© 2003 American Thoracic Society
The Major Histocompatibility Complex Gene Region and Sarcoidosis Susceptibility in African AmericansDepartment of Biostatistics and Research Epidemiology and Division of Pulmonary and Critical Care Medicine, Henry Ford Health System, Detroit, Michigan; and Division of Pulmonary and Critical Care Medicine, Mount Sinai Medical Center, New York, New York Correspondence and requests for reprints should be addressed to Benjamin A. Rybicki, Ph.D., Department of Biostatistics and Research Epidemiology, Henry Ford Health System, 1 Ford Place, 3E, Detroit, MI 48202. E-mail: brybick1{at}hfhs.org
Investigators have intensively evaluated the major histocompatibility (MHC) complex for sarcoidosis susceptibility genes with the majority of reports implicating the human leukocyte antigen (HLA)-DRB1 gene. Because most studies have been performed in white and Asian populations, we sought to determine which MHC genes might be risk factors for sarcoidosis in African Americans. We genotyped six microsatellite markers spanning 11.6 megabases that overlapped the MHC region on chromosome 6p21-22 in 225 nuclear families ascertained by African American probands with a history of sarcoidosis. Using a family-based association methods approach, we performed multiallelic tests of association between each marker and sarcoidosis. A statistically significant association was detected between sarcoidosis and the DQCAR marker (p = 0.002) less than two kilobases telomeric from the HLA-DQB1 gene. Typing two additional markers in this region revealed that DQCAR–G51152 haplotypes, spanning a 38-kilobase region across the HLA-DQB1 gene, were associated with sarcoidosis on a global level (p = 0.022). Analysis of individual DQCAR and G51152 alleles showed that the DQCAR 178 (expected = 21.0; observed = 10; p = 0.0005) and G51152 217 (expected = 25.6; observed = 14; p = 0.0009) alleles were transmitted to affected offspring less often than expected; whereas the DQCAR 182 allele was transmitted more often than expected (expected = 52.6; observed = 66; p = 0.002). Our results indicate that HLA-DQB1 and not HLA-DRB1 plays an important role in sarcoidosis susceptibility in African Americans. Identification of the specific HLA-DQB1 alleles that influence sarcoidosis susceptibility in African Americans and the study of their antigenic-binding properties may reveal why African Americans suffer disproportionately from this disease.
Key Words: family • African Americans • leukocyte antigens • human • linkage disequilibrium The search for host factors that influence sarcoidosis susceptibility has focused on genes involved in immune pathways (1). Many immune response genes are clustered in the major histocompatibility (MHC) region on chromosome 6p21-22, particularly the Class II region, which has the highest known concentration of immune-related genes (2). A recent genetic linkage study in German affected sib pairs found its strongest signal for linkage to sarcoidosis with a marker mapped to the MHC Class II region (3, 4). Reported associations of sarcoidosis susceptibility with MHC genetic polymorphisms are at best equivocal (5), but of the Class II molecules studied in sarcoidosis the majority of positive findings have been with the human leukocyte antigen (HLA)-DR–associated antigens (5–13). Most recently, a large multicenter case–control etiology study of sarcoidosis (ACCESS) in both whites and African Americans (14) extensively genotyped MHC Class II genes and reported the strongest association was with polymorphisms in HLA-DRB1. The HLA-DRB1*1501 allele conferred increased risk in whites but protection in African Americans (15). This result indicates that genetic associations in African Americans may be different than those observed in other populations. Before ACCESS only a few reports of HLA associations in African Americans existed, and these studies relied on serologic detection of HLA antigens (16, 17). A further limitation of previously reported sarcoidosis HLA associations is that nearly all these studies, including the ACCESS study, used the case–control methodology, and therefore results may have been potentially biased by population stratification (18). The results of family-based studies, which control for genetic background that may confound case–control results, add a higher level of certainty to putative regions of genetic susceptibility. To better understand the relationship of the MHC gene region to sarcoidosis susceptibility in African-American populations, we conducted an association scan over an 11.6-megabase (Mb) region overlapping the MHC region in a sample of African-American sarcoidosis families.
Study Sample and Data Collection The study protocol was approved by the Henry Ford Hospital Institutional Review Board. The majority of families were ascertained through African American individuals with a history of sarcoidosis who were seen at the Henry Ford Health System. Diagnosis was confirmed by a tissue biopsy demonstrating noncaseating granulomas in a majority of the index cases (86%). The cases without histologic confirmation of disease had radiographic evidence of bilateral hilar adenopathy and were observed for 2 years or more with no other medical condition that could explain radiographic abnormalities or clinical course. The participants described previously were first contacted by letter and then by phone. After consenting to the study, subjects underwent an interviewer-administered questionnaire. Information obtained from each participant included a medical history, occupational and environmental exposure history, and, for affected family members, a family history. Participants were also asked to donate 30 ml of blood for DNA analysis. Of the 623 eligible probands recruited for study, 359 (58%) were enrolled with one or more first-degree family members. Of these, 234 had two or more parents or sibs that donated a blood sample for DNA analysis. An additional 10 African-American sarcoidosis families were recruited outside of the Henry Ford Health System, resulting in 244 total families. Both parents were genotyped when available. When one or both parents were unavailable for genotyping, all available full sibs were genotyped. Using a panel of unlinked markers, we excluded 20 families (8.2%) with Mendelian segregation inconsistencies, and two families (0.8%) where on further follow-up the proband was found not to have sarcoidosis, from the analysis. Three of the remaining 222 families had two separate nuclear families, which were analyzed separately, resulting in 225 nuclear families or a total of 704 individuals for study. The three basic family configurations are shown in Table 1. Although the majority of family members with a sarcoidosis history were probands through whom the family was ascertained, there were also 22 sibs and four additional family members (who were treated as index cases in the analysis) with a history of sarcoidosis based on self-report of a physician diagnosis of this condition. In 18 of the 22 sib cases, and the other four additional family member cases, the family member reported that the diagnosis was confirmed by biopsy. In the remaining four sib cases, a medical record review revealed a chest radiograph report of bilateral hilar adenopathy at the time of the reported sarcoidosis diagnosis. Over 90% of the families in our study population had only one affected offspring, and in over 80% of the families additional unaffected sibs were enrolled to complete allelic transmission data where one or both parents were missing.
Genotyping We genotyped six polymorphic microsatellite markers that spanned a distance of approximately 11.6 Mb (approximately 5 cM based on the Marshfield genetic map—http://research.marshfieldclinic.org/genetics) overlapping the 3.6-Mb human MHC region on chromosome 6p21-22. Marker spacing was an average of 2.3 Mb with a spacing range of 7.5 kilobases (kb) to 4.8 Mb. High molecular weight DNA was isolated from anticoagulated blood by detergent lysis and organic extraction (19). Unlabeled and fluorescent dye–labeled microsatellite primers were purchased from Research Genetics (Huntsville, AL). Two protocols were used to amplify from genomic DNA, depending on the method of labeling. In the first case, one primer in each reaction was labeled at the 5' end with  [32P]ATP (20). The amplification reaction included 10 ng DNA; 1 to 2 pmol each primer; 50 mM Tris–hydrogen chloride (pH 8.3), 2.5 mM magnesium chloride, 0.05% Tween-20, 0.05% NP-40, 20 mM each deoxynucleotide triphosphate, and 1 U Taq DNA polymerase (Promega, Madison, WI). Samples were denatured by the addition of 10 µl 95% formatted, 20 mM ethylenediaminetetraacetic acid, and heating to 70°C for 2 minutes. Five microliters of sample were run on 4 or 6% polyacrylamide, 8 M urea gels containing 90 mM Tris–borate buffer pH 8.3, 2 mM ethylenediaminetetraacetic acid for 3 to 4 hours at 55 W. Gels were exposed to film for 18 hours at -70°C. Allele sizes were determined by running M13 DNA sequencing ladders in each gel for single base pair (bp) resolution. Individuals were genotyped according to allele size using DNA sizing markers as standards. Each DNA sample was assayed at least twice per marker, to confirm genotype. When fluorescent dye–labeled primers were used, the reaction conditions varied as follows: 2.5 to 1 pmol primer, 2.5 U Amplitaq Gold (Applied Biosystems, Foster City, CA).
Statistical Methodology
A haplotype-based test of association with disease was performed with TRANSMIT version 2.5 (23) to jointly test the association of haplotypes formed by adjacent markers with sarcoidosis. In TRANSMIT, an average score of non-Mendelian transmission is calculated for each family unit, considering all possible haplotype assignments. The family scores are used to calculate global and haplotype-wise Linkage disequilibrium (LD) between selected pairs of loci was also estimated. This was done in two stages; first, because haplotypes were not directly observable, maximum likelihood haplotype frequencies were computed using the ARLEQUIN genetic analysis software (24). These haplotype frequencies were next used to calculate the normalized measure for disequilibrium between loci, D', first proposed by Lewontin (25).
Association Scan of MHC Results Multiallelic test statistics for the six markers that we tested, spanning the 6p21-22 region, showed that only one marker, DQCAR, was associated with sarcoidosis (Table 2). This association had a level of statistical significance that was not affected by the genetic model tested (i.e., dominant model p = 0.002 versus additive model p = 0.001). On applying the Hochberg and Benjamini multiple significance testing correction method (26), both p values were still less than the corrected Type I error levels of 0.0045 and 0.0042, respectively. None of the other markers tested reached the initial 0.05 level of statistical significance. The marker with the next lowest p value was D6S1618 (p = 0.137 for the dominant model), approximately 4.6 Mb from DQCAR. Of the six markers tested, D6S1618 was the only marker where the type of genetic model (dominant versus additive) had an effect on the level of significance. Therefore, all subsequent testing was done solely under the dominant inheritance model.
Table 3 shows the statistics for individual DQCAR alleles. The most highly significant allele was the 178-bp length allele (p = 0.0005) that showed a marked under transmission to affected offspring (ratio of observed versus expected = 0.48). The next most significant allele was the 182-bp length allele (p = 0.002), which showed a slight over transmission to affected offspring (ratio of observed versus expected = 1.25). This allele was also the most prevalent allele at 49.1%. The only other allele that showed an association with sarcoidosis at the 0.05 significance level was DQCAR 190, which, like DQCAR 178, showed an under transmission to affected offspring (ratio of observed versus expected = 0.52).
Finer Mapping of HLA-DQA1–DQB1 Region To more closely investigate the strong association we observed between the DQCAR marker and sarcoidosis, we typed two additional microsatellite markers approximately 30 and 52 kb centromeric to HLA-DQB1. Figure 1 depicts the association of these two markers, and the DQCAR and DQCARII markers, in relation to their location in the HLA-DQA1–DQB1 gene region. G51152, the closest marker to HLA-DQB1 on its centromeric side, had an association with sarcoidosis that did not quite reach statistical significance (p = 0.08). The association between sarcoidosis and T16CAR, 22 kb further away from HLA-DQB1 than G51152, had a slightly lower significance level (p = 0.12). The G51152 allele that had the strongest association with sarcoidosis was 217, which was transmitted to affected offspring less often that expected (expected = 25.6; observed = 14; p = 0.0009).
Given that the DQCAR–G51152 haplotype spans the HLA-DQB1 gene, we tested whether the DQCAR–G51152 haplotype frequency distribution was associated with sarcoidosis. In our sample, 156 DQCAR–G51152 haplotypes were possible, but based on parental genotypes only 33 haplotypes had probabilities greater than zero. A global test of the association between the entire DQCAR–G51152 haplotype distribution and sarcoidosis was statistically significant (p = 0.022). Table 4 depicts the individual DQCAR–G51152 haplotype associations with sarcoidosis for those haplotypes with an expected frequency of five or greater. The DQCAR 178–G51152 217 haplotype, which was under transmitted to offspring with a sarcoidosis history, had the strongest association with the disease phenotype (p = 0.001). Alternatively, the DQCAR 182–G51152 223 haplotype, the most frequent haplotype representing 23.3% of all probable haplotypes, was over transmitted to offspring with a history of sarcoidosis (p = 0.028).
LD between the different locus pair combinations of the four microsatellite markers that spanned the HLA-DQA1–DQB1 region is depicted in Figure 2 . Overall, the D' statistic computed for six marker pairs appeared to have a negative linear relationship with physical distance. Only the G51152–T16CAR locus pair had a markedly lower LD value than would be expected based on the physical distance between these two markers. In Table 5, we depict the ten allele pairs of these four loci with the highest LD value corrected for allele frequency. The two highest LD values were both from the DQCAR–G51152 marker pair that spanned HLA-DQB1. The DQCAR 182–G51152 215 haplotype had the highest LD value and was present at a much lower frequency than expected. Conversely, the DQCAR 182–G51152 223 haplotype with the next highest LD value was present at a higher frequency than expected. Overall, the DQCAR 182 allele was in five of the ten haplotypes with the highest LD values, and DQCAR alleles were in seven of the ten haplotypes.
Our scan of the Class II MHC region on chromosome 6p21-22 showed that one marker, DQCAR, had a strong association with sarcoidosis (p = 0.002). This marker is approximately 1 to 2 kb telomeric to the HLA-DQB1 gene, which has had mixed allelic association results with sarcoidosis (10, 13, 27, 28). We typed two additional markers that are the closest microsatellite repeat markers centromeric to HLA-DQB1. The G51152 marker, 30 kb from HLA-DQB1, was marginally associated with sarcoidosis (p = 0.08). The next two closest functional genes to DQCAR and G51152, HLA-DQA1 and HLA-DRB1, have also been reported to be associated with sarcoidosis (5, 10, 13, 15, 28). Other reported associations with genes less than a megabase away from DQCAR include LMP7 (29), TAP1 and TAP2 (30), and HLA-DPB1 (28, 31, 32). LD as a gene mapping tool in the MHC region is complicated by "recombination hotspots" that lead to uneven LD across this region. For instance, much higher LD values have been observed between DQB1 and TAP2, 170 kb apart, than between TAP1 and TAP2, 15 kb apart (33). The HLA-DRB1–DQA1–DQB1 haplotype demonstrates nearly complete linkage disequilibrium in most populations (33–35). This complicates association studies of these genes because it is difficult to determine which specific HLA allele or alleles are predisposing to disease. Sarcoidosis association studies of DRB1, DQA1, and DQB1 alleles often reveal that at least one or more alleles of these Class II genes are associated with disease. Recombination between the HLA-DRB1–DQA1–DQB1 genes does occur, and it appears that historically more recombination has occurred between the smaller 25-kb region separating DQA1 and DQB1 than the 60 to 80 kb region separating DRB1 and DQA1 (33, 34). This suggests that DQA1–DQB1 haplotype associations may be easier to dissect at the genotypic level than DRB1–DQA1 haplotype associations. Furthermore, in African Americans, LD in the MHC Class II region may not be as complete as in white populations (34, 36, 37). For example, more diversity in HLA-DRB1–DQA1–DQB1 haplotypes have been found in African Americans compared with whites (37). A chromosomal crossover between DQCARII and DQCAR is the most likely explanation for this increased diversity (37). This would in part explain our positive association with DQCAR, but not DQCARII, even though the two markers are only 7.5 kb apart. In our African-American sample, we found that the LD between the DQCAR–DQCARII microsatellite markers was high (D' = 0.74) but not as high as LD values reported between the HLA-DQA1 and -DQB1 genes in white populations (33, 38), the two genes that form bookends around the DQCAR–DQCARII haplotype. Unfortunately, there is no disequilibrium measure totally independent of allele frequencies (39), which makes any comparison of D' values between locus pairs tenuous. However, Zapta recently showed that for microsatellite allele frequencies similar to those used in the present study, D' as a measure of LD is robust to variations in allele frequencies under most circumstances (40). The lack of an association between DQCARII and sarcoidosis would seem to rule out DQA1 as a possible sarcoidosis susceptibility gene in our population because DQCARII–DQA1 haplotypes are highly preserved (41). Our negative DQCARII association result also likely rules out DRB1 as a sarcoidosis susceptibility gene in African Americans, however the DQCARII–DRB1 association is not as complete as the DQCARII–DQA1 association (41). Studies of DQCAR–HLA-DQB1 haplotypes have shown that haplotypes with shorter DQCAR alleles have the least variation (42–44). We found that two of the three shortest DQCAR alleles had the strongest association with sarcoidosis: DQCAR 178 was negatively associated and DQCAR 182 was positively associated. In African Americans, DQCAR 178 and DQCAR 182 are most closely associated with DQB1*0201 and DQB1*0602, respectively (37). Two recent reports of the DQB1*0201 allele and sarcoidosis showed a strong positive association with sarcoid arthritis (45) and a strong negative association with sarcoidosis in its severe form (27). Our result of a marked under transmission of DQCAR 178 to sarcoidosis offspring is more consistent with the latter finding. The HLA-DQB1*0201 molecule is notable for missing an aspartate at residue 57, which appears to be critical for peptide binding (46) as well as T-cell recognition (47) and stability of the MHC class heterodimer on the cell surface (48). The DQB1*0602 allele has an aspartate at residue 57 and has been shown to be one of the most stable HLA-DQ molecules (49). Apart from the critical binding site at amino acid position 57, DQB1*0201 and DQB1*0602 also differ at three other amino acid positions, 30, 70, and 86, which are located in the peptide-binding groove of the molecule. Given the importance of antigen recognition and presentation in sarcoidosis, the disparate sarcoidosis risks likely conferred by DQB1*0201 and DQB1*0602, as evidenced by the DQCAR allelic associations observed, are not unexpected based on the contrasting amino acid configuration of their antigen-binding pockets. In summary, we detected a strong association between an HLA-DQ microsatellite repeat marker, DQCAR, and sarcoidosis in African American families. A finer association scan of a 75-kb region of DNA-spanning HLA-DQA1 and -DQB1 genes strongly suggests that HLA-DQB1 harbors sarcoidosis susceptibility alleles. In addition, based on reported (41, 42) and measured LD in the HLA-DRB1–DQA1–DQB1 gene region, HLA-DQA1 and -DRB1 are unlikely to be sarcoidosis susceptibility genes in our African-American study population. Allele-specific typing of HLA-DQB1 that is currently underway should help confirm whether it is the primary sarcoidosis susceptibility gene for African Americans in the MHC Class II region.
This work was supported by grant R01HL54306 from the National Heart, Lung, and Blood Institute. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Received in original form December 19, 2001; accepted in final form November 1, 2002
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
2 tests of deviance from expected transmission patterns. To protect against inflated Type I error, the global test was first checked for significance at the 0.05 level, before proceeding to individual haplotype tests of significance. 
