Background/Purpose: Genome-wide association studies have identified thousands of genetic associations with complex human diseases and traits. However, establishing the truly causal regions of risk haplotypes is complicated by the correlation of many non-causal variants to genetic associations through linkage disequilibrium (LD). This has made translating genetic findings into functional mechanisms an inefficient, resource-intensive task.

Methods: Here, we expedite causal variant discovery for autoimmune diseases by identifying genetic variants (epiQTLs) that leave an epigenomic footprint of allelic imbalance in sequencing reads recovered by chromatin immunoprecipitation of enhancer marks (H3K4me1 and H3K27ac ChIP-seq). We aligned sequencing reads using the WASP pipeline to control for reference genome bias. We used the combined haplotype test (CHT) to statistically test for robust footprints of allelic imbalance. We also characterized the three-dimensional chromatin interaction topography among H3K27ac-marked active enhancers of our cell lines by HiChIP.

Results: We discovered a total of 6261 epiQTLs across 25 patient-derived EBV-transformed B-cell lines. We overlaid these epiQTLs onto risk haplotypes from 21 autoimmune (AI) diseases and found that 145 reported AI risk haplotypes contained one or more of our epiQTLs. A total of 14 epiQTL SNPs matched AI risk SNPs reported in the NHGRI GWAS catalog, while 180 epiQTLs were proxies of these reported SNPs. We found that epiQTLs located on disease risk haplotypes disproportionately influence gene expression variance, beyond what can be expected by random chance, over non-epiQTL variants in tight linkage disequilibrium. A majority (78%) of epiQTLs were located in chromatin loop anchors. Using a generalized linear model, we identified 571 epiQTLs – not associated with autoimmune disease – that modify gene expression from 68 previously established AI disease risk haplotypes through chromatin looping at FDR ≤5%. Expanding this analysis to include both risk haplotypes and independent eQTLs increased the number of epiQTLs with significant (FDR ≤ 5%) modifier effects on gene expression to 1062. Of these gene expression modifying interactions, 717 (68%) increased eQTL-driven gene expression by the epiQTL, while 345 (32%) suppressed eQTL-driven gene expression.

Conclusion: Our data suggest that the epiQTL approach can facilitate the decomposition of risk haplotypes into specific regions that are highly likely to contain functional causal variants. Moreover, epiQTLs not associated with disease haplotypes function to modify gene expression from risk haplotypes and eQTLs through long-range allele-dependent epigenetic mechanisms.

« Back to ACR/ARHP Annual Meeting 2017

ACR Meeting Abstracts - https://acrabstracts.org/abstract/an-epigenome-guided-approach-to-causal-variant-discovery-in-autoimmune-disease/