Identifying Essential Hub Genes and circRNA-Regulated ceRNA Networks in Hepatocellular Carcinoma

Competitive endogenous RNAs (ceRNAs) absorb microRNAs and subsequently promote corresponding mRNA and long noncoding RNA (lncRNA) expression, which may alter cancer cell malignancy. Thus, dissecting ceRNA networks may reveal novel targets in cancer therapies. In this study, we analyzed differentially expressed genes (DEGs) of mRNAs and lncRNAs, and differentially expressed microRNAs (DE-miRNAs) and circular RNAs (DE-circRNAs) extracted from high-throughput sequencing datasets of hepatocellular carcinoma patients. Based on these data, we identified 26 gene modules using weighted gene co-expression network analysis (WGCNA), of which 5 were associated with tumor differentiation. In these modules, 269 genes were identified by GO and KEGG enrichment and patient's survival correlation analyses. Next, 40 DE-miRNAs, each of which potentially bound a pair of DE-circRNA and hub gene, were discovered. Together with 201 circRNAs and 24 hub genes potentially bound by these miRNAs, 1151 ceRNA networks were constructed. Among them, 75 ceRNA networks consisting of 24 circRNAs, 28 miRNAs and 17 hub genes showed a positive circRNA-hub gene correlation. For validation, we carried out experiments for 4 randomly selected circRNAs regulating 19 potential ceRNA networks and verified 5 of them. This study represents a powerful strategy to identify essential gene networks and provides insights into designing effective therapeutic strategies.