Cindy Boer

106 | Chapter 2.2 size (n~9,900)[18, 22]. This indicates that assessment of stratified phenotypes in OA may be warranted to im-prove GWAS statistical power and provide novel insight into OA biology. Using bioinformatic analysis and functional genomics datasets, we were able to identify rs1158850 as potential causal variant. This SNV is nearby meQTL CpGs and the G allele of this variant is predicted to increase RAD21 (RAD21 Cohesin Complex Compo-nent) binding affinity in a region that has chromatin interactions with the WNT9A pro-moter. RAD21 is a part of the cohesion complex, involved in the formation of chromatin loops with CTCF[45]. Both RAD21 and CTCF bind to the WNT9A promoter region. In line with these findings, WNT9A expression was significantly increased in OA lesioned cartilage compared to preserved OA cartilage. Combining all results, we postulate that rs1158850 is located in a gene regulatory element, increases RAD21 binding, and me-diated by CTCF, interacts with the WNT9A promoter to influence WNT9A expression. WNT9A (Wingless-type MMTV integration site family, member 9A) previously known as WNT14 , is a member of the WNT gene family, and has been shown to play a central role in synovial joint formation[46, 47]. Knockout WNT9A mice have severe skel-etal developmental defects, and are neonatal lethal[48]. Expression of WNT members by chondrocytes leads to the destruction of the cartilage matrix by the upregulation of Wnt/β-catenin signaling. Inhibition of WNT members has been suggested as a plausible OA therapeutic strategy, with recent success in a murine model of OA[49, 50]. However, this is the first evidence for WNT9A, a non-canonical Wnt ligand, in human OA. In addition, to identifying novel OA loci, we also provide evidence for a sub-set of generalized OA genetic risk loci: GDF5 (rs143384), TGFα (rs2862851/rs3771501), RUNX2 (rs12154055), ASTN2 (rs2480930, rs13823416), COL27A1 (rs919642), and IL11 (rs4252548) . These loci should be given priority as potential therapeutic targets since genetically supported drug targets have been shown to double the success rate of ther-apeutics in clinical development and intervention at these target loci may be beneficial regardless of which joint site is affected by OA [51]. Although collectively our findings implicate WNT9A in thumb OA, there are sever-al limitations. First, age is the most predominant risk factor for OA, yet the genetic back-ground may determine the age of onset, rather than the lifetime risk for OA. Therefore, future genetic studies may benefit from examining the age of onset rather than adjust for age[52]. Second, our functional findings are based on chondrogenic data sourced from several different tissues that did not include tissues from the hand joints. How-ever, consistent results were found across the