Develop a circular RNA-related regulatory network associated with prognosis of gastric cancer

Abstract

Background: In gastric cancer (GC), circular RNAs (circRNAs) mainly play an important role in miRNA sponge, which not only indicate long-term survival and prognosis but also increase resistance to the apoptosis. The purpose of the study is to explore new circRNAs and their underlying mechanisms in GC.

Method: Through rigorous retrieval strategies, we used the sva package to analyze and identify differentially expressed circRNAs (DECs) from three Gene Expression Omnibus microarray datasets (GSE83521, GSE89143, and GSE78092). Online website CSCD and CircInteractome were used to reveal the binding sites between miRNAs and DECs. The possible target miRNAs of the DECs identified based on miRNAs, and Cytoscape was used to create a regulatory network of circRNA-miRNA-mRNA and identified the hub genes which were further validated using The Cancer Genome Atlas database and Human Protein Atlas.

Results: Twenty-eight DECs were obtained using the sva package. A regulatory network of circRNA-miRNA-mRNA (competing endogenous RNA) containing 15 circRNAs, 24 miRNAs, and 158 genes was identified. A protein-protein interaction network based on the 158 genes was established, and further determined that 10 hub genes (SKA1, ANLN, CHEK1, SKA3, TOP2A, BIRC5, RRM2, NCAPG2, FANCI, and RAD51) were associated with some cancer-related pathways based on the functional enrichment analysis. Finally, six hub genes (BIRC5, TOP2A, FANCI, NCAPG2, RAD51, and RRM2) were proven to influence the overall survival of GC.

Conclusion: Our study established a circRNA-miRNA-mRNA regulatory network and defined six circRNA-related hub genes in GC, which could serve as potential therapeutic targets or prognostic biomarker for GC treatment.

Keywords: ceRNA; circRNA; gastric cancer; prognosis; regulatory network.

FIGURE 1

Volcano plots for differentially expressed circRNAs identified from Gene Expression Omnibus

FIGURE 2

Heatmap for the 28 differentially expressed circRNAs in three microarray datasets (GSE83521, GSE89143, and GSE78092)

FIGURE 3

Venn diagram for the intersections between DEGs of TCGA and miRNA target genes. DEG, differentially expressed gene; TCGA, The Cancer Genome Atlas

FIGURE 4

circRNA‐miRNA‐mRNA regulatory network. The network consisting of 15 circRNAs, 24 miRNAs, and 158 genes was generated by Cytoscape 3.6.1

FIGURE 5

GO and KEGG enrichment annotations of DEGs (up‐(A) and down (B)‐regulated DEGs). DEG, differentially expressed gene; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes

FIGURE 6

Identification of hub genes from the PPI network with the MCODE algorithm. The node color changes gradually from blue to red according to the log2(foldchange) of genes. The node size changes gradually from small to big according to the degree calculated by the Centiscape plug‐in. The edge size represents the strength between two node which calculated by online website STRING. A, A PPI network of the 158 target genes. B, A PPI network of the 10 hub genes. PPI, protein‐protein interaction

FIGURE 7

Images of immunohistochemical staining with the 10 hub genes

FIGURE 8

Survival curves of the 10 hub genes