Cindy Boer

Genetics of Osteoarthritis Consortium GWAS Meta-Analyses | 175 4.1 ization for 72 genes which are associated with 42 signals in at least one GTEx tissue. The GTEx dataset does not contain cartilage or bone expression data; however, 33 signals colocalized in tissues and cell types that were found to have enrichment of active gene regulatory elements at the osteoarthritis-associated loci ( Figure 3, Supplementary Table 7 ). We also assessed if any of the genes residing within 1Mb of the osteoarthri- tis-associated lead variants, showed differential gene expression and protein abun- dance in primary osteoarthritis-affected tissue, in chondrocytes extracted from osteo- arthritis patients undergoing joint replacement surgery. From each patient, two paired cartilage samples were taken; a sample with a low OARSI grade which signifies healthy or macroscopically intact cartilage tissue (intact) and a sample with a high OARSI grade, which denotes highly degenerated (degraded) cartilage tissue. We combined data from two different sources: 35 paired samples from the RAAK study and 38 paired samples from a UK cohort ( Methods ). We only considered genes that were significantly (FDR <0.05) differentially expressed in both studies with the same direction of effect. For- ty-four genes showed significant differential expression across both studies, and 34 proteins showed differential abundance (FDR <0.05) ( Supplementary Table 8 ). Sim- ilarly, we compared gene expression of subchondral bone tissue underneath the intact cartilage and degraded cartilage from two different sources: 11 paired samples from the Taiwan OA study and 24 paired samples from the RAAK study ( Methods ). Thirteen genes were significantly differential expressed across both studies, of which five were also differentially expressed in cartilage tissue ( Supplementary Table 8 ). We have used these functional genomics lines of evidence (eQTL, gene expression and protein abundance differences) for the identification of likely effector genes (See Amassing ev- idence to identify effector genes section below, Table 3 ). For example, the tenascin C ( TNC ) gene demonstrated significant evidence of both increased gene expression and increased protein abundance in degraded compared to intact cartilage. TNC resides in the vicinity (<1Mb away) of three independent signals: rs7862601 (hip osteoarthritis); rs72760655 (osteoarthritis of the hip and/or knee); and rs1330349 (total hip replace- ment), which fine mapped to 3 variants all within the transcript of TNC . TNC is a compo- nent of the extracellular matrix and is involved with inflammation and cardiovascular disease[31, 32]. Causal inference analysis We used two-sample Mendelian randomisation (MR) to investigate potential caus- al relationships between 1,640 plasma proteins measured in up to 6,000 individuals and osteoarthritis[33-37]. Following validation using orthogonal information on DNA