p53, cell cycle, PI3K-Akt, and MAPK pathways are the key pathways related to ovarian cancer chemoresistance
Through the GEO database, we retrieved ovarian cancer-related microarray GSE46169 and ovarian cancer chemoresistance-related microarray GSE15372. The microarray GSE46169 contained a total of 2906 DEGs (Fig. 2A), and the microarray GSE15372 contained a total of 1433 DEGs (Fig. 2B). Finally, 224 genes related to chemoresistance in ovarian cancer were observed in the intersection (Fig. 2C).
According to the data from KEGG enrichment analysis on 224 genes, genes were found mainly enriched in p53 pathway (hsa04115), Cell cycle (hsa04110), Axon guidance (hsa04360), Protein digestion and absorption (hsa04974), PI3K-Akt pathway (hsa04151), and MAPK signaling pathway (Hsa04010) (Fig. 2D).
Recent studies have highlighted the important involvement of p53, PI3K-Akt, and p38 MAPK pathways in the chemoresistance of ovarian cancer cells [19,20,21]. Therefore, p53, cell cycle, PI3K-Akt, and MAPK pathways are considered as crucial pathways in influencing ovarian cancer chemoresistance.
CAFs are enriched in the p53, mTORC1, and G2M checkpoint pathways
To explore specific mechanism responsible for the effect of CAFs on chemotherapy resistance in ovarian cancer, we retrieved ovarian CAF-related microarrays GSE126132 and GSE40595 in the GEO database and utilized GSEA to identify the pathways in which ovarian CAFs were primarily enriched. Based on microarray GSE126132, cells were divided into CAFs and epithelial groups, and the results showed that CAFs were mainly enriched in p53, PI3K-Akt, Notch, Wnt/β-catenin, G2M checkpoint, and apoptosis pathways (Fig. 3A). The microarray GSE40595 involved normal ovarian stroma and ovarian cancer stroma. As displayed in, Fig. 3B CAFs were enriched in the p53, mTORC1, and G2M checkpoint pathways.
276 CAF-related candidate genes exert an vital role in ovarian cancer chemoresistance
To further explore the key genes that exert an vital role related to CAFs in ovarian cancer chemoresistance, differential analysis was performed on microarray GSE40595 associated with ovarian CAFs. DEGs were screened with the “limma” package in the R software, and heat map was plotted through the “pheatmap” package. GSE40595 contained 6276 DEGs. Volcano plots and heat map of TOP50 DEGs were shown in Fig. 4A-B. Meanwhile, we searched genes related to chemoresistance through the GeneCards database, and selected the top 1000 genes according to the Relevance score. Finally, 276 candidate genes were found in the intersection of the screened DEGs in GSE40595 and top genes from GeneCards database (Fig. 4C).
Functional enrichment analysis of 276 candidate genes that may affect the resistance of ovarian cancer cells to chemotherapy
Gene function enrichment analysis was performed on the aforementioned 267 candidate genes using the “clusterProfiler” package in the R software. GO enrichment analysis (p < 0.05) exhibited the significant effect of 256 molecular functions (MF) such as histone deacetylase binding (GO:0042826), enzyme binding (GO:0019899), and transmembrane receptor protein tyrosine kinase activity (GO:0004714). Also, 200 cellular components (CC) including Cytosol (GO:0005737), Cytosol (GO:0005829), and Membrane (GO:0016020) had significant effects. In addition, 944 biological processes (BP) such as negative regulation of apoptotic process (GO:0043066), positive regulation of protein phosphorylation (GO:0001934), and negative regulation of gene expression (GO:0010629) showed significant effects (Fig. 5A).
We here presented the results of KEGG enrichment analysis regarding the previously obtained key pathways related to CAF-affected chemoresistance (p < 0.05), and plotted the circle diagram to identify the key genes. It could be seen that MYC, IGF1, HRAS, CCND1, AKT1, RAC1, KDR, FGF2, FAS and EGFR are the top 10 genes in the circle diagram (Fig. 5B).
Verification of the PPI of the top 10 key genes
To further explore the PPI of these 267 candidate genes, we imported them into the String database and restricted the species to human. As shown in Fig. 6A, nodes represented proteins and edges represented correlations between proteins, involving 259 nodes and 2641 edges (PPI enrichment p-value < 1.0e-16). Degree (core degree) was the number of other genes interacted with the candidate gene in the PPI network. We sorted the top 10 according to the Degree value, namely AKT1, MYC, EGFR, IL6, STAT3, CASP3, JUN, HSP90AA1, HRAS, and CCND1 (Fig. 6B). Based on the data of 39 samples from the ovarian CAF-related microarray GSE40595, the “Corrplot” package in R software was utilized to conduct correlation analysis for the above 10 candidate genes, providing verification of PPI (Fig. 6C).
MYC, EGFR, and CCND1 may affect the survival of ovarian cancer patients by mediating chemotherapy resistance-related pathways
Based on the ranks of genes in the enrichment analysis of the key pathways and the Degree value in PPI, top 10 genes were selected from both of which and 5 key genes (AKT1, MYC, EGFR, HRAS, CCND1) were obtained by taking the intersection (Fig. 7A). GDC TCGA Ovarian Cancer (OV) was selected in UCSC Xena database and the influence of the above 5 key gene expression on the overall survival of 758 ovarian cancer patients was analyzed. The results showed that MYC, EGFR and CCND1 were significantly correlated with overall survival, while AKT1 and HRAS were not correlated with overall survival (Fig. 7B-F).
Further, we selected the survival data of TCGA-ovarian cancer patients after platinum chemotherapy for KM survival analysis, and the results exhibited that the expression patterns of MYC, EGFR, and CCND1 were highly correlated with the survival of TCGA-ovarian cancer patients after platinum chemotherapy (Fig. 7G-I). Correlation analysis was performed based on the expression of MYC, EGFR, and CCND1 in the TCGA-ovarian cancer dataset. It was observed that CCND1 expression was significantly positively correlated with EGFR and MYC (Fig. 7J-K). Therefore, it was speculated that MYC, EGFR, and CCND1 might affect the survival of ovarian cancer patients by mediating chemoresistance-related pathways.
Protein expression patterns of the key genes (MYC, EGFR, and CCND1) in ovarian cancer samples
The expression profiles of MYC, EGFR, and CCND1 were further verified at the protein level through IHC using the Human Protein Atlas database. MYC, EGFR, and CCND1 expressed at higher protein levels in ovarian cancer patients than in normal ovarian tissues (Fig. 8A-F). To further validate the relevance of MYC, EGFR, and CCND1 to platinum chemotherapy response, we analyzed their differential expression between platinum-sensitive and platinum-resistant high-grade serous ovarian cancer cells based on GEO microarray (GSE189717). Significantly high levels of MYC, EGFR, and CCND1 were detected in platinum-resistant high-grade serous ovarian cancer cell lines (Fig. 8G-I). Therefore, we speculated that MYC, EGFR, and CCND1 might enhance resistance to platinum chemotherapy in ovarian cancer.