Abstract: BACKGROUND: Osteoarthritis is a common chronic degenerative disease with a high correlation with age. However, its specific pathogenesis is still unclear, and there are many deficiencies in early diagnosis and treatment.
OBJECTIVE: To screen the key expressed genes in the synovium of osteoarthritis by bioinformatics method, so as to lay a foundation for finding biomarkers for the diagnosis of osteoarthritis and elucidating its underlying pathogenesis.
METHODS: Four osteoarthritic synovial tissue data sets were downloaded from the Gene Expression Omnibus database (GSE32317, GSE55235, GSE55457, GSE82107), including 27 normal synovial tissue samples and 49 osteoarthritic synovial tissue samples. The intersection genes of differentially expressed genes and Weighted Co-expression Network Construction Analysis results were screened using R language. Functional annotation, protein-protein interaction network, and immune infiltration analyses of the intersection genes were performed, and cytoscape was used for network visualization. Prism was used to draw receiver operating characteristic curve for the top 10 genes for degree ranking and screen out key genes, and then another synovial data set, GSE12021, was used for verification. Finally, the key genes in the synovial samples from four osteoarthritis patients and four non-osteoarthritis patients (joint injury) were further detected by PCR.
RESULTS AND CONCLUSION: (1) In this study, 263 differentially expressed genes and 1 237 key module genes of Weighted Co-expression Network Construction Analysis were identified, with a total of 98 intersection genes. (2) Protein-protein interaction node number of the top 10 most genes included interleukin 6, JUN, ATF3, MYC, DUSP1, VEGFA, FOSB, CXCL8, PTGS2, and NR4A1. (3) According to the receiver operating characteristic curve, ATF3 and DUSP1 had a higher diagnostic accuracy (area under the curve > 0.8). (4) Verification with the data set GSE12021 and PCR detection indicated that the gene expression difference of ATF3 and DUSP1 between the osteoarthritis group and the control group was statistically significant (P < 0.01), which was consistent with the results of network analysis. (5) Through bioinformatics analysis, ATF3 and DUSP1 are considered to have high diagnostic value for osteoarthritis and may be potential biomarkers and therapeutic targets for osteoarthritis diagnosis and treatment.

Key words: osteoarthritis, synovial tissue, differentially expressed genes, weighted co-expression network analysis, cellular stress response, signaling pathway, activating transcription factor 3, dual-specificity phosphatase 1