Abstract
Aim
The aim of this study was to test the microsatellites in the Type 1 Diabetes Genetics Consortium major histocompatibility complex (MHC) data set for association with type 1 diabetes (T1D) independent of the HLA-DRB1, -DQA1 and -DQB1 genes.
Methods
The data set was edited to contain only one affected child per family, and broad ethnic subgroups were defined. Genotypes for HLA-DRB1, -DQA1 and -DQB1 were replaced by a haplotype code spanning all three loci, with phase inferred based on common haplotypes. The final data set contained 8190 samples in 2301 families, 59 microsatellites and the DRB1–DQA1–DQB1 haplotype code. Statistical analyses consisted of conditional logistic regression and haplotype estimations and linkage disequilibrium calculations.
Results
The data set was screened using a main effects test approach adjusted for DRB1–DQA1–DQB1, and significant results tested for validity. After these procedures, four markers remained significant at the Bonferroni-corrected threshold: D6S2773 (p = 0.00014), DG6S185 (p = 0.00015), DG6S398 (p = 0.00043) and D6S2998 (p = 0.00015). These results were supported by allelic tests conditioned on DRB1–DQA1–DQB1 haplotypes, except for DG6S185, which may contain artefacts.
Conclusions
We have identified three microsatellites that mark additional risk factors for T1D at highly significant levels in the MHC. Further analyses are needed to establish the relationship with other possible genetic determinants in this region.
Keywords: conditional analysis, genetic association, major histocompatibility complex, microsatellites, type 1 diabetes
Introduction
The major histocompatibility complex (MHC) is a unique genomic area containing hundreds of genes at high density, of which an unusually high amount are involved in immunological processes [1]. It therefore comes as no surprise that genetic determinants for many diseases, including type 1 diabetes (T1D), are located in this region. The main genetic risk factors for T1D are known to be conferred by certain haplotypes of the HLA-DRB1, -DQA1 and -DQB1 loci within the MHC class II region [2–4]. However, previous studies have shown that additional loci within the MHC remain unidentified [5–9]. Still, mapping of this region has proved exceedingly difficult because of strong and extensive linkage disequilibrium (LD) with the HLA-DRB1, -DQA1 and -DQB1 loci. Controlling for the effects of these loci is therefore essential in any attempt to identify additional risk loci within this region. In addition, the effect sizes of additional risk factors are expected to be relatively small as is often the case for complex diseases. Therefore, large data sets are needed to obtain significant results, which are recognized in the efforts of the Type 1 Diabetes Genetics Consortium (T1DGC). The first main project under the auspices of the T1DGC is the MHC Fine Mapping Workshop. In this study, we have tested the microsatellite data gathered for this project for association with T1D independent of the HLA-DRB1, -DQA1 and -DQB1 loci.
Materials and Methods
TheT1DGCMHCdata set (February 2007 release) consists of a large collection of multiplex families (N = 2321), mostly Caucasians from Europe, USA and Australia, typed for 2957 SNPs, 66 microsatellites and all classical HLA loci. Microsatellite marker names starting with ‘DG6S..’ are internal nomenclature to deCODE genetics, and allele names are raw data calls.
Sample Selection
The data set was edited to contain only two generations per family. Twenty families were removed because of unlikely high numbers of recombinations between windows of two markers analysed using FAMHAP [10] on the complete T1DGC MHC data set. To avoid the problem of statistical dependence between siblings in the association analyses, we edited the data set to contain only one affected child (the proband) per family. Probands were specified as the child with the youngest age of onset, if available, other-wise as the affected child with the highest analytic ID. Unaffected siblings were retained to improve haplotype inference, whereas siblings diagnosed with T1D or without genotyping data were removed. The above procedures resulted in a total of 8190 samples in 2301 families.
Ethnic Groups
Ethnic groups were specified on the basis of self-reported ethnicity. The following groups were used in analyses: All (2301 families); Caucasian (2262 families): all families (including those where no ethnicity information was available) except those of Asian, African or indigenous ancestry; Northern Europe (781 families): families from the UK and Ireland, Scandinavia, Germany and the Netherlands; Eastern Europe (228 families): families from Poland, the Czech Republic, Slovenia, Hungary, Serbia, Romania, Bosnia and Slovakia; Southern Europe (161 families): families from Italy, Spain, Portugal, France and Malta.
Recoding of Markers
Genotypes for HLA-DRB1, -DQA1 and -DQB1 were replaced by a haplotype code spanning all three loci, with phase inferred based on common haplotypes. Mendelian consistency (tested using PEDCHECK [11]) was evaluated for quality control. To limit the number of input variables in the regression analyses, we grouped alleles with frequencies below 1% (calculated among parents), with the exception of the DRB1*0403-DQA1* 0301-DQB1*0302 haplotype (frequency 0.48%). For microsatellites, alleles with frequency below 0.1% were grouped in a common allele for each marker. One microsatellite (D6S2830) was removed altogether because all minor alleles were below this frequency threshold.
Microsatellite Marker Quality Checks
Mendelian inheritance of the microsatellites was evaluated using PEDCHECK, and no errors were detected. Hardy–Weinberg equilibrium was calculated in the above-described ethnic groups in the complete T1DGC MHC data set using the exact test implemented in the program PEDSTATS [12]. Microsatellites showing disequilibrium with p < 10−5 in any one of these populations, or < 10−3 in at least two separate populations, were removed (N = 6). This resulted in a final data set containing 59 microsatellites.
Regression Analysis
Conditional logistic regression, adjusted for DRB1–DQA1–DQB1 haplotypes, was used to test the overall effect of each microsatellite modelled by genotypes. We used the conditioning strategy 4 (with a likelihood ratio test) described in Cordell & Clayton [13], implemented in STATA (STATA Corp., College Station, TX, USA).
Allelic Tests on Haplotypes and LD Analyses
Maximum-likelihood haplotype frequency estimates combining markers with DRB1–DQA1–DQB1 haplotypes were computed using FAMHAP. Based on these estimates, haplotype transmission/non-transmission (T/NT) tables were constructed as described in Becker & Knapp [10] and organized in separate groups for each DRB1–DQA1–DQB1 haplotype. Statistically significant deviations of an allele within each DRB1–DQA1–DQB1 group were tested by Pearson’s Chi-squared test, and odds ratios were calculated. LD patterns were analysed using FAMHAP and the GOLD software package [14] based on the generalized D′ measure.
Results
Main Effects Tests
Single-point main effects tests of microsatellites, adjusted for DRB1–DQA1–DQB1 genotypes, were performed in the All data set and the three European subpopulations. Tests were also performed on the Caucasian data set, but results did not differ significantly from the All data set and are therefore not shown. The likelihood ratio test is affected by how rare genotypes are handled. For the initial screen, we used joint frequency cut-offs for all markers, adjusted for population size. To ensure the validity of our results, we further tested all results p < 0.05 for each microsatellite that showed significant results below the Bonferroni-corrected threshold (p = 0.05/59 = 0.00085) in any of the populations (treating the All data set as a separate population) using genotype cut-off frequencies adjusted for each microsatellite marker and population size. Because grouping of the DRB1–DQA1–DQB1 haplotype codes are more likely to have statistically adverse effects (because of potentially divergent functional properties of different haplotypes), the cut-off frequency for this marker was kept constant at a low level (p < 0.001).
The results of these tests are presented in figure 1 (all results p < 0.05, with validity corrected p values where applicable) and table 1 (significant results after validity testing and Bonferroni correction). All validity tested results from the Southern European population were removed as counts turned out to be too small to perform meaningful tests. Also, results that initially were exclusively significant in the Northern and Eastern European populations (D6S2843, and DG6S182 and DG6S745, respectively) rose above the Bonferroni-corrected threshold after validity testing. The results in the All population, however, withstood these tests and were retained: D6S2773 (p = 0.00014), DG6S185 (p = 0.00015), DG6S398 (p = 0.00043) and D6S2998 (p = 0.00015). Except for D6S2773, which also showed nominal significance in the Northern European population (p = 0.040), results in the subpopulations for these markers were not significant.
Fig. 1.
Main effects tests of microsatellites adjusted for DRB1–DQA1–DQB1 haplotypes; all results. Only results p < 0.05 are shown. Limits for statistical significance at p = 0.05 and p = 0.05 Bonferroni corrected for the number of markers tested (p = 0.05/59 = 0.00085) are indicated. Results that showed p < 0.00085 in any of the populations in the initial screen are marked with circles and marker names. The values given for these results are adjusted with results after validity testing (DG6S745 was adjusted to >0.05 and removed). Grey areas did not contain any tested markers. Positional values are along chromosome 6, genome build 36, and locations of some HLA genes are indicated for reference.
Table 1.
Main effects test of microsatellites adjusted for DRB1–DQA1–DQB1 genotypes; significant results after Bonferroni-correction and validity testing*
| Marker name (deCODE) |
Position | Closest public microsatellite (position) |
Closest RefSeq gene (position, relative location) |
p value | |
|---|---|---|---|---|---|
| All | Northern Europe |
||||
| D6S2773 | 29899213–29899413 | — | HLA-G (29903497–29906856, 5′) | 0.00014 | 0.040 |
| DG6S185 | 31153244–31153505 | D6S2814 (31153263–31153505) | C6orf15 (31186981—31188311, 3′) | 0.00015 | — |
| DG6S398 | 32385920–32386075 | D6S2734 (32386428–32386591) | C6orf10 (32368453—32447634, intron) | 0.00043 | — |
| D6S2889 | 32442738–32442882 | — | C6orf10 (32368453–32447634, intron) | 0.00015 | — |
Only those markers that showed significant results (Bonferroni corrected for the number of markers tested; p = 0.05/59 = 0.00085) after validity testing, together with the only result in the European subpopulations with p<0.05 for these markers, are shown. For microsatellites with marker names that are internal to deCODE, the closest public microsatellite is given with positions. The closest RefSeq gene to each marker is given with positions and relative location of the marker. Positions are on chromosome 6, genome build 36.
DG6S185, DG6S398 and D6S2889 showed deviations from Hardy–Weinberg equilibrium but not enough to be excluded by our criteria (p = 0.000034, 0.00020 and 0.0012, respectively, in the Caucasian data set, DG6S185 in addition p = 0.019 in the Northern European data set, otherwise not significant (n.s.)).
Allelic Associations on Haplotypes
To further explore the results from the main effects tests, we tested the individual alleles of each of the four significant microsatellites, conditioned on haplotypes with the DRB1–DQA1–DQB1 haplotype codes. Significant results at alpha = 0.05 of these analyses are given in table 2. Because D6S2773 also showed some association in the main effects test in the Northern European data set, this marker was also tested in this population. Results generally showed tendencies for risk associations in the same direction and/or for the same alleles as in the All population (some results were not significant and are therefore not shown).
Table 2.
Allelic tests of significant microsatellites from the main effects tests, conditional on DRB1–DQA1–DQB1 haplotypes†
| Conditional haplotypes | Microsatellites | |||||
|---|---|---|---|---|---|---|
| DRB1 | DQA1 | DQB1 | D6S2773 | DG6S185 | DG6S398 | D6S2889 |
| 01 | 0101 | 0501 | — | 2.84 (275) | — | 0.16 (151) |
| 1.68 (287) | ||||||
| 0.40 (291) | ||||||
| 03 | 0501 | 0201 | — | 0.75 (268) | 1.73 (165)** | 0.56 (151)** |
| 1.57 (289) | 0.59 (169)** | 2.55 (153)* | ||||
| 1.61 (163)* | ||||||
| 04 | 0301 | 0301 | 5.04 (227)* | — | — | 0.23 (151) |
| 5.62 (153)* | ||||||
| 0404 | 0301 | 0302 | 3.11/7.50 (227)‡ | — | — | — |
| 0405 | 0301 | 0302 | — | 3.79 (268) | — | — |
| 07 | 0201 | 0201 | 0.15 (222)* | 2.25 (268) | 2.06 (165)* | 0.52 (151) |
| 0.19 (275) | 0.56 (171) | 1.94 (153)* | ||||
| 08 | 0401 | 0402 | 0.31 (212)* | 5.13 (287)** | — | — |
| 3.64 (227)* | ||||||
| 0901 | 0301 | 0303 | — | — | 3.13 (167) | 3.61 (169) |
| 1301 | 0103 | 0603 | 5.00 (227)* | 3.97 (268) | — | 3.11 (153) |
| 0.14 (155) | ||||||
| 1302 | 0102 | 0604 | 3.81 (227)‡ | 2.08 (268) | 0.21 (167) | 0.33 (153) |
| 0T/0.099NT (285)*,§ | ||||||
| 16 | 0102 | 0502 | — | 0.25 (268)* | — | — |
| 4.78 (291) | ||||||
| Grouped rare | — | 7.15 (293)** | 1.90 (165) | 0.09 (157)* | ||
| haplotypes (<1%) | ||||||
Results for All population are given. For each marker, significant odds ratio (OR) values (increased or reduced risk on the given DRB1–DQA1–DQB1 haplotype) are given together with the responsible allele (in parentheses).
Results of the Northern European population.
OR calculation not possible, values are frequencies T: transmitted and NT: non-transmitted. Cut-off for significance was set at p = 0.05, results
p < 0.01
p < 0.001 are marked. Results for cell counts <10 (T + NT haplotypes) were omitted.
Significant results at alpha = 0.01 were consistent for each of the microsatellite alleles, that is odds ratio (OR) values were either above or below 1 independent of DRB1–DQA1–DQB1 haplotypes. However, results at alpha = 0.05 showed some inconsistencies, especially for DG6S185 (e.g. DG6S185*268 and DG6S185*275 and DG6S185*291 showed both protective and predisposing effects depending on the haplotype). Also, DG6S398*167 and D6S2889*153 showed protective effects on the DRB1*1302-DQA1*0102-DQB1*0604 haplotype, in contrast to the predisposing effect seen on other haplotypes. These last two results could, however, be because of false positives as cell counts were low and results only nominally significant (T + NT = 12, p = 0.033 and T + NT = 17, p = 0.048, respectively).
DG6S185*293, DG6S398*165 and D6S2889*157 showed associations on the grouped rare haplotypes allele of DRB1–DQA1–DQB1 (p = 0.00026, 0.045 and 0.0033, respectively). This is a heterogeneous group and these associations should therefore be considered artefacts.
DG6S398 and D6S2889 are located in introns of the same gene, C6orf10, and show significant associations on some of the same DRB1–DQA1–DQB1 haplotypes (e.g., DRB1*03-DQA1*0501-DQB1*0201). We therefore tested both markers together on these haplotypes (table 3). The results only showed significant findings or tendencies for association of combinations of alleles that were associated for each marker alone (cf. table 2), with similar OR values.
Table 3.
Haplotype tests of DG6S398 and D6S2889 combined, conditional onDRB1–DQA1–DQB1 haplotypes†
| Conditional haplotypes | Microsatellite alleles | Two-locus results OR (95% CI) |
Single-locus results | ||||
|---|---|---|---|---|---|---|---|
| DRB1 | DQA1 | DQB1 | DG6S398 | D6S2889 | DG6S398 | D6S2889 | |
| 03 | 0501 | 0201 | 165 | 153 | 2.48 (1.33–4.62)* | 1.73** | 2.55* |
| 165 | 163 | 1.50 (1.09–2.06) | 1.73** | 1.61* | |||
| 169 | 151 | 0.58 (0.44–0.77)** | 0.59** | 0.56** | |||
| 07 | 0201 | 0201 | 165 | 153 | 1.95 (1.16–3.27) | 2.06* | 1.94* |
| 171 | 151 | 0.62 (0.37–1.03) | 0.56 | 0.52 | |||
| 0901 | 0301 | 0303 | 167 | 169 | 3.09 (0.97–9.81) | 3.13 | 3.61 |
| 1302 | 0102 | 0604 | 167 | 153 | 0.27 (0.05–1.29) | 0.21 | 0.33 |
Results shown are for the All population. Only DRB1–DQA1–DQB1 haplotypes where significant results were obtained for both markers alone (cf. table 2) were tested. Odds ratio (OR) values from the single marker tests are included for reference. Results for cell counts <10 (T + NT haplotypes) were omitted.
Results p < 0.01
p < 0.001 are marked.
LD Plots
LD plots (global D′) of the four markers given in table 1 for the All, Northern European and Eastern European populations are shown in figure 2. The number of families from Southern Europe that could be incorporated in the analyses was low (N = 25), and results are therefore not shown. The two remaining subpopulations show similar patterns as the All data set, indicating that these microsatellites do not mark large population specific effects. The highest LD values were seen between D6S2889 and DG6S398 (D′ = 0.76 in the All data set) and between these two markers and DRB1–DQA1–DQB1 (D′ = 0.73 and 0.70, respectively), whereas the LD between D6S2773, DG6S398 and the two other markers was low (D′ < 0.35).
Fig. 2.
LD of significant markers from the main effects tests. LD was measured as the generalized D′. The colour code is given on the right (D′ = 0 to 1 in 0.1 increments).
Discussion
In this paper, we tested 59 microsatellites from theT1DGC MHC data set for association with T1D independent of the DRB1, -DQA1 and -DQB1 genes. The tests consisted of two main steps: a screen using single-point main effects tests adjusted for DRB1–DQA1–DQB1 genotypes and mapping of allele associations on the different DRB1–DQA1–DQB1 haplotypes.
Association Analyses
In the main effects tests, the four most significantly associated microsatellites (D6S2773, DG6S185, DG6S398 and D6S2889 in the All data set) only showed nominally or non-significant findings in the subpopulations. Even though the All data set consists mainly of Caucasian families, population stratification may be a factor. This may also have contributed to the observed deviations from Hardy–Weinberg equilibrium. However, the conditional logistic regression method is not sensitive to such issues [15], and this is therefore not likely to have a large impact on the results. Moreover, the similar LD patterns in the Northern and Eastern European populations do not indicate large population specific differences for these markers. Also, the chosen test strategy, in combination with a high number of alleles, requires large populations. This was especially evident for the Southern European population, which consisted of only 161 families, and for which valid results were difficult to obtain. The absence of significant results in the subpopulations is therefore likely to be a question of statistical power.
Related to this is the fact that several markers that previously have been reported associated with T1D independent of the HLA-DRB1, -DQA1 and -DQB1 loci did not reach statistical significance in our tests. This might indicate a weakness with our approach using the some-what heterogeneous All data set and conservative cut-offs for significance. Thus, more subtle, population specific effects may have gone unnoticed. For instance, D6S2223 has repeatedly shown associations with T1D in Northern European populations [5,6,16–18]. In our tests, we did observe a tendency for this marker in the Northern and Eastern European populations (p = 0.022 and 0.023, respectively), which might be stronger if looking at more specific subgroups of the material. Also, D6S2843 and DG6S184 (with no previously reported associations) showed interesting tendencies that should be investigated further.
Several characteristics of the results for DG6S185 point to potential artefacts. First, the recurrent observation of alternating protective and predisposing effects for the same allele depending on the DRB1–DQA1–DQB1 haplotype. Some of these inconsistencies could be attributed to false positives as results at alpha = 0.01 were consistent. However, another explanation could be that the global result in the main effects test was the result of complex haplotype-specific effects, which may be attributed to underlying ancestral haplotypes and not real independent effects. Second, the highly significant association of allele 293 on the grouped rare haplotype allele of DRB1–DQA1–DQB1 means that the result from the main effects test is likely to be artificially low. Lastly, this marker showed a large deviation from Hardy–Weinberg equilibrium, which may point to problems with the genotyping. Together, these observations question the status of this microsatellite as an independent genetic marker for T1D.
DG6S398*165 and D6S2889*157 also showed some effect on the grouped rare haplotype allele of DRB1–DQA1–DQB1. However, these results were either only nominally significant (DG6S398) or involved low cell counts compared with other haplotypes (D6S2889, 1.1% T + NT of total T + NT for associated haplotypes). In these cases, therefore, the results of the main effects tests are not likely to be significantly affected. Moreover, the observed deviation from Hardy–Weinberg equilibrium for these markers can be explained by the proximity and relatively high LD with the HLA-DRB1, -DQA1 and -DQB1 loci, which are expected to show deviations [19], especially in a selected population like the present.
Judging from the LD plots, D6S2773 and DG6S185 appeared to represent largely independent effects of each other and of the two other markers. In contrast, D6S2889 and DG6S398 may reflect the same aetiological locus. These markers are located 57 kb apart in the same gene and showed similar patterns of OR values on different DRB1–DQA1–DQB1 haplotypes. Moreover, haplotype estimates showed that in most cases, predisposing or protective alleles of each of the markers were located together on the same haplotypes, and the two markers therefore appeared to convey much of the same information. However, the global D′ LD value indicates some independence, and D6S2889*153 showed a significant association on the DRB1*04-DQA1*0301-DQB1* 0301 haplotype (p = 0.0021), which was not seen for any of the DG6S398 alleles.
This last observation is related to a more general one that significant allelic associations only were seen on a subset of the DRB1–DQA1–DQB1 haplotypes. For haplotypes present at low frequencies, this is likely to be a question of statistical power. However, none of the microsatellite alleles showed significant association on the high-risk DRB1*0401-DQA1*0301-DQB1*0302 haplotype, which is common (16.5% among parents in the All data set). This could be explained by considering the high degree of conservation of haplotypes in this region [20,21] and the likely situation that the associated microsatellites are serving as genetic proxies for aetiological loci. Apparent specificity for certain DRB1–DQA1–DQB1 haplotypes could then be the result of specific LD patterns, where aetiological variants in LD with the investigated markers only are picked up on certain haplotypes. An alternative or additional explanation could be that the aetiological loci have functional properties specific to certain variants of the MHC class II molecules. Answers to these questions are only possible after further fine mapping on the various DRB1–DQA1–DQB1 haplotypes, including a greater number of markers in the immediate surrounding regions.
Location of the Most Promising Markers
Although the most promising markers are not likely to be primary aetiological loci, some remarks can be made about their immediate genetic neighbourhood. D6S2773 is located close to the HLA-G gene, which has been implicated in asthma [22]. However, this has not been replicated. Other possible candidates in this region include the HLA-A and HLA-F genes, of which the first has been implicated in T1D through functional studies [23].
DG6S398 and D6S2889 are both located in introns of the C6orf10 gene. The function of this gene is unknown, and it appears to be primarily expressed in testis (http://www.ncbi.nlm.nih.gov/UniGene). C6orf10 is, however, located 23 kb downstream of the BTNL2 gene, which recently was mapped as a susceptibility locus for sarcoidosis, independent of HLA-DRB1 [24]. Sarcoidosis is an autoimmune disease, and the BTNL2 gene has been implicated as a costimulatory molecule in T-cell activation. Although a recent study could not find an association independent of the HLA-DRB1 and -DQB1 genes with T1D in Dutch patients [25], the study population was small, and only one polymorphism was investigated. This gene may therefore still be a good candidate for further studies.
Conclusions
We have identified three microsatellites, D6S2773, DG6S185 and D6S2889, that show highly significant association with T1D independent of the HLA-DRB1, -DQA1 and -DQB1 genes. These three markers with-stand Bonferroni correction and validity testing, and results are supported by haplotype estimates. In addition, DG6S185 showed association in the conditional logistic regression analyses, but allelic tests on DRB1–DQA1–DQB1 haplotypes indicated that this, at least in part, may be because of artefacts imposed by the particular characteristics of the MHC and the grouping of rare DRB1–DQA1–DQB1 haplotypes. To our knowledge, none of these microsatellites have been reported to be associated with T1D or any other autoimmune disease earlier. Further analyses are needed to establish the relationship with other possible genetic determinants in this region.
Acknowledgements
This research utilizes resources provided by the T1DGC, a collaborative clinical study sponsored by the National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of Allergy and Infectious Diseases, National Human Genome Research Institute, National Institute of Child Health and Human Development and Juvenile Diabetes Research Foundation International (JDRF) and supported by U01 DK062418. M. C. E. was supported by JDRF grant 1-2004-793. The authors also wish to thank Marte Viken for helpful comments on the final manuscript.
Footnotes
Conflict of interest:
The authors declare no competing financial interests.
References
- 1.Horton R, Wilming L, Rand V, et al. Gene map of the extended human MHC. Nat Rev Genet. 2004;5:889–899. doi: 10.1038/nrg1489. [DOI] [PubMed] [Google Scholar]
- 2.Cucca F, Lampis R, Congia M, et al. A correlation between the relative predisposition of MHC class II alleles to type 1 diabetes and the structure of their proteins. Hum Mol Genet. 2001;10:2025–2037. doi: 10.1093/hmg/10.19.2025. [DOI] [PubMed] [Google Scholar]
- 3.Nerup J, Platz P, Andersen OO, et al. HL-A antigens and diabetes mellitus. Lancet. 1974;2:864–866. doi: 10.1016/s0140-6736(74)91201-x. [DOI] [PubMed] [Google Scholar]
- 4.Noble JA, Valdes AM, Cook M, Klitz W, Thomson G, Erlich HA. The role of HLA class II genes in insulin-dependent diabetes mellitus: molecular analysis of 180 Caucasian, multiplex families. Am J Hum Genet. 1996;59:1134–1148. [PMC free article] [PubMed] [Google Scholar]
- 5.Johansson S, Lie BA, Todd JA, et al. Evidence of at least two type 1 diabetes susceptibility genes in the HLA complex distinct from HLA-DQB1, -DQA1 and -DRB1. Genes Immun. 2003;4:46–53. doi: 10.1038/sj.gene.6363917. [DOI] [PubMed] [Google Scholar]
- 6.Lie BA, Todd JA, Pociot F, et al. The predisposition to type 1 diabetes linked to the human leukocyte antigen complex includes at least one non-class II gene. Am J Hum Genet. 1999;64:793–800. doi: 10.1086/302283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Aly TA, Ide A, Jahromi MM, et al. Extreme genetic risk for type 1A diabetes. Proc Natl Acad Sci U S A. 2006;103:14074–14079. doi: 10.1073/pnas.0606349103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Nejentsev S, Gombos Z, Laine AP, et al. Non-class II HLA gene associated with type 1 diabetes maps to the 240-kb region near HLA-B. Diabetes. 2000;49:2217–2221. doi: 10.2337/diabetes.49.12.2217. [DOI] [PubMed] [Google Scholar]
- 9.Zavattari P, Lampis R, Motzo C, et al. Conditional linkage disequilibrium analysis of a complex disease super-locus, IDDM1 in the HLA region, reveals the presence of independent modifying gene effects influencing the type 1 diabetes risk encoded by the major HLA-DQB1, -DRB1 disease loci. Hum Mol Genet. 2001;10:881–889. doi: 10.1093/hmg/10.8.881. [DOI] [PubMed] [Google Scholar]
- 10.Becker T, Knapp M. Maximum-likelihood estimation of haplotype frequencies in nuclear families. Genet Epidemiol. 2004;27:21–32. doi: 10.1002/gepi.10323. [DOI] [PubMed] [Google Scholar]
- 11.O’Connell JR, Weeks DE. PedCheck: a program for identification of genotype incompatibilities in linkage analysis. Am J Hum Genet. 1998;63:259–266. doi: 10.1086/301904. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wigginton JE, Abecasis GR. PEDSTATS: descriptive statistics, graphics and quality assessment for gene mapping data. Bioinformatics. 2005;21:3445–3447. doi: 10.1093/bioinformatics/bti529. [DOI] [PubMed] [Google Scholar]
- 13.Cordell HJ, Clayton DG. A unified stepwise regression procedure for evaluating the relative effects of polymorphisms within a gene using case/control or family data: application to HLA in type 1 diabetes. Am J Hum Genet. 2002;70:124–141. doi: 10.1086/338007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Abecasis GR, Cookson WO. GOLD – graphical overview of linkage disequilibrium. Bioinformatics. 2000;16:182–183. doi: 10.1093/bioinformatics/16.2.182. [DOI] [PubMed] [Google Scholar]
- 15.Cordell HJ, Barratt BJ, Clayton DG. Case/pseudocontrol analysis in genetic association studies: a unified framework for detection of genotype and haplotype associations, gene-gene and gene-environment interactions, and parent-of-origin effects. Genet Epidemiol. 2004;26:167–185. doi: 10.1002/gepi.10307. [DOI] [PubMed] [Google Scholar]
- 16.Lie BA, Sollid LM, Ascher H, et al. A gene telomeric of the HLA class I region is involved in predisposition to both type 1 diabetes and coeliac disease. Tissue Antigens. 1999;54:162–168. doi: 10.1034/j.1399-0039.1999.540207.x. [DOI] [PubMed] [Google Scholar]
- 17.Koeleman BV, De Groot KN, Van Der Slik AR, Roep BO, Giphart MJ. Association between D6S2223 and type I diabetes independent of HLA class II in Dutch families. Diabetologia. 2002;45:598–599. doi: 10.1007/s00125-001-0725-1. [DOI] [PubMed] [Google Scholar]
- 18.Herr M, Dudbridge F, Zavattari P, et al. Evaluation of fine mapping strategies for a multifactorial disease locus: systematic linkage and association analysis of IDDM1 in the HLA region on chromosome 6p21. Hum Mol Genet. 2000;9:1291–1301. doi: 10.1093/hmg/9.9.1291. [DOI] [PubMed] [Google Scholar]
- 19.Chen JJ, Hollenbach JA, Trachtenberg EA, et al. Hardy-Weinberg testing for HLA class II (DRB1, DQA1, DQB1, and DPB1) loci in 26 human ethnic groups. Tissue Antigens. 1999;54:533–542. doi: 10.1034/j.1399-0039.1999.540601.x. [DOI] [PubMed] [Google Scholar]
- 20.Blomhoff A, Olsson M, Johansson S, et al. Linkage disequilibrium and haplotype blocks in the MHC vary in an HLA haplotype specific manner assessed mainly by DRB1*03 and DRB1*04 haplotypes. Genes Immun. 2006;7:130–140. doi: 10.1038/sj.gene.6364272. [DOI] [PubMed] [Google Scholar]
- 21.Ahmad T, Neville M, Marshall SE, et al. Haplotype-specific linkage disequilibrium patterns define the genetic topography of the human MHC. Hum Mol Genet. 2003;12:647–656. [PubMed] [Google Scholar]
- 22.Nicolae D, Cox NJ, Lester LA, et al. Fine mapping and positional candidate studies identify HLA-G as an asthma susceptibility gene on chromosome 6p21. Am J Hum Genet. 2005;76:349–357. doi: 10.1086/427763. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Marron MP, Graser RT, Chapman HD, Serreze DV. Functional evidence for the mediation of diabetogenic T cell responses by HLA-A2.1 MHC class I molecules through transgenic expression in NOD mice. Proc Natl Acad Sci U S A. 2002;99:13753–13758. doi: 10.1073/pnas.212221199. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Valentonyte R, Hampe J, Huse K, et al. Sarcoidosis is associated with a truncating splice site mutation in BTNL2. Nat Genet. 2005;37:357–364. doi: 10.1038/ng1519. [DOI] [PubMed] [Google Scholar]
- 25.Orozco G, Eerligh P, Sanchez E, et al. Analysis of a functional BTNL2 polymorphism in type 1 diabetes, rheumatoid arthritis, and systemic lupus erythematosus. Hum Immunol. 2005;66:1235–1241. doi: 10.1016/j.humimm.2006.02.003. [DOI] [PubMed] [Google Scholar]