The miRNA-mRNA Regulatory Network in Human Hepatocellular Carcinoma by Transcriptomic Analysis From GEO

Abstract

Background: Bioinformatics analysis of hepatocellular carcinoma (HCC) expression profiles can aid in understanding its molecular mechanisms and identifying new targets for diagnosis and treatment.

Aim: In this study, we analyzed expression profile datasets and miRNA expression profiles related to HCC from the GEO using R software to detect differentially expressed genes (DEGs) and differentially expressed miRNAs (DEmiRs).

Methods and results: Common DEGs were identified, and a PPI network was constructed using the STRING database and Cytoscape software to identify hub genes. The reduced levels of tumor suppressor miRNAs or down regulated DEmiRs may be increased levels of oncogenes, the oncomirs or up regulated DEmiRs may be decreased levels of tumor suppressor genes in cancerous cells. According to this strategy, increased and decreased DEGs, also increased and decreased DEmiRs were selected. The multimir package was employed to predict target genes for DEmiRs then DEmiRs-hub gene network created. We identified approximately 1000 overlapping DEGs and 60 DEmiRs. Hub genes included RRM2, MELK, KIF11, KIF23, NCAPG, DLGAP5, BUB1B, AURKB, CCNB1, KIF20A, CCNA2, TTK, PBK, TOP2A, CDK1, MAD2L1, BIRC5, ASPM, CDCA8, and CENPF, all associated with significantly worse survival in HCC. miR-224, miR-24, miR-182, miRNA-1-3p, miR-30a, miR-27a, and miR-214 were identified as important DEmiRs with targeting more than six hub genes.

Conclusion: Generally, our findings offer insight into the interaction of hub genes and miRNAs in the development of HCC by bioinformatics analysis, information that may prove useful in identifying biomarkers and therapeutic targets in HCC.

Keywords: HCC; bioinformatics; biomarker; hepatocellular carcinoma; microRNA; microarray.

FIGURE 1

Flow diagram of data analysis GSE112790, GSE115018, GSE36376, GSE113996, GSE39791, GSE10694, and GSE36915 related HCC of the GEO (www.ncbi.nlm.nih.gov/geo).

FIGURE 2

Volcano graphs for GSE112790, GSE115018, GSE36376, GSE113996, GSE39791, GSE10694, and GSE36915 dataset indicated the significant differences in expression between HCC and control samples based on p‐value < 0.05 and |log2FC| > 1. p‐value (blue dots), only logFC (green dots), both p‐value and log2FC (red dots), or not significant in both categories (gray dots) were indicated by colored dots. FC, fold‐change; HCC, hepatocellular carcinoma.

FIGURE 3

A Venn diagram (http://bioinformatics.psb.ugent.be/webtools/Venn) was created to illustrate the overlapping DEGs in GSE112790, GSE115018, GSE36376, GSE113996, and GSE39791 related to HCC. To ensure crucial genes were not overlooked and to comprehensively examine the obtained DEGs in HCC, common DEGs were selected from at least two or more datasets. DEGs, differentially expressed genes; HCC, hepatocellular carcinoma.

FIGURE 4

Expression levels of the top DEmiRs in TCGA normal (n = 49) and LIHC tumor (n = 369) were analyzed using the UALCAN database in HCC. The difference in miRNA expression between the cancer sample and the normal sample was significant in all investigated top DEmiRs, consistent with our analysis. However, the expression chart in the UALCAN database was not available for hsa‐miR‐886‐5p in LIHC. DEmiRs, differentially expressed miRNA; LIHC, liver hepatocellular carcinoma.

FIGURE 5

The PPI network of overlapping DEGs from the investigated datasets in HCC, comprising 837 nodes and 12 286 edges, was constructed through the STRING database and Cytoscape software. Within the hub genes, the red nodes represent those with higher node scores, while the yellow nodes represent those with lower node scores in the network.

FIGURE 6

mRNA expression validation of identified hub genes from the overlapping DEGs analysis in HCC was conducted using the GEPIA database. Comparison of the expression levels of the 20 identified hub genes in HCC (red box; n = 369) and normal tissues (black box; n = 160) revealed that all hub genes except for KIF11 and DLGAP5 were significantly increased in HCC compared with normal liver tissues. A p‐value < 0.01 was considered statistically significant.

FIGURE 7

Kaplan–Meier overall survival analysis was conducted for the top 20 hub genes from the investigated datasets in HCC. High expression of RRM2, MELK, KIF23, NCAPG, BUB1B, AURKB, CCNB1, KIF20A, CCNA2, TTK, PBK, TOP2A, CDK1, MAD2L1, BIRC5, ASPM, CDCA8, CENPF, KIF11, and DLGAP5 was associated with poor overall survival of HCC patients. High expression of these hub genes was linked to unfavorable overall survival.

FIGURE 8

GO enrichment analysis was performed on the obtained overlapping DEGs from the investigated datasets in HCC to determine the most significant GO terms in (A) biological processes (BP), (B) cellular components (CC), and (C) molecular functions (MF) using FunRich software. (D) KEGG pathways for overlapping DEGs were discovered and visualized using FunRich software. Terms with a p‐value < 0.05 were considered significant.

FIGURE 9

(A) Overlapping genes between predicted target genes of upregulated DEmiRs and decreased DEGs were identified via Venn diagram. PPI networks were then constructed using the overlapping genes. Hub genes in PPI networks were identified by CytoHubba based on the MCC and degree methods. (B) Overlapping genes between predicted targets of downregulated DEmiRs and upregulated DEGs were identified using a Venn diagram, and PPI networks of the obtained overlapping DEGs were generated. Hub genes in PPI networks were identified by CytoHubba based on the MCC and degree methods. (C) The DEmiRs‐DEGs network of decreased hub genes and their related DEmiRs were visualized through Cytoscape software. (D) The DEmiRs‐DEGs network of increased hub genes and their related DEmiRs were visualized through Cytoscape software.