COPZ2 and KDELR3 are Linked to Stromal Inflammation and Metabolic Reprogramming in Osteoarthritis
Cheng Wang, Juan Xiao, Hanyu Wang, Honglin Pi, Li Wang, Qianqian LiangIntroduction:
Fibroblasts are important contributors to Osteoarthritis (OA) pathogenesis, but the molecular features associated with their pathogenic activation remain incompletely defined.
Methods:
The bulk RNA-seq and single-cell (sc) RNA-seq data of OA patients were obtained from a public database. ScRNA-seq analysis was performed using the “Seurat” package. High-dimensional Weighted Gene Co-expression Network Analysis (hdWGCNA) was applied specifically to activated fibroblasts to identify hub genes, which were then integrated with Differentially Expressed Genes (DEGs) and refined using LASSO regression and SVM-RFE algorithms. Functional enrichment and molecular docking were used to assess the biological relevance of top candidates and to explore candidate compound-target interactions.
Results:
We identified a metabolically and secretory active fibroblast subset in OA synovium. hdWGCNA identified 12 gene modules, five of which are specific to activated fibroblasts. Integrating hub genes and 407 DEGs yielded 15 candidates. Two machine learning methods identified COPZ2 and KDELR3 as key signature genes. COPZ2 was associated with adipogenesis and oxidative phosphorylation, whereas KDELR3 was associated with cell cycle regulation, EMT, and TNF-α/NF-κB signaling. Molecular docking predicted a potential interaction between Trichostatin A and KDELR3, with a binding affinity of −6.47 kcal/mol.
Discussion:
This study identified COPZ2 and KDELR3 as candidate hub genes associated with pathogenic fibroblast activation in OA. Trichostatin A was identified as a candidate KDELR3-binding compound, which requires further experimental validation
Conclusion:
Our findings suggest that COPZ2 and KDELR3 may serve as potential regulators of the pathogenic fibroblast phenotype in OA. KDELR3 warrants further investigation as a potential drug-related target, and Trichostatin A should be regarded as a computationally predicted candidate compound rather than a validated inhibitor.