Background/Purpose: Nearly 150 genetic loci are convincingly associated with lupus (SLE) or rheumatoid arthritis (RA) and underlie their incompletely understood mechanisms of pathogenesis. Since 90% of complex disease genetic loci fall in genomic regulatory regions, since we suspect that regulatory elements may be shared across loci, and since other evidence is consistent with Epstein-Barr virus (EBV) contributing to SLE and RA pathogenesis, we tested the hypothesis for both SLE and RA that transcription factors and co-factors (TFs) encoded by EBV selectively bind the human genome at disease risk loci. We also evaluated over 212 TFs in over 150 cell lines and types in 1544 datasets. We selectively evaluated examples of allele specific TF binding.

Methods: We used published results, defining disease risk loci for RA and SLE as all variants with disequilibrium r²>0.8 to the most highly associated locus variant. The location of TF binding was taken from 1544 available TF chromatin immunoprecipitation with sequencing (ChIP-seq) datasets. By assigning loci to random sites in the human genome (or to open chromatin in EBV infected B cells) while preserving their locus structure, we used simulation to generate a normal distribution to estimate probability of the observed intersections of TF ChIP-seq binding and the risk loci for 213 diseases or phenotypes. We explored individual variants in ChIP-seq datasets for distorted TF binding to alleles.

Results: After Bonferroni correcting for 1544 comparisons we found powerful associations of the Epstein-Barr virus nuclear antigen 2 (EBNA2) ChIP-seq data with the loci of European SLE (26 of 53 loci, OR=6, Pc=1x10E-25) and of RA (25 of 84 loci, OR=4, Pc=2x10E-14), but not with the other EBV TFs. A set of at least 24 human TFs were similarly strongly associated and clustered with EBNA2 at largely the same loci; many of these human TFs are known to support inflammatory, interferon, or B cell development pathways (RELA, RELB, NFKB1, NFKB2, IKZF1, IRF4, PAX5, etc.) These associations were virtually exclusive to Epstein-Barr virus (EBV) infected B cells and were not found in other tissue types nor in B cell lines that were not EBV infected. These associations were present when limiting the simulation to open chromatin in EBV transformed B cells. EBNA2 is shown to differentially immunoprecipitate specific plausibly causal variant alleles. The loci of nearly 200 other disorders evaluated did not have an association with EBNA2 ChIP-seq data; however, 212 TFs from the 1544 ChIP-seq datasets produced 2264 clearly significant associations, each Pc<10E-6, with the risk loci from 94 diseases or phenotypes, including other rheumatic disorders.

Conclusion: The indirect binding of EBNA2 to human DNA is associated with the genetic risk loci of both SLE and RA, as also are an associated set of human TFs clustering with EBNA2 at particular genetic loci, identifying different subsets of loci for RA and SLE. In the data now available these relationships in both diseases are found virtually exclusively in EBV-infected B cells. EBNA2 and human TFs bind alleles differentially at some plausibly causal variants. These results are consistent with EBV, the EBV infected B cell, and EBNA2 encoded by EBV DNA participating in SLE and RA pathogenesis.

« Back to 2017 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/a-possible-environmental-origin-for-a-proportion-of-the-genetic-risk-of-rheumatoid-arthritis-and-systemic-lupus-erythematosus/