PubMedFASEB journal : official publication of the Federation of American Societies for Experimental Biology2026-10-26
Integrative Genetic, Transcriptomic, and Single-Cell Analyses Prioritize Druggable Candidate Genes in Knee Osteoarthritis.
Zheng Yiping Y, Xuan Yingying Y, Li Yi Y, Deng Changxu C et al.
Knee osteoarthritis (KOA) is a common degenerative joint disease with a complex pathogenesis and a lack of effective molecular targets. This study aimed to prioritize druggable genes associated with KOA and investigate their expression patterns and cellular localization by integrating genetic, transcriptomic, and single-cell analyses. We first screened for potential druggable genes genetically associated with KOA using MR analysis. Subsequently, batch-corrected synovial transcriptomic data were analyzed using weighted gene co-expression network analysis (WGCNA), and the genes of the most relevant module were integrated with the MR candidates. A total of 127 machine-learning pipelines were evaluated using repeated nested cross-validation, followed by external validation and SHapley Additive exPlanations (SHAP) interpretation. The expression levels of key genes were further validated by experiments. Single-cell RNA sequencing, immune-cell deconvolution, and peripheral immune-cell-specific eQTL-MR were used to investigate the cellular contexts potentially associated with the candidate genes. Finally, potential drug-gene interactions were retrieved from DGIdb, and selected target-compound pairs were evaluated using molecular docking and molecular dynamics simulations. A total of 315 druggable genes showed nominal MR associations with KOA. After integrating KOA-associated WGCNA module, we identified 15 candidate genes. Next, the random forest was selected as the optimal machine-learning model, achieving a mean outer-cross-validation AUC of 0.929 (95% CI, 0.913-0.945). SHAP and differential-expression analyses prioritized four genes. JUN, ADAMTS1, and CRISPLD2 were downregulated in KOA synovium, whereas CX3CR1 was upregulated, and these expression patterns were supported by experimental validation. Single-cell analysis revealed that JUN, ADAMTS1, and CRISPLD2 were predominantly expressed in fibroblast-related populations and CX3CR1 was mainly expressed in myeloid cells, with their expression varying across fibroblast and macrophage states. Peripheral immune-cell-specific eQTL-MR identified nominal associations for JUN in plasma cells, ADAMTS1 in NK_R cells, and CX3CR1 in NK cells. However, none remained significant after correction for multiple testing. Molecular docking and molecular dynamics simulations showed that the ADAMTS1-pravastatin and CX3CR1-Rugocrixan complexes maintained stable protein-ligand contacts during their respective simulations. Integrative multi-omics analyses identified JUN, ADAMTS1, CRISPLD2, and CX3CR1 as prioritized candidate genes associated with KOA. These genes were associated with distinct states of synovial fibroblasts and myeloid cells. These findings provide a foundation for further mechanistic validation and experimental evaluation of candidate drug-target interactions.
