105 – 758)</td><td align=\"left\">0.23*</td></tr><tr><td align=\"left\">Median Log Viral load (IQR)</td><td align=\"left\">5.57 (5.15 – 6.08)</td><td align=\"left\">5.44 (5.11 – 6.08)</td><td align=\"left\">5.77 (5.30 – 6.06)</td><td align=\"left\">0.55*</td></tr><tr><td align=\"left\">Primary caregiver</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Mother (%)</td><td align=\"left\">107 (88)</td><td align=\"left\">79 (87)</td><td align=\"left\">21 (88)</td><td align=\"left\">0.72^§</td></tr><tr><td align=\"left\"> Not mother (%)</td><td align=\"left\">12 (10)</td><td align=\"left\">9 (10)</td><td align=\"left\">3 (13)</td><td/></tr><tr><td align=\"left\"> Unknown (%)</td><td align=\"left\">3 (2)</td><td align=\"left\">3 (3)</td><td align=\"left\">0 (0)</td><td/></tr><tr><td align=\"left\">Median age of caregiver (IQR)</td><td align=\"left\">29 (26 – 32)</td><td align=\"left\">30 (26 – 32)</td><td align=\"left\">28 (25 – 35)</td><td align=\"left\">0.58*</td></tr><tr><td align=\"left\">Father provides financial support</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes (%)</td><td align=\"left\">48 (39)</td><td align=\"left\">37 (41)</td><td align=\"left\">9 (38)</td><td align=\"left\">0.8^†</td></tr><tr><td align=\"left\"> No (%)</td><td align=\"left\">70 (57)</td><td align=\"left\">51 (56)</td><td align=\"left\">14 (58)</td><td/></tr><tr><td align=\"left\"> Unknown (%)</td><td align=\"left\">4 (3)</td><td align=\"left\">3 (3)</td><td align=\"left\">1 (4)</td><td/></tr><tr><td align=\"left\">Caregiver has secondary education</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes (%)</td><td align=\"left\">107 (88)</td><td align=\"left\">83 (91)</td><td align=\"left\">18 (75)</td><td align=\"left\">0.03^†</td></tr><tr><td align=\"left\"> No (%)</td><td align=\"left\">11 (9)</td><td align=\"left\">6 (7)</td><td align=\"left\">5 (21)</td><td/></tr><tr><td align=\"left\"> Unknown (%)</td><td align=\"left\">4 (3)</td><td align=\"left\">2 (2)</td><td align=\"left\">1 (4)</td><td/></tr><tr><td align=\"left\">Caregiver employed</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes (%)</td><td align=\"left\">27 (22)</td><td align=\"left\">23 (25)</td><td align=\"left\">4 (17)</td><td align=\"left\">0.37^†</td></tr><tr><td align=\"left\"> No (%)</td><td align=\"left\">89 (73)</td><td align=\"left\">64 (70)</td><td align=\"left\">19 (79)</td><td/></tr><tr><td align=\"left\"> Unknown (%)</td><td align=\"left\">6 (5)</td><td align=\"left\">4 (4)</td><td align=\"left\">1 (4)</td><td/></tr><tr><td align=\"left\">Caregiver/child receives social grant</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes (%)</td><td align=\"left\">69 (57)</td><td align=\"left\">55 (60)</td><td align=\"left\">11 (46)</td><td align=\"left\">0.16^†</td></tr><tr><td align=\"left\"> No (%)</td><td align=\"left\">51 (42)</td><td align=\"left\">34 (37)</td><td align=\"left\">13 (54)</td><td/></tr><tr><td align=\"left\"> Unknown (%)</td><td align=\"left\">2 (2)</td><td align=\"left\">2 (2)</td><td align=\"left\">0 (0)</td><td/></tr><tr><td align=\"left\">Formal housing</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes (%)</td><td align=\"left\">62 (51)</td><td align=\"left\">50 (55)</td><td align=\"left\">8 (33)</td><td align=\"left\">0.06^†</td></tr><tr><td align=\"left\"> No (%)</td><td align=\"left\">60 (49)</td><td align=\"left\">41 (45)</td><td align=\"left\">16 (67)</td><td/></tr><tr><td align=\"left\">Access to water and electricity</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes (%)</td><td align=\"left\">97 (80)</td><td align=\"left\">76 (84)</td><td align=\"left\">15 (63)</td><td align=\"left\">0.024^†</td></tr><tr><td align=\"left\"> No (%)</td><td align=\"left\">25 (20)</td><td align=\"left\">15 (16)</td><td align=\"left\">9 (38)</td><td/></tr><tr><td align=\"left\">Access to working refrigerator</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes (%)</td><td align=\"left\">85 (70)</td><td align=\"left\">66 (73)</td><td align=\"left\">12 (50)</td><td align=\"left\">0.036^†</td></tr><tr><td align=\"left\"> No (%)</td><td align=\"left\">37 (30)</td><td align=\"left\">15 (16)</td><td align=\"left\">12 (50)</td><td/></tr><tr><td align=\"left\">Medication</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> d4T (%)</td><td align=\"left\">110 (90)</td><td align=\"left\">81 (89)</td><td align=\"left\">22 (92)</td><td align=\"left\">1.00*</td></tr><tr><td align=\"left\"> AZT (%)</td><td align=\"left\">11 (9)</td><td align=\"left\">8 (9)</td><td align=\"left\">3 (13)</td><td align=\"left\">0.70*</td></tr><tr><td align=\"left\"> 3TC (%)</td><td align=\"left\">117 (96)</td><td align=\"left\">88 (97)</td><td align=\"left\">22 (92)</td><td align=\"left\">0.28*</td></tr><tr><td align=\"left\"> ddI (%)</td><td align=\"left\">4 (3)</td><td align=\"left\">3 (3)</td><td align=\"left\">1 (4)</td><td align=\"left\">1.00*</td></tr><tr><td align=\"left\"> Efavirenz (%)</td><td align=\"left\">67(55)</td><td align=\"left\">55 (60)</td><td align=\"left\">10 (42)</td><td align=\"left\">0.08^†</td></tr><tr><td align=\"left\"> Ritonavir (%)</td><td align=\"left\">55 (45)</td><td align=\"left\">36 (40)</td><td align=\"left\">14 (58)</td><td align=\"left\">0.10^†</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Univariate and multivariate factors associated with undetectable viral load (n = 88)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Variable</bold></td><td align=\"left\"><bold>Unadjusted OR</bold></td><td align=\"left\"><bold>95%CI</bold></td><td align=\"left\"><bold>p-value</bold></td><td align=\"left\"><bold>Adjusted OR</bold></td><td align=\"left\"><bold>95%CI</bold></td><td align=\"left\"><bold>p-value</bold></td></tr></thead><tbody><tr><td align=\"left\">Female gender</td><td align=\"left\">0.91</td><td align=\"left\">0.36 – 2.35</td><td align=\"left\">0.86</td><td align=\"left\">^†</td><td/><td/></tr><tr><td align=\"left\">Age at treatment start (months)</td><td align=\"left\">1.00</td><td align=\"left\">0.99 – 1.01</td><td align=\"left\">0.81</td><td align=\"left\">^†</td><td/><td/></tr><tr><td align=\"left\">WHO stage 2&amp;3 (vs WHO stage 4)</td><td align=\"left\">2.80</td><td align=\"left\">1.07 – 7.35</td><td align=\"left\">0.04</td><td align=\"left\">*</td><td/><td/></tr><tr><td align=\"left\">Weight-for-age z-score</td><td align=\"left\">1.29</td><td align=\"left\">0.94 – 1.77</td><td align=\"left\">0.12</td><td align=\"left\">*</td><td/><td/></tr><tr><td align=\"left\">Height-for-age z-score</td><td align=\"left\">1.07</td><td align=\"left\">0.74 – 1.54</td><td align=\"left\">0.71</td><td align=\"left\">*</td><td/><td/></tr><tr><td align=\"left\">Weight-for-height z-score</td><td align=\"left\">1.94</td><td align=\"left\">1.27 – 2.96</td><td align=\"left\">0.002</td><td align=\"left\">1.8</td><td align=\"left\">1.15 – 2.80</td><td align=\"left\">0.01</td></tr><tr><td align=\"left\">CD4 percent</td><td align=\"left\">1.04</td><td align=\"left\">0.97 – 1.12</td><td align=\"left\">0.27</td><td align=\"left\">*</td><td/><td/></tr><tr><td align=\"left\">CD4 absolute (cells/l)</td><td align=\"left\">1.44</td><td align=\"left\">0.54 – 3.84</td><td align=\"left\">0.47</td><td align=\"left\">*</td><td/><td/></tr><tr><td align=\"left\">Log viral load</td><td align=\"left\">0.87</td><td align=\"left\">0.48 – 1.60</td><td align=\"left\">0.66</td><td align=\"left\">*</td><td/><td/></tr><tr><td align=\"left\">Ritonavir-containing regimen</td><td align=\"left\">0.33</td><td align=\"left\">0.13 – 0.87</td><td align=\"left\">0.03</td><td align=\"left\">^†</td><td/><td/></tr><tr><td align=\"left\">Annual average MR adherence ≥ 90%</td><td align=\"left\">10.30</td><td align=\"left\">1.92 – 55.67</td><td align=\"left\">0.005</td><td align=\"left\">5.48</td><td align=\"left\">0.84 – 35.58</td><td align=\"left\">0.075</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Factors associated with annual average MR adherence ≥ 90%. (n = 115)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Variable</bold></td><td align=\"left\"><bold>Unadjusted OR</bold></td><td align=\"left\"><bold>95% CI</bold></td><td align=\"left\"><bold>p-value</bold></td><td align=\"left\"><bold>Adjusted OR</bold></td><td align=\"left\"><bold>95% CI</bold></td><td align=\"left\"><bold>p-value</bold></td></tr></thead><tbody><tr><td align=\"left\">Access to water and electricity</td><td align=\"left\">3.04</td><td align=\"left\">1.12 – 8.22</td><td align=\"left\">0.028</td><td align=\"left\">2.65</td><td align=\"left\">0.93 – 7.55</td><td align=\"left\">0.069</td></tr><tr><td align=\"left\">Formal housing</td><td align=\"left\">2.44</td><td align=\"left\">0.95 – 6.27</td><td align=\"left\">0.064</td><td align=\"left\">*</td><td/><td/></tr><tr><td align=\"left\">Access to working refrigerator</td><td align=\"left\">2.64</td><td align=\"left\">1.04 – 6.65</td><td align=\"left\">0.039</td><td align=\"left\">*</td><td/><td/></tr><tr><td align=\"left\">Secondary education (&gt;Std 5/Grade 7)</td><td align=\"left\">3.84</td><td align=\"left\">1.06 – 13.98</td><td align=\"left\">0.041</td><td align=\"left\">4.49</td><td align=\"left\">1.10 – 18.24</td><td align=\"left\">0.035</td></tr><tr><td align=\"left\">Taking ritonavir</td><td align=\"left\">0.44</td><td align=\"left\">0.18 – 1.11</td><td align=\"left\">0.084</td><td align=\"left\">0.37</td><td align=\"left\">0.13 – 1.02</td><td align=\"left\">0.054</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>*Wilcoxon rank sum test</p><p>‡Student's t-test</p><p>^†Chi²test</p><p>^§Fisher's exact test</p></table-wrap-foot>", "<table-wrap-foot><p>*Not included in building multivariate model</p><p>^†Not significantly associated with viral suppression after adjustment for weight-for-height z-score</p></table-wrap-foot>", "<table-wrap-foot><p>*Not included in multivariate model as only one measure of socio-economic status (access to water and electricity) used.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2431-8-34-1\"/>", "<graphic xlink:href=\"1471-2431-8-34-2\"/>", "<graphic xlink:href=\"1471-2431-8-34-3\"/>", "<graphic xlink:href=\"1471-2431-8-34-4\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The shift to family/patient-centered models of care has increased the need for patient reported outcomes. Valid and reliable health-related quality of life (HRQOL) instruments are therefore expected to be in the armamentarium of clinicians and health service researchers [##REF##15264945##1##,##REF##16464429##2##].</p>", "<p>The only HRQOL instrument that has been validated in Argentinean children is the Child's Health Questionnaire (CHQ) in children with Juvenile Rheumatoid Arthritis [##REF##11510309##3##,##REF##11510308##4##]. One of the limitations of this instrument however, is that it does not include the child's perspective for children younger than 10 years of age.</p>", "<p>The Pediatric Quality of Life Inventory™ (PedsQL™) 4.0 Generic Core Scales is a generic HRQOL instrument for children and adolescents, originally developed by Varni et al. in U.S. English and U.S Spanish [##REF##10024117##5##]. It measures four domains (physical, emotional, social, and school functioning) and has age and respondent specific versions for child self-report ages 5–18 and parent proxy-report for ages 2–18. The PedsQL™ has shown good internal consistency (α = 0.88 child, and α = 0.90 parent report)[##REF##11468499##6##,##REF##11977437##7##] and has been widely used for group comparisons. The construct validity of PedsQL is supported by results from large samples of children from the US [##REF##11977437##7##, ####REF##16468077##8##, ##REF##17201920##9##, ##REF##12777555##10####12777555##10##]and several other countries [##REF##18362824##11##, ####REF##18080786##12##, ##REF##17681083##13##, ##REF##16972987##14##, ##REF##16756527##15##, ##REF##15804349##16####15804349##16##] where the instrument has been translated using accepted cross cultural language adaptation methods[##REF##8263569##17##]. These studies have given support to the instrument's ability to discriminate between healthy children and those with chronic conditions[##REF##11977437##7##,##REF##18362824##11##,##REF##18080786##12##,##REF##16756527##15##,##REF##15804349##16##,##REF##17616837##18##] and among different chronic conditions[##REF##15804349##16##,##REF##17634123##19##, ####REF##18205028##20##, ##REF##15632338##21####15632338##21##]. Responsiveness, i.e. score change after an intervention, has been reported for specific conditions such as rheumatic diseases[##REF##11920407##22##], headaches[##REF##16107544##23##], and cancer[##REF##18391698##24##,##REF##11932914##25##] and sensitivity, i.e. ability to distinguish among severity groups, for heart disease[##REF##11977437##7##], obesity[##REF##15632338##21##] and cancer[##REF##18391698##24##,##REF##11932914##25##] has also been described. In addition, the PedsQL is able to discriminate among children from lower socioeconomic strata[##REF##16468077##8##,##REF##18362824##11##] and predict variation in health care utilization and costs[##REF##15361252##26##,##REF##14616041##27##].</p>", "<p>The aim of this study was to validate the Argentinean Spanish version of the PedsQL™ 4.0 in children and adolescents with chronic conditions. Given that families who receive care at public health settings in Argentina come from low income sectors, usually have low literacy skills, and are not used to self-reporting their health status, we specially focused on the impact of socio-demographic characteristics on overall comprehensibility and acceptability.</p>" ]

[ "<title>Methods</title>", "<title>Subjects</title>", "<p>Patients were considered eligible if they were: (1) 2–18 years old, (2) receiving outpatient care at Hospital Nacional de Pediatria Juan P Garrahan, and (3) had one of the following conditions: Allogeneic Hematopoietic Stem Cell Transplantation (SCT), Chronic Obstructive Pulmonary Disease requiring domiciliary oxygen (COPD), Human Immunodeficiency Virus infection or Acquired Immune Deficiency Syndrome (HIV/AIDS), Cancer, End Stage Renal Disease (ESRD) requiring dialysis or transplant, or a Complex Congenital Cardiopathy (CCC). Patients were excluded if they had not been clinically stable in the last month (i.e., deterioration and/or acute complication related or not to their preexisting condition), had comorbidities, or were not cognitively able to complete the questionnaire. Data were collected from July 2004 to June 2005.</p>", "<p>An additional convenience sample of healthy children and adolescents was gathered to assess comprehensibility and test discriminant validity. Eligibility criteria for this sample were: (1) 2–18 years old, (2.a) attending the \"Healthy Children Outpatient Clinic\" at one of the three pediatric hospitals in the city or (2.b) students at one elementary school in the outskirts of Buenos Aires. These recruitment sources were selected because the socio-demographic characteristics of children were similar to those of the chronically ill children cared for at Hospital Garrahan. The study was approved by Hospital Garrahan's IRB. Parents or legal guardians granted written permission and children 10 years old and above were asked for assent.</p>" ]

[ "<title>Results</title>", "<p>Among 296 eligible families of children with chronic conditions 287 (96%) enrolled. Figure ##FIG##1##2## presents the study flowchart and diagnosis of the enrolled families. In Table ##TAB##0##1## their clinical and socio-demographic characteristics are presented.</p>", "<p>The distribution of socio-demographic characteristics across the different age groups was homogenous, with slight predominance of males in all of them. Twenty-five percent of children were below the appropriate school level for their age, and 6% were not attending school; 11% of adult respondents had not completed elementary school and 3.2 % were functional illiterates. Most surveyed families lived below the poverty line (66%) and 54% had no health insurance.</p>", "<p>Out of 107 eligible families of healthy children, 105 (98%) enrolled. Healthy children were comparable to those with chronic conditions, except for gender and socioeconomic status. Healthy children were more likely to be females (55% vs. 42%, p = 0.023), have no medical insurance (74% vs. 54%, p = 0.001), and less likely to live below the poverty line (54% vs. 66%, p = 0.046).</p>", "<title>Feasibility</title>", "<p>In Table ##TAB##1##2##, we present feasibility of administering the Argentinean Spanish version of the PedsQL™ 4.0. Overall, the instrument was well understood. Median time to completion was 6 minutes for children (range 2–28') and 5 minutes for adults (range 1–16'). In 54% of the cases the age-appropriate questionnaire was completed without help and in 27.5% with minimal help. The need for help decreased with age. Among the 217 children with chronic conditions surveyed, 9 (4.1%), all aged 5 to 7 years, were not able to understand and complete the questionnaire and 7 (3.5%), all aged 8 to 12 years, needed to use the young child version for 5–7 year olds. An additional 7%, mostly 8–12 year-olds, required the PedsQL™ to be administered by the interviewer. No health condition was associated with not being able to answer. No adult questionnaire was unanswered. Main difficulty for adults was with format, 12.5% forgot to complete an item or more and needed to be prompted by the interviewer in order to complete it adequately.</p>", "<p>Poverty and a low education level were significantly associated with requiring more help to complete the PedsQL™ 4.0 for both children and parents (Table ##TAB##2##3##). When both poverty and low education level were present, 30% of children and 19% of parents required significant help whereas only 15% of children and 4% of parents required significant help if they were not in this category (p = 0.049 for children and 0.001 for parents). All but one of the children who could not complete the questionnaire lived below the poverty line.</p>", "<p>There were few missing items. Children only left 2.4% (115/4784) items unanswered whereas adults left 4.3% (218/6461). Five items from the school dimension were responsible for 78% of children's and 90% of adult's missing items, and corresponded to children that were not going to school.</p>", "<p>Almost all children (95%) and parents (96%) considered the questions relevant, a large proportion found them easy to answer (81% of children and 91% of parents), and most said the paper format was friendly (91% of children and 98% of parents).</p>", "<title>Scores Distribution</title>", "<p>In Table ##TAB##3##4##, average summary and scales scores, standard deviations and range, as well as ceiling and floor effects are presented. Children and adults used the complete range of response options for all 23 items with a slight deviation towards the uppermost end. Ceiling and floor effects were negligible for all dimensions but the social domain, where a moderate ceiling effect (20.2%) was observed in proxy respondents.</p>", "<p>Older children had significantly higher scores than younger children (Table ##TAB##4##5##), except for the emotional dimension. In contrast, parent proxy-report scores for the 2–4 year-olds were significantly higher than proxy-report scores of older children.</p>", "<title>Reliability</title>", "<p>Cronbach's alpha coefficients for the summary and scale scores for all children with chronic conditions are presented in Table ##TAB##3##4##. Table ##TAB##4##5## presents results by age group. The internal consistency of the total scores, and the physical and psychosocial subscale scores exceeded the 0.70 minimum usually accepted for group comparison for all age groups except for the 2–4 year-olds proxy-report, and the physical functioning and psychosocial subscales of the 5–7 year-olds self-report (α = 0.57 and α = 0.65 respectively). In the 2–4 year-old group, educational items were missing for 51 (72.9%) patients. When these three items were excluded, internal consistency increased markedly (total α = 0.83 and psychosocial α = 0.76)). Emotional, social, and school subscales had overall lower reliability although the proxy-reports of the 5–7, 8–12, and 13–18 year-olds, and the 13–18 year-olds self-report were close or superior to the 0.70 mark (except for the emotional subscale of the 8–12 year-old proxy-reports with an α = 0.62) and below 0.65 for the other groups (the 5–7 year-old self-reports being the lowest). Among child self-reports, internal consistency increased with age.</p>", "<title>Construct Validity</title>", "<p>As hypothesized, there was a significant and negative correlation between the primary physician's assessment of health impairment status (VASphys) and both self-report and proxy total PedsQL™ 4.0 scores (Table ##TAB##5##6##). Correlation between total PedsQL™ scores and overall self-reported/proxy health status evaluation was significant and positive in both children and adults. Total self-report and proxy-report scores were also significantly correlated. Of note, self-report global scores were significantly lower than proxy-report global scores. All correlations were in the moderate range (&lt;&gt; 0.20–0.50).</p>", "<title>Discriminant Validity</title>", "<p>As expected, child self-report and parent proxy-report total, physical, and psychosocial scores for healthy children were on average significantly higher than those of children with chronic conditions (Table ##TAB##6##7##) except for the emotional and school self-report subscales. PedsQL™ 4.0 total scores also varied significantly across health conditions for both self-reports and proxy-reports (Table ##TAB##6##7##). Patients with COPD, ESRD, or cancer reported the lowest scores.</p>", "<p>Children living below the poverty line were more likely to have lower total PedsQL™ scores (65.38 vs 70.29 respectively, p = 0.035) than their counterparts. These were mainly due to significantly lower emotional and school functioning scores. No statistically significant differences were found between PedsQL™ scores and gender.</p>", "<title>Comparison with other cross-cultural adaptations</title>", "<p>Table ##TAB##7##8## presents how results from our study compare to the original validation study and other published cross-cultural validations of PedsQL™. For most cross-cultural validation studies population characteristics differed from ours. Target population was commonly restricted to school children, and thus children were older and healthier. In addition, because of country characteristics, socioeconomic status tended to be higher compared to the Argentinean families we recruited. Our scores were overall lower than most of the other validation studies, including those that included similar age ranges and conditions. Reliability was reported in different ways across these studies, but the lower bound of internal consistencies found by our study was lower than the ones reported for most of the other validation studies. Types of validity tested and findings were similar to those reported by the other cross-cultural adaptations.</p>" ]

[ "<title>Discussion</title>", "<p>Our study results provide initial evidence towards the reliability and validity of the Argentinean Spanish version of the PedsQL™ 4.0 Generic Core Scales in the public health research setting. The Argentinean Spanish version of the PedsQL™ 4.0 has good feasibility. It was easy to administer, completed without or with minimum help by most children and parents, required a short administration time (not more than 5–6 minutes on average), and only 4.1% of children (all 5–7 year-olds) could not complete the instrument. However, our results suggest that some sort of help, albeit small, is needed for many, especially for children and parents from lower socioeconomic strata and low literacy levels. Internal consistency approached or exceeded that required for group comparisons for children over 8 years old and parents of children over 5 years old. The Argentinean Spanish version of the PedsQL™ 4.0 showed good construct and discriminant validity properties in this low-income setting, making this instrument suitable for research use. In order to expand the use of the PedsQL™ 4.0 in Argentinean children, an alternative approach to scoring for the 2–4 year-olds should be considered along with further understanding of how to increase reliability for the 5–7 year-old self-report and assessment of other instrument characteristics such as responsiveness and sensitivity to change.</p>", "<p>Our initial concern that socioeconomic status and literacy may influence people's ability to use PedsQL™ 4.0 seems to be supported by our data, although to a lesser extent than was expected. As a matter of fact, all children that could not complete PedsQL™ 4.0 lived below the poverty line and both children and parents who were poor and had low literacy levels were more likely to require help with the instrument. Nevertheless, the 14.5% of 5–7 year-olds who could not complete PedsQL™ was lower than the 38% observed in the German validation of the PedsQL™[##REF##15058802##38##], and was also within our a priori requirement of &lt; 20% unanswered questionnaires. Importantly, all the parents were able to complete the questionnaire, albeit with assistance, even those that had not completed elementary school or were functional illiterates. The main implications of these findings are that in order to use PedsQL™ in our public health setting, availability of trained interviewers during questionnaire administration needs to be assured, especially for children and parents who are poor and have low literacy levels. In addition, carefully thought training guidelines for children and parents should be developed and tested.</p>", "<p>The Argentinean Spanish PedsQL™ version had lower reliability compared to other validation studies[##REF##18362824##11##, ####REF##18080786##12##, ##REF##17681083##13####17681083##13##,##REF##16756527##15##,##REF##15804349##16##,##REF##17616837##18##,##REF##18205028##20##,##REF##15058802##38##,##REF##17569714##39##]. Given the low prevalence of school attendance among the 2–4 year olds with chronic conditions, this version of the Argentinean Spanish PedsQL™ may work better if school items are not taken into consideration for scoring purposes in this group. In addition, although Cronbach alpha represents the lower bound of the reliability of a measurement instrument, and is a conservative estimate of actual reliability[##REF##5232569##40##], scales that did not approach or meet the 0.70 standard should be used only for descriptive analyses. Self-report scores of 5–7 year-olds presented the lowest internal consistency values. Of note, these children had the most difficulty with completing PedsQL™, which may be indicating that results of the Argentinean Spanish PedsQL™ version for this age group may not be as reliable as for the older groups. Although these results are somewhat comparable to the German validation[##REF##15058802##38##], other studies in this age group [##REF##17201920##9##,##REF##18205028##20##] have showed higher alpha coefficients and less problem with instrument completion. HRQOL measurement in young children is still challenging and our results warrant further research including larger samples[##REF##10998003##41##,##REF##15264962##42##].</p>", "<p>Construct validity was assessed in a similar fashion to other validation studies[##REF##11468499##6##,##REF##18080786##12##, ####REF##17681083##13##, ##REF##16972987##14####16972987##14##,##REF##15804349##16##,##REF##18205028##20##,##REF##15058802##38##,##REF##16046915##43##] and supported by our data. The self-reported health status VAS scales had not been used before in our setting, but there is substantial evidence that VAS scales are reliable and valid tools to assess general health status [##REF##14713317##44##]. Of note, all correlations were in the moderate range which indicates that although statistically significant they are not highly predictive of one another.</p>", "<p>Our results also indicate that the Argentinean Spanish version of the PedsQL™ 4.0 has good discriminant validity. The Argentinean Spanish version of the PedsQL™ was able to distinguish between healthy and chronically ill children and between those with different chronic health conditions, as previously reported for the U.S. English version[##REF##17634123##19##]. As was found in previous studies[##REF##16468077##8##,##REF##18362824##11##], the Argentinean Spanish PedsQL™ was also able to discriminate between SES levels. Interestingly, the Total Scale Score and scale scores of the Argentinean Spanish version of the PedsQL™ were consistently lower than those reported in the original publication[##REF##11468499##6##] and almost all published cross-cultural adaptations[##REF##18362824##11##, ####REF##18080786##12##, ##REF##17681083##13##, ##REF##16972987##14##, ##REF##16756527##15##, ##REF##15804349##16####15804349##16##,##REF##17616837##18##,##REF##15058802##38##,##REF##17569714##39##,##REF##18215706##45##] for both the chronically ill and healthy samples. Our results could be reflecting the socioeconomic characteristics of our sample. Compared to the general population, our sample was poorer. National statistics for Argentina [##UREF##2##34##] indicate that 46% of children ages 0–13 and 40% of children 13 and older live below the poverty line, which is lower than the 66% found in our sample. Even more, our healthy sample was purposely selected from sources that assured a higher prevalence of poverty, and in fact these children were more likely to be poorer than the general population although significantly less poor than our ill children sample. Varni et al[##REF##16468077##8##], in a recent population study in schools found that Hispanics, compared to white and other ethnic origins, and those with lower SES, compared to higher SES, reported overall significantly lower PedsQL™ scores. Thus, the lower quality of life reported by the families interviewed in our study may be reflecting a combination of cultural (Hispanic culture may be associated with reports of lower quality of life independent of socioeconomic reasons) and socioeconomic determinants. To corroborate our hypothesis, future studies should include people from higher SE strata and results should then be compared locally and internationally.</p>", "<title>Strengths and Limitations</title>", "<p>Our study provides innovative data regarding the use of a HRQOL instrument in the Argentinean public health setting. Our very high enrollment rate (&gt; 90%) seems to indicate that the sample would be representative of the study base population. Further, we took special interest in trying to unveil potential difficulties in PedsQL™ use as we worried that our population's lower socioeconomic status and literacy would impair their ability to use such an instrument. Reports of the impact of lower socioeconomic status and literacy on pediatric HRQOL are not common despite its argued value[##REF##11530898##46##]. Our results are encouraging and show that research on quality of life topics is not only possible in low socioeconomic settings but also relevant: surveyed families showed great enthusiasm about our paying attention to aspects of their lives that seem to be neglected frequently.</p>", "<p>One of the main limitations of our study is that our sample size does not allow us to conduct thorough evaluations across illnesses and age groups. In addition, two important features of patient reported outcome instruments, test-retest reliability and sensitivity to change, were not assessed and are warranted to fully understand the applicability of PedsQL™ 4.0 in Argentinean children. However, generic instruments are better suited to compare across conditions than to assess specific interventions for a given condition[##REF##11207164##47##] and in this context, responsiveness and sensitivity to change may be less relevant characteristics. Validation of specific HRQOL modules or instruments may be more appropriate to evaluate such changes[##REF##8054098##48##]. Finally, it is also important to bear in mind that instrument validity is a concept that builds upon repeated instrument use[##REF##8452328##49##].</p>" ]

[ "<title>Conclusion</title>", "<p>Overall, the Argentinean Spanish version of the PedsQL™ 4.0 Generic Core Scales version proved to be understandable and feasible to use. It showed good reliability for children over 8 years old and parents of children over 5 years old and good construct and discriminant validity properties in this low-income setting, making this instrument suitable for research use. Steps to expand the use of this tool should include an alternative approach to scoring for the 2–4 year-olds, further understanding of how to increase reliability for the 5–7 year-old self-report, and confirmation of other aspects of validity. Having a HRQOL instrument with demonstrated reliability and validity in the Argentinean culture will allow us to start addressing the impact of chronic illness on the quality of life of children and adolescents, including those in poor districts.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>To validate the Argentinean Spanish version of the PedsQL™ 4.0 Generic Core Scales in Argentinean children and adolescents with chronic conditions and to assess the impact of socio-demographic characteristics on the instrument's comprehensibility and acceptability. Reliability, and known-groups, and convergent validity were tested.</p>", "<title>Methods</title>", "<p>Consecutive sample of 287 children with chronic conditions and 105 healthy children, ages 2–18, and their parents. Chronically ill children were: (1) attending outpatient clinics and (2) had one of the following diagnoses: stem cell transplant, chronic obstructive pulmonary disease, HIV/AIDS, cancer, end stage renal disease, complex congenital cardiopathy. Patients and adult proxies completed the PedsQL™ 4.0 and an overall health status assessment. Physicians were asked to rate degree of health status impairment.</p>", "<title>Results</title>", "<p>The PedsQL™ 4.0 was feasible (only 9 children, all 5 to 7 year-olds, could not complete the instrument), easy to administer, completed without, or with minimal, help by most children and parents, and required a brief administration time (average 5–6 minutes). People living below the poverty line and/or low literacy needed more help to complete the instrument. Cronbach Alpha's internal consistency values for the total and subscale scores exceeded 0.70 for self-reports of children over 8 years-old and parent-reports of children over 5 years of age. Reliability of proxy-reports of 2–4 year-olds was low but improved when school items were excluded. Internal consistency for 5–7 year-olds was low (α range = 0.28–0.76). Construct validity was good. Child self-report and parent proxy-report PedsQL™ 4.0 scores were moderately but significantly correlated (ρ = 0.39, p &lt; 0.0001) and both significantly correlated with physician's assessment of health impairment and with child self-reported overall health status. The PedsQL™ 4.0 discriminated between healthy and chronically ill children (72.72 and 66.87, for healthy and ill children, respectively, p = 0.01), between different chronic health conditions, and children from lower socioeconomic status.</p>", "<title>Conclusion</title>", "<p>Results suggest that the Argentinean Spanish PedsQL™ 4.0 is suitable for research purposes in the public health setting for children over 8 years old and parents of children over 5 years old. People with low income and low literacy need help to complete the instrument. Steps to expand the use of the Argentinean Spanish PedsQL™ 4.0 include an alternative approach to scoring for the 2–4 year-olds, further understanding of how to increase reliability for the 5–7 year-olds self-report, and confirmation of other aspects of validity.</p>" ]

[ "<title>Instruments</title>", "<title>The PedsQL™ 4.0 Generic Core Scales</title>", "<p>The 23-item PedsQL™ 4.0 Generic Core Scales encompass: 1) Physical Functioning (8 items), 2) Emotional Functioning (5 items), 3) Social Functioning (5 items), and 4) School Functioning (5 items), and were developed through focus groups, cognitive interviews, pre-testing, and field testing measurement development protocols[##REF##10024117##5##,##REF##11468499##6##] The instrument takes approximately 5 minutes to complete[##REF##10024117##5##,##REF##11468499##6##] The PedsQL™ Scales are comprised of parallel child self-report and parent proxy-report formats. Child self-report includes ages 5–7, 8–12, and 13–18 years. Parent proxy-report includes ages 2–4 (toddler), 5–7 (young child), 8–12 (child), and 13–18 (adolescent), and assesses parent's perceptions of their child's HRQOL. The items for each of the forms are essentially identical, differing in developmentally appropriate language, or first or third person tense. The instructions ask how much of a problem each item has been during the past one month. A 5-point Likert response scale is utilized across child self-report for ages 8–18 and parent proxy-report (0 = never a problem; 1 = almost never a problem; 2 = sometimes a problem; 3 = often a problem; 4 = almost always a problem). To further increase the ease of use for the young child self-report (ages 5–7), the response scale is reworded and simplified to a 3-point scale (0 = not at all a problem; 2 = sometimes a problem; 4 = a lot of a problem), with each response choice anchored to a happy to sad faces scale[##REF##3822493##28##,##REF##8895242##29##].</p>", "<p>Items are reverse-scored and linearly transformed to a 0–100 scale (0 = 100, 1 = 75, 2 = 50, 3 = 25, 4 = 0), so that higher scores indicate better HRQOL. Scale Scores are computed as the sum of the items divided by the number of items answered (this accounts for missing data). If more than 50% of the items in the scale are missing, the Scale Score is not computed. This accounts for the differences in sample sizes for scales reported in the Tables. Although there are other strategies for imputing missing values, this computation is consistent with the previous PedsQL™ peer-reviewed publications, as well as other well-established HRQOL measures [##REF##11468499##6##,##REF##8277801##30##,##REF##8763800##31##]. The Physical Health Summary Score (8 items) is the same as the Physical Functioning Scale. To create the Psychosocial Health Summary Score (15 items), the mean is computed as the sum of the items divided by the number of items answered in the Emotional, Social, and School Functioning Scales.</p>", "<p>The adaptation of the PedsQL™ 4.0 Generic Core Scales into Argentinean Spanish was conducted following internationally accepted guidelines for cross-cultural adaptation of patient reported outcome instruments[##REF##8263569##17##,##UREF##0##32##,##UREF##1##33##]. The forward translation into Spanish of all the PedsQL™ corresponding versions was conducted by two of the authors (VD, GM), a paediatrician and a child psychologist who are fluent in English. This first draft was reviewed by a multidisciplinary team that included the two authors, a child oncologist, and a health services researcher/clinician. After extensive discussion we ended up with a reconciled first Argentinean Spanish PedsQL™ version. The back translation was done by a native English speaker fluent in Spanish not familiar with the instrument. Some items were slightly modified to ensure semantic and conceptual equivalence of the second Argentinean Spanish PedsQL™ version. Cognitive debriefing interviews were carried out in two waves, first with 15 children and their parents. This pretest prompted changes that essentially involved spelling out both the main question and answer options more thoroughly (e.g. \"problems with running\" instead of \"running\" and \"never was a problem\" instead of \"never\") to increase comprehensibility. The second wave of cognitive interviews was carried out in 30 children and parents and confirmed that the final Argentinean Spanish PedsQL™ was understandable and conceptually equivalent to the original instrument. All changes and revisions were reviewed and accepted by JV.</p>", "<title>Overall Health Status Ratings</title>", "<p>Overall health status ratings were developed for this study (see Figure ##FIG##0##1##). Physicians were prompted to assess the child's degree of health impairment due to their disease over the past month using a 0–10 visual analogue scale (VAS) where 0 was \"no impairment at all\" and 10 \"maximum impairment\". Children 5 years old and above and their proxies were asked to independently score how they considered the child was feeling over the last month. Children 8 years old and above and adults used a 0–10 VAS, where 0 was \"very bad\" and 10 \"very well\", whereas 5 to 7 year-olds used a three-point faces scale (very bad, more or less, very well) similar to the faces scale used in the corresponding PedsQL™ version.</p>", "<title>Cognitive Debriefing/Feasibility</title>", "<p>Children and proxy's impressions about the Argentinean Spanish version of PedsQL™, including difficulty with format and understanding, easiness, and comprehensibility were asked with a semi-structured cognitive interview.</p>", "<title>Clinical and Socio-demographic variables</title>", "<p>Clinical information such as diagnosis, disease severity, and duration of disease was abstracted from the patients' medical records and, when not available, was collected from the patients' primary physicians.</p>", "<p>Age, gender, education level of the child and adult proxy, and socioeconomic status were collected from adult proxies. Socioeconomic status variables included health insurance (union health insurance/private insurance/disability allowances/uninsured), and poverty level, which was dichotomized as above or below the poverty line according to the ratio income/basic family living costs[##UREF##2##34##].</p>", "<title>Design</title>", "<p>This is a cross-sectional descriptive study. One interviewer (MR, not related to patient care) administered the PedsQL™ 4.0 and the validation questionnaire to all enrolled families.</p>", "<p>Construct validity was assessed by testing the following hypothesis: (1) PedsQL™ 4.0 scores would correlate negatively with physician's assessed impairment of health status; (2) PedsQL™ 4.0 scores would correlate positively with self/proxy-reported overall health status; (3) Child self-reported and parent proxy-reported PedsQL™ 4.0 scores would correlate significantly in the medium effect size range. In addition, we used the known-groups approach to test discriminant validity by comparing PedsQL™ 4.0 scores of healthy children with those of children with chronic health conditions, as well as scores across different chronic conditions groups. It was anticipated that children with chronic health conditions would report significantly lower PedsQL™ scores overall in comparison to healthy children[##REF##17634123##19##].</p>", "<title>Procedures</title>", "<p>For the field test, outpatient clinic rosters were reviewed with primary physicians who identified subjects that met inclusion criteria. Families were then approached in the clinic before seeing their doctor and invited to enroll in the study. After enrolling, children and proxies were asked to independently complete the PedsQL™ followed by the cognitive debriefing interview. Overall health status assessment was carried out after the PedsQL™ administration to avoid cuing. Proxies provided socio-demographic information at the end. Primary physicians were asked to report the child's overall health impairment after they saw the patient.</p>", "<p>The following variables were collected by the interviewer as patients completed the instruments: (1) mode of administration (self-administered, required interviewer-administration), (2) version used (as per PedsQL™ guidelines when a patient did not understand their age-specific version they were offered the next younger age version), (3) completion time, (4) need for help (classified in 3 categories: no help, minimal help: &lt; 4 times, and significant help: ≥ 4 times during questionnaire administration), and (5) missing items.</p>", "<title>Statistical Analysis</title>", "<p>To assess the appropriateness of the PedsQL™ administration in the Argentinean public health setting we set an <italic>a priori </italic>condition indicating that at least 80% of the questionnaires should be answered based on an empirical consideration that if more than 20% of the targeted sample was not able to complete the questionnaire, the tool would not serve the purpose of generating valid, representative data[##REF##11113959##35##]. Questionnaires were considered unanswered if they took more than 30 minutes to complete (this was considered a reasonable time for research purposes given that not everyone was expected to take so long) or if more than 50% of items were not understood despite interviewer's assistance (following the author's guidelines[##UREF##3##36##] of not scoring questionnaires with more than 50% of missing items[##REF##8763800##31##]). In addition, the association between comprehensibility and sociodemographic covariates was analyzed using T-test for independent samples and Chi Square or Fisher's exact test as appropriate. A p-value &lt; 0.05 was considered significant.</p>", "<p>Descriptive statistics of the items, average scores, as well as ceiling and floor effects are reported. Ceiling and floor effects were considered present if &gt; 15% of respondents used the extreme values[##REF##17161752##37##]. Scores were stratified by respondent, age group, and type of chronic condition. Scale reliability was evaluated using Cronbach's coefficient alpha. Construct validity was tested using Pearson's correlation coefficient. Discriminant validity was evaluated by testing differences among chronic and healthy children scores, disease subgroups, gender, and SES using t-test or ANOVA for binary and categorical variables respectively. Data analysis was conducted with SPSS 10.0 for Windows.</p>", "<title>Abbreviations</title>", "<p>HRQOL: Health-related quality of life; PedsQL™: Pediatric Quality of Life Inventory™; VAS: Visual Analogue Scale; SCT: Allogenic hematopoietic stem cell transplantation; COPD: Chronic Obstructive Pulmonary Disease with indication of home oxigenotherapy; ESRD: End Stage Renal Disease; CCC: Complex Congenital Cardiopathies.</p>", "<title>Competing interests</title>", "<p>Dr. Varni holds the copyright and the trademark for the PedsQL™ and receives financial compensation from the Mapi Research Trust, which is a nonprofit research institute that charges distribution fees to for-profit companies that use the Pediatric Quality of Life Inventory™. The PedsQL™ is available at the PedsQL™ Website[##UREF##4##50##]. The rest of the authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>All authors collaborated in the study design, MR collected the data, MR, SR, GB, and VD conducted the analysis and drafted the paper, and all authors reviewed and approved the manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>We express our appreciation to the children, parents, and physicians that participated in this study, to Dr. Sonia Iorcansky for her mentoring (MR), and to the members of the Committee on Quality of Life at Hospital de Pediatria Garrahan Drs. Julia Redondo, Carlos Figueroa, Alejandra Bordato, María Magdalena Contreras, Virginia Fano, Lidia Fraquelli, Graciela Massanti, Isabel Maza, Luis Novali, Marcela Palladino, Mercedes Pico, Lucía Salvia, Griselda Splivalo, and Rodolfo Verna for their input during the study design and help with its implementation.</p>", "<p>MR is the recipient of a fellowship from the Buenos Aires Secretary of Heath (Decreto N° 2.244). VD is the recipient of a fellowship from the Agency for Health Research and Quality (T32HP10018).</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Visual analogue scales used to measure overall health status</bold>. Visual analogue scales (VAS) used to measure overall health status. Upper panel shows the VAS presented to physician's to assess degree of heath impairment in the past month. Middle panel shows VAS presented to older children and parents to assess overall health status in the past month. Lower panel shows faces scale used to assess self-reported overall health status in children aged 5 to 7 years old.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Flowchart and Patient Diagnosis for the Argentinean Spanish Validation of PedsQL™ 4.0 in children with chronic conditions</bold>. ¹Hematopoietic Stem Cell Transplant ²Chronic Obstructive Pulmonary Disease ³Human Immunodeficiency Virus infection or Acquired Immune Deficiency Syndrome ⁴End Stage Renal Disease ⁵Complex Congenital Cardiopathies.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Characteristics of children with chronic health conditions. Argentinean Spanish Validation of the PedsQL™ 4.0 Generic Core Scales.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"4\"><bold>Age Group</bold></td><td/></tr><tr><td/><td colspan=\"4\"><hr/></td><td/></tr><tr><td/><td align=\"center\"><bold>2–4</bold><break/><bold> years old</bold><break/><bold> n = 70</bold></td><td align=\"center\"><bold>5–7</bold><break/><bold> years old</bold><break/><bold> n = 62</bold></td><td align=\"center\"><bold>8–12</bold><break/><bold> years old</bold><break/><bold> n = 90</bold></td><td align=\"center\"><bold>13–18</bold><break/><bold> years old</bold><break/><bold> n = 65</bold></td><td align=\"center\"><bold>TOTAL</bold><break/><bold> N = 287</bold></td></tr></thead><tbody><tr><td align=\"left\">Patient gender</td><td/><td/><td/><td/><td/></tr><tr><td align=\"right\">Female</td><td align=\"center\">41.5%</td><td align=\"center\">39%</td><td align=\"center\">42%</td><td align=\"center\">46%</td><td align=\"center\">42%</td></tr><tr><td align=\"right\">Male</td><td align=\"center\">58.5%</td><td align=\"center\">61%</td><td align=\"center\">58%</td><td align=\"center\">54%</td><td align=\"center\">58%</td></tr><tr><td align=\"left\">Proxy respondent</td><td/><td/><td/><td/><td/></tr><tr><td align=\"right\">Mother</td><td align=\"center\">80%</td><td align=\"center\">76%</td><td align=\"center\">70%</td><td align=\"center\">61.5%</td><td align=\"center\">72%</td></tr><tr><td align=\"right\">Father</td><td align=\"center\">14%</td><td align=\"center\">14.5%</td><td align=\"center\">20%</td><td align=\"center\">9.5%</td><td align=\"center\">15%</td></tr><tr><td align=\"right\">Other</td><td align=\"center\">6%</td><td align=\"center\">9.5%</td><td align=\"center\">10%</td><td align=\"center\">29%</td><td align=\"center\">13%</td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\"><bold>Chronic condition</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"right\">SCT¹(n = 40)</td><td align=\"center\">7%</td><td align=\"center\">8%</td><td align=\"center\">16%</td><td align=\"center\">25%</td><td align=\"center\">14%</td></tr><tr><td align=\"right\">COPD²(n = 53)</td><td align=\"center\">24%</td><td align=\"center\">29%</td><td align=\"center\">14%</td><td align=\"center\">8%</td><td align=\"center\">18%</td></tr><tr><td align=\"right\">HIV/AIDS³(n = 57)</td><td align=\"center\">19%</td><td align=\"center\">24%</td><td align=\"center\">19%</td><td align=\"center\">18%</td><td align=\"center\">20%</td></tr><tr><td align=\"right\">Cancer (n = 56)</td><td align=\"center\">21%</td><td align=\"center\">20%</td><td align=\"center\">18%</td><td align=\"center\">20%</td><td align=\"center\">20%</td></tr><tr><td align=\"right\">ESRD⁴(n = 31)</td><td align=\"center\">3%</td><td align=\"center\">3%</td><td align=\"center\">18%</td><td align=\"center\">17%</td><td align=\"center\">11%</td></tr><tr><td align=\"right\">CCC⁵(n = 50)</td><td align=\"center\">26%</td><td align=\"center\">16%</td><td align=\"center\">15%</td><td align=\"center\">12%</td><td align=\"center\">17%</td></tr><tr><td align=\"left\">Time since diagnosis in months, median (range)</td><td align=\"center\">28 (1–60)</td><td align=\"center\">63 (1–84)</td><td align=\"center\">89 (1–148)</td><td align=\"center\">95 (2–204)</td><td align=\"center\">48 (1–204)</td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\"><bold>Socio-Demographics</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">Education</td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">Child below appropriate for age</td><td align=\"center\">-------</td><td align=\"center\">11%</td><td align=\"center\">29%</td><td align=\"center\">34%</td><td align=\"center\">25%</td></tr><tr><td align=\"left\">Proxy did not complete elementary school</td><td align=\"center\">3%</td><td align=\"center\">11%</td><td align=\"center\">4.5%</td><td align=\"center\">15.5%</td><td align=\"center\">11%</td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\">Below the poverty line⁶</td><td align=\"center\">67%</td><td align=\"center\">71%</td><td align=\"center\">65%</td><td align=\"center\">60%</td><td align=\"center\">66%</td></tr><tr><td align=\"left\">No health Insurance</td><td align=\"center\">67%</td><td align=\"center\">61%</td><td align=\"center\">44.5%</td><td align=\"center\">46%</td><td align=\"center\">54%</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>PedsQL 4.0 Argentinean Spanish Administration: Difficulties, help, and time to completion in children with chronic conditions</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"4\"><bold>Children </bold></td><td align=\"center\"><bold>Adults</bold></td></tr><tr><td/><td colspan=\"5\"><hr/></td></tr><tr><td/><td align=\"center\"><bold>5–7 yo</bold><break/><bold> n = 62¹</bold></td><td align=\"center\"><bold>8–12 yo</bold><break/><bold> n = 90</bold></td><td align=\"center\"><bold>13–18 yo</bold><break/><bold> n = 65</bold></td><td align=\"center\"><bold>TOTAL</bold><break/><bold> n = 217</bold></td><td align=\"center\"><bold>Total</bold><break/><bold> n = 287</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Time to completion, minutes</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"right\">Median (range)</td><td align=\"center\">5'(3–20)</td><td align=\"center\">7'(2–28)</td><td align=\"center\">5'(2–12)</td><td align=\"center\">6'(2–28)</td><td align=\"center\">5'(1–16)</td></tr><tr><td align=\"left\"><bold>Required Help¹</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"right\">No</td><td align=\"center\">42%</td><td align=\"center\">46.5%</td><td align=\"center\">77%</td><td align=\"center\">54.5%</td><td align=\"center\">69%</td></tr><tr><td align=\"right\">Minimal</td><td align=\"center\">32%</td><td align=\"center\">29%</td><td align=\"center\">21.5%</td><td align=\"center\">27.5%</td><td align=\"center\">26%</td></tr><tr><td align=\"right\">Significant</td><td align=\"center\">11.5%</td><td align=\"center\">24.5%</td><td align=\"center\">1.5%</td><td align=\"center\">14%</td><td align=\"center\">5%</td></tr><tr><td align=\"left\"><bold>Form of Administration</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"right\">Adequate</td><td align=\"center\">100%</td><td align=\"center\">78%</td><td align=\"center\">97%</td><td align=\"center\">89,5%</td><td align=\"center\">95,5%</td></tr><tr><td align=\"right\">Administered by interviewer</td><td align=\"center\">N/A²</td><td align=\"center\">14.5%</td><td align=\"center\">3%</td><td align=\"center\">7%</td><td align=\"center\">4,5%</td></tr><tr><td align=\"right\">Previous version</td><td align=\"center\">N/A³</td><td align=\"center\">7.5%</td><td align=\"center\">0%</td><td align=\"center\">3.5%</td><td align=\"center\">N/A⁴</td></tr><tr><td align=\"left\"><bold>Difficulties with the format</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"right\">Forgot</td><td align=\"center\">N/A²</td><td align=\"center\">10%</td><td align=\"center\">3%</td><td align=\"center\">5.5%</td><td align=\"center\">12.5%</td></tr><tr><td align=\"right\">Wrote over other item</td><td align=\"center\">N/A²</td><td align=\"center\">11%</td><td align=\"center\">4.5%</td><td align=\"center\">6.5%</td><td align=\"center\">4%</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>PedsQL 4.0 Argentinean Spanish Administration: Socioeconomic status, education and requirement of help to complete PedsQL in children with chronic conditions and their parents</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"4\"><bold>Required Help¹</bold></td></tr><tr><td/><td colspan=\"4\"><hr/></td></tr><tr><td/><td align=\"center\"><bold>No help</bold></td><td align=\"center\"><bold>Minimal </bold></td><td align=\"center\"><bold>Significant </bold></td><td align=\"center\">p-value²</td></tr></thead><tbody><tr><td align=\"left\"><bold>Living below poverty line</bold></td><td/><td/><td/><td/></tr><tr><td align=\"right\">Children</td><td align=\"center\">69 (47%)</td><td align=\"center\">41 (29%)</td><td align=\"center\">32 (23%)</td><td align=\"center\">0.025</td></tr><tr><td align=\"right\">Parents</td><td align=\"center\">122 (65%)</td><td align=\"center\">53(28%)</td><td align=\"center\">14 (7%)</td><td align=\"center\">0.026</td></tr><tr><td align=\"left\"><bold>Low education</bold></td><td/><td/><td/><td/></tr><tr><td align=\"right\">Children (lower than expected)</td><td align=\"center\">22 (40%)</td><td align=\"center\">16 (29%)</td><td align=\"center\">17 (31%)</td><td align=\"center\">0.008</td></tr><tr><td align=\"right\">Parents (incomplete elementary school)</td><td align=\"center\">17 (53%)</td><td align=\"center\">10 (31%)</td><td align=\"center\">5 (16%)</td><td align=\"center\">0.030</td></tr><tr><td align=\"left\"><bold>Low income and low education</bold></td><td/><td/><td/><td/></tr><tr><td align=\"right\">Children</td><td align=\"center\">18 (42%)</td><td align=\"center\">12 (28%)</td><td align=\"center\">13 (30%)</td><td align=\"center\">0.049</td></tr><tr><td align=\"right\">Parents</td><td align=\"center\">12 (44%)</td><td align=\"center\">10 (37%)</td><td align=\"center\">5 (19%)</td><td align=\"center\">0.001</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Scale Descriptives for Argentinean Spanish version of the PedsQL 4.0 Generic Core Scales Child Self-Report and Proxy-Report</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Scale</bold></td><td align=\"center\" colspan=\"6\"><bold>Scale Descriptives</bold></td></tr><tr><td/><td colspan=\"6\"><hr/></td></tr><tr><td/><td align=\"center\"><bold>Mean ± SD¹</bold></td><td align=\"center\"><bold>Range</bold></td><td align=\"center\"><bold>Floor Effect²</bold><break/><bold>(%)</bold></td><td align=\"center\"><bold>Ceiling Effect²</bold><break/><bold>(%)</bold></td><td align=\"center\"><bold>N</bold></td><td align=\"center\"><bold>α³</bold></td></tr></thead><tbody><tr><td align=\"left\" colspan=\"7\"><bold>Self-Report</bold></td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\">Total</td><td align=\"center\">66.87 ± 16.74</td><td align=\"center\">26–99</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">177</td><td align=\"center\">0.86</td></tr><tr><td align=\"left\">Physical</td><td align=\"center\">67.76 ± 19.6</td><td align=\"center\">0–100</td><td align=\"center\">0.5</td><td align=\"center\">4.8</td><td align=\"center\">196</td><td align=\"center\">0.69</td></tr><tr><td align=\"left\">Psychosocial</td><td align=\"center\">66.36 ± 17.49</td><td align=\"center\">27–100</td><td align=\"center\">0</td><td align=\"center\">0.5</td><td align=\"center\">186</td><td align=\"center\">0.80</td></tr><tr><td align=\"left\">Emotional</td><td align=\"center\">65 ± 21.31</td><td align=\"center\">0–100</td><td align=\"center\">0.5</td><td align=\"center\">5.3</td><td align=\"center\">208</td><td align=\"center\">0.59</td></tr><tr><td align=\"left\">Social</td><td align=\"center\">69.1 ± 21.67</td><td align=\"center\">10–100</td><td align=\"center\">0</td><td align=\"center\">11.1</td><td align=\"center\">203</td><td align=\"center\">0.59</td></tr><tr><td align=\"left\">School</td><td align=\"center\">65.6 ± 21.3</td><td align=\"center\">10–100</td><td align=\"center\">0</td><td align=\"center\">5.2</td><td align=\"center\">189</td><td align=\"center\">0.62</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\" colspan=\"7\"><bold>Proxy-Report</bold></td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\">Total</td><td align=\"center\">73.36 ± 16.09</td><td align=\"center\">14–100</td><td align=\"center\">0</td><td align=\"center\">1.7</td><td align=\"center\">183</td><td align=\"center\">0.87</td></tr><tr><td align=\"left\">Physical</td><td align=\"center\">74.67 ± 20.06</td><td align=\"center\">4–100</td><td align=\"center\">0</td><td align=\"center\">10.1</td><td align=\"center\">272</td><td align=\"center\">0.78</td></tr><tr><td align=\"left\">Psychosocial</td><td align=\"center\">72.41 ± 16.45</td><td align=\"center\">18–100</td><td align=\"center\">0</td><td align=\"center\">2.4</td><td align=\"center\">189</td><td align=\"center\">0.81</td></tr><tr><td align=\"left\">Emotional</td><td align=\"center\">69.16 ± 19.6</td><td align=\"center\">5–100</td><td align=\"center\">0</td><td align=\"center\">6.3</td><td align=\"center\">285</td><td align=\"center\">0.66</td></tr><tr><td align=\"left\">Social</td><td align=\"center\">77.78 ± 20.73</td><td align=\"center\">5–100</td><td align=\"center\">0</td><td align=\"center\">20.2</td><td align=\"center\">283</td><td align=\"center\">0.71</td></tr><tr><td align=\"left\">School</td><td align=\"center\">68.74 ± 24</td><td align=\"center\">5–100</td><td align=\"center\">0</td><td align=\"center\">1.7</td><td align=\"center\">192</td><td align=\"center\">0.68</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>PedsQL 4.0 Argentinean Spanish Scores and internal consistency by age group. (Analysis of Variance – ANOVA)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"3\"><bold>2–4 yo</bold></td><td align=\"center\" colspan=\"3\"><bold>5–7 yo</bold></td><td align=\"center\" colspan=\"3\"><bold>8–12 yo</bold></td><td align=\"center\" colspan=\"3\"><bold>13–18 yo</bold></td><td/></tr></thead><tbody><tr><td align=\"left\"><bold>Score/Dimension</bold></td><td align=\"center\"><bold>N</bold></td><td align=\"center\"><bold>Mean</bold><break/><bold>Total</bold><break/><bold>Score</bold><break/><bold> (SD)</bold></td><td align=\"center\"><bold>α¹</bold></td><td align=\"center\"><bold>N</bold></td><td align=\"center\"><bold>Mean</bold><break/><bold>Total</bold><break/><bold>Score</bold><break/><bold> (SD)</bold></td><td align=\"center\"><bold>α¹</bold></td><td align=\"center\"><bold>N</bold></td><td align=\"center\"><bold>Mean</bold><break/><bold>Total</bold><break/><bold>Score</bold><break/><bold> (SD)</bold></td><td align=\"center\"><bold>α¹</bold></td><td align=\"center\"><bold>N</bold></td><td align=\"center\"><bold>Mean </bold><break/><bold>Total </bold><break/><bold>Score</bold><break/><bold> (SD)</bold></td><td align=\"center\"><bold>α¹</bold></td><td align=\"center\"><bold>Differences²</bold></td></tr><tr><td colspan=\"14\"><hr/></td></tr><tr><td align=\"left\" colspan=\"14\"><bold>Self-Report </bold></td></tr><tr><td colspan=\"14\"><hr/></td></tr><tr><td align=\"left\">Total</td><td align=\"center\">N/A</td><td align=\"center\">N/A³</td><td align=\"center\">N/A</td><td align=\"center\">43</td><td align=\"center\">60.03<break/> (15.86)</td><td align=\"center\">0.76</td><td align=\"center\">76</td><td align=\"center\">66.75<break/> (16.8)</td><td align=\"center\">0.86</td><td align=\"center\">58</td><td align=\"center\">72.6<break/> (15.41)</td><td align=\"center\">0.89</td><td/></tr><tr><td align=\"left\">Physical</td><td align=\"center\">N/A</td><td align=\"center\">N/A</td><td align=\"center\">N/A</td><td align=\"center\">49</td><td align=\"center\">62.20<break/> (20.40)</td><td align=\"center\">0.57</td><td align=\"center\">85</td><td align=\"center\">66.36<break/> (20.24)</td><td align=\"center\">0.73</td><td align=\"center\">72</td><td align=\"center\">74.24<break/> (16.24)</td><td align=\"center\">0.71</td><td align=\"center\">13–18yo &gt; 5–7yo***<break/> 13–18yo &gt; 8–12yo*</td></tr><tr><td align=\"left\">Psychosocial</td><td align=\"center\">N/A</td><td align=\"center\">N/A</td><td align=\"center\">N/A</td><td align=\"center\">45</td><td align=\"center\">58.81<break/> (16.13)</td><td align=\"center\">0.65</td><td align=\"center\">80</td><td align=\"center\">66.98<break/> (17.34)</td><td align=\"center\">0.81</td><td align=\"center\">61</td><td align=\"center\">71.67 <break/>(16.83)</td><td align=\"center\">0.86</td><td align=\"center\">8–12yo &gt; 5–7yo**<break/> 13–18yo &gt; 5–7yo***</td></tr><tr><td align=\"left\">Emotional</td><td align=\"center\">N/A</td><td align=\"center\">N/A</td><td align=\"center\">N/A</td><td align=\"center\">53</td><td align=\"center\">62.07<break/> (23.56)</td><td align=\"center\">0.45</td><td align=\"center\">90</td><td align=\"center\">63.47<break/> (21.03)</td><td align=\"center\">0.62</td><td align=\"center\">65</td><td align=\"center\">69.54<break/> (19.3)</td><td align=\"center\">0.72</td><td align=\"center\">NS</td></tr><tr><td align=\"left\">Social</td><td align=\"center\">N/A</td><td align=\"center\">N/A</td><td align=\"center\">N/A</td><td align=\"center\">49</td><td align=\"center\">57.23<break/> (20.64)</td><td align=\"center\">0.28</td><td align=\"center\">89</td><td align=\"center\">70.05<break/> (20.68)</td><td align=\"center\">0.59</td><td align=\"center\">65</td><td align=\"center\">77.61<break/> (19.64)</td><td align=\"center\">0.73</td><td align=\"center\">8–12,13–18yo &gt; 5–7yo***</td></tr><tr><td align=\"left\">School</td><td align=\"center\">N/A</td><td align=\"center\">N/A</td><td align=\"center\">N/A</td><td align=\"center\">47</td><td align=\"center\">56.80<break/> (20.86)</td><td align=\"center\">0.44</td><td align=\"center\">81</td><td align=\"center\">68.47<break/> (21.69)</td><td align=\"center\">0.65</td><td align=\"center\">61</td><td align=\"center\">68.47<break/> (19.53)</td><td align=\"center\">0.69</td><td align=\"center\">8–12yo &gt; 5–7yo***<break/> 13–18yo &gt; 5–7yo**</td></tr><tr><td colspan=\"14\"><hr/></td></tr><tr><td align=\"left\" colspan=\"14\"><bold>Proxy-Report</bold></td></tr><tr><td colspan=\"14\"><hr/></td></tr><tr><td align=\"left\">Total</td><td align=\"center\">66</td><td align=\"center\">80.15 <break/>(13.19)</td><td align=\"center\">0.62⁴</td><td align=\"center\">50</td><td align=\"center\">73.88<break/> (16.26)</td><td align=\"center\">0.89</td><td align=\"center\">80</td><td align=\"center\">69 <break/>(21.24)</td><td align=\"center\">0.84</td><td align=\"center\">53</td><td align=\"center\">71.25<break/> (17.18)</td><td align=\"center\">0.89</td><td align=\"center\">2–4yo &gt; 8–12,13–18yo***</td></tr><tr><td align=\"left\">Physical</td><td align=\"center\">66</td><td align=\"center\">82.34<break/> (14.94)</td><td align=\"center\">0.65</td><td align=\"center\">58</td><td align=\"center\">74.78<break/> (21.3)</td><td align=\"center\">0.86</td><td align=\"center\">87</td><td align=\"center\">70.12<break/> (21.24)</td><td align=\"center\">0.77</td><td align=\"center\">61</td><td align=\"center\">72.59<break/> (20)</td><td align=\"center\">0.78</td><td align=\"center\">2–4yo &gt; 8–12*** <break/>2–4yo &gt; 13–18yo*</td></tr><tr><td align=\"left\">Psychosocial</td><td align=\"center\">69</td><td align=\"center\">78.41<break/> (14.29)</td><td align=\"center\">0.30⁵</td><td align=\"center\">53</td><td align=\"center\">73.33<break/> (16.5)</td><td align=\"center\">0.83</td><td align=\"center\">81</td><td align=\"center\">68.4<break/> (15.62)</td><td align=\"center\">0.77</td><td align=\"center\">55</td><td align=\"center\">70.62<break/> (18.01)</td><td align=\"center\">0.84</td><td align=\"center\">2–4yo &gt; 8–12***<break/> 2–4yo &gt; 13–18yo*</td></tr><tr><td align=\"left\">Emotional</td><td align=\"center\">70</td><td align=\"center\">75.46<break/> (15.4)</td><td align=\"center\">0.54</td><td align=\"center\">62</td><td align=\"center\">69.68 <break/>(19.97)</td><td align=\"center\">0.73</td><td align=\"center\">90</td><td align=\"center\">66.05<break/> (19.04)</td><td align=\"center\">0.62</td><td align=\"center\">63</td><td align=\"center\">66.17<break/> (22.58)</td><td align=\"center\">0.75</td><td align=\"center\">2–4yo &gt; <break/>8–12,13–18yo*</td></tr><tr><td align=\"left\">Social</td><td align=\"center\">69</td><td align=\"center\">83.86<break/> (17.65)</td><td align=\"center\">0.65</td><td align=\"center\">61</td><td align=\"center\">77.38<break/> (20.10)</td><td align=\"center\">0.65</td><td align=\"center\">90</td><td align=\"center\">73.67<break/> (21.95)</td><td align=\"center\">0.72</td><td align=\"center\">63</td><td align=\"center\">77.30<break/> (21.54)</td><td align=\"center\">0.79</td><td align=\"center\">2–4yo &gt; 8–12yo*</td></tr><tr><td align=\"left\">School</td><td align=\"center\">20</td><td align=\"center\">73.68<break/> (13.96)</td><td align=\"center\">0.47</td><td align=\"center\">54</td><td align=\"center\">71.76<break/> (22.06)</td><td align=\"center\">0.74</td><td align=\"center\">81</td><td align=\"center\">65.72<break/> (22.4)</td><td align=\"center\">0.65</td><td align=\"center\">57</td><td align=\"center\">68.65<break/> (20.55)</td><td align=\"center\">0.64</td><td align=\"center\">NS</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T6\"><label>Table 6</label><caption><p>Construct validity of the PedsQL 4.0 Argentinean Spanish Version</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\" colspan=\"5\"><bold><underline>1^stHypothesis</underline>: Correlation between PedsQL scores and overall health impairment¹</bold></td></tr></thead><tbody><tr><td/><td align=\"center\"><bold>PedsQL</bold><break/><bold> mean (SD)</bold></td><td align=\"center\"><bold>Physician VAS</bold><break/><bold> mean (SD)</bold></td><td align=\"center\"><bold>r³</bold></td><td align=\"center\"><bold>p-value</bold></td></tr><tr><td align=\"left\">Self-Report</td><td align=\"center\">66.87 (16.74)</td><td align=\"center\">4.01(2.42)</td><td align=\"center\">-0.23</td><td align=\"center\">0.001</td></tr><tr><td align=\"left\">Proxy-Report</td><td align=\"center\">73.36 (16.09)</td><td align=\"center\">4.01(2.42)</td><td align=\"center\">-0.32</td><td align=\"center\">&lt; 0.001</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\" colspan=\"5\"><bold><underline>2nd Hypothesis</underline>: Correlation between PedsQL scores and self-reported overall health status²</bold></td></tr><tr><td/><td align=\"center\"><bold>PedsQL</bold><break/><bold> mean (SD)</bold></td><td align=\"center\"><bold>Self-report VAS</bold><break/><bold> mean (SD)</bold></td><td align=\"center\"><bold>r³</bold></td><td align=\"center\"><bold>p-value</bold></td></tr><tr><td align=\"left\">Self-Report</td><td align=\"center\">66.87 (16.74)</td><td align=\"center\">8.32 (1.82)</td><td align=\"center\">0.34</td><td align=\"center\">&lt; 0.001</td></tr><tr><td align=\"left\">Proxy-Report</td><td align=\"center\">73.36(16.09)</td><td align=\"center\">8.38 (1.51)</td><td align=\"center\">0.33</td><td align=\"center\">&lt; 0.001</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\"><bold><underline>3^rdHypothesis</underline>: Correlation between Self-report and Proxy-report PedsQL scores²</bold></td><td/><td/><td/><td/></tr><tr><td/><td align=\"center\"><bold>Self-Report</bold></td><td align=\"center\"><bold>Proxy-Report</bold></td><td align=\"center\"><bold>R³</bold></td><td align=\"center\"><bold>p-value</bold></td></tr><tr><td align=\"left\">PedsQL, mean (SD)</td><td align=\"center\">66.87 (16.74)</td><td align=\"center\">73.36 (16.09)</td><td align=\"center\">0.39</td><td align=\"center\">&lt; 0.001</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T7\"><label>Table 7</label><caption><p>Comparison of PedsQL 4.0 Argentinean Spanish scores of healthy children and children with chronic conditions</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Healthy children</bold></td><td align=\"center\"><bold>All Chronic conditions</bold></td><td align=\"center\"><bold>p-value¹</bold></td><td align=\"center\" colspan=\"7\"><bold>Scores by illness group</bold></td></tr><tr><td/><td/><td/><td/><td colspan=\"7\"><hr/></td></tr><tr><td/><td/><td/><td/><td align=\"center\"><bold>SCT²</bold><break/><bold>Mean (SD)</bold></td><td align=\"center\"><bold>COPD³</bold><break/><bold>Mean (SD)</bold></td><td align=\"center\"><bold>HIV/AIDS⁴</bold><break/><bold>Mean (SD)</bold></td><td align=\"center\"><bold>Cancer</bold><break/><bold> Mean (SD)</bold></td><td align=\"center\"><bold>ESRD⁵</bold><break/><bold>Mean (SD)</bold></td><td align=\"center\"><bold>CCC⁶</bold><break/><bold>Mean (SD)</bold></td><td align=\"center\"><bold>Differences⁷</bold></td></tr></thead><tbody><tr><td align=\"left\" colspan=\"11\"><bold>Self-Report </bold></td></tr><tr><td colspan=\"11\"><hr/></td></tr><tr><td align=\"left\">Total</td><td align=\"center\">72.72<break/> (14.21)</td><td align=\"center\">66.87<break/> (16.74)</td><td align=\"center\">0.011</td><td align=\"center\">71.64<break/> (17.13)</td><td align=\"center\">58.54<break/> (15.8)</td><td align=\"center\">71.31<break/> (17.07)</td><td align=\"center\">65.16<break/> (15.45)</td><td align=\"center\">65.29<break/> (19.96)</td><td align=\"center\">68.98<break/> (15.22)</td><td align=\"center\">SCT, HIV &gt; COPD*</td></tr><tr><td align=\"left\">Physical</td><td align=\"center\">75.42<break/> (15.93)</td><td align=\"center\">67.76<break/> (19.60)</td><td align=\"center\">0.004</td><td align=\"center\">74.24<break/> (17.08)</td><td align=\"center\">58.16 <break/>(19.84)</td><td align=\"center\">74.58<break/> (18.05)</td><td align=\"center\">60.98<break/> (19.44)</td><td align=\"center\">67.17<break/> (1.32)</td><td align=\"center\">72.03<break/> (15.84)</td><td align=\"center\">SCT, HIV &gt; Cancer* SCT &gt; COPD** HIV &gt; COPD***</td></tr><tr><td align=\"left\">Psychosocial</td><td align=\"center\">71.20<break/> (14.84)</td><td align=\"center\">66.36<break/> (17.49)</td><td align=\"center\">0.028</td><td align=\"center\">70.10<break/> (18.51)</td><td align=\"center\">58.76<break/> (16.43)</td><td align=\"center\">69.52<break/> (17.69)</td><td align=\"center\">67.57<break/> (6.46)</td><td align=\"center\">64.07<break/> (7.45)</td><td align=\"center\">67.32<break/>(7.04)</td><td align=\"center\">NS⁸</td></tr><tr><td colspan=\"11\"><hr/></td></tr><tr><td align=\"left\" colspan=\"11\"><bold>Proxy-Report*</bold></td></tr><tr><td align=\"left\">Total</td><td align=\"center\">82.19<break/> (12.97)</td><td align=\"center\">73.36<break/> (16.09)</td><td align=\"center\">&lt; 0.001</td><td align=\"center\">75.46<break/> (15.99)</td><td align=\"center\">69.61<break/> (18.29)</td><td align=\"center\">79.32<break/> (13.3)</td><td align=\"center\">71.39<break/> (14.79)</td><td align=\"center\">66.92<break/> (18.8)</td><td align=\"center\">74.61<break/> (13.52)</td><td align=\"center\">HIV &gt; COPD, ESRD*</td></tr><tr><td align=\"left\">Physical</td><td align=\"center\">86.20<break/> (12.27)</td><td align=\"center\">74.67<break/> (20.06)</td><td align=\"center\">&lt; 0.001</td><td align=\"center\">77.13<break/> (17.44)</td><td align=\"center\">67.85<break/> (23)</td><td align=\"center\">83.98<break/> (14.23)</td><td align=\"center\">69.92<break/> (18.82)</td><td align=\"center\">66.83<break/> (26.28)</td><td align=\"center\">79.49<break/> (15.78)</td><td align=\"center\">HIV &gt; COPD, Cancer, ESRD*** CCC &gt; COPD*</td></tr><tr><td align=\"left\">Psychosocial</td><td align=\"center\">79.91<break/> (14.96)</td><td align=\"center\">72.41<break/> (16.45)</td><td align=\"center\">&lt; 0.001</td><td align=\"center\">74.46 <break/>(17.89)</td><td align=\"center\">70.38<break/> (18.47)</td><td align=\"center\">76.74<break/> (14.77)</td><td align=\"center\">72.24 <break/>(16.34)</td><td align=\"center\">66.93<break/> (16.53)</td><td align=\"center\">71.57<break/> (14.02)</td><td align=\"center\">NS⁷</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T8\"><label>Table 8</label><caption><p>PedsQL 4.0 cross-cultural adaptation's reliability and validity. Comparison of published studies.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td/><td/><td align=\"center\" colspan=\"6\"><bold>Scores</bold></td><td/><td/></tr><tr><td/><td/><td/><td/><td colspan=\"6\"><hr/></td><td/><td/></tr><tr><td align=\"left\"><bold>Study</bold></td><td align=\"center\"><bold>Sample characteristics</bold></td><td align=\"center\"><bold>N</bold></td><td align=\"center\"><bold>Age groups</bold></td><td align=\"center\" colspan=\"3\"><bold>Self-Report</bold></td><td align=\"center\" colspan=\"3\"><bold>Proxy Report</bold></td><td align=\"center\"><bold>Reliability range</bold></td><td align=\"center\"><bold>Type of Validity tested</bold></td></tr><tr><td/><td/><td/><td/><td colspan=\"6\"><hr/></td><td/><td/></tr><tr><td/><td/><td/><td/><td align=\"center\">Total</td><td align=\"center\">Physical</td><td align=\"center\">Psychosocial</td><td align=\"center\">Total</td><td align=\"center\">Physical</td><td align=\"center\">Psychosocial</td><td/><td/></tr></thead><tbody><tr><td align=\"left\"><bold>Our Study</bold></td><td align=\"center\">Healthy Children</td><td align=\"center\">105</td><td align=\"center\">2–18 yo</td><td align=\"center\">72.72</td><td align=\"center\">75.42</td><td align=\"center\">71.20</td><td align=\"center\">82.19</td><td align=\"center\">86.20</td><td align=\"center\">79.91</td><td/><td align=\"center\">Known groups validity Convergent validity Self-report/Proxy correlations</td></tr><tr><td/><td align=\"center\">Chronic conditions</td><td align=\"center\">287</td><td align=\"center\">2–18 yo</td><td align=\"center\">66.87</td><td align=\"center\">67.76</td><td align=\"center\">66.36</td><td align=\"center\">73.36</td><td align=\"center\">74.67</td><td align=\"center\">72.41</td><td/><td/></tr><tr><td align=\"left\"><bold>US Original¹‡</bold></td><td align=\"center\">Well-child visits, clinic visits, children who had an admission</td><td align=\"center\">1645</td><td align=\"center\">2–18 yo</td><td align=\"center\">79.62</td><td align=\"center\">80.19</td><td align=\"center\">79.37</td><td align=\"center\">80.87</td><td align=\"center\">81.38</td><td align=\"center\">80.58</td><td align=\"center\">0.68–0.90</td><td align=\"center\">Known groups validity Predictive validity Factor analysis</td></tr><tr><td align=\"left\"><bold>Austria²</bold></td><td align=\"center\">School children</td><td align=\"center\">1412</td><td align=\"center\">8–12 yo</td><td align=\"center\">81.9</td><td align=\"center\">87.8</td><td align=\"center\">79.9</td><td align=\"center\">84.9</td><td align=\"center\">90.6</td><td align=\"center\">83.1</td><td align=\"center\">NR</td><td align=\"center\">Construct validity Predictive validity</td></tr><tr><td align=\"left\"><bold>Finland³</bold></td><td align=\"center\">School children</td><td align=\"center\">1097</td><td align=\"center\">8–12 yo</td><td align=\"center\">81.54</td><td align=\"center\">85.57</td><td align=\"center\">78.68</td><td align=\"center\">77.61</td><td align=\"center\">79.20</td><td align=\"center\">76.26</td><td align=\"center\">0.69–0.91</td><td align=\"center\">Compared to US study results</td></tr><tr><td align=\"left\"><bold>Germany⁴</bold></td><td align=\"center\">Chronic conditions</td><td align=\"center\">41 (epilepsy)</td><td align=\"center\">2–17 yo</td><td align=\"center\">78.0</td><td align=\"center\">87.3</td><td align=\"center\">NR</td><td align=\"center\">76.7</td><td align=\"center\">84.1</td><td align=\"center\">NR</td><td align=\"center\">0.72–0.91</td><td align=\"center\">Known groups validity Self-report/Proxy correlations</td></tr><tr><td/><td/><td align=\"center\">126 (cancer)</td><td/><td align=\"center\">82.6</td><td align=\"center\">86.7</td><td align=\"center\">NR</td><td align=\"center\">80.4</td><td align=\"center\">85.0</td><td align=\"center\">NR</td><td align=\"center\">0.60–0.84</td><td/></tr><tr><td align=\"left\"><bold>Greece⁵</bold></td><td align=\"center\">School children</td><td align=\"center\">645</td><td align=\"center\">8–12 yo</td><td align=\"center\">82.10</td><td align=\"center\">84.27</td><td align=\"center\">80.94</td><td align=\"center\">83.11</td><td align=\"center\">87.75</td><td align=\"center\">80.67</td><td align=\"center\">0.65–0.84</td><td align=\"center\">Factor analysis Self-report/Proxy correlations</td></tr><tr><td align=\"left\"><bold>Iceland⁶</bold></td><td align=\"center\">School children</td><td align=\"center\">480</td><td align=\"center\">10–12 yo</td><td align=\"center\" colspan=\"6\">Not reported summarized</td><td align=\"center\">NR</td><td align=\"center\">Predictive validity Known groups validity</td></tr><tr><td align=\"left\"><bold>Norway⁷</bold></td><td align=\"center\">School children</td><td align=\"center\">425</td><td align=\"center\">13–15 yo</td><td align=\"center\">85.29</td><td align=\"center\">91.12</td><td align=\"center\">82.16</td><td align=\"center\">86.10</td><td align=\"center\">88.83</td><td align=\"center\">84.66</td><td align=\"center\">0.73–0.88</td><td align=\"center\">Factor analysis Convergent correlation Self-report/Proxy correlations</td></tr><tr><td align=\"left\"><bold>UK⁸</bold></td><td align=\"center\">School children</td><td align=\"center\">1399</td><td align=\"center\">2–18 yo</td><td align=\"center\">83.89</td><td align=\"center\">88.51</td><td align=\"center\">81.84</td><td align=\"center\">84.61</td><td align=\"center\">89.06</td><td align=\"center\">82.21</td><td align=\"center\">&gt; 0.70</td><td align=\"center\">Known groups validity Self-report/Proxy correlations</td></tr><tr><td/><td align=\"center\">Chronic conditions</td><td align=\"center\">365</td><td align=\"center\">2–18 yo</td><td align=\"center\" colspan=\"6\">Scores were reported for each condition but not summarized</td><td/><td/></tr><tr><td align=\"left\"><bold>Turkey⁹</bold></td><td align=\"center\">Healthy children, children with acute and chronic conditions</td><td align=\"center\">223</td><td align=\"center\">2–4 yo</td><td align=\"center\">NA</td><td align=\"center\">NA</td><td align=\"center\">NA</td><td align=\"center\">78.17</td><td align=\"center\">79.40</td><td align=\"center\">77.25</td><td align=\"center\">0.66–0.85</td><td align=\"center\">Known groups validity Self-report/Proxy correlations</td></tr><tr><td/><td/><td align=\"center\">198</td><td align=\"center\">5–7 yo</td><td align=\"center\">71.56</td><td align=\"center\">72.66</td><td align=\"center\">70.82</td><td align=\"center\">72.92</td><td align=\"center\">69.96</td><td align=\"center\">74.76</td><td align=\"center\">0.57–0.86</td><td/></tr><tr><td align=\"left\"><bold>Japan¹⁰</bold></td><td align=\"center\">School children</td><td align=\"center\">229</td><td align=\"center\">6–13</td><td align=\"center\">76.7</td><td align=\"center\">83.4</td><td align=\"center\">73.3</td><td align=\"center\">81.4</td><td align=\"center\">92.6</td><td align=\"center\">75.8</td><td align=\"center\">0.71–0.86</td><td align=\"center\">Known groups validity Self-report/Proxy correlations Factor analysis</td></tr><tr><td/><td align=\"center\">Chronic conditions</td><td align=\"center\">100</td><td align=\"center\">5–18</td><td align=\"center\">NR</td><td align=\"center\">NR</td><td align=\"center\">NR</td><td align=\"center\">NR</td><td align=\"center\">NR</td><td align=\"center\">NR</td><td align=\"center\">NR</td><td/></tr><tr><td align=\"left\"><bold>Catalunya¹¹</bold></td><td align=\"center\">School children</td><td align=\"center\">511</td><td align=\"center\">9–17 yo</td><td align=\"center\">81.53</td><td align=\"center\">88.26</td><td align=\"center\">79.23</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">0.76–0.80</td><td align=\"center\">Known groups Convergent validity (compared with KINDL scores) Predictive validity</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>¹Stem Cell Transplant ²Chronic Obstructive Pulmonary Disease ³Human Immunodeficiency Virus infection or Acquired Immune Deficiency Syndrome ⁴End Stage Renal Disease ⁵Complex Congenital Cardiopathies ⁶Poverty line is calculated according to total income, and number and age of people in the household, <italic>as per </italic>National Institute of Statistics and Census (INDEC) guidelines.</p></table-wrap-foot>", "<table-wrap-foot><p>¹Minimal help: &lt; 4 times, Significant: ≥ 4 times during questionnaire administration.</p><p>²N/A: not applicable, always administered by interviewer</p><p>³N/A: not applicable, No existence of previous versions</p></table-wrap-foot>", "<table-wrap-foot><p>¹Minimal help: &lt; 4 times, Significant: ≥ 4 times during questionnaire administration or could not complete questionnaire. ²Chi-square test</p></table-wrap-foot>", "<table-wrap-foot><p>¹Higher mean values indicate better HRQOL (range 0–100).</p><p>²Floor and ceiling effects are considered present if &gt; 15% of extreme values were used</p><p>³Cronbach α Coefficient.</p></table-wrap-foot>", "<table-wrap-foot><p>¹Cronbach α Coefficient ²p values based on analysis of variance (ANOVA) comparing the mean scores across age groups *p &lt; .05 **p &lt; 0.01 ***p &lt; .005 with Bonferroni correction for the number of comparisons, p &lt; .005 values should be considered statistically significant. NS: non significant, p &gt; 0.05. ³N/A: not applicable ⁴If school items are excluded, α = 0.83. ⁵If school items are excluded, α = 0.76.</p></table-wrap-foot>", "<table-wrap-foot><p>¹VASphys: 0 (none) to 10 (maximum) impairment of health status</p><p>²VASc or VASp: 0 (very bad) to 10 (very well) overall feeling during the past month.</p><p>³r = Pearson's correlation coefficient interpreted as low (&lt; 0.10), moderate (0.11–0.30) and high (&gt; 0.30).</p></table-wrap-foot>", "<table-wrap-foot><p>¹Student's t test ²Stem Cell Transplant ³Chronic Obstructive Pulmonary Disease ⁴Human Immunodeficiency Virus infection or Acquired Immune Deficiency Syndrome ⁵End Stage Renal Disease ⁶Complex Congenital Cardiopathies ⁷p values based on analysis of variance (ANOVA) * p &lt; .05 **p &lt; .01 ***p &lt; .005 With a Bonferroni correction for the number of comparisons, p &lt; 0.005 should be considered statistically significant. Higher values equal better health-related quality of life. ⁸NS: non-significant</p></table-wrap-foot>", "<table-wrap-foot><p>‡ PedsQL 4.0 has undergone multiple validation studies in the US. A summary of the results and citations is provided in the introduction.</p><p>NR: Not reported</p><p>1. Varni JW, Seid M, Kurtin PS. PedsQL 4.0: reliability and validity of the Pediatric Quality of Life Inventory version 4.0 generic core scales in healthy and patient populations. <italic>Med Care </italic>2001;<bold>39</bold>(8)<bold>:</bold>800–12.</p><p>2. Felder-Puig R, Baumgartner M, Topf R, Gadner H, Formann AK. Health-Related Quality of Life in Austrian Elementary School Children. <italic>Med Care </italic>2008;<bold>46</bold>(4)<bold>:</bold>432–439.</p><p>3. Laaksonen C, Aromaa M, Heinonen OJ, Suominen S, Salantera S. Paediatric health-related quality of life instrument for primary school children: cross-cultural validation. <italic>J Adv Nurs </italic>2007;<bold>59</bold>(5)<bold>:</bold>542–50.</p><p>4. Felder-Puig R, Frey E, Proksch K, Varni JW, Gadner H, Topf R. Validation of the German version of the Pediatric Quality of Life Inventory (PedsQL) in childhood cancer patients off treatment and children with epilepsy. <italic>Qual Life Res </italic>2004;<bold>13</bold>(1)<bold>:</bold>223–34.</p><p>5. Gkoltsiou K, Dimitrakaki C, Tzavara C, Papaevangelou V, Varni JW, Tountas Y. Measuring health-related quality of life in Greek children: psychometric properties of the Greek version of the Pediatric Quality of Life Inventory(TM) 4.0 Generic Core Scales. <italic>Qual Life Res </italic>2008;<bold>17</bold>(2)<bold>:</bold>299–305.</p><p>6. Svavarsdottir EK, Orlygsdottir B. Health-related quality of life in Icelandic school children. <italic>Scand J Caring Sci </italic>2006;<bold>20</bold>(2)<bold>:</bold>209–15.</p><p>7. Reinfjell T, Diseth TH, Veenstra M, Vikan A. Measuring health-related quality of life in young adolescents: reliability and validity in the Norwegian version of the Pediatric Quality of Life Inventory 4.0 (PedsQL) generic core scales. <italic>Health Qual Life Outcomes </italic>2006;<bold>4:</bold>61.</p><p>8. Upton P, Eiser C, Cheung I, et al. Measurement properties of the UK-English version of the Pediatric Quality of Life Inventory 4.0 (PedsQL) generic core scales. <italic>Health Qual Life Outcomes </italic>2005;<bold>3:</bold>22.</p><p>9. Uneri OS, Agaoglu B, Coskun A, Memik NC. Validity and reliability of Pediatric Quality of Life Inventory for 2- to 4-year-old and 5- to 7-year-old Turkish children. <italic>Qual Life Res </italic>2008;<bold>17</bold>(2)<bold>:</bold>307–15.</p><p>10. Chen X, Origasa H, Ichida F, Kamibeppu K, Varni JW. Reliability and validity of the Pediatric Quality of Life Inventory (PedsQL) Short Form 15 Generic Core Scales in Japan. <italic>Qual Life Res </italic>2007;<bold>16</bold>(7)<bold>:</bold>1239–49.</p><p>11. Huguet A, Miro J. Development and psychometric evaluation of a Catalan self- and interviewer-administered version of the Pediatric Quality of Life Inventory version 4.0. <italic>J Pediatr Psychol </italic>2008;<bold>33</bold>(1)<bold>:</bold>63–79.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1477-7525-6-59-1\"/>", "<graphic xlink:href=\"1477-7525-6-59-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Autosomal Dominant Polycystic Kidney Disease is the most common hereditary renal disorder with a prevalence of at least 1:1000 and accounts for 8%–10% for all end-stage renal failure [##REF##8321262##1##]. The disease is characterized by the formation of large fluid-containing renal cysts that grow over time and destroy the renal parenchyma.</p>", "<p>It is believed that cysts originate from tubular epithelial cells that exhibit increased proliferation and reduced differentiation. This may happen after a second somatic hit occurs that inactivates the <italic>PKD1 </italic>or the <italic>PKD2 </italic>allele inherited from the healthy parent [##REF##10655555##2##, ####REF##9949210##3##, ##REF##7614844##4####7614844##4##]. Microdissection of cystic kidneys revealed that cyst growth is due to an increase in cell number and not to the stretching of the cyst wall. In addition, tubular epithelial cells cultured from ADPKD cysts display augmented levels of proliferation and upregulation of proliferation-associated genes such as c-Myc, Ki-67 and PCNA [##REF##3599654##5##, ####REF##9407432##6##, ##REF##9422539##7##, ##REF##9621276##8####9621276##8##]. The role of polycystin-1 (PC-1), the protein product of <italic>PKD1</italic>, in the proliferation of tubular epithelium has been documented. Polycystin-1 has been implicated in a variety of pathways tied to proliferation, including G-protein signaling, Wnt signaling and AP-1. [##REF##11786542##9##, ####REF##10339594##10##, ##REF##9988738##11##, ##REF##11912216##12####11912216##12##]. Direct evidence about the involvement of PC-1 in cell cycle regulation was demonstrated by the observation that PC-1 overexpression activates the JAK2/STAT-1 pathway, thereby up regulating p21^waf1and inducing cell cycle arrest in G0/G1 in a process requiring functional polycystin-2 (PC-2). Based on these results it was postulated that mutations in either gene could result in deregulated growth [##REF##12007403##13##].</p>", "<p>Polycystin-2 has been implicated in cell cycle regulation mainly through its calcium channel activity and its ability to activate transcription factor AP-1 [##REF##11991947##14##, ####REF##10207066##15##, ##REF##12640140##16####12640140##16##]. However, there was little direct evidence linking polycystin-2 to cellular proliferation. Recently, PC-2 was directly tied to cell cycle regulation through direct interaction with Id2, a member of the helix-loop-helix (HLH) proteins that are known to regulate cell proliferation and differentiation. Overexpression of wild-type PC-2 in kidney cell lines induced cell cycle arrest at G0/G1, through upregulation of p21 and subsequent inhibition of Cdk2 kinase activity. This process was dependent on both PC-2-Id2 interaction and PC-1-dependent phosphorylation of PC-2. Although inhibition of Id2 expression corrected the hyperproliferative phenotype of mutant cells, the contribution of p21/Cdk2 pathway on the abnormal cell proliferation was not clearly addressed [##REF##16311606##17##]. In an independent study, PC-2 was shown to regulate proliferation and differentiation of kidney epithelial cells and suggested that its calcium channel activity may play an important role in this process [##REF##16278216##18##].</p>", "<p>In this study, we examined the contribution of the JAK2/STAT-1/p21/Cdk2 pathway on PC-2-dependent kidney epithelial cell proliferation. We utilized cell lines HEK293 and NRK-52E expressing wild-type and mutant PC-2 as well as primary tubular epithelial cells from a <italic>PKD2</italic>-mutant transgenic rat [##REF##16943309##19##]. Interestingly, expression of mutant PC-2 had an effect on the aforementioned pathway only in the primary epithelial cells expressing mutant <italic>PKD2</italic>, but this was independent of p21. On the contrary multiple approaches provided unequivocal evidence that a different cyclin-dependent kinase inhibitor, p57, is reduced in these cells. These results suggest that p57 might be the end-point of an alternative pathway that regulates PC-2-induced proliferation in ADPKD.</p>" ]

[ "<title>Methods</title>", "<title>Cell culture and isolation of renal primary epithelial cells</title>", "<p>Human embryonic kidney 293 cells, PC-2 overexpressing cells and the rat epithelial cell line NRK-52E were maintained in DMEM medium supplemented with 10% (HEK293) or 5% (NRK-52E) fetal bovine serum (FBS).</p>", "<p>Renal primary epithelial cells were isolated from a 7.5 week-old <italic>PKD2 </italic>mutant trangenic rat (TGR (CMV-h<italic>PKD2</italic>/1–703)), abbreviated in the text as: <italic>PKD2</italic>(1–703) [##REF##16943309##19##]. There were two transgenic rat lines created initially, 111 and 247, expressing a truncated PC-2, owing to a STOP codon at postion 704. Of the two models 247 was chosen for further work owing to a more severe phenotype. Line 111 is not maintained at the moment [##REF##16943309##19##]. The primary cells were isolated by a sequential filtration method as follows: Normal Spraque-Dawley rats and <italic>PKD2</italic>(1–703) rats were sacrificed following standard procedures; kidneys were extracted and minced under sterile conditions. The cell mixture was passed through a 180 μm metal sieve (Retsch, Germany) followed by filtration through a 40 μm nylon cell strainer (BD Biosciences). The retained cells were collected and passed through a second 100 μm cell strainer. The filtrate of this step comprises the tubular epithelial fraction of the kidney homogenate. Tubular epithelial cells were cultured on laminin-coated tissue culture plates and maintained in Endothelial Cell Growth Medium (PromoCell, Germany) supplemented with 5%FBS, ECGS, EGF, Hydrocortison, Amphotericin B and Gentamycin. Under these conditions the cells maintained their epithelial phenotype for at least 4 passages.</p>", "<title>Antibodies</title>", "<p>The following antibodies were used in this study: mouse anti-HA, goat anti-p21, goat anti-Cdk2 (Santa Cruz, Biotechnology), rabbit-anti phospho STAT1 (Cell Signaling), rabbit anti-p57 (Santa Cruz). The rabbit polyclonal anti-PC2 (epitope corresponds to amino acids 679–742 of PC2) was previously described[##REF##16135816##20##]</p>", "<title>Plasmids</title>", "<p>HA-<italic>PKD2 </italic>(WT) was generated by cloning wild-type human <italic>PKD2 </italic>in pcDNA 3 (Invitrogen) plasmid. HA-<italic>PKD2</italic>/1–702 contains almost the entire of <italic>PKD2 </italic>(aa 1–702) and was constructed by the addition of a stop linker in the <italic>PKD2 </italic>sequence. Finally, HA-<italic>PKD2</italic>/R742X contains amino acids 1–742 of <italic>PKD2</italic>. Both, HA-<italic>PKD2</italic>/1–702 and HA-<italic>PKD2</italic>/R742X were cloned in pcDNA3 vector.</p>", "<title>Transient transfection and Western blot analysis</title>", "<p>Plasmids were transfected into HEK293 and NRK-52E cells using Lipofectamine 2000 (Invitrogen) reagent according to the manufacture's instructions. Western blot analysis was performed as mentioned before [##REF##15671247##21##]. Briefly, cells were lysed in Nonipet P-40 (NP40) buffer (0.1% NP40, 200 mM NaCl, 50 mM TrisCl (pH 7.4) and protease inhibitors). After centrifugation at 14,000 rpm for 5 minutes, the supernatants were collected. Total amount of protein was determined using the BCA kit (Pierce). Equal amount of protein was denatured by addition of equal volume of 2 × SDS loading buffer and heating for 30 min at 50°C. Proteins were separated on an SDS-PAGE gel. After transfer to a PVDF membrane, Western blots were developed by ECL following the vendor's protocol (Amersham).</p>", "<title>Cdk2 Kinase Assay</title>", "<p>Cdk2 assay was performed as previously described [##REF##14711830##22##]. Briefly, 250–500 μg of the total cellular protein was immunoprecipitated with 1 μg of Cdk2 antibody for 2 hours at 4°C. After extensive washing, the precipitate was subjected to the kinase assay in the presence of 50 mM HEPES, 7.5 mM MgCl₂, 0.5 mM EDTA, 20 mM β-glycerophosphate, 1 mM NaF, 5 mM dithiothreitol, 100 μM ATP and 10 μCi of [γ-³²P] ATP in a total volume of 30 μl. As a substrate, 2 μg of histone H1 (Calbiochem) were added to the reaction. The reaction was carried out at 30°C for 30 min. After elution, the supernatant was fractionated by SDS-PAGE, transferred onto a PVDF membrane and autoradiographed.</p>", "<title>Electrophysiology</title>", "<p>The conventional whole-cell voltage-clamp configuration was used to measure transmembrane currents in single cells as described previously [##REF##16135816##20##]. Briefly, patch-clamp recordings were obtained from single cells at room temperature using a Warner PC-505B amplifier (Warner Instrument Corp., Hamden, CT) and pClamp 8 software (Axon Instrument, Foster City, CA). Glass pipettes (plain, Fisher Scientific, Pittsburgh, PA) with resistances of 5–8 MΩ were prepared with a pipette puller and polisher (PP-830 and MF-830, respectively, Narishige, Tokyo, Japan). After the whole-cell configuration was achieved, cell capacitance and series resistance were compensated (~70%) before each recording period. From a holding potential of -60 mV, voltage steps were applied from -100 to 100 mV in 20 mV increments with 200 ms duration at 3 s intervals. Current traces were filtered at 1 kHz and analyzed off-line with pClamp 8. The pipette solution contained (in mM): 100 K-aspartate, 30 KCl, 0.3 Mg-ATP, 10 HEPES, 10 EGTA, and 0.03 GTP (pH 7.2). The extracellular solution contained (in mM): 135 NaCl, 5.4 KCl, 0.33 NaH₂PO₄, 1 MgCl₂, 1.8 CaCl₂, 5 HEPES, 5.5 glucose (pH 7.4) or 130 KCl, 1 MgCl₂, 10 HEPES, 0.1 CaCl₂, and 5 glucose (pH 7.4).</p>", "<title>Cell Cycle Analysis</title>", "<p>Cells were seeded in six-well plates in triplicates. Upon attachment, cells were synchronized by serum starvation for 24 h followed by addition of 10% serum-containing medium for the HEK293 or 2% serum-containing medium for the primary cells, for 24 hours. Cells were harvested, fixed with 80% cold ethanol followed by treatment with 25 μg/ml Ribonuclease A (SIGMA) and 50 μg/ml propidium iodide (SIGMA) for 30 min at 37°C. After incubation the cells were analyzed by FACS.</p>", "<title>Gene Expression Profiling with Microarrays</title>", "<p>Gene expression profiles of primary tubular epithelial cells (TECs) isolated from PKD2(1–703) rats and SD rats were compared. RNA isolation, cDNA and cRNA synthesis and hybridization to arrays of type Rae230A from Affymetrix (Santa Clara, CA, USA) were performed according to the recommendations of the manufacturer. Microarray data was analysed based on a mixed model analysis using JMP Genomics, version 3.0 (SAS Institute, Cary, NC, USA). Standard settings were used, except the following specifications[##REF##17620128##23##]: log-linear mixed models [##REF##11867082##24##], were fitted for values of perfect-matches, with probe and rat group considered to be constant and the array-id random. Custom CDF, [##REF##16284200##25##] with Unigene based gene/transcript definitions different from the original Affymetrix probe set definitions was used to annotate the arrays. Microarray data were submitted to NCBI GEO <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.ncbi.nlm.nih.gov/geo/query\"/>, sample number [GSE11500].</p>", "<title>Quantitative RT-PCR</title>", "<p>Total RNA (1 μg) was isolated from cultured cells using the Rneasy Mini kit (Qiagen) and was reverse transcribed with the Protoscript reverse transcription kit (New England Biolabs) using the VN(dT)₂₃primer as recommended by the manufacturer. As a standard for relative RNA quantification (Calibrator), 1 μg of all sample RNAs was pooled together and reverse transcribed as mentioned above. Quantitative RT-PCR (qRT-PCR) amplifications were performed with a LightCycler (Roche Molecular Biochemicals) using the same starting amount and LightCycler^®FastStart DNA MasterPLUS SYBR Green I reagents in a standard volume of 20 μl. Real-time detection of fluorimetric intensity of SYBR Green I, indicating the amount of PCR product formed, was measured at the end of each elongation phase. Fluorescence values measured in the log-linear phase of amplification were considered using the second-derivative-maximum method of the LightCycler Data Analysis software (Roche Molecular Biochemicals). Relative quantification was performed using serial dilutions of the Calibrator cDNA to provide a standard curve for each run. For all experiments, the standard curve had an error of below 5% and extended over the relative quantities of all individual samples.</p>", "<p>Genes whose differential expression was tested by gene-specific qRT-PCR analysis were rat p57 (forward primer: TGATGAGCTGGGAGCTGAG and reverse primer: TGGCGAAGAAGTCAGAGATG) and Cdk2 (forward primer: TGTGGCGCTTAAGAAAATCC and reverse primer: CCAGCAGCTTGACGATGTTA). Differences in the quantity of starting material were compensated by normalization with the housekeeping genes HPRT (forward primer: CTCATGGACTGATTATGGACAGGAC and reverse primer: GCAGGTCAGCAAAGAACTTATAGCC) and GAPDH (forward primer: GTATTGGGCGCCTGGTCACC and reverse primer: CGCTCCTGGAAGATGGTGATGG). Normalized fold-changes between mutant and normal samples were calculated by the REST XL software.</p>", "<title>Data Analysis and Statistics</title>", "<p>Data are reported as means ± SEM. Comparisons between multiple groups were performed using single-factor ANOVA, and secondary comparisons were performed using the Tukey test. Statistical analysis was performed using the SPSS statistical software package. For electrophysiology experiments, statistical analysis was employed with the SigmaStat (Chicago, IL) software. Data were reported as means ± SEM. Due to high variability in cells transfected with wild type <italic>PKD2</italic>, statistical significance was determined by the Mann-Whitey Rank Sum test. Differences were considered significant at p &lt; 0.05 if not stated otherwise (patch clamp and gene expression profiling. Gene expression statistical analysis is described above.</p>" ]

[ "<title>Results</title>", "<title>Generation of stable clones expressing wild-type and mutant PKD2 in HEK293 cells</title>", "<p>To test the role of <italic>PKD2 </italic>in renal cell proliferation and specifically on the p21/Cdk2 pathway, we generated a series of HEK293 cell lines with stable expression of hemaglutinin (HA)-tagged wild-type human <italic>PKD2 </italic>(WT <italic>PKD2</italic>), HA-tagged mutant <italic>PKD2 </italic>(R742X <italic>PKD2</italic>) and a selectable marker (Vector). The R742X <italic>PKD2 </italic>encodes for a truncated PC-2 lacking the polycystin-1 (PC-1) interacting region at the carboxy-terminal of the protein. R742X, is a disease-causing PC-2 mutant firstly identified in a Greek-Cypriot family with Polycystic Kidney Disease type 2 [##REF##8650545##26##, ####REF##12407099##27##, ##REF##11302751##28####11302751##28##]. Three individual clones were isolated from each transfectant and used for further experimentation. Immunoblotting of whole cell lysates from the selected clones with an HA antibody, showed good expression of HA-tagged WT <italic>PKD2 </italic>and HA-tagged R742X <italic>PKD2 </italic>(Figure ##FIG##0##1A##). The same lysates were immunoblotted with anti-PC-2 antibody to demonstrate that we indeed have PC-2 overexpression in these clones. As seen in figure ##FIG##0##1A##, endogenous PC-2 is barely detectable by Western blot analysis in vector-only and R742X <italic>PKD2 </italic>transfectants. The lower molecular weight band detected most likely represents a non-specific band detected with the anti-PC-2 antibody, since it is detected on vector-only transfectants and untransfected cells (Figure ##FIG##0##1A## and data not shown).</p>", "<p>We used these tools to test the effect of wild-type and mutant PC-2 expression on the JAK2/STAT-1/p21/Cdk2 pathway, as it was previously implicated in its regulation by showing that overexpression of wild type <italic>PKD1 </italic>activates JAK2 kinase, which in turn phosphorylates STAT-1 [##REF##12007403##13##]. Lysates from synchronized clones were immunoblotted with an anti-phospho-STAT-1 antibody, which detects the expression of serine phosphorylated STAT-1, and an anti-p21 to detect endogenous p21 expression. As shown in figure ##FIG##0##1A##, p21 levels and STAT-1 phosphorylation were unaffected by wild-type or mutant <italic>PKD2 </italic>expression. Equal loading was confirmed by re-probing the same membrane with anti-β-tubulin.</p>", "<p>Similarly, endogenous Cdk2 activity was equivalent among the different clones as judged by the kinase assay performed on Cdk2 immunoprecipitates from two selected clones of each transfectant. Western blot analysis demonstrated that similar amount of Cdk2 was precipitated from each clone (Figure ##FIG##0##1B##). Cell cycle analysis performed by propidium iodide (PI) staining revealed that expression of wild-type or mutant PC-2 does not alter the cell cycle profile of these cells (Figure ##FIG##1##2##). Furthermore, proliferating cell nuclear antigen (PCNA) levels were equal among the different clones (Figure ##FIG##0##1A##). Collectively, the results suggest that expression of wild-type and mutant <italic>PKD2 </italic>has no effect on the proliferation of HEK293 cells.</p>", "<p>To determine whether mislocalization of exogenous WT and R742X PC-2 is responsible for their inability to regulate cellular proliferation, we compared the sub-cellular localization of HA-tagged WT or R742X PC-2 with endogenous PC-2 by immunofluoresence. Both HA-tagged WT and R742X PC-2 were detected at the same subcellular compartments (endoplasmic reticulum and plasma membrane) as the endogenous PC-2 (data not shown).</p>", "<p>ER-localized PC-2 is known to function as a Ca²⁺-activated intracellular Ca²⁺release channel while plasma membrane-associated PC-2 is believed to function as a nonselective cation channel [##REF##11854751##29##, ####REF##11252306##30##, ##REF##11140688##31####11140688##31##]. Previous work has demonstrated that <italic>PKD2 </italic>overexpression augmented the amplitude of whole cell currents in renal epithelial cells [##REF##16135816##20##]. To test the effectiveness of the expressed WT <italic>PKD2 </italic>in HEK293 cells we performed whole cell current measurements in vector-only, WT <italic>PKD2 </italic>and R742X <italic>PKD2 </italic>clones. Functional expression of transfected wild type <italic>PKD2 </italic>in HEK cells has been shown [##REF##12514735##32##]. Figure ##FIG##2##3## shows that stable expression of wild type <italic>PKD2 </italic>in HEK cells resulted in a significant increase in the current amplitude of whole cell inward currents recorded either in normal extracellular tyrode solution or symmetrical K⁺(Figure ##FIG##2##3##). Outward currents were larger in WT <italic>PKD2 </italic>expressing cells compared to untransfected, mock-transfected, or R742X <italic>PKD2</italic>-transfected cells in symmetrical K⁺. <italic>PKD2</italic>-mediated K⁺currents were larger compared to Na⁺/Ca²⁺currents as was expected for <italic>PKD2 </italic>which shows higher permeability to K⁺compared to Na⁺or Ca²⁺[##REF##12640140##16##,##REF##16135816##20##]. Overexpression of R742X <italic>PKD2 </italic>did not have a significant effect on whole cell inward or outward currents in HEK293. Collectively, the electrophysiology data show that transfection of wild type <italic>PKD2 </italic>resulted in functional expression in HEK293 cells. However, <italic>PKD2 </italic>has no effect on the STAT-1/p21/Cdk2 pathway or on the proliferation status of these cells.</p>", "<title>Examination of the effect of wild-type and mutant PKD2 on the JAK2/STAT-1/p21/Cdk2 pathway in NRK-52E cells</title>", "<p>HEK293 cells were generated by transformation of human embryonic kidney cell cultures with sheared adenovirus 5 DNA [##REF##886304##33##]. The cell line has an epithelial morphology and is widely used as a kidney epithelial model. Nevertheless, there is controversy as to whether these cells are of true kidney origin, since expression studies have demonstrated that HEK293 cells have an unexpected relationship with neurons [##REF##11967234##34##]. For these reasons we decided to perform the same experiments in a different cell line system more closely resembling mature kidney epithelial cells, NRK-52E.</p>", "<p>The rat kidney epithelial cell line, NRK-52E was transiently transfected with vector-only (CT), WT <italic>PKD2</italic>, R742X <italic>PKD2 </italic>and 1–702 <italic>PKD2 </italic>(a <italic>PKD2 </italic>mutant lacking the entire carboxy-terminal region of the protein). At 48 hours after transfection, cells were synchronized by serum starvation. Whole cell lysates were immunoblotted with anti-p21 and anti-phospho-STAT-1 antibodies. Neither p21 levels nor STAT-1 phosphorylation is affected by expression of wild-type or mutant <italic>PKD2 </italic>(Figure ##FIG##3##4A##). Similarly, the levels of active Cdk2 were comparable among the four transfectants. In addition to the JAK2/STAT-1/p21/Cdk2 pathway, the proliferation capacity of NRK-52E transfected with WT, R742X and 1–702 <italic>PKD2 </italic>appeared unaltered compared to vector only transfectants as judged by PCNA Western blot analysis. Good expression of the wild-type PC-2 and of the two truncated proteins was achieved as judged by anti-HA and anti-PC2 blotting. In summary, these results duplicate the observation in HEK293 that wild-type or mutant <italic>PKD2 </italic>expression do not modify the activity of the JAK2/STAT-1/p21/Cdk2 pathway.</p>", "<title>Renal tubular epithelial cells from PKD2(1–703) transgenic rat display augmented proliferation independent of the JAK2/STAT-1/p21/Cdk2 pathway</title>", "<p>The unexpected but significant results above, prompted us to utilize primary renal epithelial cells obtained from a 7.5 week-old mutant <italic>PKD2 </italic>transgenic rat (1–703) [abbreviated <italic>PKD2</italic>(1–703)], expressing a truncated form of PC-2 lacking the C-terminal region of the protein. The transgenic animals manifest a cystic phenotype characterized by the formation of multiple cysts in the kidneys [##REF##16943309##19##]. Tubular renal epithelial cells were isolated by sequential filtration of renal cells and cultured in low serum-containing medium. The epithelial character of the isolated cells and the absence of contaminating fibroblasts were confirmed by cadherin and vimentin expression respectively (Figure ##FIG##4##5A##).</p>", "<p>In contrast to the cell lines examined, primary tubular epithelial cells (TECs) isolated from <italic>PKD2</italic>(1–703) transgenic rat, demonstrated an increase in cellular proliferation compared with their normal counterparts. Specifically, Western blot analysis on whole cell lysates demonstrated that TECs isolated from the <italic>PKD2</italic>(1–703) rat have significantly higher levels of PCNA than TECs isolated from normal Sprague-Dawley rats (Figure ##FIG##4##5A##). In addition, the percentage of cells in the G0/G1 phase of the cell cycle was lower in the mutant cells than in normal cells as judged by cell cycle analysis (90.6 ± 0.93 to 84.1 ± 1.28). In concert, the percentage of G2/M-phase mutant cells was higher than G2/M-phase normal cells (5.06 ± 0.31 to 12.9 ± 1.37) (Figure ##FIG##4##5B##). Despite the higher proliferative activity of mutant cells, p21 levels and STAT-1 phosphorylation remain unaltered (Figure ##FIG##4##5A##), suggesting that <italic>PKD2</italic>-induced proliferation is STAT-1/p21-independent.</p>", "<p>We then hypothesized that alternative pathways might be responsible for <italic>PKD2</italic>-induced proliferation in this system. To this end, we performed a genome-wide gene expression analysis on TECs isolated from two normal Sprague Dawley rats and three <italic>PKD2 </italic>(1–703) rats. Differentially expressed genes were identified with ANOVA. We concentrated only on genes involved in the cell cycle regulation (Figure ##FIG##5##6A##). From all the cell cycle genes listed in figure ##FIG##5##6A##, only two differ significantly in expression between normal and mutant cells, those being Cdk2 and cyclin-dependent kinase inhibitor 1C or p57^KIP2. On the contrary, p21 did not show any significant difference, confirming the Western blot results (Figure ##FIG##5##6A## and ##FIG##4##5A##).</p>", "<p>The chip data were verified by quantitative real-time PCR analysis after normalization using two housekeeping genes, HPRT and GAPDH. In agreement with the chip data, p57 mRNA levels were downregulated in the mutant animals as compared with their normal counterparts (normalized fold change 4.7 ± 0.19). Similarly, Cdk2 mRNA levels were augmented in the mutant cells (normalized fold change 1.2 ± 0.015) (Figure ##FIG##6##7A##). Cdk2 protein upregulation and p57 protein downregulation were also verified by immunoblotting. Consistent with the microarray data, Cdk2 protein levels were significantly elevated in mutant primary cultures (normalized fold change 2.2 ± 0.06). Similarly, p57 levels were downregulated in mutant TECs (normalized fold change 1.9 ± 0.2) (Figure ##FIG##6##7B##). On the contrary, Western blot analysis demonstrated, as expected, that p57 protein levels remain unchanged in HEK293 stable clones and NRK-52E transfectants (Figure ##FIG##6##7C##). It should be noted that p57 levels in the cell lines examined is expressed at very low levels and it was barely detectable by Western blot. Given that in the <italic>PKD2</italic>(1–703) transgenic rat the cysts originate predominantly from the proximal tubule segment of the nephron, we wanted to exclude the possibility that proximal tubule cells are overrepresented in the primary mutant TECs culture, thus confounding the interpretation of the results. In order to do that, lysates from normal and mutant TECs were immunobloted with anti-Megalin antibody, a proximal tubule marker [##REF##16652363##35##]. As shown on figure ##FIG##4##5A##, Megalin protein levels were equivalent among normal and mutant TECs suggesting that the proportion of cells of proximal origin was comparable among the different cultures and did not create a sampling bias.</p>" ]

[ "<title>Discussion</title>", "<p>Various studies on renal cystic tissues and cell lines demonstrated that altered regulation of tubular epithelial cell proliferation is a key factor in the pathogenesis of ADPKD. What remains unclear is the timing of the misregulated growth as well as the pathways involved. Recently, in an attempt to answer these questions a number of groups provided evidence for the involvement of Cdk2 in the process of cystogenesis. Progression through the cell cycle is regulated by a family of cyclin-dependent kinases (CKs) whose activities are controlled by the relative ratio of cyclins and Cdk inhibitors (CKIs) [##REF##8791491##36##,##REF##8939849##37##]. There are two classes of CKIs in mammals, the p21^CIP1and p16^INK4families. Members of the p16^NK4family bind and inhibit only Cdk4 and Cdk6 kinases [##REF##8620534##38##]. In contrast, members of the p21^CIP1family (p21^CIP1, p27^KIP1and p57^KIP2) inhibit all G1/S phase CDKs. The transition of cells from the G0/G1 into the S phase of the cell cycle involves the activities of Cdk2, Cdk4 and Cdk6 [##REF##8939849##37##].</p>", "<p>Bhunia et. al. were the first to address the role of CDKs in PKD-induced proliferation. Specifically, they demonstrated that one of the functions of the polycystin-1/2 complex is to regulate the JAK/STAT pathway and consequently control cellular proliferation. They showed that overexpression of wild-type polycystin-1 can activate JAK2/STAT-1, a process that resulted in upregulation of the CKI p21^waf1. As expected, increase in p21 levels led to inhibition of Cdk2 and cell cycle arrest. The ability of polycystin-1 to modulate Cdk2 activity was dependent on polycystin-2. These results implied that compromised polycystin-1 activity is expected to have the opposite effect, thus explaining the abnormal proliferation observed in ADPKD cystic cells.</p>", "<p>An independent study addressed directly the role of PC2 in cell cycle regulation and Cdk2 activity. It was demonstrated that PC2 could directly interact with Id2, a member of the HLH family that is known to control cell proliferation and differentiation. The direct association of PC2 with Id2 was shown to regulate the nuclear translocation of Id2 and thus modulate the cell cycle through the Id2/p21/Cdk2 pathway [##REF##16311606##17##]. Based on these results a model was proposed according to which PC1 can increase PC2 phosphorylation leading to enhanced Id2/PC2 interaction and reduced Id2 nuclear import. This in turn, prevents Id2 repression of E-box-dependent activation of transcription of genes such as p21. Increased p21 will inhibit Cdk2 activity and arrest the cells at G0/G1 phase of the cell cycle. At the same time PC1 can lead to Cdk2 inhibition independent of Id2 through the JAK/STAT pathway. Based on this model mutations in either PC1 or PC2 can disrupt these pathways leading to abnormal cell proliferation [##REF##16311606##17##]. A recent report also demonstrated reduced levels of p21 in human and animal PKD tissues as well as in affected cell lines implying a role of p21/Cdk2 in cystogenesis [##REF##17714589##39##].</p>", "<p>In this study we attempted to examine further this hypothesis. We generated stable clones expressing either wild-type or mutant R742X <italic>PKD2 </italic>in HEK293. To our surprise, overexpression of wild-type PC2 did not affect proliferation of these cells. Cell cycle profile analysis, PCNA, p21 expression levels and Cdk2 activity remained unchanged among different transfectants. The reason for this discrepancy remains unclear given that the same cell line and similar experimental conditions were used in the previous studies [##REF##12007403##13##,##REF##16311606##17##]. In order to eliminate the possibility that the exogenously expressed wild-type <italic>PKD2 </italic>was not functional, we performed whole cell current measurements in vector-only, WT <italic>PKD2 </italic>and R742X <italic>PKD2 </italic>clones. As expected, HEK293 clones expressing wild-type <italic>PKD2 </italic>displayed an increase in the current amplitude of whole cell inward and outward currents recorded either in normal extracellular tyrode solution or symmetrical K⁺. Such result excludes the possibility that an inactive PC-2 was expressed in HEK293 cells. In addition, absence of phenotype could not be attributed to the mislocalization of the expressed protein as determined by immunofluorescent analysis.</p>", "<p>In an attempt to clarify these contradictory results we utilized a different cell line system. The NRK-52E cells are \"normal\" rat tubular epithelial cells, thus we hypothesized that this is a more appropriate system to study PC-2-induced proliferation and STAT-1/p21/Cdk2 activation. Nevertheless, similar results were obtained with the NRK-52E cells (Figure ##FIG##3##4##). The disparity of our results compared to previous studies is puzzling. Li et. al [##REF##16311606##17##], observed cell cycle arrest and Cdk2 inhibition in HEK293T cells after expression of wild-type PC-2, and not in HEK293 cells used in our study. HEK293T cell line is a derivative of HEK293 that stably expresses the large T-antigen of SV40. In these cells transfected plasmids that contain the SV40 origin are replicated to a copy number of 400–1000 plasmids/cell and therefore express the transgene at higher levels. However, this is unlikely to be the reason for the discrepancy given that high expression of wild-type and mutated PC-2 was achieved in our HEK293 clones (Figure ##FIG##0##1##) and in NRK-52E cells (Figure ##FIG##3##4##) after transient transfection.</p>", "<p>One of the unwanted side effects of cellular immortalization might be the alteration of basal proliferation rate in cells. This can be highly significant in proliferation studies. As a result we decided to switch to a primary cell culture system. We examined the ability of mutated PC-2 (1–703) to activate the STAT-1/p21/Cdk2 pathway in primary renal epithelial cells isolated from <italic>PKD2</italic>(1–703) transgenic rat [##REF##16943309##19##].</p>", "<p>Isolation of TECs from the transgenic animals was performed using a sequential filtration method. Using this method we avoided any potential activation of surface receptors taking place during antibody-based isolation techniques. Purified tubular epithelial cells were cultured in low serum medium and on laminin-coated plates to avoid differentiation. The epithelial character of the cells was regularly evaluated by measuring epithelial (cadherin) and fibroblastic (vimentin) markers. TECs isolated from different animals showed augmented PCNA levels, a decrease of the G0/G1 phase cells and increase of the G2/M phase cells. This was the first time in our hands that we observed a higher proliferation activity in cells overexpressing a mutated PC-2. These results indicated that indeed PC-2 can alter cellular proliferation in renal epithelial cells, but it also suggests that such process is complicated and possibly multifactorial and can not be easily recapitulated in <italic>in vitro </italic>cell line systems [##REF##17084592##40##]. In support of this, a recent report focused on the dynamics of cyst formation by utilizing an inducible <italic>Pkd1 </italic>mouse model, demonstrated that proliferation was not appreciably higher in cystic specimens than in aged matched controls. Based on their results, the authors suggested that the relationship between cellular proliferation and cyst formation may be indirect [##REF##17965720##41##]. Similar data were obtained from Zebrafish studies where it was shown that increased cell number in cyctic phenotype is a secondary consequence of tubule dilation rather than the leading cause of cyst formation [##REF##18178183##42##]. In our study, it appears that mutated PC-2-induced proliferation in primary cells proceeds independently of the STAT-1/p21 pathway since there is no change in the levels of p21 or on STAT-1 phosphorylation. Based on these results it is clear that in the rat system we investigated, PC-2-induced proliferation proceeds through an alternative pathway other than STAT-1/p21.</p>", "<p>Using gene expression profiling we were able to identify a candidate that may mediate the PC2-induced proliferation in <italic>PKD2</italic>(1–703) rat. Among all the cell cycle related genes, only two showed misregulation in TECs isolated from diseased rats, cyclin-dependent kinase inhibitor 1C (p57kip2) and Cdk2. The p57 kip2 belongs to the p21^WAF/Cip1family. Studies have shown that p57 binds tightly to the G1 and S phase kinases, cyclin E/Cdk2, cyclin D2/Cdk4, cyclin A/Cdk2 and to a lesser extent to cyclin B/Cdc2 and effectively inhibits their activity [##REF##7729683##43##]. An important difference between p57 and the other members of the family, is that p57 is not regulated by p53 but by p73 [##REF##7729684##44##, ####REF##11891335##45##, ##REF##15985436##46####15985436##46##].</p>", "<p>We observed a downregulation of p57 at both mRNA and protein levels in mutant cells with the absence of any change in p21 levels. This possibly signifies that PC-2 might alter cellular proliferation through p57/Cdk2 in these cells. It is possible that expression of mutant PC-2 can result in p57 downregulation by augmenting Id2 nuclear import and subsequent inhibition of p57 transcription [##REF##16311606##17##]. This hypothesis is in agreement with experiments in neural cells where it was shown that Id2 could regulate cell cycle through p57 [##REF##16705184##47##].</p>", "<p>In addition to p57 downregulation, we observed an increase in Cdk2 protein level. This is interesting since it appears that Cdk2 activity might be augmented simultaneously in two different ways (downregulation of the inhibitor and upregulation of the kinase). Whether Cdk2 increase is part of a positive feedback loop is still not known. Nevertheless, this simultaneous alteration in p57 and Cdk2 levels might result in a rapid increase in Cdk2 activity and subsequently to higher proliferation rate. A concern regarding our results might arise from the possibility that the isolated TECs are not equally representative of the various nephron segments in healthy and mutant rats, a concern however that cannot easily be addressed within the scope of this work. More specifically though, we addressed the issue of over-representation of the TECs from the proximal cysts by showing similar levels of a proximal tubule marker, megalin expression in normal and mutant TECs (Fig. ##FIG##4##5A##).</p>", "<p>In conclusion, the level of p57 contribution in the PC-2-induced proliferation in renal epithelial cells is still unclear. Future experiments will focus on identifying the pathways leading to p57 reduction and whether this decrease is necessary for PC-2-induced proliferation in renal tubular epithelial cells. We consider it of particular significance that no matter how these experiments pan-out, our study introduces a new pathway in ADPKD, through which PC-2 might lead to Cdk2 activation and increase in cellular proliferation, which is independent of STAT-1/p21. Also, once again it should be emphasized that biological systems are unpredictably complex and may exert similar effects and end results through more than one pathway. Finally a word of caution should be expressed as regards the interpretation of experiments performed on genetically grossly modified established cells lines, which are far from representing the complexity of whole organs or organisms.</p>" ]

[ "<title>Conclusion</title>", "<p>We have shown that p57^KIP2, a cyclin-dependent kinase inhibitor is downregulated and cyclin-dependent kinase 2 (Cdk2) is upregulated in primary tubular epithelial cells isolated from a <italic>PKD2 </italic>transgenic rat. In addition, primary cells expressing mutant <italic>PKD2 </italic>exhibit increased proliferation compared to their normal counterparts. On the contrary, expression of mutant <italic>PKD2 </italic>in two kidney cell lines failed to alter cellular proliferation and p57^KIP2protein levels. Most importantly, although exogenous expression of mutant <italic>PKD2 </italic>ablated current activity, compared to wild-type, however in cell lines or primary TECs had no effect on the STAT-1/p21/Cdk2 pathway. In conclusion this report highlights the probable involvement of p57^KIP2on epithelial cell proliferation in ADPKD implicating a new mechanism for mutant polycystin-2 induced proliferation.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Autosomal Dominant Polycystic Kidney Disease (ADPKD) is characterized by the formation of multiple fluid-filled cysts that destroy the kidney architecture resulting in end-stage renal failure. Mutations in genes <italic>PKD1 </italic>and <italic>PKD2 </italic>account for nearly all cases of ADPKD. Increased cell proliferation is one of the key features of the disease. Several studies indicated that polycystin-1 regulates cellular proliferation through various signaling pathways, but little is known about the role played by polycystin-2, the product of <italic>PKD2</italic>. Recently, it was reported that as with polycystin-1, polycystin-2 can act as a negative regulator of cell growth by modulating the levels of the cyclin-dependent kinase inhibitor, p21 and the activity of the cyclin-dependent kinase 2, Cdk2.</p>", "<title>Methods</title>", "<p>Here we utilized different kidney cell-lines expressing wild-type and mutant <italic>PKD2 </italic>as well as primary tubular epithelial cells isolated from a PKD transgenic rat to further explore the contribution of the p21/Cdk2 pathway in ADPKD proliferation.</p>", "<title>Results</title>", "<p>Surprisingly, over-expression of wild-type <italic>PKD2 </italic>in renal cell lines failed to inactivate Cdk2 and consequently had no effect on cell proliferation. On the other hand, expression of mutated <italic>PKD2 </italic>augmented proliferation only in the primary tubular epithelial cells of a rat model but this was independent of the STAT-1/p21 pathway. On the contrary, multiple approaches revealed unequivocally that expression of the cyclin-dependent kinase inhibitor, p57^KIP2, is downregulated, while p21 remains unchanged. This p57 reduction is accompanied by an increase in Cdk2 levels.</p>", "<title>Conclusion</title>", "<p>Our results indicate the probable involvement of p57^KIP2on epithelial cell proliferation in ADPKD implicating a new mechanism for mutant polycystin-2 induced proliferation. Most importantly, contrary to previous studies, abnormal proliferation in cells expressing mutant polycystin-2 appears to be independent of STAT-1/p21.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>All authors have read and approved the manuscript. KNF and PK performed most of the experiments. KNF also helped in the conception of the experimental plan and in the writing of this manuscript. EK helped in the establishment of HEK293 stable clones. RW and NG created and maintained the transgenic PKD rat. NG provided the PKD transgenic rats. LT and C–XB provided the cDNA constructs and performed the electrophysiology experiments. NG and LL designed and performed the microarray experiments. Finally CD conceived the study, supervised the work and helped in the writing of this manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2369/9/10/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Dr Bettina Kränzlin and Dr Sigrid Hoffmann for their assistance with the transgenic rats and Maria Saile for the microarray experiments. We thank Dr Petros Petrou for critical discussions and suggestions during the course of this work. This work was funded mainly through an internal competitive grant from the University of Cyprus and a grant from the Cyprus Research Promotion Foundation ENISX/0504/12.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Expression of wild-type or mutant PC-2 does not affect proliferation or STAT-1/p21/Cdk2 activity in HEK293 cells</bold>. (A) Whole cell lysates containing equal amounts of protein from three stable individual clones of each transfectant (Vector-only, HA-WT <italic>PKD2 </italic>and HA-R742X <italic>PKD2</italic>) were analyzed by Western blotting for expression of p21, phosphorylated STAT-1, PCNA, β-tubulin, HA and PC-2. (B) Cdk2 immunoprecipitates from two clones of each transfectant were subjected into an in vitro Cdk2 kinase assay using Histone 1A as substrate. Equal amount of Cdk2 was confirmed by immunoblotting the precipitates with anti-Cdk2 antibody. Data are representative of five independent experiments.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Cell cycle profile of HEK293 clones is unaffected by expression of wild-type or mutant PC-2</bold>. Three different clones of each transfectant were synchronized and subjected to propidium iodide cell cycle analysis by flow cytometry. The percentage of cells in each phase of the cell cycle was determined. The results are presented as mean of triplicate counts for each clone ± SEM. No statistically significant difference was detected. Analysis of serum-starved cells demonstrated that more than 90% of the live cells were arrested at G0/G1 phase confirming successful synchronization of the cells cultured (data not shown).</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Functional expression of <italic>PKD2 </italic>in HEK293 cells</bold>. Whole cell step currents in vector-only (A and F), wild type <italic>PKD2</italic>- (B and G), or <italic>PKD2</italic>(742X)-stably transfected HEK293 cells (C and H) in normal extracellular tyrode solution (130K⁺:135Na⁺/2 Ca²⁺/5.4 K⁺) (A-C) or symmetrical K⁺(130K⁺:130K⁺) (F-H). Current-voltage (I–V) curves derived from a step protocol in untransfected (black squares), mock-transfected (vector control) (black circles), WT <italic>PKD2</italic>- (red circles), or <italic>PKD2 </italic>(R742X)-stably transfected HEK293 cells (green circles) in normal extracellular tyrode solution (D) or symmetrical K⁺(I). ''*'': p &lt; 0.05.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Expression of wild-type or mutant PC-2 does not affect proliferation or STAT-1/p21/Cdk2 activity in NRK-52E cells</bold>. <bold>(A)</bold>. Whole cell lysates containing equal amounts of protein from NRK-52E cells trasnsiently transfected with vector-only, HA-WT <italic>PKD2</italic>, HA-R742X <italic>PKD2 </italic>and HA-1–702 <italic>PKD2 </italic>were analyzed by Western blotting for expression of p21, phosphorylated STAT-1, PCNA, tubulin, HA and PC-2. A non-specific band is detected in vector-only and WT PKD2 lanes in the HA blot and co-migrates with mutated PC-2 in this cell line. (B) Cdk2 immunoprecipitates from each transfectant were subjected into an in vitro Cdk2 kinase assay using Histone 1A as substrate. Equal amount of Cdk2 was confirmed by immunoblotting the precipitates with anti-Cdk2 antibody. Data are representative of three independent experiments performed. No statistically significant difference was detected.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>Primary tubular epithelial cells (TECs) isolated from the kidneys of a PKD transgenic rat expressing a truncated PC-2 (<italic>PKD2</italic>(1–703)) display higher proliferative activity compared with TECs isolated from normal Sprague-Dawley rats</bold>. (A). Whole cell lysates containing equal amounts of protein from TECs isolated from normal Sprague-Dawley rat (SD) and TECs isolated from PKD transgenic rat (Mut) were analyzed by Western blotting for expression of p21, phosphorylated STAT-1, PCNA, tubulin, PC-2, cadherin, vimentin and megalin. All blots are representative of experiments performed on at least two different transgenic animals. Endogenous PC-2 can be seen on long exposures that also bring out high background (not shown) (B). Cell cycle profile of normal (SD) or mutants (Mut) TECs. The results are presented as mean of triplicate counts (three independent cultures) for each animal. ± SEM (ANOVA p &lt; 0.01. * = significant difference). The data are representative of two experiments performed using two different pairs of animals.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p><bold>Genome-wide expression analysis reveals differential expression of Cdk2 and p57 in TECs isolated from transgenic rat (<italic>PKD2</italic>(1–703))</bold>. (A). List of cell cycle-related genes examined by microarray analysis. The (*) denotes statistical significance after Bonferroni correction. (B). Each data point on the volcano plot stands for one gene. The cutoff of p-value after Bonferroni correction is shown by the red line. Only the two significantly differentially expressed genes are labeled with their gene symbol.</p></caption></fig>", "<fig position=\"float\" id=\"F7\"><label>Figure 7</label><caption><p><bold>Tubular epithelial cells (TECs) isolated from the <italic>PKD2</italic>(1–703) rat (Mut) have reduced p57 and augmented Cdk2 mRNA and protein compared with TECs isolated from normal rat (SD)</bold>. (A). Real-time PCR of p57 and Cdk2 in isolated TECs. Data represent the mean of normalized fold change from three independent samples ± SEM (ANOVA p &lt; 0.01. * = significant difference). Data were normalized using two housekeeping genes, HPRT and GAPDH. (B). Protein levels of p57 and Cdk2 represented as the mean of normalized fold change of two independent Western blot experiments ± SEM (ANOVA p &lt; 0.05. * = significant difference). Data were normalized by β-tubulin expression. (C). p57 protein levels were determined by western blot analysis in HEK293 stable clones and in NRK-52E transfectants. As expected, protein level of p57 is not perceptibly altered.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1471-2369-9-10-1\"/>", "<graphic xlink:href=\"1471-2369-9-10-2\"/>", "<graphic xlink:href=\"1471-2369-9-10-3\"/>", "<graphic xlink:href=\"1471-2369-9-10-4\"/>", "<graphic xlink:href=\"1471-2369-9-10-5\"/>", "<graphic xlink:href=\"1471-2369-9-10-6\"/>", "<graphic xlink:href=\"1471-2369-9-10-7\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Schizophrenia is among the most costly of all mental illnesses, with an estimated annual per person direct treatment cost that is approximately 2-fold higher than the cost of major depression and more than 4-fold higher than any anxiety disorder [##REF##15877774##1##]. In addition to genetic factors [##REF##16199828##2##] and medication adherence [##REF##7481573##3##, ####REF##15056516##4##, ##REF##15172945##5####15172945##5##], various clinical and demographic factors are associated with higher risk of relapse and hospitalization, events that are predictive of higher costs [##REF##14720128##6##]. These factors include male gender, lower educational level, unemployment, higher illness chronicity, higher frequency of alcohol consumption, co-occurring alcohol and substance abuse, a history of depression and/or suicide attempts, a history of violence and/or arrests, and recent hospitalization [##REF##15172946##7##, ####REF##15056580##8##, ##REF##17110813##9##, ##REF##17194273##10####17194273##10##]. The positive association of these factors with mental health treatment costs suggests that persons who are more vulnerable to crises (i.e., personal life events typically associated with an acute societal intervention) have higher treatment costs than persons who are less vulnerable to these events.</p>", "<p>The relationship between crisis events and treatment costs has not been well-studied. Most prior studies of costs are based on administrative data that provide relatively little information about vulnerabilities and recent crises [##REF##14552502##11##]. Information about the relationship of crises to costs is essential for accurate risk adjustment, the process of assigning capitation rates for enrollees of public and private health insurance plans. Capitation rates that are set too low provide plans and HMOs insufficient incentive to treat high-risk, high cost patients [##REF##11508640##12##, ####REF##10170344##13##, ##REF##8669521##14####8669521##14##]. Information on the effect of crisis events is particularly valuable at a time when there are concerted efforts to decrease patient hospitalization and manage various types of psychiatric crises in the community. In the United States, for example, the length of hospital stay has gradually declined in the past 10 years [##REF##15994711##15##], attesting to economic and policy-driven pressures to reduce psychiatric hospitalizations. Information about crisis events has also clinical utility in usual care settings, where it may help identify more vulnerable patients with more complex illness trajectories who require specialized interventions and better coordination with other social agencies, including the criminal justice system.</p>", "<p>Most prospective longitudinal data do not enable the study of whether patients diagnosed with schizophrenia who have experienced a recent crisis incur higher direct mental health costs compared to patients who have not experienced a crisis. However, the availability of comprehensive clinical, functional and economic data from a large 3-year prospective non-interventional observational study of persons treated for schizophrenia in the United States provided the opportunity to address this topic in some detail. The objective of the current investigation was to assess the relationship between recent crisis events and direct annual mental health costs in patients diagnosed with schizophrenia spectrum disorders in clinical practice settings. The crisis events examined included (1) suicide attempt, (2) psychiatric hospitalization, (3) arrest, (4) violent behavior, and (5) substance use disorder. Five separate propensity score analyses were performed (one for each crisis event) to evaluate the relationship between each crisis event and cost when compared to the group of individuals who did not experience that specific crisis event.</p>" ]

[ "<title>Methods</title>", "<title>Study design</title>", "<p>Data used in the current analysis are from the U.S. Schizophrenia Care and Assessment Program (US-SCAP), a prospective, non-interventional, non-randomized, 3-year observational study of more than 2300 persons with schizophrenia. The data used in the current analysis consist of the first full year of data available for each patient in the study. The goal of the US-SCAP was to examine the relationship of clinical and treatment variables with outcomes of persons diagnosed with schizophrenia-spectrum disorders who were receiving mental health care in outpatient and inpatient settings at 6 treatment sites. The study, conducted between July 1997 and September 2003, enrolled a total of 2327 patients at 6 health care sites chosen to provide a diverse patient sample in terms of geography, ethnicity, and clinical setting (e.g., university and community mental health centers, Veterans Affairs (VA) hospitals, and community and state hospitals). Sites were only included in the study if they offered open and unrestricted formulary access to all available antipsychotics and were not relying on any algorithms for treatment decision-making. The study protocol was reviewed and approved by the Institutional Review Board (IRB) at each site prior to study initiation, and written informed consent was obtained from all participants.</p>", "<p>A detailed description of study design and methods are provided in previous publications [##REF##15175112##16##, ####UREF##0##17##, ##REF##14770056##18####14770056##18##]. Briefly, US-SCAP study enrollment was offered to all patients who were 18 years or older who had a DSM-IV diagnosis of schizophrenia, schizoaffective, or schizophreniform disorder. Patients were enrolled regardless of psychiatric or medical comorbidity, use of concomitant medications, or presence of behavioral problems (criminal or otherwise). The goal was to obtain the broadest and most representative sample of schizophrenia-spectrum patients seen in clinical practice settings.</p>", "<p>For the purposes of the current cost analysis, a subgroup of US-SCAP enrollees (n = 1557; 67%) was identified who had a full year of information available on mental health resource utilization. If patients had &gt;1 year of mental health resource information, the patient's earliest year was used in the current analysis.</p>", "<p>Five crisis subgroups were defined based on the presence of any one of the following events at the start of the 1-year observation period: (1) a suicide attempt in 4 weeks prior to the baseline assessment per patient self report; (2) a psychiatric hospitalization in the 6 months prior to baseline per medical records; (3) an arrest in the 6 months prior to baseline per patient self report; (4) violent behavior in the 4 weeks prior to baseline per patient self report; and (5) a diagnosis of substance use disorder, based on medical record, which occurred at any time during the study period.</p>", "<title>Measures</title>", "<p>At study entry, patients completed a semi-structured interview, during which information about psychiatric history and background characteristics was collected. Patients' medical records were systematically abstracted every 6 months by examiners, trained and certified by the contract research organization, using a medical record abstraction form developed for this study that summarized mental health resource utilization during the preceding 6 months. Patients were queried about use of medical and psychiatric services outside of their usual treatment site. Study staff members regularly obtained medical records from treatment sites mentioned by patients. At 6-month intervals, patients also completed the SCAP-Health Questionnaire (SCAP-HQ [##REF##14552500##19##]), a personal interview that includes questions on recent drug and alcohol use (in the past 12 months), suicide attempt (in the past 4 weeks), arrests (in the past 6 months), and violent behavior (in the past 4 weeks). At 1-year intervals, clinical assessments of psychiatric symptoms, medication side-effects, and functioning were completed by trained clinicians.</p>", "<title>Assessment of costs</title>", "<p>The total annual direct costs were calculated as the sum of the following component charges: medication costs, including antipsychotics and other psychotropics (the cost of antipsychotics was based on Average Wholesale Price discounted by 15% for atypical antipsychotics); costs of psychiatric hospitalization (based on the actual charges); and costs of other mental health services (based on Relative Value Units [RVUs; 20,21] developed from management information systems (MIS) data at each site to help address variations across sites in available information about resource types, costs, and durations). The costs of other mental health services included the following cost components: emergency services, day treatment, outpatient medication management by a physician, individual outpatient therapy, outpatient psychosocial group interventions, and case management.</p>", "<title>Statistical analyses</title>", "<p>For comparisons of baseline characteristics, t-tests were used for continuous variables and Mantel-Haenszel chi-square tests for categorical variables. In order to correct for potential bias not attributable to membership in a crisis event subgroup, the propensity score method was used to balance the crisis event versus non-crisis event subgroups. The variables selected <italic>a priori </italic>for calculating the logit score using the propensity method were: age at enrollment, gender, race, illness duration, comorbid affective disorder, comorbid substance use, comorbid personality disorder, diagnosis of mental retardation, insurance status, inpatient status at the beginning of the 1-year observation period, site, and days from July 1 1997 to the beginning of the study. Substance use was not used in the propensity score model when substance use was being analyzed as the crisis event, and inpatient status was not used in the propensity score model when costs associated with recent psychiatric hospitalizations were estimated.</p>", "<p>The propensity score-adjusted bootstrapping method analyzed mean differences of costs between each crisis event and non-crisis event group by first calculating the logit score for each patient based on the above adjustments. Five bins of logit scores were then created for each crisis event group. The bootstrap resampling method was performed by randomly selecting an equal size of sample from each of the 10 bins (5 bins for each treatment) into 1 group and calculating the total cost difference between the 2 treatment groups. The above steps were then repeated 1000 times, generating a total of 1000 data points' distribution for testing the null hypothesis using 2-tailed p-values [##UREF##1##22##]. The propensity score-adjusted bootstrapping method was also used to test the mean differences of costs for the 2 high level crisis event groups: patients with 2 or more crisis-event variables and patients with 3 or more crisis-event variables. This method accommodates the distributional and correlational properties of the data [##UREF##2##23##]. In addition to mean total cost and cost of psychiatric hospitalization per patient for the index year, we also calculated the mean annual length of psychiatric hospital stay and the mean number of psychiatric hospital admissions per crisis event category. This information aimed at clarifying which type of crisis made a significant and unique contribution to increased costs due to hospitalization, the costliest component in the treatment of schizophrenia. SAS version 8 was used to perform all statistical analyses, with all effects tested at a 2-sided α level of 0.05 [##UREF##3##24##].</p>" ]

[ "<title>Results</title>", "<p>The total sample used in the cost analysis consisted of the 1557 patients (Table ##TAB##0##1##) for whom there were at least 1 year of complete medical information during the 3-year study period. For the cost analysis sample, the typical patient was between 30 and 50 years of age, with at least a 10-year history of illness. Almost all (94.9%) of patients were taking an antipsychotic drug at the time of enrollment; 38.3% were also taking an antidepressant, and 31.0% were also taking a mood stabilizer.</p>", "<p>Participants who had incomplete medical information for the purpose of costs calculations (N = 770) were excluded from this study. The excluded patients were similar to the analysis sample (N = 1557) in terms of gender (male: 63% vs. 60%; p = .489), but were significantly younger (40.8 ± 11.4 vs. 42.4 ± 11.1 years; p &lt; .01) and were less likely to be African-American (32% vs. 38%), but more likely to be Hispanic (17% vs. 13%) or Caucasian (51% or 49%; overall p = .013). Compared to the 1557 retained subjects, the excluded group was also more likely to have been hospitalized in the prior 6 months (15.4% vs 25.2%, respectively), to be arrested in the prior 6 months (3.8% vs 10.1%), to manifest violent behaviors in the past 4 weeks (4.2% vs 8.4%) and attempt suicide in the past 4 weeks (1.2% vs 3.8%), but had a similar proportion of persons with substance abuse diagnosis (36.1% vs 29.5%). The excluded group may have experienced a greater interface with the criminal justice system (e.g., jails), thus less likely to have complete mental health information in the present study.</p>", "<p>The mean 1-year mental health treatment costs per patient totaled $16,098 (Table ##TAB##1##2##). Te two largest costs were psychotropic medication (30%) and hospitalization (29%; Figure ##FIG##0##1##). The remaining 6 cost component categories comprised 48% of the total cost, with none contributing more than 10% (Figure ##FIG##0##1##). Among the hospitalized participants, the correlation between hospitalization costs and total mental health treatment costs during the index year was high (r = .987, p &lt; .001), reflecting the fact that hospitalization is the core driver of total costs.</p>", "<p>Sorting patients into crisis event categories yielded 5 overlapping subgroups. The largest subgroups were co-occurring substance use disorder (n = 413) and hospitalized in the past 6 months (n = 240), followed by violent in the past 4 weeks (n = 62), arrested in the past 6 months (n = 56), and attempted suicide in the past 4 weeks (n = 18). Prior to formal matching, each of the crisis event subgroups was not substantially different from each other on most baseline demographic and clinical characteristics (Table ##TAB##2##3##). The typical patient was most likely to be a male (~61%) between the ages of 30 and 50 years. As expected, the proportion of males was somewhat higher in the subgroup who had been arrested in the past 6 months (80%) and the subgroup with comorbid substance abuse (74%). Though the sample size was small (N = 18), the subgroup who had recently attempted suicide was younger (mean age, 34 years), more likely to be female (61%), and more likely to have a depressive component, with a mean MADRS score of 31, and a diagnosis of schizoaffective disorder (50%; Table ##TAB##2##3##).</p>", "<p>The five crisis event categories were not mutually exclusive (Table ##TAB##3##4##), with highest levels of overlap between substance use disorder, arrest, suicide attempt and violent behavior. A substantial proportion (46.4%) of those with arrest, with suicide attempt (44.4%), and with violent behaviors (41.9%) had also a diagnosis of substance abuse. While 21.1% of those with substance abuse diagnosis were hospitalized, about a third (36.3%) of the hospitalized had co-occurring diagnosis of substance abuse. Furthermore, participants with substance use disorders had the lowest mean annual number of hospitalization days (25.5) and a relatively low mean annual number of psychiatric admissions (0.7).</p>", "<p>The 1-year hospitalization and total 1-year mental health treatment costs were calculated for each crisis event subgroup, and compared, using a propensity score adjusted bootstrap re-sampling method (repeated 1000 times), to the non-crisis event subgroup. The results of this analysis (summarized in Table ##TAB##4##5##) found significantly higher mean 1-year hospitalization costs, and total mental health costs, for each crisis event subgroup except comorbid substance abuse. The mean total annual mental health cost for the subgroup who had experienced no crisis event was $11,739 per patient. The mean total mental health cost for patients with at least 1 crisis event was $22,704 per patient, with the highest cost observed in the subgroup of patients who attempted suicide, followed by patients with a recent psychiatric hospitalization, arrest, violent behavior, and those with comorbid substance use disorder (Table ##TAB##4##5##). Of the patients (n = 619) who qualified for inclusion in at least 1 crisis event subgroup, 30.9% had experienced only one crisis event, 8.9% had experienced 2 or more crisis events, and 1.9% had experienced 3 or more events. The presence of ≥2 and ≥3 crisis events was associated with a significant and step-wise increase in both 1-year hospitalization and total 1-year mental health costs compared to the costs in the non-crisis event subgroup, reflecting a high mean number of hospitalization days (72.7 and 101.1 days, respectively) and a high mean number of hospital admissions (1.7 and 2.7, respectively) (Table ##TAB##5##6##).</p>" ]

[ "<title>Discussion</title>", "<p>Previous research has identified various clinical variables as significant predictors of relapse and hospitalization in patients diagnosed with schizophrenia. The current study extends these findings by providing, for the first time, information on the annual mental health costs associated with experiencing specific crisis events. The use of comprehensive assessments in a prospective naturalistic study with valid and reliable instruments enabled the identification of individuals who had experienced specific crisis events, and the systematic collection of resource utilization data for different types of mental health services. Given the large and diverse sample of schizophrenia patients analyzed and the prospective, naturalistic design, the findings of the study are likely to be applicable to patients with schizophrenia treated in large systems of care across the United States.</p>", "<p>Although 60.2% of the participants did not report experiencing a crisis event, 39.8% of the patients did, with 30.9% of all participants meeting only one crisis event criterion, 8.9% meeting 2 or more criteria, and about 2% meeting three or more criteria. Thus, the base rate for experiencing a crisis event was not low.</p>", "<p>As hypothesized, patients experiencing a crisis event that took place (or at least started) before the 1-year observation period accounted for a preponderance of the total mental health costs during that period. Furthermore, a disproportionate contribution to increased mental health costs was made by the subgroup who reported experiencing 2 or more crisis events. The per patient mean total annual mental health cost for the non-crisis event subgroup was $11,739, while patients who reported at least 1 crisis event had a per patient mean of $22,704. (Note that the mean total costs shown in Table ##TAB##4##5## for each crisis event subgroup are higher, because some of the individuals in each group may have experienced 1 or more additional crisis events.)</p>", "<p>The highest annual per patient total mental health cost was in the subgroup of patients who attempted suicide, followed by patients with a recent psychiatric hospitalization. In the current investigation, patients with a concurrent substance use disorder did not appear to have significantly higher total mental health costs compared to those without a substance use disorder. This is an unexpected finding since substance abuse has previously been reported to be a predictor of medication nonadherence [##REF##12416599##25##,##REF##16889456##26##], which, in turn, is often reported to be a highly significant predictor of future relapse and hospitalization [##REF##11376229##27##, ####REF##12187177##28##, ##REF##16649833##29####16649833##29##]. One possible explanation is suggested by recent data from the Clinical Antipsychotic Trials in Intervention Effectiveness (CATIE) study which found that substance abuse was a marker for higher psychosocial functioning in schizophrenia [##REF##16870961##30##]. Another possibility is that some of the substance abusing participants – especially those with prior arrests and violent behaviors- had more extensive interface with the criminal justice system (e.g., jails), thus some of their costs may have shifted from the mental health to the criminal justice system. In addition, nonadherence with medication may also underlie the higher costs incurred by patients in the other crisis event categories, since nonadherence with antipsychotic medication was previously found to be associated not only with psychiatric hospitalizations but also with a higher risk of violent behaviors, arrests and suicide attempts [##REF##16649833##29##]. While detecting ongoing medication nonadherence may be difficult and challenging in usual care settings, the occurrence of a crisis event is apt to be more readily identifiable, thus serve as a clinical marker of patients' greater vulnerability from a clinical and functional perspective. This may turn out to be a convenient way of targeting subsets of patients with different illness profiles (e.g., with different diagnostic characteristics, illness trajectories, or vulnerability factors) and effectively treating individuals who experience crisis events.</p>", "<p>Cost differences among patients who experience a crisis event appeared to be primarily driven by the cost of psychiatric hospitalization, with a strong and significant correlation (r = .99) between hospitalization cost and total annual mental health costs. In the current analysis, hospitalization comprised, however, only 29% of the total annual mental health costs, while antipsychotic medication comprised 23%. This finding is inconsistent with prior research in which the cost attributed to hospitalization was of larger proportion- about 50%–80% – of the total cost [##REF##17828655##31##,##REF##16675759##32##]. The reason for the inconsistency is unclear. The relatively low percentage attributed to hospitalization may reflect the way that costs were calculated in the present study. Specifically, the cost of psychiatric hospitalization was based on actual charges while the cost of antipsychotics was based on Average Wholesale Price discounted by 15% and the cost of other non-hospitalization mental health services was based on Relative Value Units. It is possible, therefore, that the methods of estimating non-hospitalization costs led to their overestimation relative to hospitalization cost. Furthermore, the relative cost contribution of hospitalization and antipsychotic medication to overall total direct costs would be substantially lower in the current study if non-psychiatric medical costs and direct non-health care costs had been available for analysis.</p>", "<title>Study limitations</title>", "<p>The study has important limitations which deserve to be highlighted. First, the criteria for several of the crisis event subgroups (attempted suicide, arrested, violent) were based on a patient-reported measure, not on objective data. It should also be noted that substance abuse is not, strictly speaking, a discrete event, but an ongoing condition. Second, the sample sizes for several of the crisis event subgroups were small, most notably for the \"attempted suicide\" group. The appropriateness of the propensity score adjusted bootstrap re-sampling method with such small sample sizes may be questionable. Third, some hospitalizations may have been underreported since stays in state psychiatric hospitals may not have been reported by all patients, and such hospitalizations would have been missed by most of the MIS systems. Fourth, emergency department visits may have also be significantly underreported, since they would not have been captured by MIS or by medical record in most cases. This might help to explain why the estimate of emergency department costs was negligible in the current study. Fifth, an estimation bias may have occurred in the calculation of medication costs using average wholesale price (AWP; discounted by 15% for atypical antipsychotics). The AWP may not reflect variations in medication costs and greater discount rates across different health care systems. A final limitation is that the estimates reported in this paper are for mental health costs only and do not include non-psychiatric medical costs or direct non-health care costs (e.g., patient involvement in the criminal justice system, use of homeless shelters) – the latter alone have been estimated at $9.3 billion per year [##REF##16199828##2##]. The omission of these cost categories has likely resulted, in the current analysis, in an overestimate of the relative contribution of some cost categories. Future cost studies should include data on direct non-health care costs, especially since the mental health cost burden is being increasingly shifted to the criminal justice system.</p>" ]

[ "<title>Conclusion</title>", "<p>Mental health costs of treating patients diagnosed with schizophrenia and related disorders are highly heterogeneous. Patients who experience crisis events, particularly those with a recent suicide attempt or psychiatric hospitalization, tend to incur the highest annual mental health costs, driven primarily by the cost of psychiatric hospitalization. Patients involved in the criminal justice system (with prior arrest) also accrue relatively high costs within the mental health delivery system. More prospective research, in usual care settings, is needed to identify high risk patients and to determine which interventions are the most cost effective.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Relatively little is known about the relationship between psychosocial crises and treatment costs for persons with schizophrenia. This naturalistic prospective study assessed the association of recent crises with mental health treatment costs among persons receiving treatment for schizophrenia.</p>", "<title>Methods</title>", "<p>Data were drawn from a large multi-site, non-interventional study of schizophrenia patients in the United States, conducted between 1997 and 2003. Participants were treated at mental health treatment systems, including the Department of Veterans Affairs (VA) hospitals, community mental health centers, community and state hospitals, and university health care service systems. Total costs over a 1-year period for mental health services and component costs (psychiatric hospitalizations, antipsychotic medications, other psychotropic medications, day treatment, emergency psychiatric services, psychosocial/rehabilitation group therapy, individual therapy, medication management, and case management) were calculated for 1557 patients with complete medical information. Direct mental health treatment costs for patients who had experienced 1 or more of 5 recent crisis events were compared to propensity-matched samples of persons who had not experienced a crisis event. The 5 non-mutually exclusive crisis event subgroups were: suicide attempt in the past 4 weeks (n = 18), psychiatric hospitalization in the past 6 months (n = 240), arrest in the past 6 months (n = 56), violent behaviors in the past 4 weeks (n = 62), and diagnosis of a co-occurring substance use disorder (n = 413).</p>", "<title>Results</title>", "<p>Across all 5 categories of crisis events, patients who had a recent crisis had higher average annual mental health treatment costs than patients in propensity-score matched comparison samples. Average annual mental health treatment costs were significantly higher for persons who attempted suicide ($46,024), followed by persons with psychiatric hospitalization in the past 6 months ($37,329), persons with prior arrests ($31,081), and persons with violent behaviors ($18,778). Total cost was not significantly higher for those with co-occurring substance use disorder ($19,034).</p>", "<title>Conclusion</title>", "<p>Recent crises, particularly suicide attempts, psychiatric hospitalizations, and criminal arrests, are predictive of higher mental health treatment costs in schizophrenia patients.</p>" ]

[ "<title>Abbreviations</title>", "<p>AWP: average wholesale price; CATIE: Clinical Antipsychotic Trials in Intervention Effectiveness; MADRS: Montgomery-Åsburg Depression Rating Scale; MIS: management information system; PANSS: Positive and Negative Syndrome Scale; RVUs: Relative Value Units; SCAP-HQ: SCAP-Health Questionnaire; TD: tardive dyskinesia; US-SCAP: U.S. Schizophrenia Care and Assessment Program; VA: Veterans Affairs.</p>", "<title>Competing interests</title>", "<p>Baojin Zhu, Haya Ascher-Svanum, Douglas Faries and Xiaomei Peng are full-time employees and minor stockholders of Eli Lilly and Company, the sponsor of the study.</p>", "<p>David Salkever and Eric P. Slade have no competing financial or non-financial interests.</p>", "<title>Authors' contributions</title>", "<p>BZ performed the statistical analyses, participated in the design of the study, the analytical plan, the interpretation of the results, and assisted in drafting the manuscript. HA–S conceived of the study, participated in its design, the analytical plan, the interpretation of the results, and helped draft the manuscript. DEF participated in the design of the study, the analytical plan, the interpretation of the results, and assisted in drafting the manuscript. XP performed additional statistical analyses. DS and EPS participated in the design of the study, the analytical plan, the interpretation of the results, and assisted in drafting the manuscript. DS and EPS were also involved in preparing the resource utilization costing data of US-SCAP.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-244X/8/72/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>Financial support for this research was provided by a grant from Eli Lilly and company. The authors also wish to thank the US-SCAP site investigators and others who collaborated in the research. By site, they include Maryland: A. F. Lehman, M.D., M.S.P.H., University of Maryland School of Medicine, and G. Gallucci, M.D., M.H.S., Johns Hopkins Bayview Medical Center (previously); Colorado: C. Harding, Ph.D., University of Colorado (previously); Florida: D. Shern, Ph.D., Florida Mental Health Institute, University of South Florida (previously), and T. Saunders, M.S., Florida Mental Health Institute (previously); North Carolina: J. Swanson, Ph.D., L. A. Dunn, M.D., and M. Swartz, M.D., Duke University Medical School; California: R. L. Hough, Ph.D., and C. Barrio, Ph.D., Child and Adolescent Services Research Center and San Diego State University; Connecticut: R. A. Rosenheck, M.D., and R. Desai, Ph.D., VA Connecticut Health Care System; Medstat Group: P. Russo, Ph.D., M.S.W., R.N., (previously), L. Palmer, Ph.D., L. Torres, M.B.A., and B. Cuffel, Ph.D. (previously); Eli Lilly and Co.: D. Buesching, Ph.D., Bryan M. Johnstone, Ph.D., and T. Croghan, M.D. (previously); Consultants: D. Salkever, Ph.D., Johns Hopkins University (previously), E. Slade, Ph.D., Johns Hopkins University (previously), and W. Hargreaves, Ph.D., and M. Shumway, Ph.D., University of California, San Francisco. The authors wish to thank Paul Crits-Christoph, PhD, who was compensated by Eli Lilly and Company, for his valuable assistance with the preparation of the manuscript.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Mental health cost components as a proportion of total annual mental health costs</bold>. Of the 1-year per patient total mental health treatment cost of $16,098, the largest single contributor was the cost of hospitalization (29%), followed by antipsychotic medication (23%). Each of the remaining 6 cost component categories contributed less than 10% to the total.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Demographic and clinical characteristics of patients at enrollment</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Characteristic</bold></td><td align=\"center\" colspan=\"2\"><bold>N = 1557</bold></td></tr><tr><td/><td colspan=\"2\"><hr/></td></tr><tr><td/><td align=\"center\"><bold>N</bold></td><td align=\"center\"><bold>%</bold></td></tr></thead><tbody><tr><td align=\"left\">Male</td><td align=\"center\">948</td><td align=\"center\">60.9</td></tr><tr><td align=\"left\">Single, never married</td><td align=\"center\">938</td><td align=\"center\">60.5</td></tr><tr><td align=\"left\">Ethnicity</td><td/><td/></tr><tr><td align=\"left\"> White</td><td align=\"center\">762</td><td align=\"center\">48.9</td></tr><tr><td align=\"left\"> Black</td><td align=\"center\">589</td><td align=\"center\">37.8</td></tr><tr><td align=\"left\"> Other</td><td align=\"center\">206</td><td align=\"center\">13.2</td></tr><tr><td align=\"left\">Health insurance</td><td/><td/></tr><tr><td align=\"left\"> Medicaid/Medicare</td><td align=\"center\">1243</td><td align=\"center\">81.2</td></tr><tr><td align=\"left\"> Department of Veteran Affairs</td><td align=\"center\">97</td><td align=\"center\">6.3</td></tr><tr><td align=\"left\"> Private insurance</td><td align=\"center\">70</td><td align=\"center\">4.6</td></tr><tr><td align=\"left\"> Other coverage</td><td align=\"center\">16</td><td align=\"center\">1.1</td></tr><tr><td align=\"left\"> No health insurance</td><td align=\"center\">104</td><td align=\"center\">6.8</td></tr><tr><td align=\"left\">Educational attainment, high school or less</td><td align=\"center\">1047</td><td align=\"center\">67.9</td></tr><tr><td/><td/><td/></tr><tr><td/><td align=\"center\"><bold>Mean</bold></td><td align=\"center\"><bold>SD</bold></td></tr><tr><td align=\"left\">Age, years</td><td align=\"center\">42.4</td><td align=\"center\">11.1</td></tr><tr><td align=\"left\">Age at illness onset, years</td><td align=\"center\">20.7</td><td align=\"center\">8.9</td></tr><tr><td align=\"left\">MADRS total score</td><td align=\"center\">13.6</td><td align=\"center\">10.2</td></tr><tr><td align=\"left\">PANSS total score</td><td align=\"center\">69.1</td><td align=\"center\">18.4</td></tr><tr><td align=\"left\">Medication days in the 6 months prior to enrollment</td><td/><td/></tr><tr><td align=\"left\"> Atypical antipsychotic</td><td align=\"center\">96.5</td><td align=\"center\">87</td></tr><tr><td align=\"left\"> Typical antipsychotic</td><td align=\"center\">90.1</td><td align=\"center\">87</td></tr><tr><td align=\"left\"> Antidepressant</td><td align=\"center\">59.1</td><td align=\"center\">81.8</td></tr><tr><td align=\"left\"> Mood stabilizer</td><td align=\"center\">46.7</td><td align=\"center\">77.3</td></tr><tr><td align=\"left\"> Other psychotropic</td><td align=\"center\">75.5</td><td align=\"center\">84.9</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Mean 1-year per patient total cost and cost components (n = 1557)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Cost component</bold></td><td align=\"center\"><bold>Mean Cost ($) </bold><break/><bold>(standard deviation)</bold></td></tr></thead><tbody><tr><td align=\"left\">Total annual cost</td><td align=\"center\">16,098 (24,791)</td></tr><tr><td align=\"left\">Medications</td><td align=\"center\">4,817 (3,858)</td></tr><tr><td align=\"left\"> Antipsychotics</td><td align=\"center\">3,770 (3,244)</td></tr><tr><td align=\"left\"> Other psychotropics</td><td align=\"center\">1,047 (1,313)</td></tr><tr><td align=\"left\">Psychiatric hospitalizations</td><td align=\"center\">4,687 (23,536)</td></tr><tr><td align=\"left\">Day Treatment</td><td align=\"center\">1,571 (3,734)</td></tr><tr><td align=\"left\">Emergency Services</td><td align=\"center\">84 (196)</td></tr><tr><td align=\"left\">Psychosocial group therapy</td><td align=\"center\">1,478 (3,126)</td></tr><tr><td align=\"left\">Medication management</td><td align=\"center\">1,187 (1,331)</td></tr><tr><td align=\"left\">Individual therapy</td><td align=\"center\">1,267 (1,826)</td></tr><tr><td align=\"left\">Case management</td><td align=\"center\">1,006 (958)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Characteristics of patients in each crisis event subgroup prior to propensity score matching</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"6\"><bold>Crisis events</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Characteristic</bold></td><td align=\"center\"><bold>Hospitalized in past 6 months</bold></td><td align=\"center\"><bold>Arrested in past 6 months</bold></td><td align=\"center\"><bold>Violent in past 4 weeks</bold></td><td align=\"center\"><bold>Concurrent substance abuse</bold></td><td align=\"center\"><bold>Attempted suicide in past 4 weeks</bold></td><td align=\"center\"><bold>Non-crisis event subgroup</bold></td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\">Sample sizes</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">240</td><td align=\"center\">56</td><td align=\"center\">62</td><td align=\"center\">413</td><td align=\"center\">18</td><td align=\"center\">938</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">1317</td><td align=\"center\">1469</td><td align=\"center\">1463</td><td align=\"center\">1144</td><td align=\"center\">1507</td><td align=\"center\">619</td></tr><tr><td align=\"left\">Age, years</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">39 (10.3)</td><td align=\"center\">38.9 (9.7)</td><td align=\"center\">40.7 (10.0)</td><td align=\"center\">39.9 (9.8)</td><td align=\"center\">33.8 (11.0)</td><td align=\"center\">43.9 (11.4)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">43 (11.2)</td><td align=\"center\">42.6 (11.1)</td><td align=\"center\">42.5 (11.1)</td><td align=\"center\">43.2 (11.4)</td><td align=\"center\">42.6 (11.0)</td><td align=\"center\">40 (10.2)</td></tr><tr><td align=\"left\">Age of Illness onset, yrs</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">19.8 (8.6)</td><td align=\"center\">19.4 (6.3)</td><td align=\"center\">18.9 (7.8)</td><td align=\"center\">19.4 (8.3)</td><td align=\"center\">16.4 (5.8)</td><td align=\"center\">21.4 (9.1)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">20.8 (8.9)</td><td align=\"center\">20.8 (9.0)</td><td align=\"center\">20.8 (9.0)</td><td align=\"center\">21.1 (9.0)</td><td align=\"center\">20.8 (9.0)</td><td align=\"center\">19.6 (8.5)</td></tr><tr><td align=\"left\">MADRS total score</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">15.6 (11.0)</td><td align=\"center\">14.0 (10.5)</td><td align=\"center\">19.9 (12.0)</td><td align=\"center\">13.5 (10.8)</td><td align=\"center\">30.9 (12.5)</td><td align=\"center\">13.1 (9.8)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">13.2 (10.0)</td><td align=\"center\">13.5 (10.1)</td><td align=\"center\">13.3 (10.0)</td><td align=\"center\">13.6 (10.0)</td><td align=\"center\">13.4 (10.0)</td><td align=\"center\">14.3 (10.8)</td></tr><tr><td align=\"left\">PANSS total score</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">73.9 (18.4)</td><td align=\"center\">71.1 (15.7)</td><td align=\"center\">72.0 (18.8)</td><td align=\"center\">69.6 (19.1)</td><td align=\"center\">86.7 (16.9)</td><td align=\"center\">68.0 (17.8)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">68.2 (18.2)</td><td align=\"center\">69.1 (18.5)</td><td align=\"center\">69.0 (18.4)</td><td align=\"center\">68.9 (18.0)</td><td align=\"center\">69.0 (18.3)</td><td align=\"center\">70.8 (19.0)</td></tr><tr><td align=\"left\">Medication days in 6 mos. prior to enrollment</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Atypicals</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">96.2 (82.9)</td><td align=\"center\">61.7 (81.1)</td><td align=\"center\">71.2 (84.4)</td><td align=\"center\">91.0 (87.0)</td><td align=\"center\">89.3 (82.2)</td><td align=\"center\">98.6 (87.5)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">96.6 (87.7)</td><td align=\"center\">97.7 (86.9)</td><td align=\"center\">97.4 (86.9)</td><td align=\"center\">98.5 (86.9)</td><td align=\"center\">96.5 (87.0)</td><td align=\"center\">93.4 (86.2)</td></tr><tr><td align=\"left\"> Typicals</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">72.2 (82.0)</td><td align=\"center\">95.9 (82.9)</td><td align=\"center\">98.4 (88.8)</td><td align=\"center\">98.1 (86.5)</td><td align=\"center\">60.9 (80.8)</td><td align=\"center\">90.3 (87.6)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">93. 4 (87.4)</td><td align=\"center\">89.8 (87.1)</td><td align=\"center\">89.7 (86.9)</td><td align=\"center\">87.2 (86.9)</td><td align=\"center\">90.4 (87.0)</td><td align=\"center\">89.7 (85.8)</td></tr><tr><td align=\"left\"> Antidepressant</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">54.2 (78.9)</td><td align=\"center\">55.1 (81.0)</td><td align=\"center\">71.2 (84.4)</td><td align=\"center\">53.1 (80.0)</td><td align=\"center\">74.7 (86.6)</td><td align=\"center\">61.3 (82.6)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">60(82.3)</td><td align=\"center\">59.6 (82.0)</td><td align=\"center\">58.9 (81.8)</td><td align=\"center\">61.3 (82.4)</td><td align=\"center\">59.2 (81.9)</td><td align=\"center\">55.9 (80.6)</td></tr><tr><td align=\"left\"> Mood stabilizer</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">64.1 (82.0)</td><td align=\"center\">41.3 (74.3)</td><td align=\"center\">46.4 (77.1)</td><td align=\"center\">38.7 (71.7)</td><td align=\"center\">61.0 (84.1)</td><td align=\"center\">45.5 (77.3)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">43.5 (76.0)</td><td align=\"center\">47.0 (77.5)</td><td align=\"center\">46.8 (77.4)</td><td align=\"center\">49.6 (79.1)</td><td align=\"center\">46.6 (77.3)</td><td align=\"center\">48.6 (77.4)</td></tr><tr><td align=\"left\"> Other psychotropic</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">71.0 (80.5)</td><td align=\"center\">63.8 (79.9)</td><td align=\"center\">99.1 (82.5)</td><td align=\"center\">79.3 (85.4)</td><td align=\"center\">36.1 (63.3)</td><td align=\"center\">74.1 (85.5)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">76.4 (85.7)</td><td align=\"center\">75.7 (85.0)</td><td align=\"center\">74.3 (84.8)</td><td align=\"center\">74.2 (84.8)</td><td align=\"center\">75.7 (85.0)</td><td align=\"center\">77.7 (84.2)</td></tr><tr><td align=\"left\">Male, N (%)</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">142 (59.2)</td><td align=\"center\">45 (80.4)</td><td align=\"center\">38 (61.3)</td><td align=\"center\">304 (73.6)</td><td align=\"center\">7 (38.9)</td><td align=\"center\">527 (56.2)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">806 (61.2)</td><td align=\"center\">882 (60.0)</td><td align=\"center\">889 (60.8)</td><td align=\"center\">644 (56.3)</td><td align=\"center\">920 (61.0)</td><td align=\"center\">421 (68.0)</td></tr><tr><td align=\"left\">Single, N (%)</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">149 (62.3)</td><td align=\"center\">46 (82.0)</td><td align=\"center\">26 (41.9)</td><td align=\"center\">262 (63.8)</td><td align=\"center\">11 (61.1)</td><td align=\"center\">549 (58.8)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">789 (60.2)</td><td align=\"center\">871 (59.6)</td><td align=\"center\">891 (61.2)</td><td align=\"center\">676 (59.4)</td><td align=\"center\">906 (60.4)</td><td align=\"center\">389 (63.2)</td></tr><tr><td align=\"left\">Diagnosis, N (%)</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Schizophrenia</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">119 (49.6)</td><td align=\"center\">33 (58.9)</td><td align=\"center\">32 (51.6)</td><td align=\"center\">257 (62.2)</td><td align=\"center\">9 (50.0)</td><td align=\"center\">620 (66.1)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">861 (65.4)</td><td align=\"center\">926 (63.0)</td><td align=\"center\">927 (63.4)</td><td align=\"center\">723 (63.2)</td><td align=\"center\">950 (63.0)</td><td align=\"center\">360 (58.2)</td></tr><tr><td align=\"left\"> Schizoaffective</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">103 (42.9)</td><td align=\"center\">17 (30.4)</td><td align=\"center\">28 (45.2)</td><td align=\"center\">125 (30.3)</td><td align=\"center\">9 (50.0)</td><td align=\"center\">282 (30.1)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">393 (29.8)</td><td align=\"center\">469 (31.9)</td><td align=\"center\">458 (31.3)</td><td align=\"center\">371 (32.4)</td><td align=\"center\">477 (31.7)</td><td align=\"center\">214 (34.6)</td></tr><tr><td align=\"left\"> Other psychotic</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\">  Yes</td><td align=\"center\">19 (7.9)</td><td align=\"center\">6 (10.7)</td><td align=\"center\">2 (3.2)</td><td align=\"center\">31 (7.5)</td><td align=\"center\">0</td><td align=\"center\">38 (4.1)</td></tr><tr><td align=\"left\">  No</td><td align=\"center\">65 (4.9)</td><td align=\"center\">77 (5.2)</td><td align=\"center\">81 (5.5)</td><td align=\"center\">53 (4.6)</td><td align=\"center\">83 (5.5)</td><td align=\"center\">46 (7.4)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Proportion of participants in each crisis event category and degree of overlap between categories</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Hospitalized in prior 6 months</bold></td><td align=\"center\"><bold>Arrested in previous 6 months</bold></td><td align=\"center\"><bold>Violent behavior in previous 4 weeks</bold></td><td align=\"center\"><bold>Concurrent substance abuse diagnosis</bold></td><td align=\"center\"><bold>Attempted suicide in past 4 weeks</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Hospitalized in prior 6 months </bold><break/><bold>(N = 240)</bold></td><td align=\"center\">__</td><td align=\"center\">22 (9.2%)</td><td align=\"center\">17 (7.1%)</td><td align=\"center\">87 (36.3%)</td><td align=\"center\">11 (4.6%)</td></tr><tr><td align=\"left\"><bold>Arrested in previous 6 months </bold><break/><bold>(N = 56)</bold></td><td align=\"center\">22 (39.3%)</td><td align=\"center\">__</td><td align=\"center\">5 (8.9%)</td><td align=\"center\">26 (46.4%)</td><td align=\"center\">1 (1.8%)</td></tr><tr><td align=\"left\"><bold>Violent behavior in previous 4 weeks </bold><break/><bold>(N = 62)</bold></td><td align=\"center\">17 (27.4%)</td><td align=\"center\">5 (8.1%)</td><td align=\"center\">__</td><td align=\"center\">26 (41.9%)</td><td align=\"center\">2 (3.2%)</td></tr><tr><td align=\"left\"><bold>Concurrent substance abuse diagnosis </bold><break/><bold>(N = 413)</bold></td><td align=\"center\">87 (21.1%)</td><td align=\"center\">26 (6.6%)</td><td align=\"center\">26 (6.6%)</td><td align=\"center\">__</td><td align=\"center\">8 (2.0%)</td></tr><tr><td align=\"left\"><bold>Attempted suicide in past 4 weeks </bold><break/><bold>(N = 18)</bold></td><td align=\"center\">11 (61.1%)</td><td align=\"center\">1 (5.6%)</td><td align=\"center\">2 (11.1%)</td><td align=\"center\">8 (44.4%)</td><td align=\"center\">__</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Mean 1-year total costs, hospitalization costs and hospitalization parameters for patients with and without specific crisis event categories</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\"><bold>Patients with event</bold></td><td align=\"center\"><bold>Mean 1-year total mental health cost per patient</bold></td><td align=\"center\"><bold>Mean 1-year cost of psych hospitalization per patient</bold></td><td align=\"center\"><bold>Mean number of days hospitalized</bold></td><td align=\"center\"><bold>Mean number of hospital admissions</bold></td></tr><tr><td/><td align=\"center\"><bold>N</bold></td><td align=\"center\"><bold>%</bold></td><td/><td/><td/><td/></tr></thead><tbody><tr><td align=\"left\">Hospitalization in previous 6 months</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes</td><td align=\"center\">240</td><td align=\"center\">15.4</td><td align=\"center\">$37,329 **</td><td align=\"center\">$23,962 **</td><td align=\"center\">73.3</td><td align=\"center\">1.9</td></tr><tr><td align=\"left\"> No</td><td align=\"center\">1317</td><td align=\"center\">84.6</td><td align=\"center\">$12,229</td><td align=\"center\">$1,175</td><td align=\"center\">4.8</td><td align=\"center\">0.3</td></tr><tr><td align=\"left\">Arrested in previous 6 months</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes</td><td align=\"center\">56</td><td align=\"center\">3.7</td><td align=\"center\">$31,081 *</td><td align=\"center\">$20,334 *</td><td align=\"center\">55.8</td><td align=\"center\">1.1</td></tr><tr><td align=\"left\"> No</td><td align=\"center\">1469</td><td align=\"center\">96.3</td><td align=\"center\">$15,654</td><td align=\"center\">$4,180</td><td align=\"center\">14.1</td><td align=\"center\">0.5</td></tr><tr><td align=\"left\">Violent behavior in previous 4 weeks</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes</td><td align=\"center\">62</td><td align=\"center\">4.1</td><td align=\"center\">$18,778 **</td><td align=\"center\">$7,416 **</td><td align=\"center\">29.6</td><td align=\"center\">1.3</td></tr><tr><td align=\"left\"> No</td><td align=\"center\">1463</td><td align=\"center\">95.9</td><td align=\"center\">$16,113</td><td align=\"center\">$4,661</td><td align=\"center\">15.1</td><td align=\"center\">0.5</td></tr><tr><td align=\"left\">Concurrent substance abuse disorder</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes</td><td align=\"center\">413</td><td align=\"center\">26.5</td><td align=\"center\">$19,034</td><td align=\"center\">$7,455</td><td align=\"center\">25.5</td><td align=\"center\">0.7</td></tr><tr><td align=\"left\"> No</td><td align=\"center\">1144</td><td align=\"center\">73.5</td><td align=\"center\">$15,038</td><td align=\"center\">$3,688</td><td align=\"center\">11.7</td><td align=\"center\">0.5</td></tr><tr><td align=\"left\">Attempted suicide in past 4 weeks</td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Yes</td><td align=\"center\">18</td><td align=\"center\">1.2</td><td align=\"center\">$46,024 **</td><td align=\"center\">$30,080 **</td><td align=\"center\">61.8</td><td align=\"center\">1.3</td></tr><tr><td align=\"left\"> No</td><td align=\"center\">1507</td><td align=\"center\">98.8</td><td align=\"center\">$15,865</td><td align=\"center\">$4,471</td><td align=\"center\">15.1</td><td align=\"center\">0.5</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T6\"><label>Table 6</label><caption><p>Mean 1-year costs for patients with and without specific number of crisis event categories</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"4\"><bold>Number of types of crisis events</bold></td></tr></thead><tbody><tr><td/><td align=\"center\"><bold>None</bold></td><td align=\"center\"><bold>One only</bold></td><td align=\"center\"><bold>Two or more</bold></td><td align=\"center\"><bold>Three or more</bold></td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\">Patients with event, N (%)</td><td align=\"center\">938 (60.2%)</td><td align=\"center\">481 (30.9%)</td><td align=\"center\">138 (8.9%)</td><td align=\"center\">29 (1.9%)</td></tr><tr><td align=\"left\">Mean 1-year cost of hospitalization per patient</td><td align=\"center\">$830</td><td align=\"center\">$6,912 *</td><td align=\"center\">$23,149 *</td><td align=\"center\">$33,199 *</td></tr><tr><td align=\"left\">Mean 1-year total mental health cost per patient</td><td align=\"center\">$11,739</td><td align=\"center\">$19,066 *</td><td align=\"center\">$35,385 *</td><td align=\"center\">$44,599 *</td></tr><tr><td align=\"left\">Mean number of days hospitalized</td><td align=\"center\">3.7</td><td align=\"center\">21.8</td><td align=\"center\">72.7</td><td align=\"center\">101.1</td></tr><tr><td align=\"left\">Mean number of hospital admissions</td><td align=\"center\">0.2</td><td align=\"center\">0.8</td><td align=\"center\">1.7</td><td align=\"center\">2.7</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>PANSS, Positive and Negative Symptoms of Schizophrenia scale</p><p>MADRS, Montgomery-Åsberg Depression Rating Scale</p></table-wrap-foot>", "<table-wrap-foot><p>Means (standard deviations) unless otherwise specified. Patients may belong to more than one subgroup. Yes = with event; No = without event.</p><p>PANSS, Positive and Negative Symptoms of Schizophrenia scale</p><p>MADRS, Montgomery-Åsberg Depression Rating Scale</p><p>Atypicals = atypical antipsychotics</p><p>Typicals = typical antipsychotics</p></table-wrap-foot>", "<table-wrap-foot><p>* P &lt; .05; ** P &lt; .01; P values compare non-crisis event group cost (\"no\") vs. crisis event group cost (\"yes\") for each event.</p></table-wrap-foot>", "<table-wrap-foot><p>* P &lt; .01; P values based on non-crisis event group vs. each category: patients with only one type of crisis event, two or more types of events, and three of more types of crisis events.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-244X-8-72-1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Lesbian, gay and bisexual (LGB) people appear to be at greater risk than heterosexual people of mental disorders and suicidal behaviour [##REF##14645028##1##,##REF##12956539##2##]. LGB people are subject to institutionalised prejudice, social stress, social exclusion (even within families) and anti-homosexual hatred and violence and often internalise a sense of shame about their sexuality [##REF##14645028##1##,##REF##12956539##2##]. Lifestyle factors such as alcohol and drugs misuse also increase the risk of morbidity [##REF##14645028##1##] as well as suicide attempts [##UREF##0##3##]. Deliberate self harm (DSH) is intentional self poisoning or injury, irrespective of the apparent purpose of the act. DSH is one of the leading causes of acute medical admissions in the UK. Incidence rose steadily from the mid 1980s to the late 1990s with a peak incidence rate of 400 per 100,000 per annum [##UREF##1##4##], one of the highest in Europe. However there is evidence of a steady drop in suicide in England (and other developed countries) since 2000 [##UREF##2##5##]. The evidence on mental health of LGB people is inconclusive partly because of the difficulty of defining or recruiting samples that are representative of all non-heterosexual people. Specific methodological obstacles include variation in the definition of sexual orientation, DSH and mental illness; difficulty in achieving random samples; reliance on participants' recall; unwillingness of people to be open about their sexual orientation; lack of information on sexuality in suicide victims who are part of psychological post mortem studies; the complexity of choosing appropriate comparison groups and poor or absent adjustment for confounding influences such as substance use and personality factors.</p>", "<p>There is an urgent need to quantify the risk for mental disorder, DSH and suicide in LGB people, to understand the precipitants and to examine the efficacy of prevention efforts. There is also a need to make a judgement about the quality of the evidence available. We undertook a systematic review of the world literature on risk of mental disorder, substance misuse, DSH, suicidal ideation and suicide in LGB people. These parameters are the main ones reported in the literature and provide a comprehensive picture of mental health and well being.</p>", "<title>Hypothesis</title>", "<p>Gay, lesbian and bisexual people have higher risks than heterosexual people of mental disorder, substance misuse and dependence, suicide, suicidal ideation and DSH.</p>", "<title>Objective</title>", "<p>To undertake a systematic review of the international research literature to establish whether LGB people are at higher risk of mental disorder, substance misuse, suicide, suicidal ideation and DSH than heterosexual people and to quantify this risk.</p>" ]

[ "<title>Method</title>", "<p>We searched for studies of mental disorder, drug and alcohol misuse and dependence, DSH, suicidal ideation and/or suicide in general (community) or selected (e.g. student) populations in which sexual orientation was reported. We use the following abbreviations: GB (gay and bisexual men); LB (lesbians and bisexual women) and LGB (lesbians, gay men and bisexual men and women).</p>", "<title>Data sources</title>", "<p>We searched Medline, Embase, PsycInfo, Cinahl, the Cochrane Library Database, the Web of Knowledge, the Applied Social Sciences Index and Abstracts, the International Bibliography of the Social Sciences, Sociological Abstracts, the Campbell Collaboration and grey literature databases for articles published between January 1966 and April 2005. We searched all terms related to homosexual, lesbian and bisexual people and all terms related to mental disorders, suicide, and deliberate self harm. No language limits were imposed. A full internet search was also carried out using Google and Google Scholar and authors were contacted where necessary. We also searched the reference lists of relevant papers.</p>", "<title>Study selection</title>", "<title>Eligibility</title>", "<p>We included papers that provided valid definition of sexual orientation and mental health outcomes. Random sampling is hampered by participants' reluctance to disclose their sexual orientation and the small numbers of LGB people recruited. Thus other methods such as snowball sampling (initial LGB participants recruit other LGB people in successive waves) were regarded as acceptable if the study met other inclusion criteria. We included studies in which people defined themselves as: gay, lesbian, homosexual, bisexual and/or in which they reported levels of same sex attraction or behaviour. We excluded studies based in clinical or psychological services. We only included studies in which there was a concurrent heterosexual comparison group within either a cohort, case-control or cross sectional study. Outcomes were defined as: a) a psychiatric disorder according to the International Classification of Diseases or the American Psychiatric Association's Diagnostic and Statistical Manual (including substance misuse disorders); b) scores or a recognised threshold for psychiatric morbidity on standardised scales (including alcohol or drug dependence); c) alcohol misuse: consumption above UK Government recommended maximum weekly limits (21 units men, 14 units women); d) suicide (the intentional taking one's own life) e) suicidal ideation (i.e. thoughts of taking one's life without acting on them); f) DSH: intentional self poisoning or injury irrespective of the apparent purpose of the act [##UREF##2##5##]. These outcomes were extracted for both the LGB and heterosexual comparison groups as cumulative incidence rates in prospective cohort studies or period prevalence rates in cross sectional studies.</p>", "<title>Screening process and assessment of eligibility</title>", "<p>The titles and abstracts of citations were screened by JS and DP and those not meeting eligibility criteria, unpublished dissertation theses, case reports, letters, commentaries, or review papers were excluded. Decisions on papers included in the final review were made by pairs of authors and disagreements discussed at steering group meetings involving all authors.</p>", "<title>Data extraction</title>", "<p>At least two of the authors extracted data from each paper on study setting, study design, population and sampling details, attrition and response rate. We recorded the definition of LGB sexual orientation (same sex attraction; same sex behaviour; self identification as lesbian gay or bisexual; a score above zero on the Kinsey scale [##UREF##3##6##]) and outcome (mental disorder, substance misuse, DSH, suicidal ideation and suicide). Where appropriate we extracted prevalence estimates and/or odds ratios; for continuous data we extracted means and standard deviations. In instances of disagreement, each case was discussed by all authors.</p>", "<title>Quality of studies reviewed</title>", "<p>We used the Cochrane Handbook's general guidance on non-experimental studies to inform our choice of quality indicators (2 indicating higher quality than 1). We examined for: <italic>sampling</italic>: non random = 1, random = 2; <italic>representativeness</italic>: response rates: &lt;60% = 1, 60% or more = 2; <italic>population definition</italic>: selected sample (e.g. school students) = 1; general population = 2 and <italic>sample size</italic>: &lt;100 LB or GB people = 1, &gt;100 LB or GB people = 2.</p>", "<title>Data synthesis</title>", "<p>Studies were grouped according to lifetime or 12 month prevalence and where possible we analysed outcomes for lesbians, gay men and bisexual people separately and collectively. We calculated risk ratios and attributable risks (differences between rates in LGB and non LGB people) from extracted prevalence data. We examined suicide attempts when reported instead of or in addition to DSH. For continuous outcomes we calculated the effect size as standardised mean difference in scores between LGB people and controls.</p>", "<title>Meta-analytic approach</title>", "<p>We adopted standard methods for conducting meta-analyses where there were two or more studies with useable outcome data. We used a random effects model which used inverse variance methods to calculate the pooled effect estimate in which the weight given to each study is the inverse of the variance of the study estimate together with the common heterogeneity variance. We quantified the effect of heterogeneity [##REF##9462324##7##] by using <italic>I</italic>²which describes the percentage of total variation across studies that can be attributed to heterogeneity rather than chance [##REF##12958120##8##].</p>" ]

[ "<title>Results</title>", "<p>From 13706 citations identified, 476 papers were retrieved of which 429 were excluded (figure ##FIG##0##1##). Eighty-three of those excluded were controlled studies [##REF##11920991##36##, ####REF##11777125##37##, ##REF##11197924##38##, ##REF##9565422##39##, ##REF##10323629##40##, ##REF##15322415##41##, ##REF##10745751##42##, ##REF##8261890##43##, ##REF##12810141##44##, ##REF##15607842##45##, ##REF##15231053##46##, ##REF##12639368##47##, ##REF##15451703##48##, ##UREF##6##49##, ##REF##14499515##50##, ##REF##14651501##51##, ##REF##8932550##52##, ##REF##2051150##53##, ##UREF##7##54##, ##REF##12964921##55##, ##UREF##8##56##, ##REF##10087691##57##, ##REF##11244906##58##, ##REF##11262511##59##, ##REF##12949929##60##, ##REF##8897107##61##, ##REF##9850116##62##, ##UREF##9##63##, ##REF##11818291##64##, ##REF##12084692##65##, ##UREF##10##66##, ##REF##12243487##67##, ##REF##8201061##68##, ##REF##2734426##69##, ##REF##10641358##70##, ##REF##3266145##71##, ##UREF##11##72##, ##REF##11404525##73##, ##REF##15000974##74##, ##REF##11392949##75##, ##UREF##12##76##, ##REF##15086222##77##, ##REF##15774420##78##, ##REF##9247400##79##, ##REF##1190349##80##, ##REF##8387422##81##, ##REF##12222848##82##, ##UREF##13##83##, ##UREF##14##84##, ##UREF##15##85##, ##REF##5358533##86##, ##REF##14582576##87##, ##REF##1772888##88##, ##REF##11845647##89##, ##REF##7085588##90##, ##REF##9259045##91##, ##REF##3966852##92##, ##UREF##16##93##, ##REF##11429303##94##, ##UREF##17##95##, ##UREF##18##96##, ##UREF##19##97##, ##REF##12144160##98##, ##REF##10596508##99##, ##REF##5411361##100##, ##REF##5473144##101##, ##UREF##20##102##, ##REF##6481605##103##, ##REF##4668541##104##, ##REF##10402685##105##, ##UREF##21##106##, ##UREF##22##107##, ##REF##2681278##108##, ##UREF##23##109##, ##UREF##24##110##, ##REF##15765273##111##, ##UREF##25##112##, ##REF##3591995##113##, ##REF##15311980##114##, ##REF##9672522##115##, ##UREF##26##116##, ##REF##7825557##117##, ##REF##3798521##118##, ##REF##8553430##119##, ##REF##10530627##120####10530627##120##], 122–123]; two [##REF##9565422##39##,##REF##10323629##40##] were excluded because the data were repeated elsewhere [##REF##11929369##29##]; seven did not meet sampling criteria [##REF##11920991##36##, ####REF##11777125##37##, ##REF##11197924##38####11197924##38##,##REF##11392949##75##,##UREF##12##76##,##REF##3798521##118##,##REF##8553430##119##]; 34 did not report suicide, DSH or diagnostic outcomes [##REF##15322415##41##, ####REF##10745751##42##, ##REF##8261890##43##, ##REF##12810141##44##, ##REF##15607842##45##, ##REF##15231053##46##, ##REF##12639368##47##, ##REF##15451703##48##, ##UREF##6##49##, ##REF##14499515##50##, ##REF##14651501##51##, ##REF##8932550##52##, ##REF##2051150##53##, ##UREF##7##54##, ##REF##12964921##55##, ##UREF##8##56##, ##REF##10087691##57##, ##REF##11244906##58##, ##REF##11262511##59##, ##REF##12949929##60##, ##REF##8897107##61##, ##REF##9850116##62##, ##UREF##9##63##, ##REF##11818291##64##, ##REF##12084692##65##, ##UREF##10##66##, ##REF##12243487##67##, ##REF##8201061##68##, ##REF##2734426##69##, ##REF##10641358##70##, ##REF##3266145##71##, ##UREF##11##72##, ##REF##11404525##73##, ##REF##15000974##74####15000974##74##]; 37 involved unrepresentative populations [##REF##15086222##77##, ####REF##15774420##78##, ##REF##9247400##79##, ##REF##1190349##80##, ##REF##8387422##81##, ##REF##12222848##82##, ##UREF##13##83##, ##UREF##14##84##, ##UREF##15##85##, ##REF##5358533##86##, ##REF##14582576##87##, ##REF##1772888##88##, ##REF##11845647##89##, ##REF##7085588##90##, ##REF##9259045##91##, ##REF##3966852##92##, ##UREF##16##93##, ##REF##11429303##94##, ##UREF##17##95##, ##UREF##18##96##, ##UREF##19##97##, ##REF##12144160##98##, ##REF##10596508##99##, ##REF##5411361##100##, ##REF##5473144##101##, ##UREF##20##102##, ##REF##6481605##103##, ##REF##4668541##104##, ##REF##10402685##105##, ##UREF##21##106##, ##UREF##22##107##, ##REF##2681278##108##, ##UREF##23##109##, ##UREF##24##110##, ##REF##15765273##111##, ##UREF##25##112##, ##REF##3591995##113####3591995##113##] and three on closer inspection did not concern LGB people [##REF##15311980##114##, ####REF##9672522##115##, ##UREF##26##116####26##116##]. There were insufficient data in three studies on completed suicide to include it as an outcome in the review. One that involved suicide in a cohort of bisexual and gay men was excluded because it was uncontrolled [##REF##7825557##117##]; one study comparing clinical characteristics of a subpopulation of gay and non-gay male suicides was excluded because of sampling concerns [##REF##3798521##118##] and a psychological autopsy study carried out in 1995 [##REF##8553430##119##] was excluded as it contained only three gay male suicides.</p>", "<title>Study characteristics</title>", "<p>Twenty-eight papers [##REF##14645028##1##,##REF##9141776##9##, ####REF##11996785##10##, ##REF##15665660##11##, ##REF##10707921##12##, ##REF##15265096##13##, ##REF##10754972##14##, ##REF##12602425##15##, ##REF##11684618##16##, ##REF##9491018##17##, ##REF##10530626##18##, ##REF##11392937##19##, ##REF##11392944##20##, ##REF##10530625##21##, ##REF##11983639##22##, ##UREF##4##23##, ##REF##12084697##24##, ##REF##12766375##25##, ##REF##15369203##26##, ##REF##11026127##27##, ##REF##9584034##28##, ##REF##11929369##29##, ##REF##11499118##30##, ##REF##11146762##31##, ##UREF##5##32##, ##REF##12611836##33##, ##REF##12653422##34##, ##REF##15830911##35####15830911##35##] reporting on 25 studies [##REF##14645028##1##,##REF##9141776##9##, ####REF##11996785##10##, ##REF##15665660##11##, ##REF##10707921##12####10707921##12##,##REF##10754972##14##,##REF##12602425##15##,##REF##9491018##17##, ####REF##10530626##18##, ##REF##11392937##19##, ##REF##11392944##20##, ##REF##10530625##21##, ##REF##11983639##22##, ##UREF##4##23##, ##REF##12084697##24##, ##REF##12766375##25##, ##REF##15369203##26##, ##REF##11026127##27##, ##REF##9584034##28##, ##REF##11929369##29##, ##REF##11499118##30##, ##REF##11146762##31####11146762##31##,##REF##12611836##33##, ####REF##12653422##34##, ##REF##15830911##35####15830911##35##] met our inclusion criteria (Additional file ##SUPPL##0##1##); six papers [##REF##10707921##12##,##REF##15265096##13##,##REF##12602425##15##,##REF##11684618##16##,##REF##11146762##31##,##UREF##5##32##] reported data on three studies. Five studies could not be included in a meta-analysis because the data were not extractable or in a format that allowed comparison [##REF##11996785##10##,##REF##15665660##11##,##REF##11983639##22##,##REF##11026127##27##,##REF##12653422##34##]. Three of the four longitudinal cohorts [##REF##15665660##11##,##REF##10530626##18##,##REF##12611836##33##] presented nested cross-sectional data on sexual orientation and mental health at one time point. One cohort study, however, conducted a longitudinal analysis of cumulative incidence of suicidal attempts but did not provide extractable data [##REF##12653422##34##]. No case-control studies were identified. The studies were conducted in seven countries in North America, Europe and Australasia, with most based in the USA (17/25, 67%). The papers were published between 1997 and 2004, with two thirds published between 2000 and 2003. Participation rates ranged from 25% [##UREF##4##23##] to 95% [##REF##9584034##28##].</p>", "<title>Population</title>", "<p>The papers contained data on 214,344 heterosexual and 11,971 non heterosexual people aged 12 and over. Four studies involved people aged under 18 [##REF##11996785##10##,##REF##9491018##17##,##REF##11929369##29##,##REF##11499118##30##] and 18 involved people under 25 years. Four studies included only women [##REF##15665660##11##,##REF##11392944##20##,##REF##12084697##24##,##REF##15369203##26##], three only men [##REF##9141776##9##,##REF##10754972##14##,##REF##10530625##21##] and 18 both sexes. Eight studies [##REF##11996785##10##,##REF##9491018##17##,##REF##10530625##21##,##REF##12766375##25##,##REF##15369203##26##,##REF##9584034##28##, ####REF##11929369##29##, ##REF##11499118##30####11499118##30##,##REF##12653422##34##] concerned high school and college students. Of the 21 cross sectional studies, nine used random sampling [##REF##9141776##9##,##REF##12602425##15##,##REF##11392937##19##,##REF##11392944##20##,##REF##11983639##22##,##REF##12766375##25##,##REF##15369203##26##,##REF##11146762##31##,##REF##15830911##35##]; two multi-stage sampling [##REF##10707921##12##,##REF##10754972##14##]; two snowball sampling [##REF##14645028##1##,##REF##12084697##24##]; one systematic sampling (i.e. 26 years follow up data on a birth cohort) [##UREF##4##23##]; and seven did not specify their sampling method [##REF##11996785##10##,##REF##9491018##17##,##REF##10530625##21##,##REF##11026127##27##, ####REF##9584034##28##, ##REF##11929369##29##, ##REF##11499118##30####11499118##30##].</p>", "<title>Definition of sexuality</title>", "<p>Sexuality was defined in a number of ways even within the same study: four studies used same sex attraction [##REF##12084697##24##,##REF##11499118##30##,##REF##12611836##33##,##REF##12653422##34##]; 13 used same sex behaviour [##REF##9141776##9##,##REF##11996785##10##,##REF##10707921##12##,##REF##10754972##14##,##REF##9491018##17##, ####REF##10530626##18##, ##REF##11392937##19####11392937##19##,##REF##10530625##21##,##REF##12084697##24##,##REF##11929369##29##, ####REF##11499118##30##, ##REF##11146762##31####11146762##31##,##REF##12653422##34##,##REF##15830911##35##]; 15 used participant self identification [##REF##14645028##1##,##REF##9141776##9##, ####REF##11996785##10##, ##REF##15665660##11####15665660##11##,##REF##12602425##15##,##REF##10530626##18##,##REF##11392944##20##,##REF##11983639##22##,##UREF##4##23##,##REF##12766375##25##, ####REF##15369203##26##, ##REF##11026127##27##, ##REF##9584034##28##, ##REF##11929369##29####11929369##29##]; and three used a score above zero on the Kinsey scale [##REF##14645028##1##,##REF##9584034##28##,##REF##12653422##34##] (see Additional file ##SUPPL##0##1##). Nine studies used two definitions of sexual orientation [##REF##14645028##1##,##REF##9141776##9##,##REF##11996785##10##,##REF##10530626##18##,##REF##12084697##24##,##REF##9584034##28##, ####REF##11929369##29##, ##REF##11499118##30####11499118##30##,##REF##15830911##35##] and one used three definitions [##REF##12653422##34##]. Self-identified sexuality was based on the categories heterosexual, homosexual or bisexual [##REF##9141776##9##,##REF##12602425##15##,##REF##10530626##18##,##REF##11392944##20##,##REF##11983639##22##,##UREF##4##23##,##REF##9584034##28##] or included the choices gay or lesbian [##REF##14645028##1##,##REF##11996785##10##,##REF##15665660##11##,##REF##12766375##25##, ####REF##15369203##26##, ##REF##11026127##27####11026127##27##,##REF##11929369##29##]. Eighteen studies used a specific time frame to assess sexuality. Lifetime same sex attraction was assessed in two studies [##REF##11499118##30##,##REF##12611836##33##]; current same sex attraction assessed in four [##REF##12084697##24##,##REF##12611836##33##, ####REF##12653422##34##, ##REF##15830911##35####15830911##35##] and in one study both were assessed [##REF##12611836##33##]. Same sex behaviour was assessed as occurring 'in the last year' in two studies [##REF##10707921##12##,##REF##12084697##24##], 'in the last five years' in one study [##REF##11392937##19##] or 'ever' in nine studies [##REF##9141776##9##,##REF##11996785##10##,##REF##10754972##14##,##REF##9491018##17##,##REF##10530626##18##,##REF##10530625##21##,##REF##11929369##29##,##REF##11499118##30##,##REF##12653422##34##].</p>", "<title>Outcomes of interest</title>", "<p>Fifteen studies assessed suicide attempts or DSH [##REF##14645028##1##,##REF##9141776##9##,##REF##11996785##10##,##REF##10754972##14##,##REF##9491018##17##, ####REF##10530626##18##, ##REF##11392937##19####11392937##19##,##REF##10530625##21##,##UREF##4##23##,##REF##12084697##24##,##REF##9584034##28##, ####REF##11929369##29##, ##REF##11499118##30####11499118##30##,##REF##12611836##33##,##REF##12653422##34##] and 12 assessed suicidal ideation [##REF##10754972##14##,##REF##9491018##17##, ####REF##10530626##18##, ##REF##11392937##19####11392937##19##,##REF##10530625##21##, ####REF##11983639##22##, ##UREF##4##23##, ##REF##12084697##24####12084697##24##,##REF##15369203##26##,##REF##9584034##28##,##REF##11499118##30##,##REF##12611836##33##]. Data on mental disorder were assessed in 10 studies [##REF##14645028##1##,##REF##9141776##9##,##REF##15665660##11##,##REF##10707921##12##,##REF##10754972##14##,##REF##12602425##15##,##REF##10530626##18##,##REF##11392937##19##,##REF##11983639##22##,##REF##11146762##31##], substance dependence in six studies [##REF##10707921##12##,##REF##12602425##15##,##REF##10530626##18##,##REF##11392937##19##,##REF##11146762##31##,##REF##15830911##35##] and substance misuse in nine studies [##REF##14645028##1##,##REF##11392937##19##,##REF##11392944##20##,##REF##11983639##22##,##REF##12766375##25##, ####REF##15369203##26##, ##REF##11026127##27####11026127##27##,##REF##11146762##31##,##REF##15830911##35##]. Eighteen studies assessed more than one of these outcomes [##REF##14645028##1##,##REF##9141776##9##,##REF##10707921##12##,##REF##10754972##14##,##REF##12602425##15##,##REF##9491018##17##, ####REF##10530626##18##, ##REF##11392937##19####11392937##19##,##REF##10530625##21##, ####REF##11983639##22##, ##UREF##4##23##, ##REF##12084697##24####12084697##24##,##REF##15369203##26##,##REF##9584034##28##,##REF##11499118##30##,##REF##11146762##31##,##REF##12611836##33##,##REF##15830911##35##] and one study assessed all [##REF##11392937##19##]. Risk ratios and attributable risks were calculated for all outcomes of interest (figures ##FIG##1##2##, ##FIG##2##3##, ##FIG##3##4##, ##FIG##4##5##, ##FIG##5##6##, ##FIG##6##7##, ##FIG##7##8##, ##FIG##8##9##).</p>", "<title>Quality of cross sectional studies</title>", "<p>Nine studies were based on random populations but only seven of these were sampled from the community rather than from specific groups (e.g. schools). Only four of these reported responses of at least 60% and of these only one [##REF##11392937##19##] sampled 100 or more LGB people (table ##TAB##0##1##).</p>", "<title>Data syntheses</title>", "<title>Suicide attempts and DSH</title>", "<p>Only one cohort study [##REF##12653422##34##] reported cumulative incidence of suicide attempts over two years in 2924 Norwegian school youths. They reported an odds ratio of 4.69 (95% CI 2.29, 10.62) for LB girls after adjustment but no significant differential for BG boys.</p>", "<p>Meta-analyses of cross-sectional studies of lifetime suicide attempts demonstrated increased risk in all groups when compared to heterosexuals but there was substantial heterogeneity when these data were combined for both sexes and for men only (Figure ##FIG##1##2##). Attributable risk ranged from 0.03 to 0.25 and was higher in men than women. Studies in this analysis were limited by small samples [##REF##9141776##9##,##REF##10754972##14##,##REF##10530626##18##,##REF##12611836##33##] or selection bias [##REF##10530626##18##,##UREF##4##23##,##REF##9584034##28##,##REF##12611836##33##] (Table ##TAB##0##1##). One small study that met all but one quality criteria showed a high risk of suicide attempts in men (Figure ##FIG##1##2##) [##REF##9141776##9##]. Meta-analysis in women demonstrated 1.82 times increased risk of lifetime suicide attempts in lesbians and bisexuals compared to controls and showed little heterogeneity (Figure ##FIG##1##2##). However, all the studies failed to meet several of our quality indicators.</p>", "<p>Risk ratios for 12 month prevalence of suicide attempts ranged from 1.96 to 2.76 (men 2.23 to 2.53; women 1.94 to 2.46), while attributable risk ranged from 0.01 to 0.14 (men 0.01 to 0.03; women 0.01 to 0.07). The pooled estimate for men and women was 2.56 (Figure ##FIG##1##2##) with similar values for LB and GB people and all showed little or no heterogeneity. The highest quality study [##REF##11392937##19##], however, showed a non significant risk ratio for all groups.</p>", "<p>Only two studies reported lifetime prevalence of DSH [##REF##14645028##1##,##REF##12611836##33##] (Figure ##FIG##2##3##) and meta-analyses of these data produced equivocal results. One further study that met all but one of our quality criteria reported elevated risk of lifetime prevalence of DSH and/or suicide attempts [##REF##9141776##9##] in gay rather than bisexual men (RR: Gay = 3.61, CI 1.86, 7.01; Bisexual men = 1.95, CI 0.73, 5.19).</p>", "<title>Suicidal ideation</title>", "<p>Meta-analyses of lifetime prevalence of suicidal ideation revealed risk ratios of 2.04 for both sexes (range: both sexes 1.72 to 2.42; men 2.0 to 4.10; women 1.75 to 2.10) with considerable heterogeneity. Attributable risk ranged from 0.10 to 0.40 (Figure ##FIG##3##4##). All studies included in this analysis were limited by selection bias [##UREF##4##23##,##REF##12084697##24##] and small samples [##REF##10707921##12##,##REF##9491018##17##].</p>", "<p>The combined meta-analysis of 12 month prevalence of suicidal ideation contained some heterogeneity in both sexes and in women, but none in men. The risk ratio in both sexes was 1.71 (men 1.64; women 2.31) while attributable risk ranged from 0.02 to 0.21 (men 0.02 to 0.13; women 0.02 to 0.21). One study that met all four quality criteria [##REF##11392937##19##] demonstrated over three times the risk in women but not in men. The other studies were limited by selection of very young populations [##REF##9491018##17##,##REF##9584034##28##,##REF##11499118##30##,##REF##12611836##33##,##REF##15369203##26##] or low participation rates [##REF##15369203##26##].</p>", "<p>In summary, there were elevated risks for suicide attempts and ideation in LGB people but quality of studies was limited. Data from higher quality studies showed higher cumulative incidence of suicide in LB school girls, increased lifetime risk of suicide attempts in GB men and increased 12 months risk of suicidal ideation in LB women.</p>", "<title>Mental disorders – depression</title>", "<p>Three studies reported lifetime prevalence of depression [##REF##10754972##14##,##REF##10530626##18##,##REF##11146762##31##]. Increased risk of lifetime depression was observed in both sexes and men with little heterogeneity in the analyses (Figure ##FIG##4##5##). One of the two studies that met all but one quality criteria demonstrated a risk ratio of 2.2 in both sexes; 2.68 in men (Figure ##FIG##4##5##); and 2.21 (CI 1.57, 3.12) in women [##REF##11146762##31##].</p>", "<p>The risk of 12 months prevalence of depression in LGB people on meta-analysis was at least twice that of heterosexual controls with little heterogeneity (Figure ##FIG##4##5##). All studies in this analysis were of good quality based on general population samples with high participation rates. Risk ratios ranged from 1.57 to 3.74 (men 1.57 to 3.74; women 1.67 to 3.69) and attributable risk from two studies ranged from 0.04 to 0.20 (men 0.04 to 0.20; women 0.04 to 0.22). The only study that met the highest standard on the four quality criteria demonstrated significantly higher risk ratios and attributable risk for women but not men [##REF##11392937##19##]. Lastly, a study of 45 gay and 37 bisexual men that recorded depression on a standardised scale and met all but one of our quality criteria showed a small but positive effect size indicating more depression in gay or bisexual men (standardised mean difference in depression score 0.16) [##REF##9141776##9##].</p>", "<title>Mental disorders – anxiety</title>", "<p>Two studies reported lifetime prevalence of any anxiety disorder and both met all but one of the quality criteria [##REF##10530626##18##,##REF##11146762##31##]. Although their data could not be combined in a meta-analysis, increased risk was reported in both sexes (RR 2.28 CI 1.25, 4.21) [##REF##10530626##18##] and in men (RR 2.40, CI 1.72, 3.35) [##REF##11146762##31##], but not in women (RR 1.02, CI 0.61, 1.70) [##REF##11146762##31##]. The meta-analyses of data on 12 month prevalence of any anxiety disorder (Figure ##FIG##5##6##) resulted in a pooled RR of 1.54 for both sexes and 1.88 in men with little heterogeneity. Attributable risk ranged from 0.00 to 0.17 (men 0.01 to 0.12; women 0.00 to 0.17). The result in women was less convincing because of heterogeneity. The only study of the four in this analysis that met the highest of all four of our quality criteria demonstrated an elevated risk of 1.75 in women [##REF##11392937##19##]. All the studies were based on general population samples and were of reasonable quality.</p>", "<p>In summary, on the basis of studies of relatively good quality, there was an elevated risk of lifetime and 12 month prevalence of depression and anxiety disorders in all LGB groups compared to heterosexual controls.</p>", "<title>Alcohol misuse</title>", "<p>Data from a single study that met all but one of our highest quality criteria showed increased risk of lifetime prevalence of alcohol dependence in both sexes (RR 2.59 CI 1.62, 4.15) and women (RR 6.51, CI 2.74, 15.44) but not in men (RR 1.60, CI 0.91, 2.80) [##REF##11146762##31##]. All the studies in this analysis met at least three of our four quality criteria. Risk ratios for alcohol dependence in the previous 12 months in both sexes ranged from 1.76 to 3.05 and were higher in women (Figure ##FIG##7##8##). Attributable risk for alcohol dependence over 12 months was higher in women (Figure ##FIG##7##8##). Two studies presented data in accordance with our definition of alcohol misuse within the previous 12 months. McCabe et al (2003) [##REF##12766375##25##] reported little difference between LGB people and controls, but Gruskin et al (2001) [##REF##11392944##20##] reported higher risk of alcohol misuse (RR 3.52, CI 1.97, 6.26) in LB than heterosexual women, with an attributable risk of 7%.</p>", "<title>Drug misuse or any substance misuse disorder</title>", "<p>One study reported higher risks of lifetime prevalence of drug dependence in both sexes (RR 4.32, CI 2.14, 8.72), men (RR 2.71, CI 1.01, 7.37) and in women (RR 7.74, CI 2.88, 20.75) [##REF##11146762##31##]. Meta-analyses of data on drug dependence over the previous 12 months showed 2.73 times greater risk in both sexes, 3.5 times greater in women and 2.41 times greater in men than controls (Figure ##FIG##8##9##). Attributable risk for drug dependence in the previous 12 months ranged from 0.002 to 0.05 in both sexes, in men 0.03 to 0.05 and women 0.02 to 0.04 (Figure ##FIG##8##9##).</p>", "<p>One good quality study [##REF##11146762##31##] of lifetime prevalence of any substance use disorder showed elevated risk in women (RR 3.61 CI 2.13, 6.11, attributable risk 0.11 to 0.26) but not men (RR 1.05, CI 0.76, 1.47; attributable risk -0.08 to 0.11). Similar findings arose in the meta-analyses of data from two good quality studies on 12 months prevalence of any substance use disorder (figure ##FIG##8##9##).</p>", "<p>In summary, there was an increased lifetime and 12 month risk of alcohol and drug dependency in all groups compared with heterosexuals with markedly higher risk in lesbian and bisexual women.</p>" ]

[ "<title>Discussion</title>", "<p>LGB people are at higher risk of suicidal behaviour, mental disorder and substance misuse and dependence than heterosexual people. The results of the meta-analyses demonstrate a two fold excess in risk of suicide attempts in the preceding year in men and women, and a four fold excess in risk in gay and bisexual men over a lifetime. Similarly, depression, anxiety, alcohol and substance misuse were at least 1.5 times more common in LGB people. Findings were similar in men and women but LB women were at particular risk of substance dependence, while lifetime risk of suicide attempts was especially high in GB men.</p>", "<title>Strengths and limitations of the review</title>", "<p>We found 25 studies that met our inclusion criteria for epidemiological rather than clinical studies. Our search terms included all possible subcategories of mental disorder and substance dependence. We identified a wide range of study methods but excluded designs that provided biased or erroneous estimates. We included studies with consistent definitions of sexual orientation and with contemporaneous comparison groups. However, the lower than expected prevalence of LGB people in several of the population surveys [##REF##11026127##27##,##REF##11146762##31##,##UREF##5##32##] indicates that many studies were unable to recruit a representative sample. Thus, it is likely that a proportion of LGB people are reluctant to participate in research for all sorts of reasons, but most likely for fear of disclosure. Until it becomes less risky to identify oneself as LGB for the purposes of research we shall know little about this hidden population or how it influences the conclusions we can make here. All studies used well-described and potentially replicable mental health outcomes. However, only one study met all four of our quality criteria, while seven met all but one of our quality markers. The number of studies in each meta-analysis was relatively small and thus we were unable to interpret funnel plots to investigate sources of bias or run a meta-regression analysis to account for the variable quality of the studies identified in this review.</p>", "<p>Given the range of study design and definitions of exposure and outcome, we encountered significant heterogeneity in our meta-analyses. However, these estimates did not deviate markedly from data reported in the better quality studies. Although, in some studies reported data were weighted or shown as percentages, our calculated risk ratios were similar to unadjusted ratios reported in these papers making it unlikely that we have extrapolated beyond the studies' findings. The distinction between suicide attempt and DSH was often unclear. We followed authors' definitions of the acts and did not judge the life threatening nature of the behaviour. Finally, uncertainties inherent in defining and recruiting a representative sample of LGB people cannot be overcome in a systematic review. For example, participants may be asked about their sexuality in ways that are unfamiliar to them or it may be assumed that sexual orientation is a fixed life-time characteristic. Despite these reservations about our review, the consistent direction of our findings suggests that mental health is poorer in LGB people.</p>", "<title>Selection of studies</title>", "<p>We had to exclude otherwise well conducted research that was based in specialised populations or in health services or that selected LGB people in a particular way. We wish to highlight three studies that we eventually excluded on grounds of selection of the LGB population [##REF##11920991##36##, ####REF##11777125##37##, ##REF##11197924##38####11197924##38##]; but whose results were broadly in the direction of our findings. Russell &amp; Keel (2002) [##REF##11920991##36##] reported data on depression using the Beck Depression Inventory; van Heeringen &amp; Vincke (2000) [##REF##11197924##38##] reported data on suicide attempts and ideation and Savin-Williams (2001) [##REF##11777125##37##] reported data on suicide attempts.</p>", "<title>Explanations for our findings</title>", "<p>Our study aimed to determine whether there was unequivocal evidence for a preponderance of mental health problems in LGB people relative to heterosexuals. Thus, circumspection is required when discussing possible mechanisms which generate them [##REF##10530627##120##]. Although our evidence does not specify the nature of such mechanisms, there is no evidence to suggest that homosexuality is itself a disorder that is thereby subject to a higher co-morbidity than is found in heterosexuals [##REF##10530627##120##]. This review was strictly limited to documenting whether or not there was an excess of mental health problems in LGB people. It will take other, prospective research to investigate the components of this vulnerability. Unfortunately prospective studies were unusual among the 25 reviewed here and thus we cannot say much with certainty about the risk factors for mental disorder in LGB people. Nevertheless, it is likely that the social hostility, stigma and discrimination that most LGB people experience is at least part of the reason for the higher rates of psychological morbidity observed. This may be aggravated by easy access to alcohol and drugs in gay venues that LGB people frequent both to find the company of others who will accept them less critically and to meet potential partners. However, why LB women are at greater risk of substance misuse than GB men is not clear as most LGB commercial venues provide alcohol.</p>", "<title>Implications of our findings</title>", "<p>It is of considerable concern that sexual minorities such as LGB people suffer so many disadvantages in terms of mental health. Our findings need consideration in planning public health and clinical services, as well as in terms of international human rights. Although we cannot report on whether or not LGB people are at greater risk than heterosexuals for completed suicide, the elevated risks for all forms of mental disorder, DSH and substance misuse would suggest very strongly that this is the case. Thus, national suicide strategies need to include LGB people as a high risk group now rather than await more evidence on suicide. The hidden nature of sexual orientation makes it very unlikely that we shall be able to show definitely in post-mortem psychological studies that LGB are over-represented among suicide victims.</p>" ]

[ "<title>Conclusion and further research</title>", "<p>Besides more qualitative and case-control research, we need prospective studies as these are most likely to reveal the mechanisms involved. Although, in this review we identified four cohorts [##REF##12602425##15##,##REF##10530626##18##,##REF##12611836##33##,##REF##12653422##34##] only one collected prospective data on suicidal risk in lesbian, gay and bisexual people [##REF##12653422##34##]. Prospective studies, however, are difficult to undertake as many people cannot or will not identify themselves as LGB until late adolescence or even young adulthood when the emotional damage may already have occurred. Nevertheless, a cohort of young LGB people who are followed through as they complete education and career training and start relationships and families, would begin to address this difficult issue. We also need to address the complexities of defining sexual orientation. Most modern conceptions of sexual orientation consider personal identification, sexual behaviour and sexual fantasy [##UREF##27##121##]. Thus, we chose these parameters as the most pragmatic and commonly used definitions for this review. However, we need more detailed study of the development of sexuality across the spectrum of partner preference, its stability over time and its relationship to other preferences and behaviour.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Lesbian, gay and bisexual (LGB) people may be at higher risk of mental disorders than heterosexual people.</p>", "<title>Method</title>", "<p>We conducted a systematic review and meta-analysis of the prevalence of mental disorder, substance misuse, suicide, suicidal ideation and deliberate self harm in LGB people. We searched Medline, Embase, PsycInfo, Cinahl, the Cochrane Library Database, the Web of Knowledge, the Applied Social Sciences Index and Abstracts, the International Bibliography of the Social Sciences, Sociological Abstracts, the Campbell Collaboration and grey literature databases for articles published January 1966 to April 2005. We also used Google and Google Scholar and contacted authors where necessary. We searched all terms related to homosexual, lesbian and bisexual people and all terms related to mental disorders, suicide, and deliberate self harm. We included papers on population based studies which contained concurrent heterosexual comparison groups and valid definition of sexual orientation and mental health outcomes.</p>", "<title>Results</title>", "<p>Of 13706 papers identified, 476 were initially selected and 28 (25 studies) met inclusion criteria. Only one study met all our four quality criteria and seven met three of these criteria. Data was extracted on 214,344 heterosexual and 11,971 non heterosexual people. Meta-analyses revealed a two fold excess in suicide attempts in lesbian, gay and bisexual people [pooled risk ratio for lifetime risk 2.47 (CI 1.87, 3.28)]. The risk for depression and anxiety disorders (over a period of 12 months or a lifetime) on meta-analyses were at least 1.5 times higher in lesbian, gay and bisexual people (RR range 1.54–2.58) and alcohol and other substance dependence over 12 months was also 1.5 times higher (RR range 1.51–4.00). Results were similar in both sexes but meta analyses revealed that lesbian and bisexual women were particularly at risk of substance dependence (alcohol 12 months: RR 4.00, CI 2.85, 5.61; drug dependence: RR 3.50, CI 1.87, 6.53; any substance use disorder RR 3.42, CI 1.97–5.92), while lifetime prevalence of suicide attempt was especially high in gay and bisexual men (RR 4.28, CI 2.32, 7.88).</p>", "<title>Conclusion</title>", "<p>LGB people are at higher risk of mental disorder, suicidal ideation, substance misuse, and deliberate self harm than heterosexual people.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests</p>", "<title>Authors' contributions</title>", "<p>MK, HK, DO, IN and SST obtained funding for the study. JS and DP conducted the literature search, obtained papers and extracted data. JS, MK, HK, DO, IN and DP scanned abstracts and read papers. SST conducted the meta-analysis with input from IN and MK. MK drafted the paper and all authors contributed to the final version. All authors read and approved the final version.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-244X/8/70/prepub\"/></p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>The study was funded by a grant from the National Institute for Mental Health England. The funder had no involvement in the conduct of the research. We thank Rosalind Lai M Lib, Medical Librarian, for constructing the search strategy.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Study inclusion process.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Forest plots for lifetime and 12 month prevalence of suicide attempts.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Forest plots for lifetime prevalence of deliberate self harm.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p>Forest plots for lifetime and 12 month prevalence of suicide ideation.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p>Forest plots for lifetime and 12 month prevalence of depression.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p>Forest plots for 12 month prevalence of anxiety.</p></caption></fig>", "<fig position=\"float\" id=\"F7\"><label>Figure 7</label><caption><p>Forest plots for 12 month prevalence of alcohol dependence.</p></caption></fig>", "<fig position=\"float\" id=\"F8\"><label>Figure 8</label><caption><p>Forest plots for 12 month prevalence of drug dependence.</p></caption></fig>", "<fig position=\"float\" id=\"F9\"><label>Figure 9</label><caption><p>Forest plots for lifetime and 12 month prevalence of any substance use disorder.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Classification of quality indicators of studies included in the review</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"left\">Sampling</td><td align=\"left\">Participation rate</td><td align=\"left\">Population</td><td align=\"left\">Sample size</td></tr><tr><td/><td align=\"left\">1 = Non-random</td><td align=\"left\">1&lt;60%</td><td align=\"left\">1 = Selected</td><td align=\"left\">1&lt;100</td></tr><tr><td/><td align=\"left\">2 = Random</td><td align=\"left\">2 ≥ 60%</td><td align=\"left\">2 = General</td><td align=\"left\">2 ≥ 100</td></tr></thead><tbody><tr><td align=\"left\">Bagley 1997 [##REF##9141776##9##]</td><td align=\"left\">2</td><td align=\"left\">2</td><td align=\"left\">2</td><td align=\"left\">1</td></tr><tr><td align=\"left\">Bontempo &amp; D'Augelli 2002 [##REF##11996785##10##]</td><td align=\"left\">Not known</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Case et al. 2004 [##REF##15665660##11##]*</td><td align=\"left\">1</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Cochran &amp; Mays (2000a) &amp; Mays, Ross (2004) [##REF##10707921##12##,##REF##15265096##13##]</td><td align=\"left\">1</td><td align=\"left\">2</td><td align=\"left\">2</td><td align=\"left\">1</td></tr><tr><td align=\"left\">Cochran &amp; Mays (2000b) [##REF##10754972##14##]</td><td align=\"left\">1</td><td align=\"left\">2</td><td align=\"left\">2</td><td align=\"left\">1</td></tr><tr><td align=\"left\">Cochran et al. 2003 &amp; Mays &amp; Cochran (2001) [##REF##12602425##15##,##REF##11684618##16##]</td><td align=\"left\">2</td><td align=\"left\">2</td><td align=\"left\">2</td><td align=\"left\">1</td></tr><tr><td align=\"left\">Faulkner et al. (1998) [##REF##9491018##17##]</td><td align=\"left\">Not known</td><td align=\"left\">1</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Fergusson et al 1999 [##REF##10530626##18##]*</td><td align=\"left\">1</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">1</td></tr><tr><td align=\"left\">Gilman et al. 2001 [##REF##11392937##19##]</td><td align=\"left\">2</td><td align=\"left\">2</td><td align=\"left\">2</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Gruskin et al. 2001 [##REF##11392944##20##]</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">2</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Herrell et al. 1999 [##REF##10530625##21##]</td><td align=\"left\">Not known</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Jorm et al. 2002 [##REF##11983639##22##]</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">2</td><td align=\"left\">2</td></tr><tr><td align=\"left\">King et al. 2003 [##REF##14645028##1##]</td><td align=\"left\">1</td><td align=\"left\">NA</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Mathy 2002 [##UREF##4##23##]</td><td align=\"left\">1</td><td align=\"left\">1</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Mathews et al. 2002 [##REF##12084697##24##]</td><td align=\"left\">1</td><td align=\"left\">1</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">McCabe et al. 2003 [##REF##12766375##25##]</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Mc Cabe et al. 2004 [##REF##15369203##26##]</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">1</td><td align=\"left\">1</td></tr><tr><td align=\"left\">Nawyn et al. 2000 [##REF##11026127##27##]</td><td align=\"left\">Not known</td><td align=\"left\">1</td><td align=\"left\">1</td><td align=\"left\">1</td></tr><tr><td align=\"left\">Remafedi et al. 1998 [##REF##9584034##28##]</td><td align=\"left\">Not known</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Robin et al. 2002 [##REF##11929369##29##]</td><td align=\"left\">Not known</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Russell &amp; Joyner 2001 [##REF##11499118##30##]</td><td align=\"left\">Not known</td><td align=\"left\">Not known</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Sandfort et al. 2001 &amp; Sandfort, de Graf, Bijl (2003) [##REF##11146762##31##,##UREF##5##32##]</td><td align=\"left\">2</td><td align=\"left\">2</td><td align=\"left\">2</td><td align=\"left\">1</td></tr><tr><td align=\"left\">Skegg et al. 2003 [##REF##12611836##33##]*</td><td align=\"left\">1</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">1</td></tr><tr><td align=\"left\">Wichstrom &amp; Hegna 2003 [##REF##12653422##34##]*</td><td align=\"left\">1</td><td align=\"left\">2</td><td align=\"left\">1</td><td align=\"left\">2</td></tr><tr><td align=\"left\">Drabble et al. 2005 [##REF##15830911##35##]</td><td align=\"left\">2</td><td align=\"left\">Not known</td><td align=\"left\">2</td><td align=\"left\">2</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional File 1</title><p>Table 1: review studies.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>Key: * Longitudinal studies; 2 suggests higher quality and than 1</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-244X-8-70-1\"/>", "<graphic xlink:href=\"1471-244X-8-70-2\"/>", "<graphic xlink:href=\"1471-244X-8-70-3\"/>", "<graphic xlink:href=\"1471-244X-8-70-4\"/>", "<graphic xlink:href=\"1471-244X-8-70-5\"/>", "<graphic xlink:href=\"1471-244X-8-70-6\"/>", "<graphic xlink:href=\"1471-244X-8-70-7\"/>", "<graphic xlink:href=\"1471-244X-8-70-8\"/>", "<graphic xlink:href=\"1471-244X-8-70-9\"/>" ]

[ "<media xlink:href=\"1471-244X-8-70-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Leprosy is a chronic infectious disease caused by the bacterium, <italic>Mycobacterium leprae</italic>, which can affect all ages and both sexes. Over the last two decades, prevalence of leprosy has come down substantially at the global level. Introduction and expansion of Multi drug therapy (MDT) in leprosy control programmes have dramatically lowered the prevalence level in almost all the endemic countries. For instance, in India leprosy prevalence has come down from 51 per 10,000 in 1981 to around 2.4 per 10,000 in March 2004 [##UREF##0##1##] and further below 1 per 10,000 by December 2005 [##UREF##1##2##].</p>", "<p>In disease epidemiology, many of the infectious disease events do not occur randomly in geographical context but occur in clusters. In fact, leprosy epidemiology shows such uneven distribution between different geographic areas of the country, e.g. in China [##REF##11357209##3##] and in Indonesia where in the leprosy endemic was high, the cases were extensively clustered and not equally distributed [##REF##12225510##4##]. Hence, data analyses and interpretation should not ignore spatial dependence [##UREF##2##5##]. In India, we have observed that distribution of leprosy is uneven [##REF##17343215##6##] even within the smallest community groups such as villages, right up to the family level [##UREF##3##7##].</p>", "<p>Geographical or spatial analysis comes into play due to the existence of spatial dependence in data. Bayesian method lends itself for representing the spatial dependence during the estimation of model parameters. Application of Bayesian analysis in the field of leprosy is very limited [##REF##15840546##8##]. A large controlled, double blind, randomized, prophylactic leprosy vaccine trial was conducted in South India to assess the prophylactic efficacies of four different candidate vaccines [##REF##10189587##9##]. Within the trial area, we observed that leprosy was not randomly distributed and showed significant spatial dependence confirmed by using various measures of spatial autocorrelation like Moran's I, Geary's C and Kulldorff's SATSCAN statistics [##UREF##4##10##]. Hence, our main objective was to examine the variation in the prevalence of leprosy using four Bayesian models described by Arbyn etal [##UREF##5##11##] which was earlier proposed by Lagazio et al [##UREF##6##12##] and to explore possible factors that might have influenced these variations in the study area.</p>" ]

[ "<title>Materials and Methods</title>", "<p>Leprosy prevalence data from an endemic area, covering 148 panchayats (rural administrative units) comprising of 264 contiguous villages from Chingleput district, Tamil Nadu, South India was studied. This area was specifically identified for a leprosy vaccine trial [##REF##10189587##9##] because of the high endemicity of leprosy. The entire population of about 300,000 people was screened by house to house examinations for leprosy and cases were identified at four time schedules between January 1991 and March 2003. Few socio- economic factors like population density and economic status were also collected.</p>", "<title>Definition of leprosy</title>", "<p>A case of leprosy was defined as a person, having one or more of the following manifestations and who needed antileprosy treatment: hypopigmented or reddish skin lesion(s) with definite loss of sensation; damage to the peripheral nerves, as demonstrated by loss of sensation and or weakness of the muscles in parts supplied by these nerves; skin smear positive for acid fast bacilli.</p>", "<p>The entire population in this area was screened for leprosy before vaccination. First, paramedical workers, trained in leprosy detection, screened the population. A proportion (5%) of this population was randomly allotted to \"blinded\" senior persons – either medical officers or senior paramedical officers for quality control. All cases and suspects detected by the junior paramedical workers were examined for diagnosis by two senior persons and by a third independent examiner in case of disagreement. Skin smear examination for detecting acid fast bacilli was done for all suspects and definite cases by the senior workers. A team of independent clinicians visited the field at frequent intervals to monitor the procedures for diagnosis of leprosy [##REF##10189587##9##].</p>", "<p>The data collected was also validated in many ways with the earlier surveys [##UREF##3##7##]. Hence the quality of data collected was remarkable and comparable to world standard as certified by the independent assessment committee consisting of national and international experts.</p>", "<title>Data analysis</title>", "<p>Leprosy cases and population for each panchayat were cross-classified into 20 age groups (1–4,5–9,...,90–94,95–99) (there were no cases of leprosy under one year age) and four survey time periods (1991–93,1993–95,1997–98 &amp;1999–2003). Since the time schedule for each of the survey was of varying length (January 1991 to March 2003) the mid-point of the four surveys (1992, 1994, 1997 &amp; 2001) was considered for the time period. Cohorts were computed on the basis of survey time period and age. There were 20 overlapping or rolling birth cohorts like moving averages defined in this model, considering the mid-point, the cohorts were labeled as 1902, 1907,...,1997. The variation of prevalence of leprosy over space and time was modeled from January 1991 to March 2003 over 148 panchayats, after controlling for age. The term cohort effect refers to population born during a particular survey period identified by period of birth so that its characteristics can be ascertained as it enters successive and age strata [##UREF##7##13##].</p>", "<p>Four Bayesian models, namely</p>", "<p>(i) Space-Cohort (SC) with interactions,</p>", "<p>(ii) Space-Cohort (SC) without interactions,</p>", "<p>(iii) Space-Period (SP) with interactions and</p>", "<p>(iv) Space-Period (SP) without interactions were fitted.</p>", "<p>Models with and without interaction terms were compared using the Deviance Information Criterion (DIC) [##UREF##8##14##], being a generalization of the Akaike's Information Criterion in the Bayesian framework, i.e., lower the DIC values better the model. Posterior distributions (using the priors and the available data) of the parameters of interest were obtained using Gibbs sampling in WinBUGS [##UREF##9##15##]. The data for the SC model (observed and expected cases) were obtained by aggregating the data over the four time periods. The data for the SP model (observed and expected cases) were obtained by aggregating the data over the age groups. For the SC model the average of all the cohorts and for SP model the average of all periods was used as the reference.</p>", "<p>The spatial effects were measured using each of the models. These smoothed effects were compared using the raw standardized morbidity rate (SMR) [unsmoothed].</p>", "<p>We have dealt with subgroups like panchayats hence we used prevalence instead of incidence to get sufficiently larger number of cases to draw meaningful conclusions. Moreover if we use incidence rather than prevalence, we will have only three survey data leaving the baseline survey. Generally the second survey incidence (a mixed bag of old prevalence- missed and new cases) is not considered for vaccine efficacy and also for trend analysis. Hence we would be left with the third and fourth survey and ultimately this exercise of Bayesian model would not be possible to examine the trend over years. Hence as a pragmatic measure, we considered prevalence cases for this study.</p>", "<p>In the study area majority of leprosy cases belonged to paucibacillary variety and the multibacillary cases constituted a smaller proportion (17.4%, 3.3%, 5.4% and 1.4% in the four surveys respectively). Fixed duration of MDT for six months was practiced all through the study period. For the purpose of analysis in this paper we restricted it only for paucibacillary variety. Hence, prevalence trends also indicate trends in leprosy incidence and both go hand-in-hand. Any change in beneficial effect that happens during the survey period decreases incidence as well as prevalence. Since we had limited data collected specifically by panchayat on socio-economic factors and there could be other important factors apart from the two mentioned above, we did not include any of these factors directly in the model. Though the period of study was very short, in the region, major remarkable changes have taken place both in the leprosy control programme and improvement in the socio-economic status. According to the World Development Indicators 2005 [##UREF##10##16##] the rural poverty in India declined from 53% to 27% between 1977–78 and 1999–2000 and the Indian social structure was transformed between 1991 and 2001 like increased literacy rate, urbanization, industrialization, new economic liberalization etc.</p>", "<p>The formulae used are described in Appendix-1</p>" ]

[ "<title>Results</title>", "<p>There were 6,601 cases, 4,731 cases 3,342 cases and 2,098 cases of leprosy respectively at the four time periods. The DIC value for the space cohort model with and without interaction were compared. Since the models with the interaction term out- performed with smaller DIC values (Table ##TAB##0##1## and Table ##TAB##1##2##). Further analyses were carried out with interaction term included into the SC and SP model. The Markov Chain Monte Carlo Simulation (MCMC) scalar parameters and their 95% credible intervals for the SC and SP models with interactions are shown in Table ##TAB##2##3## and Table ##TAB##3##4## respectively.</p>", "<title>Cohort effect using the SC model with interaction</title>", "<p>The median cohort effects declined from 1.86 to 0.08 over the successive cohorts. The cohort effect using SC model (Table ##TAB##4##5##) measured in terms of relative risk, was significantly higher than the average in the older cohorts C ≤ 11, i.e. persons born before 1957. The cohort effect steadily decreased up to 1922–1931 (C ≤ 6) and increased substantially during 1927–1936,1932–1941 (C = 7 &amp; C = 8) and again significantly declined to a higher risk of 36% at C = 11 and finally reached to a lower risk of 92% (C = 20) than the average for persons born after 1996. There were four major jumps (difference between a cohort and the preceding one) observed in the risk pattern of the cohorts i.e. 1927–1936 (higher risk of 45%), 1942–51(reduced risk of 31%), 1987–1996 (reduced risk of 33%) and 1992–2001(reduced risk of 40%).</p>", "<title>Period effect using the SP model with interaction</title>", "<p>The period effect over four time points using SP model showed a significantly higher risk of 1.032 than the average, i.e. 3.2% to a lower risk 1.8% (Table ##TAB##5##6##).</p>", "<title>Space effect using SC and SP model with interaction</title>", "<p>The spatial effect values (smoothed Bayesian) using SC and SP models using interaction terms were similar as observed in the Belgium study [##UREF##5##11##]. The spatial effects of different panchayats are listed in Table ##TAB##6##7## and it can be further visualized in the choropleth map (Figure ##FIG##0##1##). The spatial effects varied between 0.59 and 2. Bayesian model identified 26 panchayats that had a significantly higher risk of leprosy. There was a higher risk of leprosy (50% or more) found in 10 panchayats and most of them lay close to each other towards the North-Eastern end of the study area. The lowest risk (relative risk 0.59) was observed in two panchayats. Raw SMR identified 43 panchayats at a higher risk, which was statistically significant (Figure ##FIG##1##2##); whereas it was 26 panchayats by the Bayesian models.</p>" ]

[ "<title>Discussion</title>", "<p>We examined variation in prevalence of leprosy using Bayesian methods over 20 rolling cohorts and four time periods from a meticulously collected dataset of a vaccine trial conducted in South India. Observing the cohort effects it neither showed a steady decreasing nor an increasing effect. It showed an intricate effect, whereas leprosy prevalence trends in period effects has decreased continuously over four time points.</p>", "<p>The steady decreasing period effect reflects that the control programme had reached all the target population whereas the intricate cohort effects showed that the effects captured were not similar in all the cohorts.</p>", "<p>The higher difference in the risk in successive <bold>cohorts </bold>could be attributable to persons coming forward for seeking treatment, gradual awareness in the community that leprosy is curable, slowly combating the social stigma, early screening programmes, better case detection methods, availability of therapies and elimination programmes. This is particularly true in the vaccine trial setting where health systems operations and leprosy programme have been implemented in total [##REF##17343215##6##].</p>", "<p>The turning point in the risk pattern (from high risk to low risk) occurred during the cohort 1952–1961 (C = 12). This change could be due to the impact of Dapsone based National Leprosy Control Programme introduced during 1955. Gradual reduction of risk in the latter cohorts (C = 13, C = 14) may be due to continuing effect of programs and possibly improvement in socio-economic conditions the transition that has been taking place in India. Further reduction in risk (C = 19) could be due to more effective and intensified treatment programmes like Multi-Drug Therapy in 1991, that has changed the face of leprosy [##REF##9401483##17##] and introduction of effective prophylactic vaccines through the leprosy vaccine trial [##REF##10189587##9##].</p>", "<p>Our data indicate 92% reduction in the leprosy prevalence for persons born after 1996 (C = 20). The case detection and treatment activity in the community has brought down the new infection rate in the younger age group as a secondary effect of MDT in addition to the primary effect of prophylactic vaccines. The older age group might have been infected in the past i.e. before the introduction of Dapsone and MDT as well as before the introduction of the vaccination programme. They might already be harbouring the infection and the break down could result in fresh disease. This is similar to the endogenous reactivation observed in tuberculosis [##REF##823803##18##].</p>", "<p>We observed slow decline in the estimated <bold>period effects </bold>of leprosy in the study area. It agrees with the fact that the prevalence of leprosy has come down over years globally as well as in India [##UREF##11##19##]. Leprosy being a chronic disease, temporal changes within endemic regions are slow [##UREF##3##7##]. This phenomenon is observed in the study area. Moreover, the decline in the risk of leprosy over the time periods in the study area could be due to better understanding on nutrition, hygiene, and increased public understanding of the disease (socio-economic factors) which limit the spread of leprosy or increased resistance to leprosy.</p>", "<p>The <bold>spatial effect </bold>using Bayesian model was compared with the raw SMR gives the variation in the geographical distribution of the leprosy prevalence. The use of Bayesian smoothing approach accounts for the variability in the population at risk and clustering effect. Observing the spatial Bayesian effect, there was a strong pattern of clustering towards the North-Eastern region of the study area. Though the effects obtained through the models showed the reduction in the risk of prevalence of leprosy still a few pockets of high prevalence exist. The situation was similar in Tuscany [##UREF##5##11##] where the epidemic of lung cancer when analyzed using birth cohorts showed a decline but the spatial pattern was evident and strong towards the north-west/south-east gradient.</p>", "<p>In the state of Ceara North- East Brazil [##REF##15654421##20##], the spatial pattern of the leprosy disease was heterogeneous and municipalities with very high prevalence were clustered towards the North-South axis. Surprisingly the region with the highest incidences were most urbanized and economically developed. According to the authors the reasons for spatial clustering of disease rates might be related to a heterogeneous distribution of the factors such as crowding, social inequality and environmental characteristics which by themselves determine the transmission of Mycobacterium leprae. It could also be due to more efficient health system present in these regions to detect new cases of leprosy more efficiently. We tried to explore the authors' above perception and observed that the pockets identified by the Bayesian model had a population density of about 1,427 per sq km which is two times higher than the district and three times higher than the state population density [##UREF##12##21##]. Hence, possibly as result of this, environmental factors, such as urbanization and overcrowding due to inadequate housing, could have led to more frequent close contact with the source of infection and favoured the spread of leprosy. We also observed that nearly 37% of the people from this pocket belong to the economically poorer strata. Generally people from economically poorer strata are more prone to infectious diseases like leprosy as they live in close proximity to one another resulting in higher risk of contracting the disease [##UREF##13##22##]. Dharmendra [##UREF##14##23##] emphasizes that one cannot control, eliminate or eradicate leprosy without improving the socio-economic status or changing the in sanitary habits of the common people. Hence these few pockets or strata need greater care to bring down the leprosy prevalence to much greater extent and to make the study area free from leprosy.</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>In leprosy endemic areas, patients are usually spatially clustered and not randomly distributed. Classical statistical techniques fail to address the problem of spatial clustering in the regression model. Bayesian method is one which allows itself to incorporate spatial dependence in the model. However little is explored in the field of leprosy. The Bayesian approach may improve our understanding about the variation of the disease prevalence of leprosy over space and time.</p>", "<title>Methods</title>", "<p>Data from an endemic area of leprosy, covering 148 panchayats from two taluks in South India for four time points between January 1991 and March 2003 was used. Four Bayesian models, namely, space-cohort and space-period models with and without interactions were compared using the Deviance Information Criterion. Cohort effect, period effect over four time points and spatial effect (smoothed) were obtained using WinBUGS. The spatial or panchayat effect thus estimated was compared with the raw standardized morbidity (leprosy prevalence) rate (SMR) using a choropleth map. The possible factors that might have influenced the variations of prevalence of leprosy were explored.</p>", "<title>Results</title>", "<p>Bayesian models with the interaction term were found to be the best fitted model. Leprosy prevalence was higher than average in the older cohorts. The last two cohorts 1987–1996 and 1992–2001 showed a notable decline in leprosy prevalence. Period effect over 4 time points varied from a high of 3.2% to a low of 1.8%. Spatial effect varied between 0.59 and 2. Twenty-six panchayats showed significantly higher prevalence of leprosy than the average when Bayesian method was used and it was 40 panchayats with the raw SMR.</p>", "<title>Conclusion</title>", "<p>Reduction of prevalence of leprosy was 92% for persons born after 1996, which could be attributed to various intervention and treatment programmes like vaccine trial and MDT. The estimated period effects showed a gradual decline in the risk of leprosy which could be due to better nutrition, hygiene and increased awareness about the disease. Comparison of the maps of the relative risk using the Bayesian smoothing and the raw SMR showed the variation of the geographical distribution of the leprosy prevalence in the study area. Panchayat or spatial effects using Bayesian showed clustersing of leprosy cases towards the northeastern end of the study area which was overcrowded and population belonging to poor economic status.</p>" ]

[ "<title>Appendix-1</title>", "<p>The step by step algorithms and winbug codes can be downloaded from annexes of Arbyn etal [##UREF##5##11##].</p>", "<p>Let <italic>O</italic>_{<italic>ijt </italic>}denote the observed count of leprosy cases in panchayat i (i = 1, 2,...,148) in the j^thcohort (j = 1, 2 ..... 20) and during the time point t (t = 1, 2, 3 &amp; 4);</p>", "<p> be the observed number of leprosy cases in the i^thpanchayat and j^thcohort;</p>", "<p> be the expected number of leprosy cases in the i^thpanchayat and j^thcohort;</p>", "<p><italic>O</italic>_{<italic>ij </italic>}~ Poisson (<italic>α</italic>_{<italic>ij</italic>}),</p>", "<p>with <italic>α</italic>_{<italic>ij </italic>}= <italic>rr</italic>_{<italic>ij </italic>}. X_{<italic>ij</italic>}, where i = 1, 2,...,148 panchayats,</p>", "<p>j = 1, 2 ..... 20 cohorts,</p>", "<p></p>", "<p>where</p>", "<p><italic>ξ</italic>_{<italic>ij </italic>}is a linear predictor,</p>", "<p><italic>rr</italic>_{<italic>ij </italic>}is the relative risk of the i^thpanchayat and the j^thcohort.</p>", "<p> is the estimated cohort effects and similar to the standardized cohort morbidity ratios described by Beral [##REF##4133714##24##].</p>", "<p><italic>ξ</italic>_ij, the linear predictor can be specified in two ways.</p>", "<p>Model without interactions: <italic>ξ</italic>_{<italic>ij </italic>}= a + <italic>β</italic>_i^str+ <italic>β</italic>_i^unstr+ <italic>β</italic>_j^coh,</p>", "<p>Model with interactions: <italic>ξ</italic>_ij= a + <italic>β</italic>_i^str+ <italic>β</italic>_i^unstr+ <italic>β</italic>_j^coh+ SS_ij^ac,</p>", "<p>Where SS_ij^ac= S_ij^ac- S_1j^ac- S_i1^ac+ S₁₁^acand</p>", "<p>The first term S_ij^acrepresents the interaction term of panchayat and cohort and SS_ij^acis the centering used to improve a convergence of the Markov Chain Monte Carlo Simulation (MCMC) same as discussed in detail by Arbyn et al [##UREF##5##11##].</p>", "<p>Where <italic>β</italic>_i^strrepresents structured spatial variability;</p>", "<p><italic>β</italic>_i^unstrrepresents unstructured spatial variability;</p>", "<p><italic>β</italic>_j^cohrepresents the effect of the j^thcohort;</p>", "<p>Prior distribution for the model</p>", "<p></p>", "<p></p>", "<p></p>", "<p></p>", "<p>The priors are multivariate normals.</p>", "<p></p>", "<p></p>", "<p></p>", "<p></p>", "<p>The structured spatial term <italic>β</italic>^strand cohort effect <italic>β</italic>^cohare assigned the Gaussian Conditional Autoregression (CAR) prior distribution. They followed the closer specification of matrices <italic>σ</italic>_{<italic>str</italic>}, <italic>σ</italic>_{<italic>coh </italic>}and <italic>σ</italic>_{<italic>ac </italic>}as mentioned by Lagazio [##UREF##6##12##]. They are implemented using WinBUGS' function car.normal().</p>", "<p>The above function constraints, the random effects to add up to zero, so that the following constraints are satisfied in the model.</p>", "<p></p>", "<p>The prior for the interaction vector S^acis a Markov random field.</p>", "<p>The intercept term 'a' was given a flat prior through WinBUGS function dflat().</p>", "<p>The precision terms <italic>μ</italic>_{<italic>str </italic>}<italic>and μ</italic>_{<italic>coh </italic>}and <italic>μ</italic>_{<italic>unstr </italic>}were given Gamma priors.</p>", "<p>The interaction precision parameter <italic>μ</italic>^acdiscussed in detail [##UREF##5##11##] was also given the gamma prior closely to Lagazio [##UREF##6##12##].</p>", "<p>The space period model is similar as above instead of cohorts, periods used (t = 4). The algorithmic steps for the four models using WINBUGS with and without interactions similar to the space cohort model are discussed in detail elsewhere [##UREF##5##11##].</p>", "<p>Deviance Information Criterion (DIC) is defined as</p>", "<p> where</p>", "<p> – the posterior expectation of the deviance and summarizes the fit of the model,</p>", "<p> – the deviance evaluated at the posterior expectations of parameters.</p>", "<p> – the effective number of parameters.</p>", "<p>In the model the spatial effect (autocorrelation) depends on</p>", "<p>(i) whether any two panchayats share a common boundary and</p>", "<p>(ii) the number of shared neighbours (panchayats).</p>", "<p>For fitting the models with and without interactions, two chains, with 1:10 thinning was used to obtain a sample of 10 000 values.</p>", "<p>In case of the model without interactions a burn-in of 50,000 iterations and an additional 50,000 iterations were used. For fitting the models with interactions, a burn-in of 100 000 iterations and an additional 50,000 were used.</p>", "<p>Convergence was checked using procedures mentioned by Gelman and Geweke [##UREF##15##25##,##UREF##16##26##] test and partial correlation plots to check for achieved convergence, of relative risks and hyper-parameters.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>VJ conceived the ideas, performed the statistical analysis and drafted the manuscript. MDG conceived the study, coordinated and participated in the trial, gave critical and intellectual comments for the improvement of the manuscript. MB contributed to data analysis and gave critical comments for the improvement of the manuscript. All the authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Dr. P Manickam^@and Dr. V Selvaraj^@for their critical comments and reading the manuscript. We thank Mr. A Elangovan^@for digitizing the map.</p>", "<p>^@National Institute of Epidemiology.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>The median relative risk of leprosy prevalence across 148 panchayats estimated using the Bayesian model in two taluks of Tamil Nadu, South India (1991–2003).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Geographical distribution of raw SMR across 148 panchayats in two taluks of Tamil Nadu, South India (1991–2003).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Space-cohort model in two taluks of Tamil Nadu, South India.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Model without interactions</bold></td><td align=\"center\"><bold>Model with interactions</bold></td></tr></thead><tbody><tr><td align=\"center\"></td><td align=\"center\">12097.8</td><td align=\"center\">10605.9</td></tr><tr><td align=\"center\"></td><td align=\"center\">11954.7</td><td align=\"center\">10474.6</td></tr><tr><td align=\"center\">DIC</td><td align=\"center\">12240.9</td><td align=\"center\">10737.2</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Space-period model in two taluks of Tamil Nadu, South India.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Model without interactions</bold></td><td align=\"center\"><bold>Model with interactions</bold></td></tr></thead><tbody><tr><td align=\"center\"></td><td align=\"center\">3880.5</td><td align=\"center\">3774.2</td></tr><tr><td align=\"center\"></td><td align=\"center\">3757.4</td><td align=\"center\">3649.1</td></tr><tr><td align=\"center\">DIC</td><td align=\"center\">4003.7</td><td align=\"center\">3899.4</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Summary of Markov Chain Monte Carlo scalar parameters. Space Cohort Bayesian model with interactions.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\"><bold>node</bold></td><td align=\"center\"><bold>mean</bold></td><td align=\"center\"><bold>sd</bold></td><td align=\"center\"><bold>MC error</bold></td><td align=\"center\"><bold>Median RR</bold></td><td align=\"center\"><bold>95% CI</bold></td></tr></thead><tbody><tr><td align=\"center\"><bold><italic>α</italic></bold></td><td align=\"center\">0.004</td><td align=\"center\">0.056</td><td align=\"center\">0.004</td><td align=\"center\">0.011</td><td align=\"center\">-0.139, 0.096</td></tr><tr><td align=\"center\"><bold><italic>τ</italic></bold>^{<italic>struct</italic>}</td><td align=\"center\">19.21</td><td align=\"center\">10.14</td><td align=\"center\">0.55</td><td align=\"center\">16.86</td><td align=\"center\">6.74,46.11</td></tr><tr><td align=\"center\"><bold><italic>τ</italic></bold>^{<italic>unstruct</italic>}</td><td align=\"center\">24.49</td><td align=\"center\">8.87</td><td align=\"center\">0.44</td><td align=\"center\">22.70</td><td align=\"center\">14.29,46.33</td></tr><tr><td align=\"center\"><bold><italic>τ</italic></bold>^{<italic>cohort</italic>}</td><td align=\"center\">4.53</td><td align=\"center\">1.53</td><td align=\"center\">0.02</td><td align=\"center\">4.35</td><td align=\"center\">2.06, 8.06</td></tr><tr><td align=\"center\"><bold><italic>τ</italic></bold>^{<italic>ac</italic>}</td><td align=\"center\">5926</td><td align=\"center\">1746</td><td align=\"center\">131</td><td align=\"center\">6352</td><td align=\"center\">1651, 8460</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Summary of Markov Chain Monte Carlo scalar parameters. Space period Bayesian models with interactions.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\"><bold>node</bold></td><td align=\"center\"><bold>Mean</bold></td><td align=\"center\"><bold>sd</bold></td><td align=\"center\"><bold>MC error</bold></td><td align=\"center\"><bold>Median RR</bold></td><td align=\"center\"><bold>95% CI</bold></td></tr></thead><tbody><tr><td align=\"center\"><bold><italic>α</italic></bold></td><td align=\"center\">-0.136</td><td align=\"center\">0.029</td><td align=\"center\">0.001</td><td align=\"center\">-0.136</td><td align=\"center\">-0.194, -0.081</td></tr><tr><td align=\"center\"><bold><italic>τ</italic></bold>^{<italic>struct</italic>}</td><td align=\"center\">26.45</td><td align=\"center\">19.11</td><td align=\"center\">1.01</td><td align=\"center\">21.35</td><td align=\"center\">6.78, 78.25</td></tr><tr><td align=\"center\"><bold><italic>τ</italic></bold>^{<italic>unstruct</italic>}</td><td align=\"center\">25.38</td><td align=\"center\">28.45</td><td align=\"center\">1.61</td><td align=\"center\">21.46</td><td align=\"center\">13.91, 48.71</td></tr><tr><td align=\"center\"><bold><italic>τ</italic></bold>^{<italic>period</italic>}</td><td align=\"center\">1490</td><td align=\"center\">1382</td><td align=\"center\">21</td><td align=\"center\">1084</td><td align=\"center\">125, 5282</td></tr><tr><td align=\"center\"><bold><italic>τ</italic></bold>^{<italic>ac</italic>}</td><td align=\"center\">5563</td><td align=\"center\">810</td><td align=\"center\">35</td><td align=\"center\">5513</td><td align=\"center\">4128, 7274</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Median Relative risk (cohort effects) and 95% credible intervals in two taluks of Tamil Nadu, South India.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\"><bold>Rolling Cohort</bold></td><td align=\"center\"><bold>Mid year</bold></td><td align=\"center\"><bold>Index (C)</bold></td><td align=\"center\"><bold>Median RR</bold></td><td align=\"center\"><bold>95% CI</bold></td></tr></thead><tbody><tr><td align=\"center\">-1906</td><td align=\"center\">1902</td><td align=\"center\">1</td><td align=\"center\">1.86</td><td align=\"center\">1.01, 4.22</td></tr><tr><td align=\"center\">1902–1911</td><td align=\"center\">1907</td><td align=\"center\">2</td><td align=\"center\">1.71</td><td align=\"center\">1.08, 2.98</td></tr><tr><td align=\"center\">1907–1916</td><td align=\"center\">1912</td><td align=\"center\">3</td><td align=\"center\">1.65</td><td align=\"center\">1.10, 2.41</td></tr><tr><td align=\"center\">1912–1921</td><td align=\"center\">1917</td><td align=\"center\">4</td><td align=\"center\">1.51</td><td align=\"center\">1.14, 1.97</td></tr><tr><td align=\"center\">1917–1926</td><td align=\"center\">1922</td><td align=\"center\">5</td><td align=\"center\">1.34</td><td align=\"center\">1.09, 1.63</td></tr><tr><td align=\"center\">1922–1931</td><td align=\"center\">1927</td><td align=\"center\">6</td><td align=\"center\">1.28</td><td align=\"center\">1.10, 1.49</td></tr><tr><td align=\"center\">1927–1936</td><td align=\"center\">1932</td><td align=\"center\">7</td><td align=\"center\">1.73</td><td align=\"center\">1.51, 1.96</td></tr><tr><td align=\"center\">1932–1941</td><td align=\"center\">1937</td><td align=\"center\">8</td><td align=\"center\">1.74</td><td align=\"center\">1.54, 1.96</td></tr><tr><td align=\"center\">1937–1946</td><td align=\"center\">1942</td><td align=\"center\">9</td><td align=\"center\">1.66</td><td align=\"center\">1.48, 1.87</td></tr><tr><td align=\"center\">1942–1951</td><td align=\"center\">1947</td><td align=\"center\">10</td><td align=\"center\">1.35</td><td align=\"center\">1.20, 1.51</td></tr><tr><td align=\"center\">1947–1956</td><td align=\"center\">1952</td><td align=\"center\">11</td><td align=\"center\">1.36</td><td align=\"center\">1.21, 1.52</td></tr><tr><td align=\"center\"><bold>1952–1961</bold></td><td align=\"center\"><bold>1957</bold></td><td align=\"center\"><bold>12</bold></td><td align=\"center\"><bold>0.99</bold></td><td align=\"center\"><bold>0.88, 1.11</bold></td></tr><tr><td align=\"center\">1957–1966</td><td align=\"center\">1962</td><td align=\"center\">13</td><td align=\"center\">0.97</td><td align=\"center\">0.87, 1.09</td></tr><tr><td align=\"center\">1962–1971</td><td align=\"center\">1967</td><td align=\"center\">14</td><td align=\"center\">0.76</td><td align=\"center\">0.67, 0.85</td></tr><tr><td align=\"center\">1967–1976</td><td align=\"center\">1972</td><td align=\"center\">15</td><td align=\"center\">0.81</td><td align=\"center\">0.72, 0.91</td></tr><tr><td align=\"center\">1972–1981</td><td align=\"center\">1977</td><td align=\"center\">16</td><td align=\"center\">0.66</td><td align=\"center\">0.59, 0.74</td></tr><tr><td align=\"center\">1977–1986</td><td align=\"center\">1982</td><td align=\"center\">17</td><td align=\"center\">0.71</td><td align=\"center\">0.64, 0.80</td></tr><tr><td align=\"center\">1982–1991</td><td align=\"center\">1987</td><td align=\"center\">18</td><td align=\"center\">0.81</td><td align=\"center\">0.73, 0.91</td></tr><tr><td align=\"center\">1987–1996</td><td align=\"center\">1992</td><td align=\"center\">19</td><td align=\"center\">0.48</td><td align=\"center\">0.42, 0.55</td></tr><tr><td align=\"center\">1992–2001</td><td align=\"center\">1997</td><td align=\"center\">20</td><td align=\"center\">0.08</td><td align=\"center\">0.06, 0.10</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T6\"><label>Table 6</label><caption><p>Median period effects and 95% credible intervals using Bayesian models in two taluks of Tamil Nadu, South India.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\"><bold>Period</bold></td><td align=\"center\"><bold>Median RR</bold></td><td align=\"center\"><bold>95% CI</bold></td></tr></thead><tbody><tr><td align=\"center\">April 1992</td><td align=\"center\">1.032</td><td align=\"center\">1.010, 1.070</td></tr><tr><td align=\"center\">May 1994</td><td align=\"center\">1.002</td><td align=\"center\">1.001, 1.026</td></tr><tr><td align=\"center\">November 1997</td><td align=\"center\">0.985</td><td align=\"center\">0.959, 1.007</td></tr><tr><td align=\"center\">March 2001</td><td align=\"center\">0.982</td><td align=\"center\">0.949, 1.013</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T7\"><label>Table 7</label><caption><p>Panchayat effect and 95% credible intervals of prevalence of leprosy in two taluks of Tamil Nadu, South India.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\"><bold>Panch. No.</bold></td><td align=\"center\"><bold>Median RR</bold></td><td align=\"center\"><bold>95% CI</bold></td><td align=\"center\"><bold>Panch. No.</bold></td><td align=\"center\"><bold>Median RR</bold></td><td align=\"center\"><bold>95% CI</bold></td></tr></thead><tbody><tr><td align=\"center\"><bold>072</bold></td><td align=\"center\"><bold>2.00</bold></td><td align=\"center\"><bold>1.64, 2.41</bold></td><td align=\"center\">140</td><td align=\"center\">1.00</td><td align=\"center\">0.85, 1.18</td></tr><tr><td align=\"center\"><bold>006</bold></td><td align=\"center\"><bold>1.88</bold></td><td align=\"center\"><bold>1.71, 2.07</bold></td><td align=\"center\">142</td><td align=\"center\">0.98</td><td align=\"center\">0.80, 1.19</td></tr><tr><td align=\"center\"><bold>010</bold></td><td align=\"center\"><bold>1.73</bold></td><td align=\"center\"><bold>1.46, 2.05</bold></td><td align=\"center\">050</td><td align=\"center\">0.97</td><td align=\"center\">0.75, 1.24</td></tr><tr><td align=\"center\"><bold>009</bold></td><td align=\"center\"><bold>1.66</bold></td><td align=\"center\"><bold>1.48, 1.86</bold></td><td align=\"center\">043</td><td align=\"center\">0.97</td><td align=\"center\">0.85, 1.10</td></tr><tr><td align=\"center\"><bold>015</bold></td><td align=\"center\"><bold>1.55</bold></td><td align=\"center\"><bold>1.35, 1.78</bold></td><td align=\"center\">105</td><td align=\"center\">0.97</td><td align=\"center\">0.77, 1.20</td></tr><tr><td align=\"center\"><bold>012</bold></td><td align=\"center\"><bold>1.54</bold></td><td align=\"center\"><bold>1.38, 1.72</bold></td><td align=\"center\">055</td><td align=\"center\">0.95</td><td align=\"center\">0.72, 1.23</td></tr><tr><td align=\"center\"><bold>017</bold></td><td align=\"center\"><bold>1.54</bold></td><td align=\"center\"><bold>1.37, 1.73</bold></td><td align=\"center\">057</td><td align=\"center\">0.95</td><td align=\"center\">0.76, 1.18</td></tr><tr><td align=\"center\"><bold>077</bold></td><td align=\"center\"><bold>1.54</bold></td><td align=\"center\"><bold>1.23, 1.91</bold></td><td align=\"center\">085</td><td align=\"center\">0.95</td><td align=\"center\">0.75, 1.19</td></tr><tr><td align=\"center\"><bold>088</bold></td><td align=\"center\"><bold>1.51</bold></td><td align=\"center\"><bold>1.23, 1.85</bold></td><td align=\"center\">021</td><td align=\"center\">0.95</td><td align=\"center\">0.77, 1.15</td></tr><tr><td align=\"center\"><bold>045</bold></td><td align=\"center\"><bold>1.50</bold></td><td align=\"center\"><bold>1.20, 1.86</bold></td><td align=\"center\">005</td><td align=\"center\">0.95</td><td align=\"center\">0.81, 1.10</td></tr><tr><td align=\"center\"><bold>053</bold></td><td align=\"center\"><bold>1.49</bold></td><td align=\"center\"><bold>1.26, 1.76</bold></td><td align=\"center\">137</td><td align=\"center\">0.94</td><td align=\"center\">0.83, 1.06</td></tr><tr><td align=\"center\"><bold>014</bold></td><td align=\"center\"><bold>1.48</bold></td><td align=\"center\"><bold>1.26, 1.72</bold></td><td align=\"center\">099</td><td align=\"center\">0.94</td><td align=\"center\">0.77, 1.13</td></tr><tr><td align=\"center\"><bold>013</bold></td><td align=\"center\"><bold>1.47</bold></td><td align=\"center\"><bold>1.27, 1.71</bold></td><td align=\"center\">135</td><td align=\"center\">0.93</td><td align=\"center\">0.76, 1.13</td></tr><tr><td align=\"center\"><bold>087</bold></td><td align=\"center\"><bold>1.47</bold></td><td align=\"center\"><bold>1.25, 1.73</bold></td><td align=\"center\">119</td><td align=\"center\">0.93</td><td align=\"center\">0.76, 1.12</td></tr><tr><td align=\"center\"><bold>016</bold></td><td align=\"center\"><bold>1.44</bold></td><td align=\"center\"><bold>1.29, 1.61</bold></td><td align=\"center\">066</td><td align=\"center\">0.92</td><td align=\"center\">0.69, 1.21</td></tr><tr><td align=\"center\"><bold>100</bold></td><td align=\"center\"><bold>1.44</bold></td><td align=\"center\"><bold>1.18, 1.73</bold></td><td align=\"center\">027</td><td align=\"center\">0.92</td><td align=\"center\">0.73, 1.15</td></tr><tr><td align=\"center\"><bold>007</bold></td><td align=\"center\"><bold>1.43</bold></td><td align=\"center\"><bold>1.25, 1.62</bold></td><td align=\"center\">102</td><td align=\"center\">0.92</td><td align=\"center\">0.75, 1.12</td></tr><tr><td align=\"center\"><bold>122</bold></td><td align=\"center\"><bold>1.42</bold></td><td align=\"center\"><bold>1.18, 1.69</bold></td><td align=\"center\">123</td><td align=\"center\">0.92</td><td align=\"center\">0.70, 1.18</td></tr><tr><td align=\"center\"><bold>011</bold></td><td align=\"center\"><bold>1.41</bold></td><td align=\"center\"><bold>1.21, 1.63</bold></td><td align=\"center\">097</td><td align=\"center\">0.92</td><td align=\"center\">0.65, 1.27</td></tr><tr><td align=\"center\"><bold>138</bold></td><td align=\"center\"><bold>1.38</bold></td><td align=\"center\"><bold>1.25, 1.54</bold></td><td align=\"center\">033</td><td align=\"center\">0.91</td><td align=\"center\">0.74, 1.11</td></tr><tr><td align=\"center\">065</td><td align=\"center\">1.35</td><td align=\"center\">0.98, 1.86</td><td align=\"center\">134</td><td align=\"center\">0.90</td><td align=\"center\">0.76, 1.07</td></tr><tr><td align=\"center\"><bold>091</bold></td><td align=\"center\"><bold>1.34</bold></td><td align=\"center\"><bold>1.09, 1.64</bold></td><td align=\"center\">039</td><td align=\"center\">0.89</td><td align=\"center\">0.63, 1.22</td></tr><tr><td align=\"center\">094</td><td align=\"center\">1.33</td><td align=\"center\">1.00, 1.60</td><td align=\"center\">038</td><td align=\"center\">0.89</td><td align=\"center\">0.70, 1.11</td></tr><tr><td align=\"center\"><bold>075</bold></td><td align=\"center\"><bold>1.32</bold></td><td align=\"center\"><bold>1.15, 1.51</bold></td><td align=\"center\">067</td><td align=\"center\">0.89</td><td align=\"center\">0.67, 1.14</td></tr><tr><td align=\"center\"><bold>081</bold></td><td align=\"center\"><bold>1.32</bold></td><td align=\"center\"><bold>1.14, 1.52</bold></td><td align=\"center\">062</td><td align=\"center\">0.88</td><td align=\"center\">0.70, 1.09</td></tr><tr><td align=\"center\"><bold>145</bold></td><td align=\"center\"><bold>1.29</bold></td><td align=\"center\"><bold>1.16, 1.44</bold></td><td align=\"center\">144</td><td align=\"center\">0.87</td><td align=\"center\">0.73, 1.04</td></tr><tr><td align=\"center\"><bold>069</bold></td><td align=\"center\"><bold>1.29</bold></td><td align=\"center\"><bold>1.13, 1.47</bold></td><td align=\"center\">133</td><td align=\"center\">0.87</td><td align=\"center\">0.70, 1.08</td></tr><tr><td align=\"center\">056</td><td align=\"center\">1.29</td><td align=\"center\">1.00, 1.56</td><td align=\"center\">052</td><td align=\"center\">0.87</td><td align=\"center\">0.68, 1.11</td></tr><tr><td align=\"center\">058</td><td align=\"center\">1.25</td><td align=\"center\">0.97, 1.59</td><td align=\"center\">048</td><td align=\"center\">0.86</td><td align=\"center\">0.63, 1.17</td></tr><tr><td align=\"center\"><bold>030</bold></td><td align=\"center\"><bold>1.24</bold></td><td align=\"center\"><bold>1.12, 1.38</bold></td><td align=\"center\">115</td><td align=\"center\">0.86</td><td align=\"center\">0.64, 1.14</td></tr><tr><td align=\"center\">071</td><td align=\"center\">1.23</td><td align=\"center\">1.00, 1.47</td><td align=\"center\">136</td><td align=\"center\">0.86</td><td align=\"center\">0.68, 1.07</td></tr><tr><td align=\"center\">008</td><td align=\"center\">1.23</td><td align=\"center\">0.95, 1.57</td><td align=\"center\">084</td><td align=\"center\">0.85</td><td align=\"center\">0.69, 1.05</td></tr><tr><td align=\"center\">018</td><td align=\"center\">1.23</td><td align=\"center\">1.00, 1.48</td><td align=\"center\">131</td><td align=\"center\">0.84</td><td align=\"center\">0.67, 1.05</td></tr><tr><td align=\"center\">042</td><td align=\"center\">1.22</td><td align=\"center\">1.00, 1.40</td><td align=\"center\">037</td><td align=\"center\">0.84</td><td align=\"center\">0.68, 1.03</td></tr><tr><td align=\"center\">029</td><td align=\"center\">1.21</td><td align=\"center\">1.00, 1.34</td><td align=\"center\">034</td><td align=\"center\">0.84</td><td align=\"center\">0.67, 1.03</td></tr><tr><td align=\"center\">061</td><td align=\"center\">1.20</td><td align=\"center\">1.00, 1.43</td><td align=\"center\">022</td><td align=\"center\">0.83</td><td align=\"center\">0.65, 1.05</td></tr><tr><td align=\"center\">107</td><td align=\"center\">1.20</td><td align=\"center\">0.96, 1.49</td><td align=\"center\">120</td><td align=\"center\">0.83</td><td align=\"center\">0.63,1.07</td></tr><tr><td align=\"center\">049</td><td align=\"center\">1.18</td><td align=\"center\">0.93, 1.49</td><td align=\"center\">139</td><td align=\"center\">0.82</td><td align=\"center\">0.67, 1.00</td></tr><tr><td align=\"center\">092</td><td align=\"center\">1.18</td><td align=\"center\">1.00, 1.35</td><td align=\"center\">124</td><td align=\"center\">0.81</td><td align=\"center\">0.65, 1.01</td></tr><tr><td align=\"center\">004</td><td align=\"center\">1.18</td><td align=\"center\">1.00, 1.38</td><td align=\"center\">026</td><td align=\"center\">0.81</td><td align=\"center\">0.64, 1.01</td></tr><tr><td align=\"center\">003</td><td align=\"center\">1.17</td><td align=\"center\">1.00, 1.36</td><td align=\"center\">093</td><td align=\"center\">0.80</td><td align=\"center\">0.60, 1.05</td></tr><tr><td align=\"center\">078</td><td align=\"center\">1.17</td><td align=\"center\">0.91, 1.47</td><td align=\"center\">023</td><td align=\"center\">0.80</td><td align=\"center\">0.60, 1.05</td></tr><tr><td align=\"center\">089</td><td align=\"center\">1.16</td><td align=\"center\">0.99, 1.36</td><td align=\"center\">060</td><td align=\"center\">0.80</td><td align=\"center\">0.58, 1.09</td></tr><tr><td align=\"center\">096</td><td align=\"center\">1.16</td><td align=\"center\">0.91, 1.45</td><td align=\"center\">082</td><td align=\"center\">0.80</td><td align=\"center\">0.64, 0.98</td></tr><tr><td align=\"center\">070</td><td align=\"center\">1.15</td><td align=\"center\">1.01, 1.31</td><td align=\"center\">147</td><td align=\"center\">0.80</td><td align=\"center\">0.64, 0.98</td></tr><tr><td align=\"center\">098</td><td align=\"center\">1.15</td><td align=\"center\">0.91, 1.44</td><td align=\"center\">148</td><td align=\"center\">0.79</td><td align=\"center\">0.62, 1.00</td></tr><tr><td align=\"center\">146</td><td align=\"center\">1.14</td><td align=\"center\">0.99, 1.30</td><td align=\"center\">032</td><td align=\"center\">0.79</td><td align=\"center\">0.63, 0.99</td></tr><tr><td align=\"center\">054</td><td align=\"center\">1.14</td><td align=\"center\">0.94, 1.36</td><td align=\"center\">127</td><td align=\"center\">0.79</td><td align=\"center\">0.62, 1.00</td></tr><tr><td align=\"center\">025</td><td align=\"center\">1.12</td><td align=\"center\">0.91, 1.38</td><td align=\"center\">129</td><td align=\"center\">0.78</td><td align=\"center\">0.60, 1.01</td></tr><tr><td align=\"center\">118</td><td align=\"center\">1.12</td><td align=\"center\">0.91, 1.36</td><td align=\"center\">080</td><td align=\"center\">0.78</td><td align=\"center\">0.64, 0.95</td></tr><tr><td align=\"center\">111</td><td align=\"center\">1.12</td><td align=\"center\">0.90, 1.37</td><td align=\"center\">116</td><td align=\"center\">0.77</td><td align=\"center\">0.58, 1.01</td></tr><tr><td align=\"center\">141</td><td align=\"center\">1.12</td><td align=\"center\">0.86, 1.43</td><td align=\"center\">035</td><td align=\"center\">0.77</td><td align=\"center\">0.59, 0.98</td></tr><tr><td align=\"center\">040</td><td align=\"center\">1.11</td><td align=\"center\">0.88, 1.39</td><td align=\"center\">020</td><td align=\"center\">0.76</td><td align=\"center\">0.61, 0.94</td></tr><tr><td align=\"center\">002</td><td align=\"center\">1.11</td><td align=\"center\">0.90, 1.35</td><td align=\"center\">019</td><td align=\"center\">0.75</td><td align=\"center\">0.59, 0.95</td></tr><tr><td align=\"center\">068</td><td align=\"center\">1.11</td><td align=\"center\">0.94, 1.31</td><td align=\"center\">126</td><td align=\"center\">0.75</td><td align=\"center\">0.62, 0.89</td></tr><tr><td align=\"center\">106</td><td align=\"center\">1.10</td><td align=\"center\">0.82, 1.46</td><td align=\"center\">112</td><td align=\"center\">0.74</td><td align=\"center\">0.55, 0.97</td></tr><tr><td align=\"center\">044</td><td align=\"center\">1.09</td><td align=\"center\">0.94, 1.26</td><td align=\"center\">117</td><td align=\"center\">0.73</td><td align=\"center\">0.62, 0.85</td></tr><tr><td align=\"center\">024</td><td align=\"center\">1.09</td><td align=\"center\">0.90, 1.31</td><td align=\"center\">114</td><td align=\"center\">0.73</td><td align=\"center\">0.55, 0.96</td></tr><tr><td align=\"center\">083</td><td align=\"center\">1.08</td><td align=\"center\">0.93, 1.25</td><td align=\"center\">128</td><td align=\"center\">0.70</td><td align=\"center\">0.57, 0.86</td></tr><tr><td align=\"center\">001</td><td align=\"center\">1.07</td><td align=\"center\">0.92, 1.24</td><td align=\"center\">130</td><td align=\"center\">0.70</td><td align=\"center\">0.54, 0.89</td></tr><tr><td align=\"center\">073</td><td align=\"center\">1.07</td><td align=\"center\">0.87, 1.30</td><td align=\"center\">121</td><td align=\"center\">0.70</td><td align=\"center\">0.53, 0.90</td></tr><tr><td align=\"center\">095</td><td align=\"center\">1.07</td><td align=\"center\">0.88, 1.28</td><td align=\"center\">047</td><td align=\"center\">0.70</td><td align=\"center\">0.55, 0.87</td></tr><tr><td align=\"center\">051</td><td align=\"center\">1.06</td><td align=\"center\">0.85, 1.31</td><td align=\"center\">113</td><td align=\"center\">0.69</td><td align=\"center\">0.54, 0.88</td></tr><tr><td align=\"center\">031</td><td align=\"center\">1.06</td><td align=\"center\">0.87, 1.29</td><td align=\"center\">109</td><td align=\"center\">0.69</td><td align=\"center\">0.54, 0.87</td></tr><tr><td align=\"center\">074</td><td align=\"center\">1.06</td><td align=\"center\">0.91, 1.22</td><td align=\"center\">059</td><td align=\"center\">0.67</td><td align=\"center\">0.52, 0.86</td></tr><tr><td align=\"center\">108</td><td align=\"center\">1.05</td><td align=\"center\">0.85,1.29</td><td align=\"center\">103</td><td align=\"center\">0.64</td><td align=\"center\">0.52, 0.79</td></tr><tr><td align=\"center\">104</td><td align=\"center\">1.04</td><td align=\"center\">0.83, 1.30</td><td align=\"center\">090</td><td align=\"center\">0.64</td><td align=\"center\">0.49, 0.83</td></tr><tr><td align=\"center\">046</td><td align=\"center\">1.04</td><td align=\"center\">0.87, 1.24</td><td align=\"center\">125</td><td align=\"center\">0.64</td><td align=\"center\">0.50, 0.81</td></tr><tr><td align=\"center\">110</td><td align=\"center\">1.04</td><td align=\"center\">0.89, 1.20</td><td align=\"center\">086</td><td align=\"center\">0.63</td><td align=\"center\">0.51, 0.76</td></tr><tr><td align=\"center\">041</td><td align=\"center\">1.04</td><td align=\"center\">0.83, 1.30</td><td align=\"center\">063</td><td align=\"center\">0.62</td><td align=\"center\">0.46, 0.82</td></tr><tr><td align=\"center\">132</td><td align=\"center\">1.03</td><td align=\"center\">0.91, 1.17</td><td align=\"center\">079</td><td align=\"center\">0.62</td><td align=\"center\">0.51, 0.74</td></tr><tr><td align=\"center\">064</td><td align=\"center\">1.02</td><td align=\"center\">0.82, 1.26</td><td align=\"center\">076</td><td align=\"center\">0.60</td><td align=\"center\">0.46, 0.77</td></tr><tr><td align=\"center\">101</td><td align=\"center\">1.00</td><td align=\"center\">0.80, 1.25</td><td align=\"center\">143</td><td align=\"center\">0.59</td><td align=\"center\">0.47, 0.74</td></tr><tr><td align=\"center\">028</td><td align=\"center\">1.00</td><td align=\"center\">0.83, 1.21</td><td align=\"center\">036</td><td align=\"center\">0.59</td><td align=\"center\">0.39, 0.85</td></tr></tbody></table></table-wrap>" ]

[ "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1476-072X-7-40-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>D</mml:mi><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M2\" name=\"1476-072X-7-40-i2\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>D</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M3\" name=\"1476-072X-7-40-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>D</mml:mi><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M4\" name=\"1476-072X-7-40-i2\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>D</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M5\" name=\"1476-072X-7-40-i3\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>O</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>t</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mn>4</mml:mn></mml:munderover><mml:mrow><mml:msub><mml:mi>O</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M6\" name=\"1476-072X-7-40-i4\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mtext>X</mml:mtext><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>t</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mn>4</mml:mn></mml:munderover><mml:mrow><mml:msub><mml:mtext>X</mml:mtext><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M7\" name=\"1476-072X-7-40-i5\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mi>log</mml:mi><mml:mo>⁡</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>α</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>X</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mo>=</mml:mo><mml:mi>log</mml:mi><mml:mo>⁡</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mi>r</mml:mi><mml:msub><mml:mi>r</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mo>=</mml:mo><mml:msub><mml:mi>ξ</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub><mml:mo>,</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M8\" name=\"1476-072X-7-40-i6\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>r</mml:mi><mml:msub><mml:mi>r</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:msup><mml:mi>e</mml:mi><mml:mrow><mml:msub><mml:mi>ξ</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:msup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula><italic>β</italic>^str= (<italic>β</italic>₁^str,......, <italic>β</italic>₁₄₈^str)^T,</disp-formula>", "<disp-formula><italic>β</italic>^unstr= (<italic>β</italic>₁^unstr,......, <italic>β</italic>₁₄₈^unstr)^T,</disp-formula>", "<disp-formula><italic>β</italic>^coh= (<italic>β</italic>₁^coh,......, <italic>β</italic>₂₀^coh)^T,</disp-formula>", "<disp-formula>S^ac= (S^ac_1,1,......, S^ac_148,20)^T.</disp-formula>", "<disp-formula><italic>β</italic>^str~<italic>N</italic>(0,(<italic>μ</italic>_{<italic>str</italic>}<italic>σ</italic>_{<italic>str</italic>})^-1),</disp-formula>", "<disp-formula><italic>β</italic>^unstr~<italic>N</italic>(0,(<italic>μ</italic>_{<italic>unstr</italic>}<italic>I</italic>₁₄₈)^-1),</disp-formula>", "<disp-formula><italic>β</italic>^coh~<italic>N</italic>(0,(<italic>μ</italic>_{<italic>coh</italic>}<italic>σ</italic>_{<italic>coh</italic>})^-1),</disp-formula>", "<disp-formula>S^ac~<italic>N</italic>(0,(<italic>μ</italic>_{<italic>ac</italic>}<italic>σ</italic>_{<italic>ac</italic>})^-1).</disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M9\" name=\"1476-072X-7-40-i7\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mrow><mml:mn>148</mml:mn></mml:mrow></mml:munderover><mml:mrow><mml:msubsup><mml:mi>β</mml:mi><mml:mi>i</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mi>t</mml:mi><mml:mi>r</mml:mi></mml:mrow></mml:msubsup></mml:mrow></mml:mstyle><mml:mo>=</mml:mo><mml:mn>0</mml:mn><mml:mo>,</mml:mo></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>j</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mrow><mml:mn>20</mml:mn></mml:mrow></mml:munderover><mml:mrow><mml:msubsup><mml:mi>β</mml:mi><mml:mi>j</mml:mi><mml:mrow><mml:mi>c</mml:mi><mml:mi>o</mml:mi><mml:mi>h</mml:mi></mml:mrow></mml:msubsup></mml:mrow></mml:mstyle><mml:mo>=</mml:mo><mml:mn>0.</mml:mn></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M10\" name=\"1476-072X-7-40-i8\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>D</mml:mi><mml:mi>I</mml:mi><mml:mi>C</mml:mi><mml:mo>=</mml:mo><mml:mover accent=\"true\"><mml:mi>D</mml:mi><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover><mml:mo>+</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mover accent=\"true\"><mml:mi>D</mml:mi><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover><mml:mo>−</mml:mo><mml:mi>D</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M11\" name=\"1476-072X-7-40-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>D</mml:mi><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M12\" name=\"1476-072X-7-40-i2\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>D</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M13\" name=\"1476-072X-7-40-i9\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mo stretchy=\"false\">[</mml:mo><mml:mover accent=\"true\"><mml:mi>D</mml:mi><mml:mo stretchy=\"true\">¯</mml:mo></mml:mover><mml:mo>−</mml:mo><mml:mi>D</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>¯</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo><mml:mo stretchy=\"false\">]</mml:mo></mml:mrow></mml:semantics></mml:math></inline-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>Panchayats are arranged in descending order of median RR values for sake of comparison. Significant panchayats and RR greater than one are highlighted.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1476-072X-7-40-1\"/>", "<graphic xlink:href=\"1476-072X-7-40-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Cholera has claimed many lives throughout history and it continues to be a global threat [##REF##9583431##1##], especially in countries in Africa. Between 1999 and 2005, there were over 1 million reported cholera cases and over 28,000 reported deaths worldwide. Africa alone accounted for about 90% of the cases and 96% of the deaths worldwide [##UREF##0##2##, ####UREF##1##3##, ##UREF##2##4##, ##UREF##3##5##, ##UREF##4##6##, ##UREF##5##7##, ##UREF##6##8####6##8##]. Cholera has gained both global and public health attention due to its mode of transmission and severity. For instance it has become one of the most researched communicable diseases. The disease is also listed as one of three internationally quarantinable diseases by the World Health organization (WHO), along with plague and yellow fever [##UREF##7##9##]. In addition to human suffering and lives loss, cholera outbreak causes panic, disrupts the social and economic structure and can impede development in the affected communities [##UREF##4##6##].</p>", "<p>Cholera reached West Africa and Ghana during the seventh pandemic [##REF##5083668##10##, ####REF##4561957##11##, ##REF##1235746##12##, ##REF##7256448##13####7256448##13##]. The disease has been endemic in Ghana since its introduction in the 1970's [##UREF##8##14##]. From 1999 to 2005, a total of 26,924 cases and 620 deaths were officially reported to the WHO [##UREF##0##2##, ####UREF##1##3##, ##UREF##2##4##, ##UREF##3##5##, ##UREF##4##6##, ##UREF##5##7##, ##UREF##6##8####6##8##]. Although the disease is transmitted mainly through contaminated water and food, several demographic and geographic factors can predispose an individual or groups of individuals to infection. For example, increase in population density can strain existing sanitation systems, thus putting people at increased risk of contracting cholera [##REF##9004375##15##,##REF##1500643##16##]. Once the bacterial, <italic>V. cholerae</italic>, are present in water in sufficient dose, an outbreak can trigger and propagate depending on demographic factors such as population density [##REF##12135643##17##,##REF##12220086##18##], urbanization, and overcrowding [##UREF##9##19##]. In developing countries like Ghana, high incidence of cholera seems to predominate in the urban communities, and this is primarily due to high overcrowding and unsanitary living conditions in urban communities. While cholera is prevalent in low urban communities in certain geographical areas like Mexico [##UREF##9##19##], the disease has predominated in urban and overcrowded communities in Ghana. Intermittent water supply coupled with indiscriminate sanitation practices in urban communities in Ghana puts inhabitants at risk of contracting cholera.</p>", "<p>Studies on diarrhea related diseases in Ghana [##UREF##10##20##] so far have focused solely on the biological factors and characteristics of the individuals affected. Although such studies are very useful, they omit the spatial and regional variations of the critical risk factors. Such studies also fail to define territories at high risk. Since health levels vary substantially between different regions, it is essential to characterize these regional variations and identify areas with an accumulation of health problems for epidemiological research, and to allow appropriate public health policy decisions [##UREF##11##21##,##REF##10098991##22##]. Advances in Geographical Information Systems (GIS) technology provide new opportunities for environmental epidemiologist to study associations between demographic and environmental exposures and the spatial distribution of diseases [##UREF##12##23##]. GIS has been used in the surveillance and monitoring of vector-borne diseases [##REF##7943544##24##,##REF##7604918##25##], water-borne diseases [##UREF##13##26##], in environmental health [##REF##8017545##27##, ####UREF##14##28##, ##UREF##15##29####15##29##], analysis of disease policy and planning [##REF##8098147##30##]. Several cholera studies [##REF##12135643##17##, ####REF##12220086##18##, ##UREF##9##19####9##19##,##UREF##16##31##, ####UREF##17##32##, ##UREF##18##33##, ##UREF##19##34####19##34##] have also employed GIS technologies. This study focuses on the application of a GIS based spatial analyses and statistical technology to study the spatial patterns of cholera, identify territories of high risk, and determine demographic risk factors that contribute to high rates of cholera. No study so far has looked at the spatial patterns of cholera in Ghana. There is therefore no information and/or knowledge about its spatial patterns and demographic correlates in Ghana. Studying the spatial and demographic patterns of cholera in Ghana will prove useful for health officials and policymakers to make appropriate planning and resource allocation.</p>" ]

[ "<title>Materials and methods</title>", "<title>Research methodology</title>", "<p>An important part of health-needs assessment is the identification of high risk areas for a disease because understanding the characteristics of high risk areas may indicate what is needed to improve health care provision [##UREF##20##35##]. Several disease clustering techniques have been developed to define territories of high risk [##UREF##21##36##, ####REF##7644860##37##, ##UREF##22##38####22##38##]. However, using clustering detection technique to define high risk territories is only an exploratory technique to locate clusters, but does not establish a relation between the disease and risk factors. In this study, Moran's I for spatial autocorrelation was computed to ascertain evidence of cholera clustering. A global Bayesian smoothing technique was employed to smooth the crude rates of cholera, and then mapped to determine the spatial distribution of cholera. The districts in the region were classified into strata of districts based on demographic indicators. Population-based incidence rate ratios were then computed for each stratum to determine territories of high risk. The Extended Mantel-Haenszel <italic>Chi Square </italic>for trend analyses and associated <italic>P values </italic>(one degree of freedom) were also computed to determine the trend between the demographic factors and <italic>V. cholerae </italic>infection [##UREF##23##39##].</p>", "<title>The study area</title>", "<title>History of Cholera in Ghana</title>", "<p>On 1^stSeptember, 1970 a Togolese national in transit at the Kotoka International Airport from Conakry, Guinea, collapsed and was found to have cholera [##UREF##8##14##]. This was the announcement of the arrival of the seventh pandemic of cholera in Ghana. However, an outbreak did not begin from then until it was smuggled into the country through fishing [##UREF##24##40##]. At that time, some Ghanaians went for fishing in the waters of Togo, Liberia and Guinea. One of the fishermen died and although a sanitary cordon had been placed on our borders, his family smuggled the corpse into his home town, and the usual burial rites were performed. It was after this that cholera began to spread along the shores of Ghana. The disease swept through many coastal villages in epidemic proportions. It kept on spreading and by July 1971, Ashanti region began to report cases, indicating that cholera was spreading across the country [##UREF##24##40##]. During those periods, reported outbreaks were investigated, treatment camps were set, people were vaccinated against cholera, and the population was also educated on measures to prevent the spread of the disease. However, all these attempts to prevent cholera from taking root in Ghana failed. Since then cholera has existed in both epidemic and endemic forms in Ghana.</p>", "<title>Location and demography of the study area</title>", "<p>The Ashanti Region is centrally located in the middle belt of Ghana. It lies between longitudes 0° 9'W and 2° 15'W, and latitudes 5° 30'N and 7° 27'N. The region shares boundaries with four of the ten political regions, Brong-Ahafo region in the north, Eastern region in the east, Central region in the south and Western region in the South-west (See Figure ##FIG##1##1##). Ashanti region occupies a total land area of 24,389 km²representing 10.2% of the total land area of Ghana. Ashanti region has a population density of 148.1 persons per km², compared with a national average of about 80 persons per km². The region consists of 18 administrative districts. Kumasi, which is the capital, is the most populous district, and the only district that has gained a metropolitan status in the region. The 2000 census recorded the region's population as about 3.5 million people, representing 19.1 per cent of the country's population. The urban population (51.3%) in the region exceeds that of the rural population (48.7%). The region is currently the second most urbanized in the country after Greater Accra (87.7%), the national capital. The housing stock in the region is 329,478, of which about 37% are in urban areas and 63% in rural areas. This is in contrast to the 17.4% of houses in urban, and 82.6% in rural areas in 1970. The total stock also represents an increase of 86.8% over the stock in 1984. The relative increase in the proportion of urban housing is a reflection of the increase in urbanization, and perhaps overcrowding. Due to the high housing cost within the urban districts in the region, lots of slummy and/or squatter settlements are created. However, such areas have poor sanitation systems, and perceived to be niches where cholera outbreaks begin.</p>", "<title>Case definition of cholera</title>", "<p>In Ghana, a case definition of cholera is based on the WHO's definition which depends on whether or not the presence of cholera has been demonstrated in the area. According to the WHO [##UREF##25##41##] guidance on formulation of national policy on the control of cholera, in an area where the disease is not known to be present a case of cholera should be suspected, when a patient, 5 years of age or older develops severe dehydration or dies from acute watery diarrhea, or where an epidemic is occurring, a patient, 5 years of age or older develops acute watery diarrhea, with or without vomiting. The first case of cholera however was confirmed by bacteriological tests (personal communication with KMHD director). In this study, only cholera cases made known to the Disease Control Unit (DCU) through reporting facilities such as community volunteers, community clinics, and hospitals were used. In Ghana it is mandatory for all reporting facilities (i.e. hospitals, clinics, and community volunteers) to report weekly cholera cases to the DCU. Although hospitals are scarcely found in many communities in the districts, almost all communities in the districts have access to clinics, and community volunteers who monitor all communicable diseases. The communicable diseases surveillance network is purposely established from community level to district level to ensure effective surveillance of all communicable diseases (personal communication with head of DCU, Ashanti region).</p>", "<title>Spatial data preparation and cartographic display of cholera</title>", "<p>Topographic map of the study area at a scale of 1:2500 obtained from the planning unit of the Kumasi Metropolitan Assembly was digitized using ArcGis version 9.2. This software was developed by Environmental System Research Institute (ESRI). Before digitizing, the map was georeferenced (by defining the X and Y coordinates of corner points of the map) into a UTM coordinates system. The main boundary and the 18 districts within the study area were digitized as polygon features. Reported cases of cholera over the period 1997–2001, obtained from the DCU, Ashanti region, were entered as attributes of the districts. The cumulative incidence rates of cholera were calculated for each of the 18 districts by including all cases over the period 1997–2001. The population database was obtained from the 2000 Population and Housing Census of Ghana [##UREF##26##42##]. This census was conducted by the National Statistical Service of Ghana.</p>", "<p>Disease mapping is useful for initial exploration of relationships between exposure and the disease. The raw cumulative incidence rates were smoothed using global Empirical Bayesian Smoothing (EBS) technique. This was to get rid of variance instability as result of heterogeneity in cholera cases and population data (small number problem). The EBS technique consists of computing a weighted average between the raw rate for each district and the regional average, with weights proportional to the underlying population at risk [##UREF##27##43##]. In effect, districts with relatively small population will tend to have their rates adjusted considerably, whereas for districts with relatively large population the rates will barely change. The resulting smoothed rates were then mapped using GIS. The cutoff points for classification were based on the Jenk's classification technique as employed in ArcGis version 9.2.</p>", "<title>Statistical analyses</title>", "<p>In this study, spatial autocorrelation statistic was used to measure the correlation among neighboring observations in a pattern and the levels of spatial clustering among neighboring districts [##UREF##28##44##,##UREF##29##45##].</p>", "<p>Global Moran's I statistic, which is similar to the Pearson correlation coefficient [##UREF##30##46##], was calculated as:</p>", "<p></p>", "<p>where <italic>N </italic>is the number of districts, <italic>w</italic>_{<italic>ij </italic>}is the element in the spatial weights matrix corresponding to the observation pair <italic>i</italic>, <italic>j</italic>,. Also <italic>x</italic>_{<italic>i </italic>}and <italic>x</italic>_{<italic>j </italic>}are observations for areas <italic>i </italic>and <italic>j </italic>with mean u.</p>", "<p></p>", "<p>Since the weights were row-standardized ∑<italic>w</italic>_{<italic>ij </italic>}= 1. The first step in the analysis of spatial autocorrelation is to construct a spatial weights file that contains information on the neighborhood structure for each location. A first order rook continuity weight file was constructed according to districts who share common boundaries. All spatial autocorrelation analyses were computed using GeoDa 0.9.5-i software [##UREF##31##47##,##UREF##27##43##]. A significance test against the null hypothesis of no spatial autocorrelation through a permutation procedure of 999 Monte Carlo replications was used to test for the significance of the statistic. Spatial autocorrelation was calculated for the years 1998, 1999, 2001 and 1997–2001. Because very few cholera cases were reported in 1997 and 2000, autocorrelation was not calculated for these years.</p>", "<p>Population based rate ratios were computed for strata of districts grouped by the following variables;</p>", "<title>Proximity to Kumasi, the most urbanized city</title>", "<p>Proximity to Kumasi Metropolis (the most urbanized district in Ashanti region) was categorized into three strata based on the order of spatial neighborhood and/or adjacency. First order neighbors were defined as districts sharing common boundaries with the Kumasi Metropolis. Second order neighbors were defined as districts sharing common boundaries with the first order neighbors, whereas third order neighbors were defined as district sharing common boundaries with the second order neighbors. All spatial neighbors were defined using ArcGIS version 9.2. Population based rate ratios were computed for each strata by taken the stratum of third order neighbors as reference (baseline).</p>", "<title>Urbanization level</title>", "<p>The indicator for urbanization was population based. Each district within the region is made up of localities and/or communities. A locality with a population of 5,000 or more was classified as urban, and less than 5,000 as rural. This is the criteria given by the Ghana Statistical Service [##UREF##26##42##]. The urbanization level for each district was then computed as the proportion of a district's population residing in localities and/or communities of ≥ 5,000 people in the year 2000. Three urbanization strata were determined, each representing a quartile of districts. Each quartile was composed of six districts. Population-based incidence rate ratios were calculated for each stratum by taking that of lower urbanization as reference.</p>", "<title>Overcrowding</title>", "<p>The indicator for overcrowding was based on four variables: (1) Population density; (2) population per house; (3) single room occupancy (i.e., percentage of households living in single rooms) and; (4) households per house. A household was defined as \"a person or group of persons who live together in the same house or compound, sharing the same house-keeping arrangements and are catered for as one unit\". Each variable was standardized to have a mean of zero and a standard deviation of one. The variables were combined to form a single index of risk, called overcrowding index (OI). The OI for each district was computed as the mean of the algebraic sum of the standardized values of the four variables. The assumption was that the variables carry equal weights. The Jenk's method of classification was used to classify OI into three strata of districts. Population-based incidence rate ratios were calculated for each stratum by taking that of lower OI as reference (baseline).</p>", "<p>Taking into account urbanization stratum and overcrowding level, neighborhood population-based double stratification analyses were performed to explore whether cholera incidence rate was associated with the order of neighborhood with Kumasi Metropolis.</p>", "<p>The Extended Mantel-Haenszel <italic>Chi Square </italic>(<italic>χ</italic>²) for trend analysis and 95% confidence intervals for rate ratios were computed to explore the relationship between cholera incidence rates and the variables under study [##UREF##22##38##]. Given a series of proportions representing increasing or decreasing exposure (risk factor) and numbers of affected and non affected people in each stratum (group), the Mantel-Haenszel's extension tests whether the rates in successive groups increase or decrease when compared to the baseline (reference). The results of such test include the rate ratios of successive exposure levels, and <italic>χ</italic>²and <italic>P</italic>-value (one degree of freedom). A <italic>P</italic>-value less than 0.05 may be taken as reasonable indication of trend in the rates of successive levels compared with the baseline (reference) [EpiInfo version 6.03 help manual]. EpiInfo version 6.03 was used for the computation of the rate ratios, 95% confidence intervals, and the <italic>χ</italic>²for trend analysis.</p>" ]

[ "<title>Results and analyses</title>", "<p>The extent to which neighboring values are correlated was measured using Global Moran's index. A significance assessment through a permutation procedure was implemented to determine the significance of the computed Moran's index. There is a positive and statistically significant spatial autocorrelation for cumulative incidence rate of cholera from 1997 to 2001 (Moran's I = 0.271, <italic>P-value </italic>= 0.0009). Moreover, a spatial autocorrelation statistic computed for each of the periods 1998, 1999 and 2001 were statistically significant (<italic>P-value </italic>&lt; 0.05) for Moran's I (See Table ##TAB##0##1##). This reflects clustering of high rates of cholera at the central part of the region (See Figure ##FIG##0##2##).</p>", "<p>Figure ##FIG##0##2## also shows the Empirical Bayesian smoothed rates of cholera. A visual inspection reflects clustering of high rates of cholera at areas surrounding Kumasi Metropolis and its adjoining neighbors (See Figure ##FIG##0##2d##). Moreover, clustering of high rates of cholera was persistent at Kumasi Metropolis and its adjoining neighbors in the years 1998, 1999, and 2001 (See Figures ##FIG##0##2a, 2b, 2c##).</p>", "<p>The rate ratio within each stratum was computed using EpiInfo, and the results shown in Tables ##TAB##1##2##, ##TAB##2##3##, ##TAB##3##4##, ##TAB##4##5##, ##TAB##5##6##. The cumulative incidence rate of cholera was 22 times higher in the first order neighborhood stratum than the third order neighborhood stratum. The cumulative incidence rate in the second order neighborhood (with Kumasi Metropolis) was not very high (1.4 times) compared to the third order neighborhood (See Table ##TAB##1##2##). Cholera incidence rate in the most urban stratum was about 20 times higher than the least urban stratum (Table ##TAB##2##3##), while the incidence rate in the most overcrowded stratum was about 30 times higher than the less overcrowded stratum (Table ##TAB##3##4##). A <italic>Chi square </italic>for trend analysis reflected a direct spatial relationship between cholera and urbanization (Table ##TAB##2##3##: <italic>χ</italic>²= 2995.5, <italic>P </italic>= 0.000001), overcrowding (Table ##TAB##3##4##: <italic>χ</italic>²= 1757.2, <italic>P </italic>= 0.000001), and an inverse relationship between cholera and order of neighborhood (Table ##TAB##1##2##: <italic>χ</italic>²= 831.38, <italic>P </italic>= 0.000001).</p>", "<p>The cumulative incidence rate of cholera was higher in the adjoining (first order neighborhood) stratum than in the non-adjoining stratum (second and third order neighborhoods) within each urbanization stratum (Table ##TAB##4##5##). Similar pattern was observed in the high and medium overcrowding strata (Table ##TAB##5##6##). No adjoining district was within the least overcrowded stratum.</p>" ]

[ "<title>Discussion</title>", "<p>This study takes advantage of the advancements in Geographic information system such as spatial disease mapping and smoothing, exploratory spatial data analysis such as spatial autocorrelation, and spatial statistical techniques to identify demographic risk factors of cholera. The extent to which neighboring values are correlated was measured using Global Moran's index for spatial autocorrelation. All autocorrelation analyses suggest significant spatial clustering of cholera with positive Moran's index (see Table ##TAB##0##1##). This non random distribution also suggests spatial clustering of high rates of cholera incidence at the central part of the region, and low rates at the peripheries (See Figure ##FIG##0##2##). This is also shown by the high rate ratio of 22 times in the first order neighborhood stratum (i.e. direct neighbors with Kumasi Metropolis, See Table ##TAB##1##2##). Visual inspections of the EBS maps also suggest possible sustained transmission of cholera at districts within the central part of the region (see Figures ##FIG##0##2a, 2b, 2c, 2d##). These patterns are plausible largely because of two main reasons. (1) <italic>Demographic status</italic>: Kumasi is the most urbanized and highly commercialized district in Ashanti region, and therefore there is always a high daily influx of traders and civil workers from neighboring districts to Kumasi Metropolis. Such a high daily influx strain existing sanitation systems, thereby putting people at increased risk of cholera transmission. Also, the rural poor most often migrate to city centers with the hope of better life. However, due to the high cost of housing, such migrants settle at slummy and/or squatter areas where environmental sanitation is poor. This largely explains the high <italic>northern population </italic>(inhabitants from the northern sector of Ghana; which is the most deprived sector) within Kumasi Metropolis (2) <italic>Geographic location</italic>: Kumasi Metropolis is the central nodal district of Ghana, and therefore, all road networks linking the northern sector and the southern sector of Ghana pass through Kumasi. There is the high probability of stoppage and transit by travelers, resulting in a high daily population increase and overcrowding at city centers.</p>", "<p>This study has also shown that high urbanization and overcrowding are the most important predictors of cholera in Ashanti region, Ghana (See Tables ##TAB##2##3##, ##TAB##3##4##, ##TAB##4##5##, ##TAB##5##6##). Although cholera is transmitted mainly through contaminated water or food, sanitary conditions in the environment play an important role since the <italic>V. cholerae </italic>bacterium survives and multiplies outside the human body and can spread rapidly where living conditions are overcrowded and water sources unprotected and where there is no safe disposal of solid waste, liquid waste, and human faces [##UREF##7##9##]. These conditions are met in highly urbanized communities in Ashanti Region. The high rate of urbanization has led to the high level of overcrowding, which necessarily results in shorter disease transmission path. This is shown by the very high rate ratios within the high urban (RR = 19.86) and high OI (RR = 31.11) stratum (See Tables ##TAB##2##3## and ##TAB##3##4##). In fact, the DCU has attributed outbreaks of cholera in urban communities to poor waste management and sanitation systems. In Ghana, urban population growth rate of about 4.3% has outstripped the overall national population growth rate of about 2.7%. The proportion of the population residing in urban areas rose from 32% in 1984 to 43.8% in 2000 [##UREF##32##48##]. Such high urbanization rate strain existing resources meant for providing better sanitation systems and potable water. Inadequate sanitation systems coupled with intermittent supply of pipe borne water in urban communities puts the population at risk of cholera. Surface water pollution is particularly found to be worse where rivers pass through urban and overcrowded cities, and the commonest contamination is from human excreta and sewage [##UREF##26##42##]. Due to the cosmopolitan (multi-ethnic) nature of the urban cities in Ghana [##UREF##33##49##], the traditional laws which were used to protect water bodies form fecal pollution are no longer adhered (Traditionally, it is a taboo to defecate or dispose waste in a water body). Therefore, defecating and dumping of waste in and at the banks of surface water bodies has become a common practice in most urban communities. However, urban inhabitants resort to such polluted water bodies for various household activities like cooking and washing during periods of water shortages.</p>", "<p>Further, the rate of slums and/or squatter formation in urban communities is high due to the high rate of migration and population redistribution. Inhabitants living at slums and/or squatter settlements are generally poor, and face problems including access to potable water and sanitation. The urban poor (slums and squatter settlers), are worse off than their rural counterparts in terms of access and affordability to safe drinking water and sanitation. In many cases public utilities providers (e.g. Water distribution) legally fail to serve the urban poor living in slums due to factors regarding land tenure system, technical and service regulations, and city development plans. Most slums and/or squatter settlements are also located at low lying areas susceptible to flooding. Unfavorable topography, soil, and hydro-geological conditions make it difficult to achieve and maintain high sanitation standards among populations living in these territories [##UREF##9##19##].</p>", "<p>This study has shown the capabilities of spatial analysis and GIS in analyzing geographically referenced health data in Ghana. Moreover, the study has also proven that the demographic risk factors of cholera may not be the same in every geographical area or country. For example, Barroto and Martnez-Piedra [##UREF##9##19##] identified low urbanization as one of the most important ecologic predictors of cholera in Mexico, a Latin American country. However, the results of this study show that high urbanization positively correlates with high cholera incidence.</p>", "<p>Although some findings of this research reaffirms the already known hypothesis of cholera, we present the possibility of using GIS and spatial statistical tools for health research in this study area where GIS application in the health sector has not been extensively utilized.</p>", "<title>Limitations of this research</title>", "<p>The results of the Extended Mantel-Haenszel <italic>Chi Square </italic>for trend analyses should be interpreted with caution. The number of cholera cases reported to the DCU may only be a fraction of cases that actually occurred, especially in lowly urbanized districts (or rural areas) of the country where level of education is extremely low. It has been suggested that educational level indirectly determines a person's healthcare seeking behavior [##REF##12220086##18##].</p>", "<p>The spatial scale of the data may invariably affect the results of the spatial analysis. The areas identified as high risk of cholera are generally large areas defined by administrative boundaries. In such a large spatial scale, it is difficult to demonstrate the actual risk of cholera within a smaller group of people. A more detailed study at a smaller spatial scale is therefore required to assess the accurate individual or smaller groups of individuals' exposure levels. The spatial autocorrelation analysis should be interpreted with caution due the different shapes and sizes of the districts [##UREF##30##46##].</p>" ]

[ "<title>Conclusion</title>", "<p>This study has demonstrated the use of spatial statistical analysis and GIS to map hotspots, and the spatial dependency of cholera distribution within a population. Through spatial statistical procedures, non-randomness in the distribution of cholera and the identification of unusual concentration of cholera incidence has been defined. This therefore prompts health planners in the country to take a critical look at these risk areas, and make appropriate health planning and resource allocation. In conclusion, the results of this research suggest that high urbanization, high overcrowding, and neighborhood with Kumasi Metropolis are the most important predictors of cholera in Ashanti region-Ghana. It is therefore necessary that health officials and policy makers reasonably improve their surveillance systems to prepare for the possibility of sustained transmission should an infection be introduced. Since this research is the first of its kind in Ghana, a more detailed research is required to consider factors like access to safe drinking water, and availability of waste disposal systems to thoroughly evaluate the risk of cholera in the region.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Cholera has claimed many lives throughout history and it continues to be a global threat, especially in countries in Africa. The disease is listed as one of three internationally quarantinable diseases by the World Health organization, along with plague and yellow fever. Between 1999 and 2005, Africa alone accounted for about 90% of over 1 million reported cholera cases worldwide. In Ghana, there have been over 27000 reported cases since 1999. In one of the affected regions in Ghana, Ashanti region, massive outbreaks and high incidences of cholera have predominated in urban and overcrowded communities.</p>", "<title>Results</title>", "<p>A GIS based spatial analysis and statistical analysis, carried out to determine clustering of cholera, showed that high cholera rates are clustered around Kumasi Metropolis (the central part of the region), with Moran's Index = 0.271 and <italic>P </italic>&lt; 0.001. Furthermore, A Mantel-Haenszel <italic>Chi square </italic>for trend analysis reflected a direct spatial relationship between cholera and urbanization (<italic>χ</italic>²= 2995.5, <italic>P </italic>&lt; 0.0001), overcrowding (<italic>χ</italic>²= 1757.2, <italic>P </italic>&lt; 0.0001), and an inverse relationship between cholera and order of neighborhood with Kumasi Metropolis (<italic>χ</italic>²= 831.38, <italic>P </italic>&lt; 0.0001).</p>", "<title>Conclusion</title>", "<p>The results suggest that high urbanization, high overcrowding, and neighborhood with Kumasi Metropolis are the most important predictors of cholera in Ashanti region.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>FBO carried out the research and drafted the manuscript. AAD guided the research and reviewed the manuscript. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>We extend our sincere appreciation to Mr. Douglas Brenya, from the Disease Control Unit-Ashanti Region, for providing all the necessary data and background information for this research.</p>" ]

[ "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>EBS Smoothed Rates of Cholera for 1998 (2a), 1999(2b), 2001(2c), and 1997–2001(2d)</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>District Map of Ashanti region, Ghana.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Moran's I for spatial autocorrelation computed for cumulative incidence of cholera (1997–2001), and the specific years of cholera outbreaks (1998, 1999, and 2001).</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\"><bold>Year</bold></td><td align=\"center\"><bold>Moran's I</bold></td><td align=\"center\"><bold><italic>P</italic></bold><bold>-value</bold></td></tr></thead><tbody><tr><td align=\"center\">1997–2001</td><td align=\"center\">0.271</td><td align=\"center\">0.0001</td></tr><tr><td align=\"center\">1998</td><td align=\"center\">0.331</td><td align=\"center\">0.004</td></tr><tr><td align=\"center\">1999</td><td align=\"center\">0.181</td><td align=\"center\">0.040</td></tr><tr><td align=\"center\">2001</td><td align=\"center\">0.240</td><td align=\"center\">0.010</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Cholera incidence rate and population-based rate ratios by strata of districts classified according to order of neighborhood and/or adjacency to Kumasi Metropolis, 1997–2001.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Order of neighborhood</bold></td><td align=\"left\"><bold>Cholera</bold><break/><bold> cases</bold></td><td align=\"left\"><bold>Sub</bold><break/><bold> Population</bold></td><td align=\"left\"><bold>Rate (per 10000)</bold></td><td align=\"left\"><bold>Rate ratio (95%CI)</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>First order</bold></td><td align=\"left\">562</td><td align=\"left\">672482</td><td align=\"left\">8.357101008</td><td align=\"left\">22 (14.14–35.27)</td></tr><tr><td align=\"left\"><bold>Second order</bold></td><td align=\"left\">69</td><td align=\"left\">1307843</td><td align=\"left\">0.527586262</td><td align=\"left\">1.4 (0.84–2.36)</td></tr><tr><td align=\"left\"><bold>Third order</bold></td><td align=\"left\">21</td><td align=\"left\">558356</td><td align=\"left\">0.376104134</td><td align=\"left\">Reference</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Cholera incidence rate and population-based rate ratios by strata of districts classified according to level of urbanization, 1997–2001.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Urbanization (%)</bold></td><td align=\"left\"><bold>Cholera</bold><break/><bold> cases</bold></td><td align=\"left\"><bold>Sub</bold><break/><bold> Population</bold></td><td align=\"left\"><bold>Rate (per 10000)</bold></td><td align=\"left\"><bold>Rate ratio(95%CI)</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Low</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">0.0–16.5</td><td align=\"left\">126</td><td align=\"left\">1065586</td><td align=\"left\">1.182447968</td><td align=\"left\">Reference</td></tr><tr><td align=\"left\"><bold>Medium</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">19.2–35.6</td><td align=\"left\">357</td><td align=\"left\">848903</td><td align=\"left\">4.205427475</td><td align=\"left\">3.56 (2.89–4.38)</td></tr><tr><td align=\"left\"><bold>High</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">38.9–100</td><td align=\"left\">2748</td><td align=\"left\">1170270</td><td align=\"left\">23.48176062</td><td align=\"left\">19.86 (16.55–23.840</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Cholera incidence rate and population-based rate ratios by strata of districts classified according to level of overcrowding, 1997–2001.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Overcrowding index</bold></td><td align=\"left\"><bold>Cholera</bold><break/><bold> cases</bold></td><td align=\"left\"><bold>Sub</bold><break/><bold> Population</bold></td><td align=\"left\"><bold>Rate (per 10000)</bold></td><td align=\"left\"><bold>Rate ratio(95%CI)</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Low</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">(-4.5 &lt; OI &lt; -1.66)</td><td align=\"left\">51</td><td align=\"left\">1048225</td><td align=\"left\">0.49</td><td align=\"left\">Reference</td></tr><tr><td align=\"left\"><bold>Medium</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">(-0.73 &lt; OI &lt; -0.37)</td><td align=\"left\">310</td><td align=\"left\">764556</td><td align=\"left\">4.05</td><td align=\"left\">8.33 (6.14–11.34)</td></tr><tr><td align=\"left\"><bold>High</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">(-0.25 &lt; OI &lt; 3.20)</td><td align=\"left\">2870</td><td align=\"left\">1896170</td><td align=\"left\">15.14</td><td align=\"left\">31.11 (23.40–41.46)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Cholera incidence rate and population-based rate ratios by strata of districts classified according to urbanization level and order of neighborhood and/or adjacency to Kumasi Metropolis, 1997–2001.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Urbanization (%)</bold></td><td align=\"left\"><bold>Neighborhood</bold></td><td align=\"left\"><bold>Cholera cases</bold></td><td align=\"left\"><bold>Population</bold></td><td align=\"left\"><bold>Rate(per10000)</bold></td><td align=\"left\"><bold>Rate ratio(95% CI)</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Low</bold></td><td align=\"left\">Adjoining</td><td align=\"left\">84</td><td align=\"left\">146028</td><td align=\"left\">5.75</td><td align=\"left\">12.59 (8.57–18.55)</td></tr><tr><td align=\"left\">0.0–16.5</td><td align=\"left\">Non-adjoining</td><td align=\"left\">42</td><td align=\"left\">919558</td><td align=\"left\">0.46</td><td align=\"left\">Reference</td></tr><tr><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>Medium</bold></td><td align=\"left\">Adjoining</td><td align=\"left\">330</td><td align=\"left\">361786</td><td align=\"left\">9.12</td><td align=\"left\">16.46 (10.96–24.88)</td></tr><tr><td align=\"left\">19.2–35.6</td><td align=\"left\">Non-adjoining</td><td align=\"left\">27</td><td align=\"left\">487117</td><td align=\"left\">0.55</td><td align=\"left\">Reference</td></tr><tr><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>High</bold></td><td align=\"left\">Adjoining</td><td align=\"left\">148</td><td align=\"left\">164668</td><td align=\"left\">8.99</td><td align=\"left\">19.67 (12.22–31.95)</td></tr><tr><td align=\"left\">38.9–100</td><td align=\"left\">Non-adjoining</td><td align=\"left\">21</td><td align=\"left\">459524</td><td align=\"left\">0.46</td><td align=\"left\">Reference</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T6\"><label>Table 6</label><caption><p>Cholera incidence rate and population-based rate ratios by strata of districts classified according to overcrowding level and order of neighborhood and/or adjacency to Kumasi Metropolis, 1997–2001.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Overcrowding</bold></td><td align=\"left\"><bold>Neighborhood</bold></td><td align=\"left\"><bold>Cholera cases</bold></td><td align=\"left\"><bold>Population</bold></td><td align=\"left\"><bold>Rate(per 10,000)</bold></td><td align=\"left\"><bold>Rate ratio(95% CI)</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Low</bold></td><td align=\"left\">Adjoining</td><td align=\"left\">0</td><td align=\"left\">0</td><td align=\"left\">0.00</td><td align=\"left\">0</td></tr><tr><td align=\"left\">(-4.5 &lt; OI &lt; -1.66)</td><td align=\"left\">Non-adjoining</td><td align=\"left\">51</td><td align=\"left\">1048225</td><td align=\"left\">0.49</td><td align=\"left\">*</td></tr><tr><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>Medium</bold></td><td align=\"left\">Adjoining</td><td align=\"left\">286</td><td align=\"left\">237610</td><td align=\"left\">12.04</td><td align=\"left\">26.43 (17.16–41.05)</td></tr><tr><td align=\"left\">(-0.73 &lt; OI &lt; -0.37)</td><td align=\"left\">Non-adjoining</td><td align=\"left\">24</td><td align=\"left\">526946</td><td align=\"left\">0.46</td><td align=\"left\">Reference</td></tr><tr><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>High</bold></td><td align=\"left\">Adjoining</td><td align=\"left\">276</td><td align=\"left\">434872</td><td align=\"left\">6.35</td><td align=\"left\">12.31 (7.17–21.51)</td></tr><tr><td align=\"left\">(-0.25 &lt; OI &lt; 3.20)</td><td align=\"left\">Non-adjoining</td><td align=\"left\">15</td><td align=\"left\">291028</td><td align=\"left\">0.52</td><td align=\"left\">Reference</td></tr></tbody></table></table-wrap>" ]

[ "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1476-072X-7-44-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>I</mml:mi><mml:mo>=</mml:mo><mml:mfrac><mml:mi>N</mml:mi><mml:mrow><mml:msub><mml:mi>S</mml:mi><mml:mi>o</mml:mi></mml:msub></mml:mrow></mml:mfrac><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>j</mml:mi></mml:msub><mml:mrow><mml:msub><mml:mi>w</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle><mml:mfrac><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mi>u</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>j</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mi>u</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mi>u</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mstyle></mml:mrow></mml:mfrac><mml:mo>,</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula>S_{<italic>o </italic>}= ∑_{<italic>i</italic>}∑_{<italic>j </italic>}<italic>w</italic>_{<italic>ij</italic>}</disp-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>Chi-square for linear trend = 831.375, P-value = 0.000001</p></table-wrap-foot>", "<table-wrap-foot><p>Chi-square for linear trend = 2995.5, P-value = 0.000001</p></table-wrap-foot>", "<table-wrap-foot><p>Chi-square for linear trend = 1757.2, P-value = 0.000001</p></table-wrap-foot>", "<table-wrap-foot><p>*undefined</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1476-072X-7-44-2\"/>", "<graphic xlink:href=\"1476-072X-7-44-1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Residential areas are proximal to everyday life and are therefore likely to influence health of local populations through the possibility they provide for leading healthy lives [##UREF##0##1##,##REF##12137182##2##]. An accumulating body of research shows evidence for variation in health across residential areas and the significance of area context for explaining this variation, independently of the characteristics of individuals [##UREF##1##3##, ####REF##11154250##4##, ##REF##17873220##5####17873220##5##].</p>", "<p>Different scales, or spatial units, may be relevant to specific contextual conditions and to specific heath outcomes [##REF##17590377##6##,##REF##11684601##7##], as illustrated by studies reporting varying strength and magnitude of area effects on health according to the operational definition of areas [##REF##11820682##8##, ####REF##15820583##9##, ##REF##14499510##10##, ##REF##15652687##11##, ##REF##12196317##12##, ##REF##10827914##13##, ##UREF##2##14##, ##REF##18177988##15####18177988##15##] or to contextual conditions [##REF##12908718##16##, ####REF##15230507##17##, ##REF##15965140##18##, ##REF##15571894##19####15571894##19##]. Nonetheless current approaches for delimiting areas mostly rely on spatial units \"of convenience\" such as census tracts, boroughs, or wards [##UREF##1##3##,##REF##17873220##5##]. These spatial units are certainly useful because they can easily be linked to data from censuses and other surveys that can be used for measuring contextual conditions. Also, they are often designed to be homogeneous along socioeconomic conditions of populations, thus being appropriate spatial units to operationalise the socioeconomic context of areas [##REF##15246176##20##] (this may not hold for other administrative units, e.g. postal code areas which are design for postal delivery purposes and may be very heterogeneous in terms of population composition). However it is to be considered that through time, the composition of the units may change leading to modification of the socioeconomic conditions which may become more heterogeneous.</p>", "<p>Yet, other contextual dimensions relevant for health may not be optimally defined within administrative spatial units. For example, conduciveness of areas to physical activity or geographic accessibility to health services may operate on different scales than socioeconomic factors. Operationalising relevant spatial units for studying area effects on health remains a conceptual and methodological challenge [##REF##11154250##4##,##REF##17873220##5##,##REF##11684601##7##,##REF##11446289##21##, ####REF##10884951##22##, ##UREF##3##23##, ##REF##12505885##24##, ##UREF##4##25##, ##REF##15020011##26####15020011##26##] giving rise to issues of validity and soundness of areal units as units of analysis [##UREF##5##27##].</p>", "<title>Operationalising small areas: issues of validity and soundness units of analysis</title>", "<p>Construct validity refers to whether or not the measurement instrument operationalises the concept of interest. In area effects on health research, construct validity is a matter of establishing 1) the soundness of units of analysis, i.e., whether or not area boundaries are aetiologically meaningful for studying the association between area characteristics and a given health indicator, and 2) whether or not data constitute appropriate operationalisations of exposure variables, i.e. the characteristics of areas [##UREF##5##27##]. Ideally, definitions of areas, the characteristics of these areas, and the health outcome(s) being studied should be coherent with one another [##REF##11684601##7##].</p>", "<p>Measures of area characteristics derived from population censuses and other surveys, e.g. socioeconomic position, although easily accessible, provide only partial information on the context of areas and may in fact be endogenous to the composition of the areas as they are determined by individual characteristics of residents [##REF##15774331##28##]. Collecting and measuring \"true\" or \"integral\" area data, i.e. data only measurable at the area level through procedures such as ecometrics and spatial analyses has been underscored as critical for measuring unbiased area-level variables [##REF##12137182##2##,##REF##11684601##7##,##REF##15774331##28##,##REF##15694520##29##]. Likewise, defining aetiologically meaningful areas in coherence with the specific purposes of the study, either in terms of health outcomes, characteristics of environment, or associations between the two [##UREF##3##23##,##REF##15774331##28##,##REF##16452106##30##,##REF##16707654##31##] is important for understanding the significance of residential areas for health. Measurement errors can result if the spatial patterning of environmental characteristics does not correspond to the spatial units chosen for operationalising areas and their context [##UREF##5##27##].</p>", "<p>Defining relevant geographic areas becomes salient in light of the modifiable areal unit problem, i.e. the fact that analytical results are sensitive to the definition of spatial units at which data are aggregated [##UREF##6##32##,##UREF##7##33##]. In other words, area effects may be observed only at certain scales, i.e. scales at which data are collected and aggregated and may vary or be absent when observed at other scales. Imposing arbitrary spatial units on a continuous spatial process, e.g. characteristics of environments, may lead to the delineation of artificial spatial patterns. In such cases, environment characteristics may be measured with error. As a result the internal validity of the study, i.e. whether or not observed associations are unbiased, may be threatened.</p>", "<p>In addition, as per spatial autocorrelation, areas will share similar contextual conditions as a function of their proximity in space [##UREF##8##34##]. By using spatial units of convenience, it is assumed that contextual conditions within one area are different and influence health independently of conditions in neighbouring areas [##REF##11154250##4##,##REF##17873220##5##,##REF##11446289##21##,##REF##10884951##22##,##REF##12505885##24##, ####UREF##4##25##, ##REF##15020011##26####15020011##26##,##REF##17706331##35##], when in fact these conditions are clustered in space. Furthermore, for any area effects to be detected there must be variation in the exposure being studied [##UREF##9##36##]. Yet the variation of environment characteristics may be smoothed out by the definition of area units used to measure them. For example, if spatial units encompass environments that are both conducive to walking and others that are less so, averaging values of conduciveness over census tracts could potentially lead to mismeasurment of exposures. Within area homogeneity along the contextual conditions under examination is thus required for minimising measurement error. Correspondingly, for inferring about area effects on health, between area differences must be maximised: if data are collected in contiguous and heterogeneous areas, variations in both characteristics of environments and health outcomes, and their association, may be misestimated. As area effects on health have been observed to be stronger in more homogeneous areas [##REF##17369075##37##,##REF##10984584##38##], homogeneity of areas may thus influence the estimation of area effects and therefore the validity of conclusion.</p>", "<p>In Figure ##FIG##0##1##, we propose a template that could be useful for establishing the soundness of spatial units \"of convenience\" to operationally define areas for specific research questions. For example, the template could be used to guide the decision as to whether or not census tracts are the most appropriate spatial units of analysis for measuring associations between area-level socioeconomic position (SEP) and obesity. That is, if they allow for measuring indicators of SEP without bias (and ultimately for estimating non-biased association with health outcomes) by showing homogeneity in the distribution of indicators and optimising their spatial patterning. In the methods section, we propose an approach for achieving this end. Intuitively, it can be expected that census tracts are appropriate units for undertaking such a study as they are, as mentioned above, initially designed to be homogeneous along socioeconomic conditions. But across time, the socioeconomic composition of census tracts may change as people migrate in and out of areas, potentially introducing heterogeneity in the socioeconomic make-up of the area. This could result in a \"dilution\" of the true level of deprivation. Averaging indicators of SEP over census tracts thus may mask \"pockets\" of poverty. The exercise of establishing the soundness of census tracts as units of analysis would be important here, as it would allow to measure with less error indicators of SEP and their association with health outcomes. In multilevel studies, mismeasurement of environment characteristics may influence the strength of the observed association between environment characteristics and health indicators [##UREF##10##39##]. As such, associations may not be detected or may be spurious, therefore limiting the precision of research findings for informing public health and public policy actions to tackle social and geographical inequalities in health.</p>", "<p>Establishing the soundness of spatial units of analysis chosen for operationalising area boundaries and measuring area context is an important methodological consideration, but it is often overlooked. Alternatively, designing spatial units of analysis maximising homogeneity of selected environment characteristics may prove to be a viable strategy for advancing the understanding of processes linking place to health [##REF##11684601##7##].</p>" ]

[ "<title>Methods</title>", "<p>The methodology section includes two parts. First, we present criteria and methods for designing homogeneous areas (henceforth designated as \"zones\"). Second, we present analyses undertaken to assess the soundness of census tracts as units of analysis for measuring the active living potential of residential areas.</p>", "<title>Designing optimal, homogeneous zones</title>", "<p>Zone design refers to the placement of areal unit boundaries [##REF##15820583##9##]. It can be achieved discursively (manually) by grouping basic spatial units into larger ones [##REF##15038149##57##, ####REF##12639596##58##, ##REF##10077280##59####10077280##59##], by combining social, statistical, and spatial analysis methods [##REF##15886144##60##,##REF##17615065##61##], and automatically through computationally intensive automated zoning software [##REF##15820583##9##,##REF##18177988##15##,##REF##17369075##37##,##UREF##16##62##, ####UREF##17##63##, ##UREF##18##64##, ##REF##12346252##65####12346252##65##].</p>", "<p>Three criteria guided the choice of the method for zone design. First, we wanted to design zones based on the spatial distribution of environmental characteristics related to active living potential, namely population density, land use mix, and geographic accessibility to selected proximity services. We had no requirement regarding population and area sizes as zones were defined on the basis of the spatial distribution of these characteristics. Second, the method for zone design had to be optimal, i.e. to maximize variation between zones and to minimize variation within zones in the selected characteristics. In other words, the aim was to design zones that were internally homogeneous on the three indicators of active living potential, but different (heterogeneous) amongst themselves. Finally, we wanted a method that was rigorous but comprehensible and easy to implement. We opted for an approach that combined a statistical classification method, <italic>K</italic>-means clustering, to mapping applications in geographic information system. This three-step approach is described in greater details in the following sections.</p>", "<title>Step 1: Measuring environment characteristics at the smallest area level</title>", "<p>The study area is the Island of Montreal, Canada, an urban centre with 1 812 723 residents. As of January 2006, on the Island of Montreal, there are 15 municipalities, in addition to the municipality of Montreal which includes 19 boroughs [##UREF##19##66##]. The Island of Montreal is further divided into 521 census tracts and 3222 dissemination areas. Dissemination areas (DAs) were used as basic spatial units for designing zones because they are the smallest standard geographic areas for which Canadian census data are available (population size between 400 and 700 residents) [##UREF##11##40##]. On the Island of Montreal, their average size is 0.15 km²(ranging between 364 m²and 18 km²) with an average population of 562 individuals (ranging between 44 and 2138 residents). DA values for population density, land use mix, and accessibility to services were computed in a geographical information system (ArcGIS 9.2) [##UREF##20##67##].</p>", "<p><italic>Population density </italic>refers to number of individuals per unit area. It was computed by dividing the total number of residents of a DA by its area size (km²) [##UREF##21##68##].</p>", "<p><italic>Land use mix </italic>relates to the diversity or variety of land uses within an area. It was computed using an entropy index [##REF##15694519##47##,##UREF##22##69##,##UREF##23##70##] which measures the homogeneity or diversity of land uses within a spatial unit. The index is defined as follow:</p>", "<p></p>", "<p>Where A_{<italic>ij </italic>}is the surface area of land use <italic>i </italic>in dissemination area <italic>j</italic>, D_{<italic>j </italic>}is the surface area of dissemination area <italic>j</italic>, and <italic>n </italic>is the total number of possible land uses which in the current case corresponds to 16, the number of different land uses characterising the Island of Montreal [##UREF##24##71##]. The index values range between 0 and 1, where 1 corresponds to a highly mixed area, and 0 to a homogeneous area, that is an area characterised by only one type of land use (e.g. low density housing). This index has been used in many studies to measure land use mix [##REF##15694519##47##,##UREF##25##72##].</p>", "<p><italic>Geographic accessibility to proximity services </italic>refers to geographic distance to or from destinations, here to supermarkets, pharmacies, banks, and libraries. These services were selected because they are most likely to be used on a regular basis, conveying the idea of proximity services potentially accessible through walking. There are many measures of geographic accessibility [##UREF##26##73##,##UREF##27##74##]. In this study, geographic accessibility was defined in terms of the number of the selected services within an area, conferring the notion of the offer of services provided by the immediate surroundings. Supermarkets, pharmacies, banks, and libraries were geocoded at the parcel level [##UREF##28##75##]. In order to minimise aggregation errors [##UREF##26##73##,##REF##18282284##76##], accessibility was measured by computing distances of services located within a one kilometre (network distance) radius [##UREF##29##77##] from the centroid of census blocks (n = 14 527) comprised within any one DA; the distances were than averaged and weighted by the total population of each census blocks.</p>", "<p>Characterisation of DAs along the three indicators resulted in a sample of 3206 DAs. Measures of land use mix and accessibility to services were normally distributed; population density was normalized using a LOG10 transformation [##REF##16705786##78##]. Population density was significantly and positively correlated to accessibility to services (r = 0.45, p &lt; 0.001), and negatively to land use mix (r = -0.32, p &lt; 0.001). Land use mix and accessibility to services were not significantly correlated (r = 0.03, p &gt; 0.500). Prior to cluster analyses, these variables were standardized to a mean of 0 and a standard deviation of 1, higher values representing greater levels of population density, land use mix, and accessibility to services.</p>", "<title>Step 2: Classifying smallest areas into clusters, e.g. \"types of environments\", using K-means clustering</title>", "<p><italic>K</italic>-means statistical clustering techniques using SAS (version 9.1) for Windows [##UREF##30##79##] was applied to classify DAs into <italic>k </italic>number of optimal clusters homogeneous in terms of active living potential. In social sciences, notably in geography, <italic>K</italic>-means is largely employed to classify areas (e.g. geodemographics [##UREF##31##80##]). The method uses an allocation/re-allocation algorithm to optimally reassign objects, here DAs, to the nearest cluster centroid [##UREF##32##81##, ####UREF##33##82##, ##UREF##34##83####34##83##]. The goal is to maximize between cluster variations and to minimize within cluster variations. The aim of this second step was to group DAs with similar values of population density, land use mix, and accessibility to services into <italic>k </italic>types of environments that are internally homogeneous but different among them. These types of environments correspond to different levels of active living potential. For <italic>K</italic>-means clustering, the number of clusters (<italic>k</italic>) must be determined at the onset of analyses; as we had no a priori for such number, we conducted analyses for <italic>k </italic>= 4 to <italic>k </italic>= 20.</p>", "<title>Step 3: Mapping the clusters to create optimal and homogeneous zones</title>", "<p>In a final step, the <italic>k </italic>types of environments were imported into ArcGIS 9.2 and mapped out. This lead to the delineation of <italic>n </italic>homogeneous zones i.e., units of analysis, characterised by one of <italic>k </italic>active living potential.</p>", "<title>Statistical analyses: Assessing the soundness of census tracts as units of analysis for operationalising active living potential</title>", "<p>The soundness of census tracts for operationalising indicators of active living potential was assessed through three series of analyses.</p>", "<p>First, to assess the homogeneity of census tracts, variation in indicators of active living potential was estimated and decomposed between and within areas. Population density, land use mix, and accessibility of services were measured continuously at the DAs level (level 1: n = 3206). In separate two-level multilevel models, DAs were nested into zones (n = 898) and into census tracts (n = 506 with valid population and socioeconomic data). Between-area variation in indicators of active living potential was estimated using the intraclass correlation coefficient (ICC) from unconditional (null) multilevel models using HLM software Version 6.04 [##UREF##35##84##]. The ICC indicates the proportion of variation in a dependent variable that is attributable to differences between area units. Greater ICC values indicate that variation of a variable is greater between units than within, i.e. units are different among them but internally homogeneous. Using the same analytical approach, homogeneity of zones and census tracts along indicators of socioeconomic position was assessed and compared. DA-level data on the proportion of low-income households, of people with less than high school education, and of people with a university degree were obtained from the 2001 Canadian census.</p>", "<p>Second, analysis of variance was performed to examine the proportion of variation across zones in socioeconomic variables explained by the <italic>k </italic>types of environment. Indicators of SEP at the DA-level were aggregated (weighted by population) at the zone-level. These analyses were performed to examine whether or not socioeconomic and active living indicators follow a similar spatial distribution as is implicitly assumed when measured within the same area unit of analysis.</p>", "<p>Finally, descriptive statistics were employed to assess the extent to which the spatial distribution of the different types of environment coincides with the boundaries of census tracts. These analyses were conducted to examine if census tracts encompassed environments with differing levels of active living potential. The numbers of zones straddling over one or more census tracts, and the number of types of environment encompassed within census tracts were computed. To examine whether or not the spatial distribution of more mixed or more homogeneous census tracts (i.e. the number of types of environments encompassed within census tracts) was structured in space, global values of spatial autocorrelation were computed using Moran <italic>I </italic>with a first-order contiguity matrix [##UREF##36##85##,##UREF##37##86##]. Values for Moran <italic>I </italic>vary between -1 and 1, where negative values indicate negative spatial autocorrelation, i.e. neighbouring spatial units have different values, and positive values indicate positive spatial autocorrelation, i.e. neighbouring units have similar values. The covariance in Moran <italic>I </italic>is the covariance over space for neighbouring spatial units, and will not be computed unless two units are contiguous (first order); also, only one variable is considered [##UREF##36##85##], here the number of types of environments included in census tracts.</p>" ]

[ "<title>Results</title>", "<title>Description of types of environment and zones</title>", "<p>Figure ##FIG##1##2## illustrates results of the <italic>K</italic>-means clustering, which show that the 3206 DAs were optimally classified into 7 clusters or \"types of environments\" as indicated by peaks [##UREF##38##87##] in both the Pseudo-F statistic [##UREF##39##88##] and the Cubic clustering criterion [##UREF##40##89##]. These clusters explain 72.8% of the total variation in the three indicators of active living potential. Thus, differences among the seven clusters and similarity of DAs comprised within the same cluster, i.e. within-cluster homogeneity, were both maximized. The seven types of environments correspond to seven different levels of active living potential. They encompassed more suburban to more central urban types of environments defined by different values of population density, land use mix, and accessibility to services. The types of environments are described in Figure ##FIG##1##2## and Figure ##FIG##2##3##.</p>", "<p>Low-density and mid-density suburban areas are characterised by lower values of population density and accessibility to services. Diverse central urban areas and central urban areas with high accessibility are more densely populated and have greater access to services than any other types of environment. Although population density and accessibility to services follow to some extent an increasing gradient from more suburban to more urban areas, the pattern of land use mix is more complex: there are low values in urban areas and high values in suburban areas. Dissemination areas are designed to be similar in population size (among other characteristics); thus the area size required to reach the set population threshold (i.e. between 400 to 700 residents [##UREF##11##40##]) will be larger in less densely populated areas and smaller in more urban areas. As a consequence, larger dissemination areas are more likely to encompass different land use than are smaller dissemination areas located in urban areas.</p>", "<p>Figure ##FIG##1##2## also presents the statistical proximity (Euclidian distance) of the centroids of clusters (cluster mean values), i.e. types of environment, in a three dimensional graph where the axes correspond to the three indicators of active living potential. With respect to their spatial distribution, the types of environment are positively correlated in space indicating that contiguous zones were characterised by similar types of environment.</p>", "<p>Mapping of the clusters into the GIS led to the delineation of 898 zones or units of analysis characterised by one of the seven types of environments, i.e. active living potential, as illustrated in Figure ##FIG##2##3##. Zones are significantly smaller than census tracts, an average of 0.54 km²(SD = 3.50) compared to 0.96 km²(SD = 1.98) (t = -2.46; p &lt; 0.05), but the variation of their area size is not statistically different (F = 0.68; p = 0.409). Zones are significantly smaller than census tracts in population size, an average of 1960 (SD = 3867) residents compared to 3554 (SD = 1647) (t = -10.73; p &lt; 0.001), and there is significantly greater variability in population size across zones than across census tracts (F = 11.40; p &lt; 0.01). Zones characterised by more suburban contexts are on average larger and have relatively smaller population counts than urban zones.</p>", "<title>Soundness of census tracts as units of analysis for measuring active living potential</title>", "<title>Homogeneity of census tracts along active living potential indicators</title>", "<p>Results of homogeneity of zones and census tracts along active living indicators appear in Figure ##FIG##3##4##. The variation in indicators is not uniform across census tracts. A greater proportion (83%) of variation in accessibility to services is attributable to differences between census tracts, as indicated by a higher ICC value, suggesting within census tract homogeneity along this indicator. Yet about half of the variation in population density is between census tracts (52%), whereas there is greater variation in land use mix within census tracts (85%), indicating greater heterogeneity of tracts along these indicators. The degree of homogeneity of tracts therefore varies according to the indicator examined. For population density and land-use mix, but not for accessibility of services, variation between zones is greater than variation between census tracts. This shows that the method was successful in designing areas or units of analysis that were more homogeneous than census tracts along dimensions of active living potential.</p>", "<p>The degree of homogeneity of census tracts and zones along socioeconomic indicators shows that for the selected variables, variability is larger between census tracts than between zones (Figure ##FIG##3##4##). Census tracts are relatively homogeneous areas in terms of the socioeconomic environment, especially for proportion of population with a university education.</p>", "<title>Spatial distribution of active living potential and socioeconomic indicators</title>", "<p>Examining variation in socioeconomic indicators across zones shows that they follow a different spatial distribution than that of active living potential indicators. Results of analyses of variance (results not shown) revealed that 15.2% of the variation in the proportion of low-income households was explained by the seven types of environment whereas these proportions were 5.2% for the proportion of people with less than high school and 3.8% for the proportion of people with a university education.</p>", "<title>Types of environments encompassed within census tracts boundaries</title>", "<p>Overall, zones are not well contained within census tracts. As shown in Figure ##FIG##4##5##, only 30.5% of zones are completely located within the boundaries of one census tract. Forty-eight percent of zones straddle two or three census tracts whereas, 21.5% spread over more than four tracts. Correspondingly, there is considerable variability in types of environment within census tracts.</p>", "<p>As illustrated in Figure ##FIG##5##6##, 11.2% of census tracts encompass only one type of environment and 34.3% encompass two types. About 28% of census tracts are characterised by three different types of environment, whereas 26.3% comprise 4 or more different types. Among census tracts encompassing two types of environment (n = 175), about two-thirds (66.3%) comprise types that are statistically similar as indicated by distances between their centroids (two or less distance lag as indicated in the distance matrix in Figure ##FIG##1##2##; results not shown). For example, census tracts often comprise a combination of low-density suburban and suburban/urban axial zones (26.3%), or a grouping of diverse and high accessibility central urban areas (35.4%). Globally, the number of types of environment encompass within census tracts is positively correlated in space (Moran <italic>I </italic>= 0.26; p &lt; 0.001), suggesting that more homogeneous or more mixed census tracts are often contiguous in space (Figure ##FIG##5##6##). More heterogeneous census tracts are located mainly on the periphery of central urban areas and in the eastern part of the Island of Montreal, and to a lesser extend in the west-end suburbs.</p>" ]

[ "<title>Discussion</title>", "<p>The objective of this study was to assess the soundness of census tracts as units of analysis, i.e. their degree of homogeneity in terms of the active living potential of residential environments associated with walking. In order to do so, homogeneous zones that optimised the spatial patterning of active living potential indicators hypothesised to be associated with greater involvement in walking, namely population density, land-mix use, and accessibility to services, were successfully designed. This was done through the application of an easy-to-use method combining a classification method called <italic>K</italic>-means clustering with basic mapping applications of geographical information systems. The degree of soundness of census tracts as units of analysis was established through a series of analyses comparing them to the newly-designed zones.</p>", "<p>First the distribution of the three active living indicators between and within census tracts was assessed. Although census tracts were homogeneous in terms of accessibility to services, they were less homogenous in population density; for this indicator within and between census tracts variations were about equal. Census tracts were clearly not homogeneous in terms of land use mix as the variability within tracts largely exceeded the variability between tracts. In contrast, census tracts were homogeneous along socioeconomic variables. These results suggest that the spatial patterning of the active living potential of environments do not neatly follow in the delineation of census tracts, which may be more suitable as units of analysis for operationalising socioeconomic contexts.</p>", "<p>Then, findings revealed that the spatial distribution of active living and socioeconomic indicators followed different spatial distribution. At the zone-level, types of environment explained a small proportion of variation of socioeconomic variables. This indicates that processes underlying the distribution of active living and SEP indicators, although potentially linked [##REF##12137182##2##,##REF##17590377##6##], operate at different scales and thus require different units of analysis.</p>", "<p>In the final set of analyses, within tract variability in terms of what we labelled \"types of environment\" was examined. This allowed for the assessment of whether or not census tracts encompassed environments that were substantively different among them in terms of their active living potential. Census tracts comprising two different types of environments (34.3%) were not considered necessarily as problematic, given that some types of environment were more similar than others and were often contiguous in space. For example, diverse and high accessibility central urban zones were often contiguous in space and were statistically most similar (as indicated by statistical distances between clusters; Figure ##FIG##1##2##). However, census tracts comprising three or more types of environment raised concerns; such a situation was observed in more than half of census tracts. These tracts encompass environments that are simultaneously most conducive to walking and others that are least so. Averaging values of conduciveness to walking could potentially lead to significant errors when measuring active living potential at the census tract-level.</p>", "<p>The approach for defining areas or units of analysis differs from those involving the definition of strictly \"ecologically meaningful\" or \"natural\" neighbourhoods, i.e. neighbourhoods imbued with meaning for residents [##REF##11446289##21##] or as consisting of a group of homes sharing a commonly defined residential area often having name [##REF##15246176##20##]. Defining such units of analysis is important when the notion of commonly shared territory is related to the contextual condition of interest, for example social capital or collective efficacy [##REF##11684601##7##,##UREF##5##27##]; this notion is not conjured up by active living potential. Designing zones based on the spatial distribution of active living indicators empirically linked to greater involvement in walking leads to the definition of areas that are more appropriate units of analysis and increases the internal validity of study design examining the environmental determinants of walking.</p>", "<p>Future studies are needed to assess the impact of the choice of other environmental characteristics for designing zones relevant to other health indicators, and to other geographical areas. For example, areas relevant for studying the social and environmental determinants of overweight and obesity may be delimited according to the distribution of active living variables and food provision (accessibility of both healthy and non-healthy food). For studying mental health outcomes, social dimensions of area context such as social support and opportunities for social participation, may be more relevant. It is to be expected that designing zones using other indicators of contextual conditions associated with other health outcomes will lead to different spatial configuration of area units of analysis.</p>", "<p>Homogeneous zones are designed with the aim of optimising the study of a phenomenon or for the purpose of uncovering the aetiology underlying associations between area context and health. As such, the configuration of zones should not be viewed as other \"spaces\" of actions for public health and policy interventions. Rather, they may be useful for informing on viable interventions and policy strategies that may be health promoting.</p>", "<title>Limitations</title>", "<p>Results of this study should be considered in light of some limitations. First, there is a seven year time lag (2000 to 2006) between the dates of creation of the different datasets used to characterise dissemination areas in terms of their active living potential and socioeconomic position. Although changes in the built environment may have taken place during this period, the speed at which changes occur is not well documented; however over a seven-year period, changes in the built environment can be expected to be modest.</p>", "<p>Other indicators of active living potential could be examined in designing homogeneous areas, such as street connectivity, safety, and accessibility to other services or resources such as parks. In this study, the measurement of land use mix was dependent on the size of disseminations areas which are defined in part by a population size threshold: because of lower population density in suburban areas, DAs are likely to span a greater territory and therefore encompass more types of land use. Other scales for measuring land use mix could be considered [##REF##15694519##47##].</p>" ]

[ "<title>Conclusion</title>", "<p>For studies concerned with the social and environmental determinants of health and more specifically of physical activity, results of this study have several implications. Delimiting areas is a key conceptual and methodological challenge in research on health and place. In this paper, we developed an easy-to-use method for establishing homogeneous units of analysis in terms of specific environmental characteristics hypothesised to be linked to a specific health indicator. The focus was on active living potential of areas and walking behaviours. Using these homogeneous zones as comparison, the objective was to assess the soundness of spatial units \"of convenience\", i.e. census tracts, to operationalise contexts for which they were not purposely developed. The methods developed in this study add to the growing literature on alternative ways to conceptualise and define the boundaries of area units for studying the determinants of health.</p>", "<p>Findings showed that although census tracts may be homogeneous along independent indicators of active living potential, they were most often characterised by a combination of types of environment that were substantively different in terms of their active living potential. For this reason, census tracts should be used with caution as units of analysis when operationalising active living potential for studying determinants of walking. But census tracts or other administratively defined areas may be appropriate area units, i.e. may be homogeneous enough, when processes hypothesised to be operating on health are linked to the socioeconomic context of an area, for example affluence or poverty.</p>", "<p>In this study, zones were delimited for methodological and aetiological purposes with the aim of minimising measurement errors of environmental characteristics and increasing internal validity of study design for measuring area effects on health. As can be expected, the zones are context-specific and cannot be exported to other geographic areas. Rather they are representations of the local realities of processes relating environmental characteristics to health. As suggested by others, the geographical aspects of the study design should be considered prior to conducting analyses [##REF##18177988##15##]. Establishing the soundness of spatial units \"of convenience\" for representing the environmental and spatial processes under investigation should be part of the empirical approach for conceptualising, operationalising, and measuring area effects on health.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>In health and place research, definitions of areas, area characteristics, and health outcomes should ideally be coherent with one another. Yet current approaches for delimiting areas mostly rely on spatial units \"of convenience\" such as census tracts. These areas may be homogeneous along socioeconomic conditions but heterogeneous along other environmental characteristics. This heterogeneity can lead to biased measurement of environment characteristics and misestimation of area effects on health. The objective of this study was to assess the soundness of census tracts as units of analysis for measuring the active living potential of environments, hypothesised to be associated with walking.</p>", "<title>Results</title>", "<p>Starting with data at the smallest census area level available, zones homogeneous along three indicators of active living potential, i.e. population density, land use mix, and accessibility to services were designed. Delimitation of zones ensued from statistical clustering of the smallest areas into seven clusters or \"types of environment\". Mapping of clusters into a GIS led to the delineation of 898 zones characterised by one of seven types of environment, corresponding to different levels of active living potential. Homogeneity of census tracts along indicators of active living potential varied. A greater proportion (83%) of variation in accessibility to services was attributable to differences between census tracts suggesting within-tract homogeneity along this variable. However, census tracts were heterogeneous with respect to population density and land use mix where a greater proportion of the variation was attributable to within-tract differences. About 55% of tracts were characterised by a combination of three or more \"types of environment\" suggesting substantial within-tract heterogeneity in the active living potential of environments.</p>", "<title>Conclusion</title>", "<p>Soundness of census tracts for measuring active living potential may be limited. Measuring active living potential with error may lead to misestimation of associations with walking, therefore limiting the correctness of inference about area effects on walking. Future studies should aim to determine homogeneity of spatial units \"of convenience\" along environment characteristics of interest prior to examining their association with health. Further evidence is needed to assess the extent of this methodological issue with other indicators of environment context relevant to other health indicators.</p>" ]

[ "<title>Objectives</title>", "<p>The aim of this investigation is to assess the soundness of census tracts as units of analysis for studying associations between a specific exposure and a specific health outcome, namely the active living potential of residential environments and walking behaviours. Active living potential refers to the conditions of areas that encourage the likelihood of integrating physical activity into daily routines [##REF##15694520##29##]. Census tracts were selected as spatial units \"of convenience\" because of extensive use of this spatial unit of analysis in current research on health and place [##REF##11154250##4##,##REF##17873220##5##]. In Canada, census tracts are small and relatively stable geographic areas with populations ranging in size between 2500 and 8000 inhabitants; at the time of their creation, census tracts were homogeneous in terms of socioeconomic characteristics, e.g. economic status and social living conditions [##UREF##11##40##].</p>", "<p>To establish the soundness of census tracts as a unit of analysis, we developed and tested a comprehensible method for designing optimal and homogeneous spatial units espousing the spatial distribution of selected environment characteristics linked to the concept of density of destinations that is the physical and social characteristics of residential areas related to land use pattern [##REF##15694520##29##]. Three indicators were used to operationalise the construct of active living potential: population density, land use mix, and geographic accessibility to proximity services. The specific objectives of the study are to examine whether or not: 1) census tracts are homogeneous units of analysis along indicators of active living potential; 2) active living potential and socioeconomic indicators follow a similar spatial distribution; and 3) census tracts encompass smaller areas with different (or similar) levels of active living potential.</p>", "<p>Active living potential was chosen because of increasing research reporting associations between this environmental construct and walking [##REF##15571894##19##,##REF##18227097##41##, ####UREF##12##42##, ##REF##13677966##43##, ##REF##13677962##44##, ##REF##15211020##45##, ##UREF##13##46##, ##REF##15694519##47##, ##REF##15694525##48##, ##REF##12460528##49##, ##REF##12948984##50##, ##UREF##14##51##, ##REF##16679987##52##, ##REF##12704009##53####12704009##53##], an important public health indicator [##REF##12972873##54##, ####REF##12165679##55##, ##UREF##15##56####15##56##]. This choice was also motivated by availability of spatial datasets allowing for the operationalisation of integral measures of land use mix and geographic accessibility to services in geographical information systems, and by the availability of individual-level data on walking behaviours (to be examined in future analyses).</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>MR conceptualised the study. She carried out spatial and statistical analyses, mapping of results, and drafted the manuscript. PA, LG and JMB participated in the conceptualisation the study, and in data analyses. All authors critically revised the paper, and read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>At the time of data analyses and write-up, MR was the recipient of a Canada Graduate Scholarships Doctoral Award from the Canadian Institutes of Health Research (grant # CGD-76386). Data collection was supported in part by the Social Science and Humanities Research Council of Canada (SSHRC) and by the Canadian Institutes of Health Research grant 200203 MOP 57805. LG holds a Canadian Institute for Health Research / Centre de Recherche en Prévention de l'Obésité Applied Public Health Chair in Neighborhoods, Lifestyle, and Healthy Body Weight. The GRIS receives infrastructure funding from the Fonds de la recherche en santé du Québec (FRSQ), the Léa-Roback Research Center is funded through a Research Center development initiative by the Canadian Institutes of Health Research, and AnÉIS receives funding from Canadian Institutes of Health Research. </p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Template for deciding upon the soundness of \"spatial units of convenience\" to operationally define small area units of analysis</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Statistical proximity of the seven types of environment (clusters)</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Description of types of environment (clusters) and zones</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p>Decomposition of variation in indicators of active living potential and socioeconomic position across zones and across census tracts.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>Proportion of zones straddling different numbers of census tracts across the Island of Montreal</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p>Number of types of environment encompassed within census tract boundaries.</p></caption></fig>" ]

[]

[ "<disp-formula id=\"bmcM1\"><label>(1)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1476-072X-7-43-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>E</mml:mi><mml:mi>j</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mo>−</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>n</mml:mi></mml:munderover><mml:mrow><mml:mrow><mml:mo>[</mml:mo><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msub><mml:mi>A</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub><mml:mo>/</mml:mo><mml:msub><mml:mi>D</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msub><mml:mi>A</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mi>j</mml:mi></mml:mrow></mml:msub><mml:mo>/</mml:mo><mml:msub><mml:mi>D</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mo>]</mml:mo></mml:mrow></mml:mrow></mml:mstyle><mml:mo>/</mml:mo><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>n</mml:mi></mml:mrow></mml:semantics></mml:math></disp-formula>" ]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1476-072X-7-43-1\"/>", "<graphic xlink:href=\"1476-072X-7-43-2\"/>", "<graphic xlink:href=\"1476-072X-7-43-3\"/>", "<graphic xlink:href=\"1476-072X-7-43-4\"/>", "<graphic xlink:href=\"1476-072X-7-43-5\"/>", "<graphic xlink:href=\"1476-072X-7-43-6\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Diseases of the cardiovascular system play an important role in health care. They have a great impact on the burden of disease. This burden of disease is defined as the impact of a health care problem in an area measured by financial cost, mortality, morbidity, or other indicators. Quantification is often performed using Disability-adjusted life years (DALYs) [##REF##7923545##1##] or Quality-adjusted life years (QALYs) [##REF##10304223##2##,##REF##3149143##3##]. These measures combine the burden due to both death and morbidity into one index. With regard to the different disorders listed in global and national burden of disease rankings, also diseases of the respiratory play a prominent role [##REF##12477209##4##]. In this respect, four out of the ten most common causes of death are respiratory diseases [##REF##16731270##5##]. In view of the enormous socio-economic burden, it should be anticipated that a large proportion of health research funding is allocated to this field of medicine.</p>", "<p>In contrast to these features, health system resources and research funding policy are often debated as being disproportional. Numerous publications discussed this issue i.e. in the field of neurosciences [##REF##15332083##6##], cardiovascular medicine [##REF##9751669##7##], gastroenterology [##REF##11454809##8##], genetics [##REF##10604449##9##] or stem cell research [##REF##15895815##10##, ####REF##15199940##11##, ##REF##16279757##12####16279757##12##]. These areas are heavily funded by governmental and non-governmental sources and there are various statements concerning policy guidelines available [##REF##1957164##13##, ####REF##11659431##14##, ##REF##11651693##15##, ##REF##10120435##16##, ##REF##10300296##17##, ##REF##1367092##18####1367092##18##].</p>", "<p>For the high income country Germany, especially the fields of respiratory medicine and cardiology are interesting areas for health and research funding allocation policy. In this respect, the present study aimed to 1) identify and compare different output figures 2) relate these figures to selected input figures 3) provide data in relation to geographical information.</p>" ]

[ "<title>Methods</title>", "<title>Output benchmarking data source</title>", "<p>Data for output benchmarking (published items and citations) was retrieved from the biomedical database Web of Science (Thomson Institute for Scientific Information, ISI) [##REF##15253331##19##,##REF##15719884##20##].</p>", "<title>Search strategies</title>", "<p>For the different searches, phrases joined together with Boolean operators, i.e. AND, OR and NOT were used.</p>", "<title>Time frame</title>", "<p>A time frame was set and all entries between the years 2002 and 2006 were analysed.</p>", "<title>Input benchmarking data source</title>", "<p>Data for input benchmarking was retrieved from internet searches and the German Lung White Book [##UREF##0##21##]. All full professorships/chairs (W3/C4 salary level) of medical school departments of cardiology and respiratory medicine were identified (begin of analysis 2007-08-01, last update 2008-4-30) and related to the respective German states. In this respect, the numbers of full professorships/chairs were calculated for each of the 16 German states (i.e. 3 chairs for cardiology in the state of Berlin and 0 for the state of Brandenburg) and density equalizing mapping performed. The numbers were related to each German state since German medical schools are financed directly by the single states and not by the Federal Republic of Germany. Using the calculation for each state, an input analysis was possible for cardiology and respiratory medicine.</p>", "<title>Output quantity analysis: Comparison of number and origin of publications in relation to the field</title>", "<p>To perform a comparison between respiratory medicine and cardiology, published items were screened. All full professors/chairmen of medical school departments of cardiology and respiratory medicine were identified by name and their publications between 2002 and 2006 were recorded (begin of analysis 2007-08-01, last update 2008-4-30). For density equalizing procedures, only the publication type \"article\" was used. The entries of all full professors/chairmen for each of the 16 German states were added in order to establish a formula for each of the 16 German states for geographical distribution.</p>", "<title>Output quality analysis: Comparison of citations in relation to the field</title>", "<p>To perform a qualitative comparison between respiratory medicine and cardiology, published items were related to their citations. Parallel to the quantity analysis, all full professors/chairmen of medical school departments of cardiology and respiratory medicine were identified by name and the numbers of citations of their publications between 2002 and 2006 were recorded (begin of analysis 2007-08-01, last update 2008-4-30). For density equalizing procedures, only the publication type \"article\" was used. The citations of all full professors/chairmen of each state were added in order to establish a formula for each of the 16 German states for geographical distribution.</p>", "<title>Density-equalizing mapping</title>", "<p>The method of density-equalizing mapping was used following a recently described method [##REF##18315838##22##] basing on Gastner and Newman's algorithm [##REF##15136719##23##]. In brief, territories were re-sized from the original size (additional file ##SUPPL##0##1##) according to a particular variable, i.e. the number of published items or the citations. For the re-sizing procedure the area of each state was scaled in proportion to its total number of published items or citations.</p>" ]

[ "<title>Results</title>", "<title>Input benchmarking: Spatial distribution of full professorships/chairs</title>", "<p>Large differences were found in the input benchmark \"number of full professorships/chairs\" (highest level – W3/C4) of cardiology versus respiratory medicine. For cardiology, it was found that the total number of 34 medical school/faculties in Germany have established 36 full professorships/chairs (W3/C4 salary level) of cardiology. In this respect, the medical faculty of the Charité in Berlin has three independent departments of cardiology that are directed by three separate full professorships/chairs of cardiology. Also, Munich has two full professorships/chairs of cardiology which belong to two separate medical faculties. Every other university has an own department of cardiology (Fig. ##FIG##0##1a##). Density equalizing mapping calculations visualizes the geographical distribution. Every state apart from Brandenburg and Bremen has financed a medical school department for cardiology. In these states, there are no medical schools (Fig. ##FIG##0##1a##).</p>", "<p>In contrast to these input figures, the area of respiratory medicine is represented by only 8 independent clinical full professorships/chairs (W3/C4 level) of respiratory medicine (fig. ##FIG##0##1b##). Two out of the eight are situated in Hesse at the same faculty of medicine (Justus-Liebig-University Giessen). The following states do not finance an independent full clinical professorship/chair of respiratory medicine: Berlin, Brandenburg, Bremen, Hamburg, Bavaria, Saxony, Saxony-Anhalt, Thuringia which leads to a major distortion in the density equalizing map (fig. ##FIG##0##1b##).</p>", "<title>Quantity output benchmarking: Total numbers and spatial distribution of published items per state</title>", "<p>The comparison of respiratory medicine and cardiology concerning the benchmark of total numbers of published items of full professors of each German state demonstrated large quantitative differences between the two fields of medicine and the different states.</p>", "<p>For the German full professorship of clinical cardiology, an overall number of 2708 published items was found. The ranking was headed by North Rhine-Westfalia (#1 with 610 published items), followed by Bavaria (#2 with 413), Baden-Württemberg (#3 with 369), Berlin (#4 with 292), Saxony (#5 with 219), Lower Saxony (#6 with 163), Hesse (#7 with 134), Mecklenburg-Western Pomerania (#8 with 122), Hamburg (#9 with 88), Rhineland-Palatinate (#10 with 86), Schleswig-Holstein (#11 with 72), Thuringia (#12 with 69), Saxony-Anhalt (#13 with 60) and Saarland (#14 with 11). The states Brandenburg and Bremen do not have a medical faculty. Density equalizing mapping approaches were used to analyse the distribution and it was found that the states Lower Saxony, Saxony-Anhalt, Bremen and Brandenburg were distorted (Fig. ##FIG##1##2a##) in comparison to their natural shape (additional file ##SUPPL##0##1##).</p>", "<p>For respiratory medicine, lower numbers were recorded in general with an overall number of 453 published items for all 8 chairs. In this field, Hesse led the field with a total number of 255, followed by Lower Saxony (66), North Rhine-Westfalia (52), Mecklenburg-Western Pomerania (36), Baden-Württemberg (25), and Schleswig-Holstein (19). The states Bavaria, Berlin, Hamburg, Rhineland-Palatinate, Saarland, Saxonia, Saxonia-Anhalt and Thuringia do not have an independent full professorship at the C4/W3 salary level despite having cardiology professorships. Bremen and Brandenburg do not have medical faculties and therefore neither cardiology nor respiratory full professorships. Density equalizing mapping calculations led to a strong distortion with Hesse and Lower Saxony dominating the map (Fig. ##FIG##1##2b##) in comparison to the natural shape (additional file ##SUPPL##0##1##).</p>", "<title>Quality output benchmarking: Citation numbers and spatial distribution of published items per state</title>", "<p>Large differences were also present between the two fields with regard to output quality benchmarking. In this respect, the total number of citations of cardiology articles was 48337 versus 7290 citations of respiratory medicine articles. The average citation per item also differed with 17.85 for cardiology articles and 4.34 for respiratory articles.</p>", "<p>To perform a detailed spatial comparison between respiratory medicine and cardiology, citations were related to the states of origin.</p>", "<p>In the field of cardiology, the citation ranking was partly different from the publication number ranking: Parallel to the publication number ranking, North Rhine-Westfalia headed this analysis with 10825 citations, followed by Saxony (#2 with 8078 citations), Bavaria (#3 with 6003), Baden-Württemberg (#4 with 5499), Lower Saxony (#5 with 4486), Berlin (#6 with 4351), Mecklenburg-Western Pomerania (#7 with 1826), Rhineland-Palatinate (#8 with 1695), Hesse (#9 with 1653), Schleswig-Holstein (#10 with 1585), Hamburg (#11 with 1066), Saxony-Anhalt (#12 with 469), Thuringia (#13 with 464) and Saarland (#14 with 337). The states Brandenburg and Bremen do not have a medical faculty.</p>", "<p>Density equalizing mapping calculations to led slight distortions of the map (Fig. ##FIG##2##3a##) in comparison to the natural shape (additional file ##SUPPL##0##1##).</p>", "<p>For respiratory medicine, lower numbers of citations were recorded in general with an overall number of 7290 citations for all 8 chairs.</p>", "<p>In specific, Hesse ranked #1 with 5442 citations, followed by Lower Saxony (#2 with 682), North Rhine-Westfalia (#3 with 357 citations), Baden-Württemberg (#3 with 335), and Schleswig-Holstein (#4 with 172). The other states Bavaria, Berlin, Hamburg, Rhineland-Palatinate, Saarland, Saxonia, Saxonia-Anhalt and Thuringia do not have an independent full professorship at the C4/W3 salary level despite having cardiology professorships. Density equalizing mapping calculations again led to strong distortions in the map with Hesse dominating (Fig. ##FIG##2##3b##) in comparison to the natural shape (additional file ##SUPPL##0##1##) and to the shape in the cardiology citation density-equalizing map (Fig. ##FIG##2##3a##).</p>", "<title>Quality output benchmarking: Spatial distribution of average citations per item</title>", "<p>Density equalizing mapping approaches were also used to assess differences in the spatial distribution of average citations per item in both fields. These calculations based on the number of publications and citations in relation to the states of origin.</p>", "<p>In the field of cardiology, the calculations led to stronger distortions in the density-equalizing map than in the publication and citation number analysis (Fig. ##FIG##3##4a##). In this respect, the average citation per item analysis listed Saxony at the first position (36.89 citations per published item) followed by Saarland (30.64), Lower Saxony (27.52), Schleswig-Holstein (22.01), Rhineland-Palatinate (19.71), North Rhine-Westfalia (17.75), Mecklenburg-Western Pomerania (14.97), Berlin (14.9), Baden-Württemberg (14.9), Bavaria (14.54), Hesse (12.34), Hamburg (12.11), Saxony-Anhalt (7.82), Thuringia (6.72). Thus, the density equalizing map was dominated by Saxony and Saarland (Fig. ##FIG##3##4a##) in comparison to the natural shape (additional file ##SUPPL##0##1##).</p>", "<p>For respiratory medicine, Hesse was also ranked on first position in this analysis (Fig. ##FIG##3##4b##) with an average citation per published item of 21.34. Hesse was followed by Baden-Württemberg (#2 with 13.4 citations per item), Lower Saxony (#3 with 10.33), Schleswig-Holstein (#4 with 9.05), Mecklenburg-Western Pomerania (#5 with 8.39) and North Rhine-Westfalia (#6 with 6.87).</p>" ]

[ "<title>Discussion</title>", "<p>Numerous publications indicate that current settings for health system and research funding need review. Reasons are potential imbalances in the existing policy for funding allocation. The present study addressed this issue using Germany as a model high income country and the two socio-economic important fields of cardiovascular and respiratory medicine.</p>", "<p>Methodologically, we used both output and input benchmarking. Output benchmarking was divided into the quantitative measure of total number of published items (publications type \"article\") and the qualitative measure of citations. The later feature is partly debated as not being a very good tool to assess research quality [##REF##18469571##24##,##REF##18202622##25##] but other tools such as the H-index also bear limitations [##REF##18367289##26##].</p>", "<p>In terms of input parameters, the present study is limited to the number of full professorships/chairs for cardiology vs. respiratory medicine per state. For Germany, this can be used as an indicator for governmental funding since the states are responsible for the financial support of the medical schools. In this respect, the state ministers for research and education are usually also responsible to establish full professorships/chairs. A further useful figure would have been to assess the funding for the two fields by federal funding institutions such as the German Research Council (DFG), the federal ministry for Education and research (BMBF) or the European Union and the industry [##REF##12370198##27##,##REF##16300096##28##]. However, the precise funding from these sources is not accessible since some institutions and departments do not uncover these figures. In specific, industry funding is often not published as demonstrated by the tobacco industry funding policies [##REF##17523993##29##,##REF##17565125##30##]. Therefore, the present study was limited to monitor only the state financial input in terms of established independent full professorships/chairs at medical faculties.</p>", "<p>A further potential bias within the methodology of the present study is related to the issue of linguistic differences as previously discussed [##REF##18554379##31##]. In this respect, the present analyses encompassed all languages included in the data bases. The majority of publications is published in English and it is difficult for non-English journals to get included in the data bases. Therefore, numerous scientific publications in languages other than English are not accessible. However, the major German cardiovascular (Z Kardiol – Clinical Research in Cardiology) and respiratory journals (Pneumologie) are included in the data base. Also, it is generally accepted that German scientists publish their high quality research in scientific journals that use English as language.</p>", "<p>Large differences were present between the two fields: All medical faculties had chairs for cardiology. At the Berlin medical faculty, three cardiology chairs were present but not an independent single chair for respiratory medicine. This field was subordinated and headed by a full professorship for cardiology and a full professorship for infectious diseases. The presence of three independent cardiology chairs in Berlin is most probably due to historical reasons since this faculty was divided into two faculties during period of the Berlin wall and reunified in 2002/2003 [##UREF##1##32##].</p>", "<p>In striking contrast to the high number of cardiology chairs, only 8 chairs for respiratory medicine were present in Germany. The regional distribution as assessed by density equalizing mapping demonstrated a focus in North Rhine-Westfalia and Hesse. The largest state Bavaria did not have a chair of respiratory medicine.</p>", "<p>After the demonstration of an imbalance in the financial input (36 cardiology versus 8 respiratory medicine state-financed clinical university departments), the present study aimed to analyze potential imbalances in output figures. Therefore, the quantitative measure \"number of publications\" and the qualitative measures of \"overall numbers of citations\" and \"average citations per published item\" were used. In general, the imbalance in financial input is paralleled by an imbalance in overall quantitative output figures. I.e. the 36 cardiology full professorships published 2708 articles in comparison to 453 articles published by the 8 respiratory medicine full professorships. This is a ratio of 75.2 articles per cardiology chair and 56.62 articles per respiratory medicine chair. A similar trend is also present in the qualitative measures. Here, the 2708 cardiology professorship publications were cited 48337 times which is an average citation of 17.85 per publication. The average number of citations per cardiology chair was 1342.69. For respiratory medicine, the 453 publications were cited 7290 times. This is an average citation number of 16.09 per publication and a ratio of 911.25 citations per respiratory medicine chair.</p>", "<p>Interestingly, the citations per state and the number of publications per state varied to a large extend between the different states and the two fields of internal medicine. For respiratory medicine, the maximal number of publications per state was 255 for the 2 chairs in Hesse. These 255 publications were cited 5442 times which is an average citation per published item of 21.34. This is a ratio of 127.5 publications per professorship in Hesse and a ratio of 2721 citations per professorship in Hesse. By contrast, the best ratios in the field of cardiology were found for Saxony. Here, the ratio of publications per professorship was 109.5 and the ratio of citations per professorship was 4039.</p>", "<p>Closer analysis revealed that the most cited publications for the Saxony cardiology professorships were articles in which the full professor was co-author [##REF##12050336##33##] whereas the most cited publications for the Hesse respiratory medicine professorships were senior authorships [##REF##12151469##34##, ####REF##12354470##35##, ##REF##11926786##36####11926786##36##].</p>", "<p>The reasons for the presently analyzed imbalances are numerous: I.e. the high income country Germany is known to have an extremely low ratio of respiratory physicians in comparison to other European countries (as indicated in the European Lung White Book [##UREF##2##37##]. Therefore, a lower number of respiratory specialists may lead to a lower research activity. 2) The number of full professorships and department chairs for respiratory medicine at the highest level (C4/W3) is disproportional in Germany in comparison to other countries since there are 36 chairs for cardiology but only 8 for respiratory medicine. This imbalance leads to a lower research activity with a lower number of publication entries in the database.</p>", "<p>An important issue is related to the reason for this difference of 8 vs. 36 university chairs at German medical schools. Two major reasons may account for the imbalance: 1) History: in the times of tuberculosis at the beginning of the 20^thcentury, respiratory disorders were treated in remote hospitals but not in university medical schools. When the faculties started to create new chairs for internal medicine after the second world war, respiratory medicine capacities was not present at the medical faculties but in remote hospitals and the denomination of the chairs was directed towards cardiology. As a result, respiratory medicine is now underrepresented at German faculties in comparison to i.e. the UK. 2) Economics: Interventional and diagnostic procedures in cardiology such as left heart catheter offer a larger financial benefit to the faculties than respiratory interventional and diagnostic procedures [##REF##12053682##38##,##REF##12076573##39##]. Therefore, economic features may direct the faculties to the direction of cardiology professorships. Future studies should analyse these imbalances in closer detail.</p>", "<p>It is difficult to interpret how input imbalance affects on the output ratios. I.e. the allocation of public and private funding to a specific field such as cardiology and the consecutive concentration of financial resources in specific fields may lead to an increase of research actors or promotion of networking between outside institutes in this area. This may then lead to increased research activities resulting in production of higher-impact publications, eventually, obtaining more funding. Eventually a circle structure may appear that leads to the phenomenon that the rich areas automatically get richer [##REF##15244892##40##].</p>", "<p>In conclusion, the present study used input and output benchmarking in combination with density equalizing mapping to assess differences in the two important fields of cardiovascular and respiratory medicine. Germany was used a model high income country. A major imbalance in the state financial input was present with 8 respiratory medicine versus 36 cardiovascular state-financed full clinical university departments at the C4/W3 salary level. This difference in the state financial input was paralleled by large differences in overall quantitative output figures with 2708 published cardiology articles in comparison to 453 respiratory medicine in the period between 2002 and 2006. However, there was also a difference between the two fields in the qualitative citation analysis. Here, cardiology publications had an average citation of 17.85 per publication whereas the respiratory medicine publication had an average citation of 16.09 per publication. This small difference might be due to the fact that a larger number of professorships lead to a larger number of networking collaborations and citations. Despite a high significance of both cardiovascular and respiratory diseases for the burden of disease, large differences are present in Germany. This should be realized for health policy and research funding allocation.</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Historical, social and economic reasons can lead to major differences in the allocation of health system resources and research funding. These differences might endanger the progress in diagnostic and therapeutic approaches of socio-economic important diseases. The present study aimed to assess different benchmarking approaches that might be used to analyse these disproportions. Research in two categories was analysed for various output parameters and compared to input parameters. Germany was used as a high income model country. For the areas of cardiovascular and respiratory medicine density equalizing mapping procedures visualized major geographical differences in both input and output markers.</p>", "<title>Results</title>", "<p>An imbalance in the state financial input was present with 36 cardiovascular versus 8 respiratory medicine state-financed full clinical university departments at the C4/W3 salary level. The imbalance in financial input is paralleled by an imbalance in overall quantitative output figures: The 36 cardiology chairs published 2708 articles in comparison to 453 articles published by the 8 respiratory medicine chairs in the period between 2002 and 2006. This is a ratio of 75.2 articles per cardiology chair and 56.63 articles per respiratory medicine chair. A similar trend is also present in the qualitative measures. Here, the 2708 cardiology publications were cited 48337 times (7290 times for respiratory medicine) which is an average citation of 17.85 per publication vs. 16.09 for respiratory medicine. The average number of citations per cardiology chair was 1342.69 in contrast to 911.25 citations per respiratory medicine chair. Further comparison of the contribution of the 16 different German states revealed major geographical differences concerning numbers of chairs, published items, total number of citations and average citations.</p>", "<title>Conclusion</title>", "<p>Despite similar significances of cardiovascular and respiratory diseases for the global burden of disease, large input and output imbalances have been revealed in the present study which point to a need for changes in funding policies. The present study supplies data that could be used for decision making in the field of health systems funding.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>BGK, AF, TW, DQ, and CS contributed to the conception and design of the study, BGK and CK performed the analysis, BGK and CS prepared the first draft and all authors contributed to the writing of the manuscript.</p>", "<title>Supplementary Material</title>" ]