Fig. 5. Association between GDF5 expression and phenotype severity.

a Linear regression analysis showing a strong linear relationship between levels of Gdf5 expression and extent of bony shape changes in four distinct mouse lines, two of which harbor separate enhancer deletions (i.e., of R4 and GROW1) and two of which harbor separate, single-base-pair replacements in these enhancers (i.e., at rs6060369 and rs4911178). Dark circles = homozygous, open circles = heterozygous. See the Supplementary Information for related analyses. b Average fold change in anatomy between homozygous and heterozygous (enhancer null and single base-pair) mutations in R4 and GROW1 enhancers. Summary data are presented as mean and standard deviations. c Mechanistic model showing two separate non-coding regulatory variants on the same risk GDF5 haplotype, i.e., rs4911178 “A” variant in the GROW1 enhancer and rs6060369 “T” variant in the R4 enhancer control different pathologies (i.e., DDH and knee OA, respectively) in different joints, mediated by differential binding of PITX1 on each respective enhancer variant in each specific joint.