Imputation to Discover Additional non-HLA Susceptibility Loci We carried out whole-genome imputation using as a reference 96 Turkish healthy controls genotyped on Illumina HumanOmni1M-Quad SNP chips. Imputation was conducted using MACH v1.0.15 providing 779,465 SNPs for analysis in 1,209 Turkish BD cases and 1,278 healthy controls. Sequenom iPLEX assays were used to validate the imputation results and to fine map the associated region. Two independent replication sets comprising 838 Turkish cases and 630 controls, as well as 612 Japanese cases and 740 controls (if the SNP was polymorphic in the Japanese collection), were genotyped for the most significant SNPs. By Cochran-Mantel-Haenszel (CMH) meta-analysis, we identified associations at CCR1 (rs7616215, P = 4.30 times 10 to the negative thirteenth, OR = 0.72), encoding a chemokine receptor expressed on myeloid cells;KLRC4 (rs2617170, P = 1.34 times 10 to the negative ninth, OR = 0.78), encoding a receptor found on natural killer (NK) cells and gamma-delta T-cells;and STAT4 (rs7574070, P = 1.29 times 10 to the negative ninth, OR = 1.27), encoding a transcription factor expressed in T-lymphocytes. The protective CCR1 C-allele was associated with higher mRNA expression and increased leukocyte chemotaxis. The STAT4 risk allele was also associated with increased mRNA expression. It is not in linkage disequilibrium with rs7574865, a SNP associated with rheumatoid arthritis and several other autoimmune diseases, although both reside in the third intron of STAT4. These data were published in Nature Genetics during the current reporting period. Imputation to Discover Additional HLA-Associated Susceptibility Loci Genotypes from 2832 MHC SNPs were extracted from a genome-wide SNP array for 1190 Turkish BD cases and 1257 controls, and these data were used as the basis of SNP imputation with reference data from the 1000 Genomes Project. HLA-B types were determined using a sequence specific oligonucleotide-based assay. We used imputation to infer classical HLA types and the identities of polymorphic amino acid residues in each classic HLA molecule using as a reference the T1DGC panel of 5225 individuals with SNP and HLA types. Stepwise logistic regression and conditional analyses were performed using genotype probabilities produced by imputation. SNP mapping identified two regions, the HLA-B/MICA region and the region between HLA-F and HLA-A, as independently associated with BD after regression analyses (P less than 1.7 times 10 to the negative eighth). Haplotype analysis of the HLA-B/MICA region identified a common BD-associated haplotype that included B*51 and 48 BD-associated SNPs, yet an identical SNP haplotype that lacked B*51 conferred no risk of BD. We found that HLA-B*51, -B*15, -B*27, -B*57, and A*26 each contributed to BD risk independently, while HLA-B*49 and -A*03 were each protective against the development of BD. Regression analyses of polymorphic amino acid residues identified independent associations between BD and 9 MHC class I amino acid residues, including one that conferred risk at both HLA-A and HLA-B. The majority of the BD-associated residues clustered around the MHC class I antigen-binding groove. One associated residue was located on the signal peptide of HLA-B, and has previously been shown to produce enhanced NK cell activation and cytotoxicity. We are preparing a manuscript describing these findings. Clinical Subset Analyses and Gene-Gene Interactions In the subset of BD patients with uveitis, we found that 2 coding SNPs in ERAP1 conferred risk for BD under a recessive model. By CMH meta-analysis of the 420 case BD uveitis discovery collection and 1,249 controls, and a 370 case uveitis replication group with 630 Turkish controls, the overall P value for the R725Q coding variant was 4.73 times 10 to the negative eleventh, with an OR of 4.56. Moreover, under the recessive model a CMH meta-analysis of this SNP in the GWAS and replication collections (i.e., including both uveitis and non-uveitis BD patients) found significant association of the homozygous R725Q genotype with BD susceptibility (P = 4.35 times 10 to the negative eighth, OR = 3.08). ERAP1 is an endoplasmic reticulum-expressed amino peptidase, which functions to trim peptides and load them onto MHC Class I. Of note, the R725Q ERAP1 variant confers protection from ankylosing spondylitis and psoriasis, but only in patients who are HLA-B*27 positive in AS (Evans et al., Nature Genet 2011) or HLA-C*06 positive in psoriasis (Strange et al., Nature Genet 2010). Similarly, by logistic regression we found that ERAP1 R725Q preferentially conferred risk for BD in HLA-B*51 positive individuals (P = 0.0009). These data suggest that the disease-associated peptidase variant contributes to BD susceptibility through an interaction, perhaps mediated by specific peptides produced by the variant peptidase, with the HLA-B*51 protein, and suggest a mechanistic explanation for the long-observed but poorly understood HLA-B*51 association with BD. Our findings add to an emerging body of data delineating common pathogenic mechanisms for BD, ankylosing spondylitis, and psoriasis. All 3 are inflammatory disorders affecting the skin, eyes, and joints, all 3 have important HLA Class I associations, variants of the IL-23 receptor are associated with susceptibility in all 3, and now there is convincing evidence in all 3 diseases for epistasis between Class I MHC and ERAP1. These data were published in Nature Genetics during the current reporting period. Deep Resequencing of Targeted Genes Nonsynonymous variants (NSVs) identified by deep exonic resequencing of 10 genes found by GWAS (IL10, IL23R, CCR1, STAT4, KLRK1, KLRC1, KLRC2, KLRC3, KLRC4, and ERAP1) and 11 genes selected for their role in innate immunity (IL1B, IL1R1, IL1RN, NLRP3, MEFV, TNFRSF1A, PSTPIP1, CASP1, PYCARD, NOD2, and TLR4) were evaluated for BD association. A differential distribution of the rare and low-frequency NSVs of a gene in 2461 BD cases compared with 2458 controls indicated their collective association with disease. By stringent criteria requiring at least a single burden test with study-wide significance and a corroborating test with at least nominal significance, rare and low-frequency NSVs in one GWAS-identified gene, IL23R (P = 6.9 times 10 to the negative fifth), and one gene involved in innate immunity, TLR4 (P = 8.0 times 10 to the negative fourth), were associated with BD. Damaging or rare damaging NOD2 variants were nominally significant across all 3 burden tests applied (P = 0.0063 to 0045). Carriage of the familial Mediterranean fever gene (MEFV) mutation M694V, which is known to cause recessively inherited familial Mediterranean fever, conferred BD risk in the Turkish population (OR, 2.65;P = 1.8 times 10 to the negative twelfth). These data implicate innate immune and bacterial sensing mechanisms in BD pathogenesis, and were published in the PNAS during the current reporting period. High-Density Genotyping of Immune-Related Genes We have genotyped 2014 Turkish BD patients and 1826 controls with the Immunochip, a custom array with 196,524 markers in 186 loci selected from analysis of 12 autoimmune diseases. Using the basic allele test, we confirmed association at IL10 and CCR1, and identified 4 novel loci, IL1A-IL1B, SCHIP1-IL12A, IRF8, and PTPN1, as exceeding a genome-wide significance threshold of 5 times 10 to the negative eighth. For novel loci with an association test P value of less than 5 times 10 to the negative sixth, additional SNPs in the region were imputed using 1000 Genomes data as the reference, thus identifying one additional locus, EGR2, with genome-wide significance. In addition, FUT2 was significant under a dominant model of inheritance.
Showing the most recent 10 out of 14 publications
