Allelic expression imbalance in articular cartilage and subchondral bone refined genome-wide association signals in osteoarthritis.

Abstract

Osteoarthritis (OA) is an age-related joint disease with a strong and complex genetic component. Genome-wide association studies (GWAS) discovered a large number of genomic regions associated with OA. Nevertheless, to link associated genetic variants affecting the expression of OA-risk genes in relevant tissues remains a challenge. Here, we showed an unbiased approach to identify transcript single nucleotide polymorphisms (SNPs) of OA risk loci by allelic expression imbalance (AEI). We used RNA sequencing of articular cartilage (N = 65) and subchondral bone (N= 24) from OA patients. AEI was determined for all genes present in the 100 regions reported by GWAS catalog. The count fraction of the alternative allele ({varphi}) was calculated for each heterozygous individual with the risk-SNP or with the SNP in linkage disequilibrium (LD) with it. Furthermore, a meta-analysis was performed to generate a meta-{varphi} (null hypothesis median {varphi}=0.49) and P-value for each SNP. We identified 30 transcript SNPs subject to AEI (28 in cartilage and 2 in subchondral bone). Notably, 10 transcript SNPs were located in genes not previously reported in the GWAS catalogue, including two long intergenic non-coding RNAs (lincRNAs), MALAT1 (meta-{varphi}=0.54, FDR=1.7x10-4) and ILF3-DT (meta-{varphi}=0.6, FDR=1.75x10-5). Moreover, 14 drugs were interacting with 7 genes displaying AEI, of which 7 drugs has been already approved. By prioritizing proxy transcript SNPs that mark AEI in cartilage and/or subchondral bone at loci harboring GWAS signals, we present an unbiased approach to identify the most likely functional OA risk-SNP and gene. We identified 10 new potential OA risk genes ready for further, translation towards underlying biological mechanisms.