In collaboration with Professor Ahmet Gul from Istanbul University, we have performed a large genome-wide association study to identify additional variants contributing to Behcet's disease. In this study we genotyped patients with Behcet's disease and geographically matched controls from Turkey using a high-throughput SNP genotype bead array (Illumina, San Diego) that interrogated approximately 370,000 polymorphisms, including copy number variants. After curation for relatedness, a total of 1215 individuals with Behcet's disease (cases) and 1278 healthy controls from Turkey were included in the initial study. Single nucleotide polymorphisms (SNPs) were filtered for a call rate of greater than 95%, minor allele frequency of greater than 1%, and Hardy-Weinberg equilibrium P value of greater than 10 to the negative fifth power, resulting in 311,459 SNPs in the final analysis. A principal components method was used to evaluate population stratification in the cases and controls. After correction for six principal components, lambda GC, a measure of genomic inflation, was 1.05, and after removal of SNPs from the strongly associated MHC region it was 1.04. We used a P value of 5 times 10 to the negative eighth as the threshold for genome-wide significance. We also typed the HLA-B alleles in 1190 cases and 1257 controls using a reverse sequence-specific oligonucleotide method. We confirmed the known association of Behcet's disease with HLA-B*51. Occurrence (meaning the presence of either one or two copies) of the HLA-B*51 variant was found in 59.1% of cases and 29.3% of controls (OR = 3.49, 95% CI 2.95 4.12, P = 5.47 times 10 to the negative fiftieth power). Strong linkage disequilibrium (LD) was observed between HLA-B*51 and all the SNPs located in the region from HLA-B to more than 62-kb centromeric to MICA, despite the fact that this region spans several blocks of LD. While this extended haplotype occurred at a frequency of 0.321 in cases and 0.144 in controls, the identical SNP haplotype lacking HLA-B*51 occurred in cases and controls equally (frequency = 0.04), indicating that HLA-B*51 is required for disease association in the HLA-B region. Conditional logistic regression of the 294 disease-associated SNPs within the HLA region, using HLA-B*51 as a covariate, identified 3 SNPs within the HLA-A region that retained genome-wide significance in the conditional analysis. The most strongly associated SNP, rs9260997, located 50-kb centromeric to HLA-A, had a regressor P of 5.49 times ten to the negative ninth power, suggesting a highly significant association independent of the HLA-B*51 association. In the association data, we identified near genome-wide significance for a SNP, rs936551 (P = 5.29 times 10 to the negative eighth) located within the promoter region of CPLX1, which encodes complexin-1, a regulator of exocytosis during vesicle membrane fusion. Fine-mapping with additional SNPs from this genomic region increased coverage of HapMap Phase II SNPs with minor allele frequency greater than 5% from 57% to 93% in CEU samples, but failed to identify a SNP more significantly associated than rs936551. Using meta-analysis with genotypes from our discovery cohort and an additional five cohorts from Turkey, the Middle East, Europe, and Asia, comprising a total of 2430 cases and 2660 controls, we failed to establish genome-wide significance for CPLX1, although it is possible that statistical significance might be achieved with even larger sample sizes. In the initial association data with 1215 cases and 1279 controls, we also identified a SNP on chromosome 1 within the first intron of IL10, rs3024490, that was strongly suggestive for association, with a P value of 2.22 times 10 to the negative seventh. Fine mapping of the IL10 region with 27 additional SNPs resulted in coverage of 100% of the HapMap Phase II SNPs with greater than 5% minor allele frequency in CEU samples, and one SNP, rs1518111 (located in the second intron of IL10), achieved genome-wide significance (P = 1.88 times 10 to the negative eighth power). Meta-analysis with the discovery cohort and the above-noted additional cohorts from Turkey, the Middle East, Europe, and Asia resulted in a P value of 3.54 times 10 to the negative eighteenth power (OR = 1.45, 95% CI 1.34 1.58). The Behcet's-associated SNP was different from a SNP previously found to be associated with ulcerative colitis, type 1 diabetes, and systemic lupus erythematosus, but in strong LD with a SNP associated with severe juvenile rheumatoid arthritis. In order to determine the effect of the Behcet's associated SNP on gene expression, we used real-time PCR to analyze pre-mRNA expression of the disease-associated haplotype and other haplotypes in monocytes from healthy controls. In these studies expression of the disease-associated haplotype was 35% of the expression of the alternate haplotype. We also tested for in vitro stimulated IL-10 production by peripheral blood leukocytes from both healthy Turkish donors and American donors. In both cases, leukocytes derived from donors homozygous for the Behcet's risk allele produced significantly less IL-10 than heterozygotes or individuals homozygous for the non-risk allele, whereas there was no significant difference in tumor necrosis factor alpha production. IL-10 is a known anti-inflammatory cytokine, and a variant associated with decreased production is consistent with the inflammatory phenotype of Behcets disease. In the initial association data with 1215 cases and 1279 controls we identified a SNP on chromosome 1 in the IL23R-IL12RB region, rs924080, that was suggestive for association (P = 5.35 times 10 to the negative sixth power). We genotyped 11 additional SNPs, thereby increasing coverage of the HapMap Phase II SNPs with greater than 5% minor allele frequency in the CEU samples from 58% to 88%. No SNPs were found to have a stronger association than rs924080. We identified a recombination hot spot located in the intergenic region between IL23R and IL12RB2, which is also seen in the HapMap European descent and Asian populations. The Behcet's disease-associated variants were located on the IL23R side of the hot spot, suggesting the disease association is more likely to be linked to IL23R than IL12RB. Variants within IL23R have been associated with ankylosing spondylitis, psoriasis, and inflammatory bowel disease, but neither of the two IL23R coding variants associated with these seronegative diseases was associated with Behcet's disease. Meta-analysis as noted above resulted in a P value of 6.69 times 10 to the negative ninth power (OR = 1.28, 95% CI 1.18 1.39). In summary, these genome-wide association data strongly implicate variants at the IL10 and IL23R loci in susceptibility to Behcets disease. Both of these findings suggest novel therapies for Behcet's disease. In the case of IL10, potential therapies include recombinant IL10 itself (since the disease-associated variant appears to cause decreased IL10 production), as well as interferon alpha, which induces IL10 production, and inhibitors of IL-1beta, which is negatively regulated by IL10. The association of variants at the IL23R locus with Behcet's disease suggests the possibility of therapy with ustekinumab, a monoclonal antibody against both IL-23 and IL-12, as well as biologics that target Th17 signaling downstream of the IL-23 receptor. Future studies will be aimed at (1) studying the role of copy-number variants in Behcet's disease susceptibility, (2) using deep-resequencing technologies to identify rare variants associated with disease susceptibility, and (3) developing therapeutic strategies that target the immunologic pathways identified by these genetic studies.
| Remmers, Elaine F; Cosan, Fulya; Kirino, Yohei et al. (2010) Genome-wide association study identifies variants in the MHC class I, IL10, and IL23R-IL12RB2 regions associated with Behçet's disease. Nat Genet 42:698-702 |
| Kastner, Daniel L; Aksentijevich, Ivona; Goldbach-Mansky, Raphaela (2010) Autoinflammatory disease reloaded: a clinical perspective. Cell 140:784-90 |
| Masters, Seth L; Simon, Anna; Aksentijevich, Ivona et al. (2009) Horror autoinflammaticus: the molecular pathophysiology of autoinflammatory disease (*). Annu Rev Immunol 27:621-68 |
