Integrated bioinformatics and tissue-based validation reveal the oncogenic role of hsa_circ_0043256 and hsa_circ_0004789 in gastric cancer

Abstract

Circular RNAs (circRNAs) play a key role in gastric cancer (GC) pathogenesis. This study hsa_circ_0043256 and hsa_circ_0004789, and their interactions with miR-28-5p/Cyclin B1 (CCNB1) and miR-5683/CCNB1, via bioinformatic and experimental methods. We retrieved expression data for circRNAs, miRNAs, and mRNAs in GC from Gene Expression Omnibus and The Cancer Genome Atlas. Using online databases and R tools, we identified downstream miRNAs and target mRNAs to build a competing endogenous RNA (ceRNA) network. After identification of hub genes and performing functional enrichment, we defined two regulatory axes: hsa_circ_0043256/miR-28-5p/CCNB1 and hsa_circ_0004789/miR-5683/CCNB1. We studied 32 paired tumor and adjacent tissues to assess all genes and CCNB1 protein expression, along with correlations, histopathological associations, ROC curves, and survival outcomes. We identified 58 circRNAs, 123 miRNAs, and 2126 mRNAs, and, further, by novelty checking and downstream RNA analysis, identified two axes. The expression of hsa_circ_0043256, hsa_circ_0004789, and CCNB1 mRNA and protein levels was elevated, while miR-28-5p and miR-5683 levels were reduced. Correlations observed among axis components supported the ceRNA hypothesis. The hsa_circ_0004789/miR-5683/CCNB1 axis showed an AUC value for the combined ROC curve near 1, suggesting strong diagnostic potential. Lower CCNB1 and higher miR-5683 levels were correlated with better survival. Both circ_0043256 and circ_0004789 were associated with histological grade, lymphatic invasion, perineural invasion, and lymph node involvement. This study highlights circ_0043256/miR-28-5p/CCNB1 and circ_0004789/miR-5683/CCNB1 as promising axes for GC diagnosis and treatment strategies.

Keywords: Circular RNA; Computational biology; Cyclins; MicroRNA; Stomach neoplasms.

Graphical abstract

Fig. 1

Flowchart illustrating the sequential steps and methods used in the bioinformatics analysis.

Fig. 2

The box plots demonstrate the overall expression profiles of the two datasets: (A) before removing batch effect, and (B) after removing the batch effect. Volcano plots for (C) differentially expressed circRNAs (DECs), (D) differentially expressed miRNAs (DEMIs), and (E) differentially expressed mRNAs (DEGs).

Fig. 3

The schematic representations of (A) hsa_circ_0043256 and (B) hsa_circ_0004789 were obtained from the CSCD database. Box plots show the expression levels of (C) hsa_circ_0043256 and (D) hsa_circ_0004789 in both tumor and normal samples.

Fig. 4

ceRNA network. The network consisting of two DECs (hsa_circ_0043256 and hsa_circ_0004789), three FImiRNAs (hsa-miR-28–5p, hsa-miR-5683, and hsa-miR-145–3p), and 38 FImRNAs. Diamond represents circRNA, V represents miRNA, and Ellipse represents mRNA.

Fig. 5

PPI analysis. The PPI networks of FImRNAs associated with hsa_circ_0043256 (A) and hsa_circ_0004789 (B) are shown. Genes that have no relationships with others are placed outside the network. Hub gene networks of the first (C) and second axis (D) are also shown. The importance of hub genes is determined based on a color scale, with red indicating higher importance and yellow indicating lower importance.

Fig. 6

Functional enrichment analysis. The results of GO analysis for hsa_circ_0043256 (A) and hsa_circ_0004789 (D) are demonstrated in bar plots, which include three categories: biological process (BP), cellular component (CC), and molecular function (MF). The results of KEGG analysis are shown as bar plots for hsa_circ_0043256 (B) and hsa_circ_0004789 (E). A guide is located adjacent to each bar plot, indicating that greater red color intensity is associated with smaller adjusted p-values. Cnetplots show the relationship between each individual protein and its function in the BP category of GO analysis for hsa_circ_0043256 (C) and hsa_circ_0004789 (F).

Fig. 7

The expression levels of circ_0043256 (A), circ_0004789 (B), miR-28–5p (C), miR-5683 (D), and CCNB1 (E) were evaluated by RT-qPCR in gastric cancer and non-tumor margin tissues. Representative Western blot bands (F) and quantitative densitometry measurement of CCNB1 protein expression (G) are also demonstrated. The uncropped Western blot images can be found in Supplementary Fig. 1. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗∗p < 0.0001.

Fig. 8

The correlation between the components of axes in GC tissues. Correlation analysis between hsa_circ_0043256 and miR-28–5p expression levels by Spearman correlation coefficient (A). Correlation analysis between CCNB1 and hsa_circ_0043256 expression levels through Pearson method (B). Correlation analysis between CCNB1 and miR-28–5p expression levels (Spearman) (C). Correlation analysis between hsa_circ_0004789 and miR-5683 expression levels by using Spearman correlation analysis (D). Correlation analysis between CCNB1 and hsa_circ_0004789 expression levels through Pearson test (E). Correlation analysis between CCNB1 and miR-5683 expression levels (Spearman) (F).

Fig. 9

ROC curve analysis of circ_0043256 (A), circ_0004789 (B), miR-28–5p (C), miR-5683 (D), CCNB1 (E), the combination of hsa_circ_0043256/miR-28–5p/CCNB1 (F), and the combination of hsa_circ_0004789/miR-5683/CCNB1 (G).

Fig. 10

Overall survival curves of circ_0043256 (A), circ_0004789 (B), miR-28–5p (C), miR-5683 (D), and CCNB1 (E) in GC patients, evaluated by Kaplan–Meier analysis.