点击显示 收起
1 From the University of California Davis, Davis, CA (CHH, DHW, JMP, and CHW); the University of Oxford, Oxford, United Kingdom (HR and ADS); and the University of Bergen, Bergen, Norway (OKN, PMU, SEV, and GST)
2 HR is currently at the Department of Nutrition, Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway. 3 Supported by research grants DK56085 and DK 35747 from the US National Institutes of Health to CHH and by a grant from the Norman Collisson Foundation, UK, to ADS. Data on the subject volunteers were collected as part of the Hordaland Health Study 1997–1999 in collaboration with the Norwegian National Health Screening Service and with financial support from the Norwegian Research Council and the Advanced Research Program of Norway. 4 Reprints not available. Address correspondence to CH Halsted, Room 6323, Genome and Biomedical Sciences Facility, University of California Davis, Davis, CA, 95616. E-mail: chhalsted{at}ucdavis.edu.
ABSTRACT
Background: Glutamate carboxypeptidase II (GCPII) encodes for intestinal folate hydrolase and brain N-acetylated -linked acidic dipeptidase. Previous studies provided conflicting results on the effect of the GCPII 1561CT polymorphism on folate and total homocysteine (tHcy) concentrations.
Objective: We aimed to determine the potential effects of 2 polymorphisms of GCPII on plasma folate and tHcy concentrations, cognition, anxiety, and depression in a large aging cohort of Norwegians enrolled in the Hordaland Homocysteine Study.
Design: DNA samples were genotyped for the GCPII 1561CT and 484AG polymorphisms, and the results were linked to plasma folate and tHcy concentrations and to scores for cognition, anxiety, and depression.
Results: The 2 polymorphisms were in linkage disequilibrium and were associated with concentrations of tHcy. After adjustment for covariates, persons in the CT or combined CT and TT groups of the 1561CT polymorphism had higher plasma folate concentrations and lower tHcy concentrations than did those in the CC group. Subjects with the TT genotype had lower Symbol Digit Modalities Test (SDMT) scores than did subjects with the CC genotype. Compared with abstainers, moderate alcohol drinkers had higher plasma folate concentrations and higher scores on the Mini Mental State Examination. However, women abstainers with the CT genotype had lower SDMT scores than did abstainers with the CC genotype or moderate drinkers with the CT genotype.
Conclusions: The 1561CT polymorphism is associated with higher plasma folate and lower tHcy concentrations and with lower SDMT cognitive scores in women who abstain from alcohol.
Key Words: Glutamate carboxypeptidase II • GCPII • folate • homocysteine • cognition • gene polymorphisms
INTRODUCTION
Glutamate carboxypeptidase II (GCPII; EC 3.4.17.21) is located at chromosome 11p11.2, incorporates 19 exons, and encodes for 751 amino acids (1, 2). Collectively, GCPII expresses 2 separate enzymes: intestinal folate hydrolase, which regulates folate absorption by the cleavage of glutamates from dietary polyglutamyl folates, and brain N-acetylated -linked acidic dipeptidase, which produces glutamate and N-aspartyl acetate (NAA) and regulates neurotransmission (3, 4). Our original report (5) found a functional polymorphism at 1561CT in human intestinal GCPII that gave a frequency of 0.08 for CT heterozygotes. This polymorphism was associated with lower folate and higher total homocysteine (tHcy) blood concentrations in 75 healthy elderly persons residing in Oxford, United Kingdom, and the activity of folate hydrolase was lower by one-half in mammalian COS-7 cells that were transfected with a homozygous mutant (5). However, this finding of the effects of the 1561CT polymorphism on folate and tHcy concentrations was not confirmed by subsequent studies of larger populations residing in the United States or in northern Europe (6-9), including several that found that its presence had the opposite association of lower serum or red cell folate concentrations (10-13).
The Hordaland Homocysteine Study evaluated apparently healthy adults residing in Bergen and surrounding communities in western Norway, of whom 18 044 were screened initially in 1992–1993 and 7074 of the same participants were screened a second time in 1998–1999 (14, 15). The first Hordaland Homocysteine Study established separate ranges of tHcy concentrations for men and women and found that dietary folate intake, smoking status, and consumption of coffee were major determinants of elevated tHcy (14). The second Hordaland Homocysteine Study found that tHcy concentrations were associated positively with cardiovascular events and with higher serum creatinine concentrations but were significantly lowered by vitamin supplementation and quitting smoking (15). Of interest, tHcy concentrations were marginally lower in smokers who consumed moderate amounts of alcohol (16).
The goals of the present study were first to determine the potential relation of the 1561CT polymorphism in GCPII to an additional polymorphism at nucleotide 484 (484AG) in DNA samples that were available from subjects in the second Hordaland Homocysteine Study cohort. Second, we determined the relations of each polymorphism and their haplotypes with folate and tHcy concentrations and with scores of cognition, depression, and anxiety.
SUBJECTS AND METHODS
Subjects
The second round of the Hordaland Homocysteine Study was conducted as part of the Hordaland Health Study from 1997 to 1999 as a collaboration between the National Health Screening Service of Norway, the University of Bergen, and local health services. Each subject provided blood samples for DNA extractions and measurements of plasma folate and tHcy concentrations. We also collected lifestyle data and results of neuropsychiatric testing. Of the 7074 participants, 3341 were born between 1925 and 1927 (16). From this group, we selected 2734 subjects who provided comprehensive data on lifestyle variables, folate and tHcy values, and neuropsychiatric scores, as well as high-quality DNA samples. Among the final group, 43.4% were men and 56.6% women aged between 70 and 72 (Table 1). The study protocol was approved by the Regional Committee for Medical Research Ethics, and each participant provided informed consent to participate.
View this table:
TABLE 1. Characteristics of the subjects selected for study1
DNA preparation
DNA was extracted from frozen whole-blood samples in 96-well plates with the use of a NucleoSpin Multi 96 blood extraction kit (Macherey-Nagel, Düren, Germany) and was subsequently shipped to the University of California, Davis, laboratory for genotyping. Among the DNA samples, 2734 provided satisfactory genotyping patterns for at least one polymorphism, and 2471 samples provided satisfactory patterns for both polymorphisms.
Genotyping
The 1561CT (H475Y) polymorphism that was initially located in exon 13 of the catalytic region of GCPII (5) is now identified at C1684T in Ensembl (FOLH1, ENSG00000086205). In keeping with prior studies (5-8, 10-13), however, this polymorphism will be referred to as 1561CT in the present report. The second polymorphism at 484AG (H75Y) was identified through the Celera Database and is now identified through Ensembl as a missense mutation in exon 2 of the structural transmembrane region. Both polymorphisms were analyzed on an ABI 7900 genotyping machine using the 5' nuclease (Taqman) assay and a high-throughput genotyping method that uses allele-specific fluorogenic probes (Applied Biosystems, Foster City, CA). The primers and probes for genotyping of 484AG and 1561CT were provided by the Assays by Design service of Applied Biosystems (Foster City, CA). The fluorescence detection system relies on 2 allele-specific probes that contain different fluorescent reporter dyes (FAM and VIC) to differentiate the amplification of each allele when the probes are cleaved by the 5' nuclease activity of Taq DNA polymerase. Each probe consisted of an oligonucleotide with a 5' reporter dye (FAM or VIC) and a 3' nonfluorescent minor-groove-binding quencher dye that anneals specifically to complementary sequences between the forward and reverse primer sites (Applied Biosystems). The primers for 484AG included 5'AACTTGTCCATATAAACTTTCGAG GATGT3' (sense) and 5'GCATTTTTGGATG AATTGAAAGCTGAGA3' (antisense), and the probes were 1) VIC-CATCAA GAAGTTCTT ATAGTAAG and 2) 6FAM-TCAAGA AGTTCTTACAG TAAG. The primers for 1561CT included 5'GAGTTGATTGTA CACCGCTGATG3'(sense) and 5'CCACCTATGTTTA ACATAATACCTCAAG3' (antisense), and the probes were 1) 6FAM-CTTGG TACACAACCTAA and 2) VIC-AGCTTGGT ATACAACCT. PCR reactions were performed with the following thermal cycler conditions on a GeneAmp PCR System 9700 or ABI PRISM 7900HT Sequence Detection System: 95 °C for 10 min, followed by 40 cycles of 92 °C for 15 s and 60 °C for 1 min. The reaction components were as follows: 1X Taqman Universal PCR Master Mix, 900 nmol/L of each primer, 200 nmol/L of the probe for allele 1, 200 nmol/L of the probe for allele 2, and 25 ng genomic DNA. A post-PCR plate read on the 7900HT was used to determine genotype. The 7900HT Sequence Detection System collects fluorescence data on the samples for 5 s, and SDS software version 2.1 (Applied Biosystems) analyzes the fluorescence signals, which can be visualized in graph form. Samples that could not be determined were repeated, and unreadable results on the second run were scored as missing. All DNA genotyping was performed and analyzed at the University of California, Davis, laboratory, and the results were linked to phenotypic variables provided from the Hordaland Homocysteine Study data bank.
Measurement of plasma folate and tHcy and lifestyle variables
All subjects provided plasma samples for measurements of tHcy by automated HPLC with fluorescence detection (17), for measurement of folate and vitamin B-12 by microbiological assays (18, 19), and for measurement of serum creatinine concentrations. Lifestyle data were obtained for smoking as cigarettes per day, for alcohol consumption as drinks per day where one drink was estimated to be equal to 12.8 g ethanol, and for the use of supplemental multivitamins. Additional information was obtained on levels of education and on the incidence of cardiovascular events during the prior 6 y. Cognitive function was tested according to a shortened form of the Mini Mental State Examination (s-MMSE) (20), the symbol digit modality test (SDMT) (21), the Kendrick Object Learning Test (22), and the block design test, for which better cognition is signified by higher scores on each test, and by the trail making test, in which high scores indicate poor cognitive performance. Anxiety and depression were assessed by the Hospital Anxiety and Depression Scale (23) in which higher scores signify worse results.
Statistical analyses
Analyses were conducted with SAS for WINDOWS release 9 (Cary, NC). Continuous variables were assessed for conformance to the normal distribution and were transformed if appropriate. Relations between the 2 polymorphisms were determined by chi-square analysis. The Hardy-Weinberg equilibrium for each polymorphism and linkage disequilibrium between the 2 polymorphisms were determined with the 3.31 version of HAPLOVIEW (Cambridge, MA; http://www.broad.mit.edu/mpg/haploview/), where P > 0.05 indicates compliance with Hardy-Weinberg equilibrium and P < 0.05 represents significant linkage disequilibrium. The strength of the linkage disequilibrium was provided by the D' score and 95% confidence boundaries, for which a value of 1 indicates complete and 0 indicates no linkage disequilibrium. Associations between continuous outcomes and the 2 GCPII polymorphisms were first tested with analysis of variance, followed by analyses of covariance to control for age, sex, smoking, alcohol use, educational level, coffee consumption, serum vitamin B-12 and creatinine concentrations, and multivitamin use, as well as to test for interactions between these covariates and genotypes. Tukey's test was used for post hoc comparisons.
RESULTS
The characteristics of the 2734 subjects studied in the second Hordaland Homocysteine Study (1998–1999) whose phenotypic data and DNA genotypes were entered into the present study are provided in Table 1. Within the cohort, 10.5% reported a cardiovascular event within the previous 6 y; such events included myocardial infarction, stroke, thrombosis, or cardiovascular surgery. Alcoholic beverage intake was categorized as none (abstainer) in 51.0%, 1 to 14 drinks/wk (2 drinks/d; moderate) in 47.5%, and >14 drinks/wk (>2 drinks/d; excessive) in 1.5%. As shown, there were fewer abstainers and more drinkers among the men than among the women. Furthermore, the patterns of beverage consumption differed by sex (P = 0.0004): men consumed 28% of their drinks as beer, 40% as wine, and 32% as spirits; the values for women were 9%, 77%, and 14%, respectively. Smoking was categorized as never smoked in 36.1%, former smoker in 48.4%, and current smoker in 15.5%. High school education was completed by 11.4% and university education was completed by 7.9% of the subjects. All but 7% of the study population drank coffee, and one-third drank between 4 and 10 cups/d. Supplemental multivitamins were used by 28.1% of the subjects.
The allele pair distributions of each genotype and the frequencies of their combinations among 2471 samples for which both polymorphisms were identified are shown in Table 2. The allele distributions for each polymorphism were in Hardy-Weinberg equilibrium, and the D' value indicated strong linkage disequilibrium between the 2 polymorphisms. Of 9 possible haplotype pairs, 6 were present in frequencies 0.5% (0.005) and are represented in the subsequent tables.
View this table:
TABLE 2. Combined genotype distributions of the glutamate carboxypeptidase II 484A6 and 1561CT polymorphisms, 1998–19991
The associations of plasma concentrations of folate and tHcy with each of the 2 GCPII polymorphisms at 484AG and 1561CT and their haplotypes are shown in Table 3. All analyses were adjusted for age, sex, smoking, education level, alcohol use, coffee consumption, serum vitamin B-12 and creatinine concentrations, and supplemental vitamin use. The numbers of subjects available for accurate analysis of each polymorphism in separate DNA samples (Table 3) exceeded the numbers of subjects available for accurate analysis of both polymorphisms in the same DNA samples (Table 2). Consistent with prior observations (14), serum creatinine concentrations were highly associated with plasma tHcy but not with plasma folate concentrations (not shown). The main effect of the 1561CT polymorphism was significant for both plasma folate and plasma tHcy concentrations (P = 0.0005, P = 0.0004). Compared with those with the 1561CC wild type and with control for serum vitamin B-12 and creatinine and all other variables, plasma folate concentrations were higher and tHcy concentrations were lower in the 10.6% of individuals who were CT heterozygotes (P = 0.0003 for both comparisons) and in the 11.2% of individuals who were carriers of the T allele in the combined population of 1561CT heterozygotes and TT homozygotes (P < 0.0001 for both comparisons). The main effect of the 484AG polymorphism was significant for tHcy (P = 0.026), with values in AG heterozygotes greater than those in the GG homozygotes (P < 0.05). There were no significant interactions between the 2 polymorphisms, and there were no significant pair-wise differences between the haplotype combinations. There was no association of either genotype with cardiovascular disease risk as determined by the documented occurrence of angina, myocardial infarction, stroke, or vascular surgery in the 6-y interval between the first and second Hordaland Homocysteine Studies.
View this table:
TABLE 3. Associations of glutamate carboxypeptidase II (GCPII) genotypes with plasma concentrations of folate and total homocysteine (tHey)1
After the analysis was controlled for lifestyle variables, education level, age, sex, and serum concentrations of vitamin B-12 and creatinine, there were no associations of the GCPII 484AG or 1561CT polymorphisms with selected cognitive test scores, including scores for the Kendrick Object Learning Test, block design, trail making, and the s-MMSE. However, there was a significant main effect of the 1561CT polymorphism on SDMT scores (P = 0.036), which were lower in the small group of 8 subjects with the TT genotype than in those with either the CC or CT genotype of 1561CT (P < 0.027 and P < 0.031, respectively; Table 4). There was no effect of either genotype on total anxiety, whereas the depression score was marginally higher in subjects with the TT genotype of 1561CT than in those with the CC or CT genotype (P < 0.09 and P < 0.10, respectively).
View this table:
TABLE 4. Associations of glutamate carboxypeptidase II (GCPII) genotypes with cognition, anxiety, and depression1
Because the Hordaland Homocysteine Study had previously shown an inverse association between moderate alcohol consumption and plasma tHcy concentrations in nonsmokers (16), we looked for interactions among alcohol intake, sex, and 1561CT effects on folate, tHcy, and cognition while controlling for the other variables. As shown in Table 5, we found a main effect of alcohol use on plasma folate concentrations (P = 0.0015) and on the s-MMSE score (P = 0.0053), and the moderate use of alcohol was associated with higher plasma folate and s-MMSE scores than those in abstainers (P = 0.0012 and P = 0.0045, respectively). The effect of alcohol use on the s-MMSE score remained significant after controlling for folate and tHcy concentrations.
View this table:
TABLE 5. Effects of moderate alcohol use on folate and homocysteine concentrations and cognitive scores1
As shown in Table 6, we found a significant three-factor interaction among sex, alcohol use, and the 1561CT genotype on the SDMT score (P = 0.021). Among the 804 women who abstained from alcohol, the 92 women with the CT genotype had a lower mean score for the SDMT cognition test than did the 707 women in the CC group (P = 0.042). At the same time, among women with the CT genotype, the 92 women abstainers had a lower mean SDMT score than did the 38 women who consumed moderate amounts of alcohol (P = 0.016). These effects on the SDMT score remained significant after controlling for folate and tHcy concentrations.
View this table:
TABLE 6. Effects of moderate alcohol consumption, sex, and 1561TCT genotype on Symbol Digit Modalities Test score1
DISCUSSION
The present analysis of 2734 DNA samples from the 1997–1999 Hordaland Homocysteine Study had sufficient statistical power to compare the distributions of 2 GCPII polymorphisms with metabolic or lifestyle phenotypes. The major finding of the present study relates to the effect of the 1561CT polymorphism on plasma folate and tHcy concentrations (Table 3). Although the actual differences were small, the data in this large population of subjects were highly significant in showing that either CT heterozygosity that occurred in 10.6% of the population or the presence of the T allele that occurred in 11.2% of the population who carried the CT or TT genotype imparted higher plasma folate and lower tHcy concentrations, irrespective of age, sex, alcohol consumption, smoking, education level, coffee consumption, serum vitamin B-12 or creatinine concentrations, or multivitamin use. In addition, our study showed a weaker but significant main effect of the 484AG polymorphism on raising tHcy concentrations, although there were no interactions or haplotype effects of the 2 genotypes (Table 3).
Except for one report (8), all previous studies of the GCPII 1561CT polymorphism found population distributions of the CC, CT, and TT genotypes that were similar to those shown in Table 2. As summarized in Table 7, the heterozygous CT genotype was shown to have no effect (6-9), to be associated with higher plasma or red cell folate concentrations (10-13), or, in our original study, to be associated with lower folate and higher tHcy concentrations (5). One of the studies found that the bioavailability of polyglutamyl folate was not significantly different among carriers of the heterozygote or wild-type forms of GCPII (12). The present and largest study is the only one to show a lowering effect of the 1561CT polymorphism on plasma tHcy concentrations along with an elevation of plasma folate concentrations (Table 3). From these comparisons, we conclude that the findings of our original report (5) were based on an insufficient number of subjects for genetic comparisons and on an in vitro transfection experiment that excluded other physiologic effects on genetic expression.
View this table:
TABLE 7. Summary of published studies of the effects of the glutamate carboxypeptidase II (GCPII) 1561CT polymorphism on concentrations of folate and total homocysteine (tHcy)
The observed relations of the CT and combined CT and TT genotypes of the 1561CT polymorphism with higher plasma folate and lower tHcy concentrations (Table 3) suggest a positive effect of the T allele on increasing the efficiency of intestinal folate hydrolase. Expressed as folate hydrolase in the brush border surface of the small intestine, GCPII regulates the initial digestion of dietary polyglutamyl folates, which is an essential step for subsequent transport of monoglutamyl folates across the intestine and which regulates plasma folate concentrations (24). Subsequently, tHcy concentrations are regulated by circulating folate, which provides substrate for the transmethylation of tHcy to methionine in the liver and other tissues (25).
The secondary findings of the present study suggest that cognitive function is affected by the 1561CT polymorphism in a way that may be influenced by moderate alcohol consumption. Although folate concentrations were higher and tHcy concentrations were lower in the CT heterozygotes and in the combined CT and TT groups than in the CC homozygotes (Table 3), the 1561CT polymorphism had a main effect on the SDMT score, which was found to be significantly lower in the small number of individuals with the TT genotype than in those with the CC genotype, even when the analysis was controlled for plasma concentrations of folate and tHcy (Table 4). First described in 1968 (21), the SDMT assesses attention, visual scanning, and motor speed and is not significantly affected by age or sex (26).
When evaluating the effects of alcohol consumption and controlling for other variables, on the one hand we found no association of moderate drinking with tHcy concentrations (Table 5). On the other hand, we found main effects of moderate drinking on both plasma folate concentrations and cognition according to the s-MMSE score, whereby moderate drinkers had higher folate concentrations and s-MMSE scores of cognition than did abstainers across all groups (Table 5). Although a cross-sectional study of 801 adults aged >65 y found no effect of the level of alcohol consumption on the MMSE score (27), several prospective studies found that age-related changes in the MMSE score are favorably influenced by moderate alcohol consumption (28-30). For example, a 7-y study of elderly subjects found that light to moderate drinkers had lesser declines in the MMSE score than did nondrinkers (29), an 11.5-y study of 1488 subjects of all ages found less cognitive decline in women who drank moderately than in nondrinkers or heavy drinkers (30), and a 2-y Chinese study of elderly subjects found that light to moderate drinking of wine but not beer was associated with less risk of cognitive decline (28). Other studies have shown differing effects of moderate to excessive alcohol consumption on folate and tHcy concentrations. For example, plasma tHcy concentrations were found to be elevated in excessive alcohol users (31, 32), potentially because of an inhibitory effect of alcohol and its metabolite acetaldehyde on methionine synthase, which regulates the transmethylation and reduction of tHcy (33). Larger epidemiologic studies, including the Framingham Offspring study (34) and the third National Health and Nutrition Examination Survey (35), found an association of progressive alcohol use with elevated tHcy. However, a Dutch study found that that alcohol use was associated with lower tHcy concentrations (36), and a French study concluded that beer drinking lowered tHcy, whereas it was elevated in wine drinkers (37). Others described a protective effect of beer consumption on tHcy concentrations, presumably because of its high folate content (38). A study in Wales suggested a protective effect on ischemic heart disease of the high folate content of beer, the preferred alcoholic beverage (39). However, the present findings on the permissive effect of moderate alcohol intake on plasma folate concentrations appears to be unrelated to beverage choice, because wine accounted for most of the alcohol consumption and beer for only about one-quarter of reported alcohol intake overall. Contrasting with previous data from the Hordaland Homocysteine Study (40), the positive effect of moderate alcohol consumption on improved s-MMSE score persisted after the analysis was controlled for plasma concentrations of folate and tHcy (Table 5).
After removing a potential protective effect of moderate alcohol consumption, we found also that SDMT cognition scores were significantly lower in the abstinent women in the CT group than in those in the CC group or in moderate drinkers in the CT group (Table 6). This finding is consistent with a prior prospective study and suggests that the positive effect of moderate drinking on overall cognition according to the MMSE score occurs in women but not in men (30), while indicating an additional genotype effect. The observed association of the CT genotype with lower tHcy concentrations (Table 3) and with lower SMTP scores of cognition in a subset of women who abstained from alcohol (Table 6) conflicts with previous findings from the second Hordaland Homocysteine Study and others that reduced cognition is associated with elevated tHcy (40, 41). Although the P values for the association of sex, alcohol status, and the 1561CT genotype shown in Table 6 were significant after the Tukey test, it is possible that these are spurious findings due to the type 1 error that could result from the increasing numbers of comparisons.
Assuming that the finding of reduced SDMT scores in abstinent women with the CT genotype despite unchanged tHcy is true (Tables 3 and 6), the data suggest that the effect of the CT genotype on cognition in abstinent women is unrelated to its effect on intestinal folate hydrolase and subsequent folate and tHcy concentrations. Alternatively, this observation could reflect the influence of the T allele on the expression of GCPII as N-acetylated -linked acidic dipeptidase, a brain enzyme that regulates the cleavage of N-acetyl-L-aspartyl-L-glutamate to NAA and the neurotransmitter glutamate (42). Recent data indicate that N-acetylated -linked acidic dipeptidase is widely distributed in astrocytes in the human brain, whereas its higher activity is associated with glutamate neurotoxicity in various neurodegenerative disorders (43). It is possible that regulation of the amount of glutamate produced and released from astrocytes (44) may be influenced by the GCPII genotype. Other data indicate that lower concentrations of NAA, which can be measured by in vivo magnetic resonance spectrophotometry, correlate directly with reduced cognitive function in aging individuals (45, 46) and that NAA is a more specific indicator of intelligence in women than in men (47). The present study points to a need for additional investigations on the effects of the 1561CT polymorphism of GCPII on the expression of brain N-acetylated -linked acidic dipeptidase and on its regulation of cognitive function.
ACKNOWLEDGMENTS
We are grateful to D Warden for DNA isolation from blood samples, K Engedal and HA Nygaard for assistance with the cognitive assessments, and AA Dahl for permission to use the data on anxiety and depression.
All authors participated sufficiently in the work to take public responsibility for appropriate portions of its content. The individual contributions of the authors were as follows—CHH: provided the original concept of the study, oversaw all aspects of the study and data analysis, and wrote the manuscript; DHW: performed all the genotyping, participated in its analysis, and wrote portions of the manuscript; JMP: provided all statistical analyses and a portion of the manuscript; CHW: provided consultations on genotyping and its analysis; ADS: participated in study design and data analysis, provided oversight of DNA collections, and provided critique of the manuscript; HR: participated in HUSK data collection, participated in study design and data analysis, and provided critical evaluation of the manuscript; PMU, SEV, and OKN: participated in HUSK data collection and provided critical evaluation of the manuscript; GST: supervised the HUSK data collection, participated in study design, and provided critique of the manuscript. None of the authors had any undisclosed financial or personal interest in any company or organization connected in any way to this article.
REFERENCES