Background Rheumatoid arthritis (RA) is a genetically complex disease of immune dysregulation. Genome-wide association scans (GWAS) have shown common variants at approximately 100 genetic loci to be associated with RA. Single nucleotide polymorphisms (SNPs) are typically intergenic and causative disease genes remain ill-defined. Examining how RA-associated SNPs influence gene expression in relevant biological contexts will begin to address this.
Methods Patients naïve to immunomodulatory therapy attending the Newcastle Early Arthritis Clinic donated RNA and DNA, extracted from purified peripheral blood CD4+ T- and B-lymphocytes within 4 hours of blood draw. Detailed baseline and longitudinal clinical data were recorded for all participants, each followed up for >1 year. Genotyping and global gene expression measurement was carried out using the Illumina Human CoreExome array, and either HT12v4 or WG6v3 BeadChip arrays, respectively, and expression data was batch-corrected and normalised using standard algorithms. Data from both cell types were analysed using the R package. Variants in linkage disequilibrium (LD) with 100 confirmed RA- SNPs (r2 >0.8) were included in analyses, seeking evidence of cis- and trans- eQTLs according to whether associated probes were or were not within 4 MB of these LD blocks.
Results Cell-type specific gene expression data from 351 genotyped white Caucasian early arthritis patients were available for analysis (including paired CD4+/B lymphocyte data for 160 of these). Genes subject to cis eQTL effects common to both CD4+ and B-lymphocytes at RA risk loci included FADS1, FCRL3, PIPL3, ORMDL3 and GSDMB. Cis eQTLs acting on BLK, IKZF3 and PADI4 were, by contrast, unique to CD4+ lymphocytes in this population, and equivalent B-lymphocyte-specific effects were seen for IRF5 and FAM167. Evidence emerged that the 12q13 RA risk variant regulates STX1B gene expression on chromosome 7 of B-lymphocytes, but no trans eQTLs achieved experiment-wide significance thresholds in CD4+ lymphocytes. Linear modelling could not identify a significant influence of biological co-variates (diagnosis, systemic inflammation, age) upon eQTL effect sizes.
Conclusions We present the first detailed eQTL analysis in the pathophysiologically relevant setting of early arthritis lymphocytes, free of confounding immunomodulatory treatment. The findings should help refine understanding of candidate causal genes in RA pathogenesis.