Functional analysis of the osteoarthritis susceptibility loci marked by the polymorphisms rs10492367 and rs9350591

Abstract

Approximately 8.5 million people in the UK are affected by osteoarthritis (OA), a multifactorial, polygenic disease characterised by articular cartilage loss. In 2012, the arcOGEN Consortium reported on the largest OA genome-wide association scan (GWAS) to date, in which five regions of the genome were significantly associated with the disease in Europeans. I aimed to characterise two of these regions: rs10492367 is intergenic between PTHLH and KLHL42, while rs9350591 is intergenic between FILIP1 and SENP6. MYO6, TMEM30A, COX7A2 and COL12A1 also surround rs9350591. There are no non-synonymous polymorphisms within either association region that could account for the signals. I first confirmed that the genes were expressed throughout in vitro chondrogenesis and osteoblastogenesis. Using quantitative real-time polymerase chain reaction, I identified the differential expression of PTHLH, KLHL42, SENP6, MYO6, COX7A2 and COL12A1 in articular cartilage stratified by disease state, joint and/or sex. There were no differences between the genotypic groups of either signal, corroborated by pyrosequencing which quantified allelic outputs. Expression quantitative trait loci, irrespective of rs9350591 genotype, acted upon MYO6 and COL12A1. I used data from an Illumina BeadChip array to identify the hypermethylation of cg26466508 in rs9350591 risk allele carriers relative to non-risk allele homozygotes. In luciferase reporter assays, the alleles of polymorphisms in high linkage disequilibrium with rs10492367 displayed differential enhancer activity. Electrophoretic mobility shift assays were used to investigate protein binding to four of these polymorphisms, with RELA, SUB1 and TCF3 binding to rs10492367: chromatin immunoprecipitation confirmed these findings. Finally, I used silencing RNAs to knockdown the transcription factors in human articular chondrocytes, with SUB1 depletion resulting in a downregulation of PTHLH. Overall, I have highlighted the complexity of characterising GWAS signals. The data suggest functional roles for the regions, perhaps by mediating OA susceptibility during joint development rather than in end-stage diseased cartilage.