Integrative Proteome-Wide Mendelian Randomization and Multi-Omics Analysis Identify ADM and CFH as Candidate Genes for Osteoarthritis.

Abstract

Background: Osteoarthritis (OA) is a prevalent degenerative joint disease lacking effective disease-modifying therapies, which necessitates the discovery of key genes for mechanistic exploration and therapeutic development. Methods: We integrated three large-scale cis-protein quantitative trait locus datasets and two OA genome-wide association study summary statistics to screen candidate proteins by two-stage proteome-wide Mendelian randomization (MR). Causal association reliability was validated via summary-data-based Mendelian randomization (SMR) and Bayesian colocalization analyses. A phenome-wide association study (PheWAS) was performed to evaluate potential pleiotropic effects of the candidates. Subsequently, transcriptomic and single-cell RNA sequencing datasets were employed to evaluate the candidate genes' expression stability, classification efficacy in the in vitro models of OA, cell-specific enrichment, and pseudotime expression dynamics in cartilage. Finally, drug repurposing potential was explored by integrating drug-gene interaction database searches and molecular docking. Results: Two-stage cis-pQTL MR combined with cis-eQTL-based SMR analysis identified 14 plasma proteins with consistent effects at the protein and transcript levels. RNA-seq revealed that adrenomedullin (ADM) and complement factor H (CFH) were upregulated in two in vitro models of OA, and both genes exhibited favorable classification efficacy in these models. Bayesian colocalization analysis provided evidence of shared causal variants for ADM, and PheWAS did not detect significant pleiotropic associations for ADM or CFH across the tested phenotypes. Single-cell analysis indicated that ADM was enriched in pre-fibrocartilage chondrocytes with biphasic pseudotime expression, whereas CFH was widely expressed across chondrocyte subsets. Database screening identified 15 potential drugs for ADM and 6 for CFH. Conclusions: Combining MR, multi-omics and pharmacological evidence, we prioritized ADM and CFH as OA candidate genes.