Significance of Identifying Key Genes Involved in HBV-Related Hepatocellular Carcinoma for Primary Care Surveillance of Patients with Cirrhosis

Abstract

Cirrhosis is frequently the final stage of disease preceding the development of hepatocellular carcinoma (HCC) and is one of the risk factors for HCC. Preventive surveillance for early HCC in patients with cirrhosis is advantageous for achieving early HCC prevention and diagnosis, thereby enhancing patient prognosis and reducing mortality. However, there is no highly sensitive diagnostic marker for the clinical surveillance of HCC in patients with cirrhosis, which significantly restricts its use in primary care for HCC. To increase the accuracy of illness diagnosis, the study of the effective and sensitive genetic biomarkers involved in HCC incidence is crucial. In this study, a set of 120 significantly differentially expressed genes (DEGs) was identified in the GSE121248 dataset. A protein-protein interaction (PPI) network was constructed among the DEGs, and Cytoscape was used to extract hub genes from the network. In TCGA database, the expression levels, correlation analysis, and predictive performance of hub genes were validated. In total, 15 hub genes showed increased expression, and their positive correlation ranged from 0.80 to 0.90, suggesting they may be involved in the same signaling pathway governing HBV-related HCC. The GSE10143, GSE25097, GSE54236, and GSE17548 datasets were used to investigate the expression pattern of these hub genes in the progression from cirrhosis to HCC. Using Cox regression analysis, a prediction model was then developed. The ROC curves, DCA, and calibration analysis demonstrated the superior disease prediction accuracy of this model. In addition, using proteomic analysis, we investigated whether these key hub genes interact with the HBV-encoded oncogene X protein (HBx), the oncogenic protein in HCC. We constructed stable HBx-expressing LO2-HBx and Huh-7-HBx cell lines. Co-immunoprecipitation coupled with mass spectrometry (Co-IP/MS) results demonstrated that CDK1, RRM2, ANLN, and HMMR interacted specifically with HBx in both cell models. Importantly, we investigated 15 potential key genes (CCNB1, CDK1, BUB1B, ECT2, RACGAP1, ANLN, PBK, TOP2A, ASPM, RRM2, NEK2, PRC1, SPP1, HMMR, and DTL) participating in the transformation process of HBV infection to HCC, of which 4 hub genes (CDK1, RRM2, ANLN, and HMMR) probably serve as potential oncogenic HBx downstream target molecules. All these findings of our study provided valuable research direction for the diagnostic gene detection of HBV-related HCC in primary care surveillance for HCC in patients with cirrhosis.

Keywords: HBV-encoded oncogene X protein (HBx); bioinformatic analysis; co-immunoprecipitation/mass spectrometry (CO-IP/MS); hepatitis B virus (HBV); hepatocellular carcinoma (HCC).

Figure 1

The workflow diagram of data acquisition, preprocessing, analysis, and validation. Two distinct techniques were used to identify DEGs in the GSE121248 dataset. Among the DEGs, a PPI network was constructed, and Cytoscape was used to extract hub genes. The TCGA database validated the expression levels, correlation analysis, and predictive performance of hub genes. Using the GSE10143, GSE25097, GSE54236, and GSE17548 datasets, the expression pattern of hub genes during the progression from cirrhosis to HCC was analyzed. A prediction model was then developed using Cox regression analysis. In both cell models, CDK1, RRM2, ANLN, and HMMR interacted specifically with HBx, as determined by Co-IP/MS. DEGs, differentially expressed genes; Co-IP/MS, co-immunoprecipitation coupled with mass spectrometry; PPI, protein–protein interaction.

Figure 2

Exploration of potential DEGs in the GSE121248 dataset. The PCA plot (A) and UMAP clustering plot (B) based on samples from the para-cancer liver tissue group and the HBV-infected HCC tissue group. The distance of sample point clusters indicates that they come from different groups. The Venn diagram of up-regulated DEGs (C) and down-regulated DEGs (D) between the two methods. Volcano plots of DEGs (E). Blue and red indicate down-regulated overlapping DEGs and up-regulated overlapping DEGs, respectively. The expression profiles of overlapping DEGs between HBV-infected HCC tissues and para-cancer liver tissues were visualized in a heatmap (F).

Figure 3

GO and KEGG enrichment analyses of differentially expressed genes in HBV-related HCC. GO term enrichment results (A,B). KEGG enrichment results (C,D). The blue nodes represent the entries, the red nodes represent the numerators, and the connecting lines represent their relationship.

Figure 4

PPI network and module analysis. PPI network construction for candidate genes using STRING11.0 was visualized using Cytoscape (A). The size of the node refers to the degree standard, and the color of the node represents the values of log₂FC. The network directly associated with the top 20 hub genes identified using CytoHubba (B). The color depth of the nodes represents their importance in the network.

Figure 5

Expression validation of candidate hub genes in TCGA database. Heatmaps of hub genes expressed in tumors and adjacent normal tissues in patients with HCC (A). Validation of the expression level of hub genes in patients with HCC (B). The correlation matrix of interaction in hub genes. Correlation coefficients are plotted with negative correlation (blue) and positive correlation (red) (C). Functional enrichment analysis of the up-regulated hub genes (D). GO 1901990: regulation of mitotic cell cycle phase transition; GO 1901987: regulation of cell cycle phase transition; GO 0010389: regulation of G2/M transition of the mitotic cell cycle; GO 0005819: spindle; GO 0030496: midbody; GO 0070938: contractile ring; GO 0004674: protein serine/threonine kinase activity; GO 0008353: RNA polymerase II CTD heptapeptide repeat kinase activity; hsa04110: cell cycle; hsa04115: p53 signaling pathway; hsa04512: ECM–receptor interaction. *** p < 0.001.

Figure 6

Expression validation of 15 up-regulated hub genes between liver cirrhosis and HCC tissues in the GEO database. HCC tissue group vs. cirrhosis tissue group in GSE10143 (A), HCC tissue group vs. cirrhosis tissue group in GSE54236 (B), HCC tissue group vs. cirrhosis tissue group in GSE25097 (C), and HBV-related HCC tissue group vs. HBV-related cirrhosis tissue group in GSE17548 (D). Green and red represent cirrhosis and HCC samples, respectively. GEO, Gene Expression Omnibus. ** p <0.01, **** p < 0.0001.

Figure 7

Multifaceted clinical value analysis of up-regulated hub genes in HCC. (A) Kaplan-Meier plot of OS between high-expression and low-expression groups. The red lines and blue lines, respectively, represent patients with high gene and low gene expression. (B) Risk score in the cohort of TCGA. The distribution of patients with an increased risk score into high- and low-risk groups, as well as scatter plots depicting patient survival with an increased risk score. (C) Forest plot showing OS after univariate Cox regression analysis in HCC. (D) Nomogram for predicting the probability of 1-, 3-, and 5-year OS for patients with HCC. OS, overall survival.

Figure 8

Assessment of precision, sensitivity, and clinical utility of the predictive model. (A) Calibration curve for predicting the probability of 1-, 3-, and 5-year OS for patients with HCC. Calibration curve of the nomogram in HCC from TCGA data. (B) DCA plots of the nomogram by calculating the C-index. (C) ROC curve analysis of 15 up-regulated hub genes to distinguish HBV-related HCC tissues from normal tissues. ROC, receiver operating characteristic; DCA, decision curve analysis.

Figure 9

Identification of HBx-interacting proteins. (A) The co-immunoprecipitated mixture was separated using SDS-PAGE and stained with Coomassie blue. (B) Venn diagram of overlapping HBx-interacting proteins and hub genes. Identification of CDK1, RRM2, ANLN, and HMMR from the HBx-interacting protein complex extracted from LO2-HBx cells (C) and Huh-7-HBx cells (D) using Co-IP/MS analysis.