The bioinformatics and experimental analysis of the novel roles of virus infection-associated gene CDC20 for prognosis and immune infiltration in hepatocellular carcinoma

Identification of differentially expressed genes

To screen the differentially expressed genes (DEGs) in hepatocellular carcinoma, we compared HCC tumor tissues and normal liver tissues in four publicly available datasets including GSE101685, GSE112790, GSE45267, and GSE84402. Using the filtering criteria of p-value < 0.05 and |log2 Fold Change|>2, 165 genes in GSE101685, 121 genes in GSE112790, 78 genes in GSE45267, and 208 genes in GSE84402 were significantly up-regulated in HCC tumor tissues. In addition, we also identified 351, 206, 232, and 308 down-regulated genes in GSE101685, GSE112790, GSE45267 and GSE84402, respectively (Supplementary Table 1). Next, using the Venn diagram, 54 co-DEGs were identified to be significantly up-regulated, and 125 co-DEGs were found to be down-regulated in these four datasets. These co-DEGs were hypothesized to have a significant impact on the tumorigenesis and progression of hepatocellular carcinoma.

Virus infections take a significant part in the tumorigenesis and development of hepatocellular carcinoma [9–11]. Next, we investigate the possible roles of HTLV-1 infections on the progression of HCC. The HTLV-1 infection-related gene datasets acquired from the MalaCards database were shown in Supplementary Table 2. We applied the Venn diagram to identify the overlapping genes between HTLV-1 infection-related genes and the aforementioned co-DEGs. As shown in Figure 1, five HTLV-1 infection-related genes, MAD2L1, CCNB2, CDC20, PTTG1 and BUB1B were highly expressed, and two genes, FOS and EGR1, were lowly-expressed (Figure 1). These seven HTLV-1 infection-related genes were presumed to play critical roles on virus infection-associated development of hepatocellular carcinoma.

CDC20 shows the greatest prognostic value in HCC

HTLV-1 coinfection has been reported to have an influence on the development of HCC in patients with hepatitis virus infection [12]. Here, we used the Kaplan–Meier plotter to explore the prognostic values of mRNA expression of the aforementioned seven HTLV-1 infection-related genes in HCC patients with hepatitis virus infection. As shown in Figure 2A and Supplementary Figure 1A, higher mRNA expression levels of MAD2L1, CDC20, PTTG1 and FOS were significantly associated with shorter overall survival (OS). Meanwhile, higher mRNA expression levels of CDC20 was obviously associated with shorter disease specific survival (DSS) (Figure 2B, Supplementary Figure 1B). Based on the above results, CDC20, as the only gene that exhibited prognostic value on OS and DSS at the same time, was presumed as a more potential prognostic target and chose for further investigation.

Furthermore, we conducted a multivariate analysis to assess the correlation between overall survival and multivariable characteristics, and the results showed that CDC20 expression was an independent factor for patient prognosis (Figure 2C, Supplementary Table 3). Figure 2D exhibited the distribution of CDC20 expression, the survival time of patients with HCC, and risk score profiles. Moreover, the ROC curve indicated that CDC20 expression possesses promising prognostic power. The AUC of CDC20 expression was 0.731 and 0.645 for predicting 1-year and 5-year overall survival, respectively (Figure 2E). These results suggest that CDC20 is a promising prognostic biomarker in patients with HCC.

CDC20 expression and its association with clinicopathological features in HCC

The expression profiles of CDC20 were further confirmed using several databases and in vitro experiments. First, four aforementioned datasets, GSE101685, GSE112790, GSE45267, and GSE84402, exhibited that the transcriptional level of CDC20 was significantly up-regulated in HCC tumor tissues than in normal liver tissues (Figure 3A). Then, the TNM plot and THPA database showed that the mRNA and protein expression levels of CDC20 were significantly higher in HCC tissues (Figure 3B and 3C). Furthermore, the up- regulation of CDC20 transcriptional and protein expression was confirmed in hepatocellular carcinoma cells HepG2 and SK-Hep1 compared with the normal liver cell L02 through qPCR and Western Blot. These results showed that CDC20 expression was significantly increased in hepatocellular carcinoma tissues and cell lines (Figure 3D and 3E).

Moreover, the relationship between the CDC20 expression and clinicopathological features of HCC was further investigated. According to the UALCAN database, the mRNA expression of CDC20 was identified to be significantly correlated with the tumor grade and the cancer stage of HCC. The results showed that the higher tumor grade, the higher CDC20 expression in HCC. Patients who had higher cancer stages (stage 1–3) tended to have higher CDC20 expression, while patients with stage 4 showed low expression of CDC20, which may be due to the small sample size and other factors (Figure 3F and 3G). In addition, we also used the XianTao tool to analyze the clinical data of HCC patients. Consist with the above results, CDC20 expression was found to be significantly correlated with histologic grade and pathologic stage. And beyond that, CDC20 expression was also assessed to be significantly correlated with tumor status and patient age (Table 1). These findings indicated that patients with higher CDC20 expression were more prone to having hepatocellular carcinoma that was more advanced in grade, stage, and status.

The generation of CDC20-associated network and the functional enrichment analysis of CDC20 in HCC

Here, cBioPortal and Cytoscape were enrolled to identify the network of 134 CDC20-associated genes in HCC patients (Supplementary Table 4). Among them, CYP3A4, CYP2E1, CYP2C9, and UGT2B15 were the top 4 genes that had the strongest correlation with CDC20 (Figure 4A). Then we presented the functional enrichment analysis of CDC20 in patients with HCC. Gene Ontology (GO) annotation revealed that CDC20 mainly involved the biological processes such as metabolic process, biological regulation, response to stimulus, and multicellular organismal process. Moreover, cell membrane, endomembrane system, nucleus, and membrane-enclosed lumen were the top 4 cellular components that were associated with the location and function of CDC20. Then the molecular functions correlated with CDC20 were evaluated. Consequently, protein binding, ion binding, hydrolase activity, and nucleic acid binding were on the list (Figure 4B). In addition, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis filtered 5 signaling pathways which were the most relevant to CDC20. They were the omega-hydroxylase P450 pathway, epoxygenase P450 pathway, long-chain fatty acid metabolic process, xenobiotic metabolic process, and steroid metabolic process (Figure 4C).

Correlation between CDC20 expression and tumor infiltrating immune cells

We explored the possible correlations between the mRNA expression of CDC20 and levels of immune infiltration in HCC using the XianTao tool. As exhibited in Figure 5, the expression of CDC20 showed significantly positive correlations with the infiltration levels of several different immune cells including T helper 2 cells (Th2), follicular helper T cell (TFH), NK CD56bright cells, and activated dendritic cells (aDC) (Figure 5).

Moreover, we also explored possible correlations between CDC20 expression and other immune signatures by using the TISIDB database. As shown in Supplementary Figure 2A and 2B, CDC20 expression showed significantly positive correlations with the levels of several immunostimulators including TNFRSF18 and MICB. Supplementary Figure 2C and 2D exhibited that CDC20 expression was negatively correlated with several immunoinhibitors including KDR and PDCD1LG2. In addition, the correlations between CDC20 expression and chemokines and chemokine receptors were further assessed. The results showed that CDC20 expression was significantly associated with several chemokines such as XCL1 and CCL26, and several chemokine receptors such as CCR10 and CXCR3 (Supplementary Figure 3). These results indicated that CDC20 could play a critical role in the infiltration and functions of immune cells in HCC.

Use of bioinformatics tools for acquiring data and reanalysis

The bioinformatics tools and public databases used in this study was summarized in Supplementary Table 5. Briefly, four GSE datasets were obtained from the gene expression omnibus (GEO) database for the identification of co-differentially expressed genes (co-DEGs) in HCC. GSE101685 included 10 normal liver tissues and 24 HCC tissues. GSE112790 included 15 normal liver tissues and 183 HCC tissues. GSE45267 included 39 normal liver tissues and 48 HCC tissues. GSE84402 included 14 paired normal liver and HCC tissues. HTLV-1 infection associated genes were downloaded as a dataset from the MalaCards database. A Venn diagram was used for filtering overlapping genes among the aforementioned datasets.

The prognostic role of CDC20 in HCC was measured via Kaplan–Meier plotter, a public online resource containing information on gene expressions and survival in all kinds of cancers [28–30]. Overall survival and disease specific survival were enrolled here as the two primary endpoints for prognostic analysis. Cox regression was used to perform the multivariate analysis to evaluate the independent risk factors of OS in HCC patients by the XianTao tool.

TNM plot [31] and THPA databases [32] were employed to exhibit the mRNA and protein levels of CDC20 in patients with HCC. A cutoff with statistical significance was set at the point that the p-value equals to 0.05. The UALCAN database and XianTao tool was used to collect the clinicopathological features (tumor grades and cancer stages) and CDC expression distribution for correlation analysis [33].

The network of co-expression genes of CDC20 was generated by the cBioPortal database [34], then WebGestalt was recruited to fulfill the functional enrichment analysis including Gene Ontology (GO) analysis and the Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis [35].

TISIDB database, an integrated portal for tumor-immune analysis, was used in this study to illustrate the correlation between CDC20 and immunomodulators [36]. The XianTao tool was used to explore the correlation between CDC20 and immune cells.

Cells and reagents

Normal human hepatocyte cell line L02 (HL7702), human hepatocellular carcinoma cell lines HepG2 and SK-Hep1 were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Cells were cultured under the normal condition with the 5% CO₂.

qPCR and Western blot

qPCR and Western blot were performed following the protocol we previously described [37]. The primer of CDC20 was as followed: F: 5′-ACGGTTTTGATGTAGAGGAAGC-3′; R: 5′-GATACGGTCTGGCAGGGAAG-3′. The primary antibodies of CDC20 (10252-1-AP, Proteintech) and β-tubulin (2146, Cell Signaling Technology) were applied.

Statistical analysis

Statistical analyses were performed with SPSS 18.0 software (IBM Analytics, USA). All experiments were performed triplicate for statistical analysis, with mean ± SD. Kaplan-Meier analysis was performed to analyze survival rates. The differential mRNA expression between HCC and hepatic cell lines were analyzed using Student’s t-test. Cox regression analysis was used for multivariate survival analysis. p < 0.05 were defined as statistically significant. ^* means p < 0.05; ^** means p < 0.01; ^*** means p < 0.001.