Construction and validation of a senescence-related gene signature for early prediction and treatment of osteoarthritis based on bioinformatics analysis

Abstract

The aim of this study is to screen key target genes of osteoarthritis associated with aging and to preliminarily explore the associated immune infiltration cells and potential drugs. Differentially expressed senescence-related genes (DESRGs) selected from Cellular senescence-related genes (SRGs) and differentially expressed genes (DEGs) were analyzed using Gene Ontology enrichment, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and protein-protein interaction networks. Hub genes in DESRGs were selected based on degree, and diagnostic genes were further screened by gene expression and receiver operating characteristic (ROC) curve. CIBERSORTx and ssGSEA algorithms were then used to assess immune cell infiltration and to analyse the correlation between key DESRGs and immune infiltration. Finally, a miRNA-gene network of diagnostic genes was constructed and targeted drug prediction was performed. Combined with the DEGs and SRGs, we screened 19 DESRGs for further study. Five diagnostic genes were ultimately identified: CDKN1A, VEGFA, MCL1, SNAI1 and MYC. ROC analysis showed that the area under the curve (AUC). Correlation analysis showed that the five hub genes were closely associated with neutrophil, plasmacytoid dendritic cell, activated CD4 T-cell and type 2 T-helper cell infiltration in the development of Osteoarthritis (OA). Finally, we found that drugs such as lithium chloride, acetaminophen, curcumin, celecoxib and resveratrol could be targeted for the treatment of senescence-related OA. The results of this study indicate that CDKN1A, VEGFA, MCL1, SNAI1, and MYC are key biomarkers that can be used to predict and prevent early aging-related OA. Lithium chloride, acetaminophen, curcumin, celecoxib, and resveratrol can be used for personalized treatment of aging-related OA.

Keywords: Bioinformatics; Immune Infiltration; Osteoarthritis; Senescence; Senescence-related genes.

Fig. 1

The flow chart of this study.

Fig. 2

Identification of DEGs and DESRGs in OA and control samples in the training set. (A) Volcano plot, (B) heatmap, and (C) difference ranking plot of DEGs between the OA and control samples. (D) Venn diagram of overlapping genes between the training set and CellAge database. DEGs, differentially expressed genes; SRGs, senescence-related genes.

Fig. 3

Expression and correlation analysis of 19 DESRGs. (A) Expression levels of DESRGs in the OA and control samples in the training set were visualized by box plot. *p < 0.05, **p < 0.01, ***p < 0.001. (B) Spearman correlation among 19 DESRGs.

Fig. 4

Dot plots of 19 DESRG-enriched GO terms and KEGG pathways. Panels (A–B) represent BP, CC, MF, and KEGG, respectively. Chord plots of 19 DESRG-enriched GO terms and KEGG pathways. Panels (C–F) represent the top five enriched items and their enriched genes of BP, CC, MF, and KEGG, respectively. GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; BP, biological process; CC, cellular component; MF, molecular function; DESRGs, differentially expressed senescence-related genes.

Fig. 5

Protein–protein interaction network and hub genes. (A) Protein–protein interaction network constituted with the DESRGs; (B) top 9 hub genes. Darker colors indicate a higher value. DESRGs, differentially expressed senescence-related genes.

Fig. 6

Diagnostic value of 7 hub genes in the training set. ROC curves of CDKN1A (A), FOS (B), MCL1 (C), MMP9 (D), MYC (E), SNAI1A (F), VEGFA (G).

Fig. 7

Diagnostic values of 7 diagnostic genes in the validation dataset. ROC curves of CDKN1A, FOS, MCL1, MMP9, MYC, SNAI1A, VEGFA (A). Comparison of the expressions of 7 diagnostic genes in the validation dataset. Expression levels of CDKN1A (B), FOS (C), MCL1 (D), MMP9 (E), MYC (F), SNAI1A (G), VEGFA (H). The mRNA relative expression levels of hub genes in theOA (n = 3) and normal groups (n = 3) were verified by RT-PCR operated (I). ns, no significance; *p < 0.05; **p < 0.01; ***p < 0.001.

Fig. 8

Evaluation of immune cell infiltration. The proportions of 22 kinds of immune cell types between OA and control samples (A). Histogram showed the differences of infiltrating immune cells between OA and control samples (B). The correlations between five hub genes and immune cells (C).

Fig. 9

Coexpression network of diagnostic genes and target miRNAs. Red ellipses represent diagnostic genes, and blue diamonds represent target miRNAs (A). The potential drugs of the 5 diagnostic genes of OA (B).