Work by the International IBD Genetics Consortium (preprint by Huang et al) has used dense genotyping on the Immunochip and Bayesian statistical analysis to elucidate 18 loci that could be fine-mapped to single SNPs with greater than 95% posterior probability of causality. Not surprisingly, 8 of these SNPs were coding variants due to their strong phenotypic impact. However, 10 other loci were identified in which the causal SNP has no known motif or feature that suggests its mechanism of action, such as being contained within a consensus transcription factor binding sequence, a chromatin modification region, or a cis expression quantitative trait locus (cis-eQTL). We used functional experiments to determine if any of these causal SNPs could show allelic differences in expression.Methods:
We took 8 loci—including LRRK2, HNF4A, IL2RA, IKZF1, GPR35, NKX2-3, JAK2, and PRDM1—of approximately 700 bp in size centered on the credible SNP, and cloned both the reference and alternate allele into a luciferase reporter vector with a minimal promoter. Expression was assayed by the Dual Luciferase kit (Promega) in HEK 293T cells. We also assessed the full-length JAK2 promoter containing variant rs1887428 in the dual luciferase assay. Electrophoresis mobility shift assay (EMSA) was used to determine if sequence-specific binding proteins could recognize these SNPs.Results:
Six of the loci did not show any expression of luciferase above the background level of the empty vector. However, both JAK2 and PRDM1 region SNPs showed significantly above-baseline expression. The JAK2 SNP lies within the promoter region 556 bp from the transcriptional start site and showed a 20% allelic difference in expression using a promoter fragment but no significant difference using the full length promoter. Both JAK2 and PRDM1 probes containing the causal SNPs could be bound by sequence-specific DNA-binding proteins.Conclusions:
These results suggest that credible set SNPs can mediate differences in gene expression even when known functional annotations are lacking or the mechanism of action is unknown. Disease-associated SNPs that confer allelic differences in gene expression are more likely to contribute to disease pathogenesis.