Identification of Hub Genes Correlated With Poor Prognosis for Patients With Uterine Corpus Endometrial Carcinoma by Integrated Bioinformatics Analysis and Experimental Validation

Abstract

Uterine Corpus Endometrial Carcinoma (UCEC) is one of the most common malignancies of the female genital tract and there remains a major public health problem. Although significant progress has been made in explaining the progression of UCEC, it is still warranted that molecular mechanisms underlying the tumorigenesis of UCEC are to be elucidated. The aim of the current study was to investigate key modules and hub genes related to UCEC pathogenesis, and to explore potential biomarkers and therapeutic targets for UCEC. The RNA-seq dataset and corresponding clinical information for UCEC patients were obtained from the Cancer Genome Atlas (TCGA) database. Differentially expressed genes (DEGs) were screened between 23 paired UCEC tissues and adjacent non-cancerous tissues. Subsequently, the co-expression network of DEGs was determined via weighted gene co-expression network analysis (WGCNA). The Blue and Brown modules were identified to be significantly positively associated with neoplasm histologic grade. The highly connected genes of the two modules were then investigated as potential key factors related to tumor differentiation. Additionally, a protein-protein interaction (PPI) network for all genes in the two modules was constructed to obtain key modules and nodes. 10 genes were identified by both WGCNA and PPI analyses, and it was shown by Kaplan-Meier curve analysis that 6 out of the 10 genes were significantly negatively related to the 5-year overall survival (OS) in patients (AURKA, BUB1, CDCA8, DLGAP5, KIF2C, TPX2). Besides, according to the DEGs from the two modules, lncRNA-miRNA-mRNA and lncRNA-TF-mRNA networks were constructed to explore the molecular mechanism of UCEC-related lncRNAs. 3 lncRNAs were identified as being significantly negatively related to the 5-year OS (AC015849.16, DUXAP8 and DGCR5), with higher expression in UCEC tissues compared to non-tumor tissues. Finally, quantitative Real-time PCR was applied to validate the expression patterns of hub genes. Cell proliferation and colony formation assays, as well as cell cycle distribution and apoptosis analysis, were performed to test the effects of representative hub genes. Altogether, this study not only promotes our understanding of the molecular mechanisms for the pathogenesis of UCEC but also identifies several promising biomarkers in UCEC development, providing potential therapeutic targets for UCEC.

Keywords: hub gene; protein-protein interaction network; tumor differentiation; uterine corpus endometrial carcinoma; weighted gene co-expression network analysis.

Figure 1

Flowchart of the study. 548 patients with UCEC and corresponding clinical information were obtained from the TCGA database. According to the exclusion criteria, a total of 24 cases were excluded, and 524 UCEC patients were selected in this study. DEGs (including mRNAs and lncRNAs) were screened based on the gene expression data of 23 paired UCEC tissues and adjacent non-tumor endometrial tissues. Further, WGCNA algorithm was applied to construct the co-expression network of DEGs. Combined with PPI network analysis, highly correlated gene modules and key genes that were mostly associated with the clinical traits of UCEC were identified. Besides, the lncRNA-miRNA-mRNA and lncRNA-TF-mRNA networks were constructed to investigate the molecular mechanisms of UCEC-related lncRNAs. Finally, after a range of screening, 6 mRNA strands and 3 lncRNA strands with prognostic predictive potential were identified, of which expression patterns were experimentally validated. Moreover, AURKA and DUXAP8 were selected as representative hub genes for functional verification.

Figure 2

Identification of DEGs associated with UCEC. (A, B) The volcano plots and heatmaps of DEmRNAs and DElncRNAs. 2569 DEmRNA strands (1295 upregulated and 1274 downregulated) and 1457 DElncRNA strands (733 upregulated and 724 downregulated) were identified at the threshold of FDR < 0.05 and |log2FC| > 2. (C, D) PCA plots of DEmRNAs and DElncRNAs showed that 23 tumors were clustered separately from their paired adjacent non-cancerous tissues.

Figure 3

Construction of weighted co-expression network of DEGs in UCEC. (A, B) Examination of the scale−free fit index for distinct soft-thresholding powers (β), and average network connectivity under different weighting coefficients. β = 3 was selected as a soft threshold to construct the co-expression modules. (C) Heatmap depicting the TOM among DEGs based on the co-expression modules. (D) 8 modules were characterized due to a dissimilarity measure. Within the 8 modules, the grey module was a gene set in which the genes were not clustered into any modules.

Figure 4

Identification of key modules related to the clinical traits. (A) Module−trait heatmap of the correlation between module eigengenes and 5 clinical traits of UCEC. The P-values of each module’s correlation with the corresponding clinical trait were shown in parentheses. (B) Hierarchical clustering dendrogram of MEs and neoplasm histologic grade (labeled by asterisk). Heatmap of the adjacencies in the eigengene network, including neoplasm histologic grade. (C, D) GO term enrichment and KEGG pathway analysis for DEGs in the Blue and Brown modules.

Figure 5

Investigation of hub genes in the Blue and Brown modules. (A) Scatterplots of GS for neoplasm histologic grade (y-axis) versus MM (x-axis) in the two modules. (B) Key nodes were analyzed by CytoHubba using the following five methods: Closeness, Degree, edge percolated component (EPC), Maximum neighborhood component (MCC), and Maximum Neighborhood Component (MNC). (C) The most significant module obtained from the PPI network (MCODE score = 105.433). (D) Venn diagram demonstrated overlapping genes of the WGCNA and PPI network.

Figure 6

Analysis of the expression patterns of candidate genes (AURKA, BIRC5, BUB1, CCNA2, CCNB1, CDCA8, DLGAP5, KIF2C, NCAPG and TPX2). (A) The expression levels of candidate genes between poorly and well-differentiated UCEC in the TCGA dataset. (B) The protein expressions of candidate genes between UCEC tissues and non-tumorous tissues based on the Human Protein Atlas database. Most of these genes were higher in UCEC tissues compared to non-tumorous tissues, except that no protein expression was detected for KIF2C gene and there was no expression data for BUB1 gene in this database.

Figure 7

Associations between the expression levels of candidate genes (AURKA, BIRC5, BUB1, CCNA2, CCNB1, CDCA8, DLGAP5, KIF2C, NCAPG and TPX2) and the 5-year OS for 524 patients with UCEC based on the TCGA dataset. 6 of the 10 genes were significantly negatively related to prolonged patient survival time (AURKA, BUB1, CDCA8, DLGAP5, KIF2C and TPX2).

Figure 8

Exploration of the possible pathogenesis of 6 hub genes (AURKA, BUB1, CDCA8, DLGAP5, KIF2C and TPX2) in UCEC by using the GSEA algorithm. 5 common functional gene sets, ‘Cell cycle’, ‘DNA replication’, ‘Mismatch repair’, ‘Homologous recombination’ and ‘Oocyte meiosis’, were significantly enriched in UCEC samples with high gene expression.

Figure 9

Exploration of the molecular mechanism of UCEC-related lncRNA in the Blue and Brown modules. (A) Construction of lncRNA-transcription factor (TF)-mRNA networks. (B) 3 lncRNA strands (AC015849.16, DUXAP8 and DGCR5) were significantly negatively related to the 5-year OS of patients with UCEC, with higher expression levels in UCEC tissues compared to non-tumorous tissues. (C) The expression levels of AC015849.16, DUXAP8 and DGCR5 between poorly and well-differentiated UCEC tissues in the TCGA dataset.

Figure 10

Expression and functional validation of the selected hub mRNAs and lncRNAs in vitro. (A) The expression patterns of these genes in two endometrial cancer cell lines, including Ishikawa (histological grade 1; G1) and KLE (histological grade 3; G3). (B–E) The biological impacts of suppressing AURKA with Alisertib (MLN8237) in UCEC cells in vitro. Alisertib decreased Ishikawa cells viability in a dose-dependent manner, induced G2/M phase arrest and enhanced cellular apoptosis. Furthermore, Alisertib limited the long-term clonogenic survival (1 μM). (F) DUXAP8-shRNA/GFP lenti-virus (versus control virus) was used to knock down DUXAP8. (G, H) Downregulation of DUXAP8 inhibited the colony formation and impaired cells growth. Data are the mean ± SD of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.