DNA methylation biomarkers for hepatocellular carcinoma

Hepatocellular carcinoma (HCC), an inflammation-driven disease, is the third deadliest cancer worldwide, and HCC prevalence is predicted to continue to rise in coming years, serving as a major economic burden [1, 2]. Most cases of HCC occur in developing countries, such as China, and the leading cause of HCC is infection with hepatitis B virus (HBV); in contrast, the main cause in developed countries, such as the USA, is infection with hepatitis C virus (HCV) [3, 4]. Other risk factors for developing HCC include exposure to aflatoxin, alcohol intake, smoking, and diabetes [5]. The best curative treatments to date in early stage HCC patients involve surgical resection, tumor ablation, and potentially liver transplantation [6, 7]. However, the prognosis after curative therapy for HCC remains unsatisfactory because of a high postoperative recurrence rate. An improved understanding of the basic biology of HCC is needed to enhance prognostic predictions and to enhance therapeutic efficacy against this deadly disease.

The term epigenetics refers to heritable gene expression alterations no associated with DNA sequence changes [8]. The DNA methylation is closely related to embryonic development [9], regulation of gene expression [10], X-chromosome inactivation [11], genomic imprinting [12], and genomic stability [13]. Altered DNA methylation such as tumor suppressor gene hypermethylation or oncogene hypomethylation is thought to promote tumorigenesis. Genes including P15, P16, Ras association domain family 1 isoform A (RASSF1A), and Retinoblastoma 1 are inactivated in HCC due to promoter hypermethylation of these genes [14‐17]. Given that methylation is potentially reversible, detection of such aberrant DNA methylation of tumor suppressors and oncogenes in HCC could be useful as a therapeutic target.

While altered methylation of many genes has been demonstrated to date in the context of HCC, a complete interaction network documenting the relationship between said genes remains to be produced. The comprehensive analysis of multiple datasets offers the power needed to properly identify and assess pertinent pathways and genes mediating the biological processes associated with HCC. To this end, we used datasets from microarrays examining gene expression (GSE84402, GSE46408) and gene methylation (GSE73003, GSE57956) to assess HCC gene and epigenetic signatures, allowing for identification of genes and pathways that were both abnormally methylated and differentially expressed. Using a protein–protein interaction network we were also able to identify key so-called “hub” genes central to these signaling events. Through this analysis, we believe it is possible to identify novel differentially methylated genes associated with HCC, offering key insights into the molecular mechanisms governing HCC development and progression.

Exploring the mechanisms underlying HCC development and progression not only has prognostic implications, but may also be helpful in monitoring treatment response, surveillance for tumor recurrence, and guidance of clinical decisions. Modern advances in sequencing technologies and microarray development have provided ample high-throughput opportunities to study disease-related biology, allowing for simultaneous assessment of gene methylation and expression for thousands of genes in the human genome. In the present study, we identified 19 hypomethylated, highly-expressed genes and 14 hypermethylated, low-expression genes, using bioinformatic analysis. Functional enrichment of these genes revealed that aberrant methylation indeed affects certain pathways and hub genes. These results can provide novel insight into the explanation of HCC pathogenesis.

The GO enrichment analysis revealed that the primary biological processes of the hypomethylated/highly-expressed genes were the regulation of cell cycle processes, chromosome segregation, and mitotic nuclear division while the hypermethylated/low-expression genes were involved mainly in controlling cell proliferation, gene expression, and the mitotic G1/S transition. This is expected given that the chromosome segregation process occurs during mitosis, which is a part of the cell cycle. The G1/S cell cycle transition is tightly controlled. Deregulation of this key checkpoint can allow cells to undergo transformation, thereby permitting tumorigenesis. This finding is consistent with the fundamental role played by cell cycle regulators in cell proliferation, invasion, and metastasis in HCC. KEGG pathway analysis of hypomethylation/high-expression genes revealed that they were linked to the cell cycle, oocyte meiosis, and ubiquitin-mediated proteolysis. The cell cycle and oocyte meiosis are vital for cell proliferation in tumor cells, and the ubiquitin proteasome pathway functions to regulate cell cycle control and the DNA damage response in tumor genesis [22]. The KEGG pathway analysis of hypermethylation/low-expression genes suggested that methylation may be linked to HCC development through the p53 and MAPK signaling pathways. p53 is a tumor suppressor to conserve genome stability by preventing mutations caused by cellular stress or DNA damage. Together, these results suggest that hypermethylation and hypomethylation are key mediators of cancer development and progression.

The PPI network of hypomethylated/highly-expressed genes provides insight into the functional associations between them, and from this, the top three hub genes were selected: MAD2L1, CDC20, and CCNB1. Mitotic Arrest Deficient 2 Like 1 (MAD2L1) and Cell Division Cycle 20 (CDC20) are two key mitotic checkpoint genes. Both MAD2L1 and CDC20 were more highly expressed in higher grade tumors than in low-grade tumors. High MAD2L1 or CDC20 levels may allow for the development of multi-organ aggressive tumors, including those affecting the breasts, lungs, liver, and stomach [23‐25]. Collectively, these findings suggest that MAD2L1 and CDC20 may be key regulators of tumor severity, ultimately dictating patient survival. Cyclin B1 (CCNB1), complexing with CDC2, is a G2/M-phase checkpoint regulator that is vital for regulation of proliferation and DNA synthesis. CCNB1 overexpression has been found to occur in HCC [26] and many other cancer, often being linked to progression, recurrence, and to poor prognoses [27]. Therefore, MAD2L1, CDC20, and CCNB1 may all be abnormally methylated genes that modulate the cell cycle and proliferation in HCC. With regard to the hypermethylated/low-expression genes, the most prominent hub genes were CCND1, AR, and ESR1. Cyclin D1 (CCND1) is a proto-oncogene regulating G1 to S phase progression; it participates in the Wnt/β-catenin pathway [28, 29]. Androgen receptor (AR) is a steroid hormone receptor superfamily member that is involved in human hepatocarcinogenesis [30]. It alters the AR-dependent transcriptome and stimulates oncogenic growth. Estrogen receptor 1 (ESR1) functions as a transcription factor, regulating cell cycle, cell proliferation, apoptosis, and inflammation-associated gene expression [31]. Research has shown that estrogen-depleted postmenopausal women undergo more rapid progression from HCV-infection to HCC development [32]. Aberrant expression of ESR subtypes may contribute to the progression of HCC. These three genes are related to prognosis, tumorigenesis, and metastasis of HCC. Furthermore, survival analysis of hub genes revealed that MAD2L1, CDC20 and CCNB1 play an oncogenic role, while CCND1, AR, and ESR1 genes were associated with favorable patient survival in HCC.

The core PPI network module for hypomethylated/highly-expressed genes was linked to the cell cycle, oocyte meiosis, and ubiquitin-mediated proteolysis, indicating that these pathways are key targets of hypermethylation. The top two modules of the hypermethylated/low-expression gene PPI network were those linked to the p53 signaling pathways, viral carcinogenesis and neurotrophin signaling pathways. p53 signaling conserves the stability of the genome. The leading cause of HCC is infection with HBV or HCV. It is reasonable that viral carcinogenesis is involved in the development of HCC. Neurotrophins and neurotrophin receptors are found on tumor and stromal cells, and are linked to many kinds of tumor development. Brain-derived neurotrophic factor and nerve growth factor are both neurotrophins linked to tumor development, promoting proliferation, angiogenesis, and invasion.

While previously groups have assessed arrays cataloging gene expression or methylation, the two have not been examined simultaneously. Furthermore, single studies generally lack the power needed to identify critical regulatory genes and signaling pathways. Our research used a bioinformatics workflow to jointly analyze extant gene expression and gene methylation profiling microarrays, allowing for more powerful and precise insights into these screening results. However, we only validated candidate abnormally methylated genes that were differentially expressed using the TCGA database. Further experiments will be necessary in order to confirm that these genes and pathways are linked to HCC.

In summary, this study provides a comprehensive bioinformatics analysis of aberrantly methylated DEGs that may be involved in the progression and development of HCC. In addition, six mostly changed hub genes were identified, including MAD2L1, CDC20, CCNB1, CCND1, AR, and ESR1. These novel findings may contribute to the unraveling of the pathogenesis of HCC, and these candidate genes may be optimal abnormal methylation-based biomarkers that can be used to accurately diagnose and treat HCC.