OBJECTIVES: To reveal causal level osteoarthritis (OA) disease biology by targeting regulatory interactions at GDF5. METHODS: By investigating different GDF5 regulatory regions (R2, R3-5, R7-R9, R18-20, GROW1) we explored their functional impacts on gene expression and joint morphology in vivo and in vitro. We additionally modeled OA variants in said enhancers in vitro and in vivo mouse models for expression and disease effects. RESULTS: For all regulatory regions we found evidence of activation/repression between or within said regions that impacted patterns of joint-specific expression. Examples are: (1) the R4 enhancer, whilst considered to be activating, has dual roles repressing expression in adjacent tissues and sites
and (2) Growth plate-specific expression patterns by the GROW1 regulatory region are confined by adjacent sequences to restrict its expression to the perichondrium. We next targeted different regions/variants in vivo. Testing the R2de region resulted in ~40% reduction in Gdf5 expression, joint morphology changes, but no increase in OA risk
likewise, modeling the most cited OA risk (rs143384) variant in mice had no impact on expression, joint morphology, or disease. However, we identified epistatic interactions between this rs143384 risk variant and downstream disease risk variants lying within regulatory regions subject to repression, that compound to impact expression. CONCLUSIONS: These findings, at the best studied OA locus to date, serve as lessons on the nature of how gene regulatory interactions and local epistasis work in the etiology of OA disease risk, and that assessment of individual variants of high GWAS significance need not alone be considered causal.