Interplay of Genetic and Epigenetic Alterations in Hepatocellular Carcinoma

Abstract

Genetic and epigenetic alterations play prominent roles in hepatocarcinogenesis and their appearance varies depending on etiological factors, race and tumor progression. Intriguingly, distinct patterns of these genetic and epigenetic mutations are coupled not only to affect each other, but to trigger different types of tumorigenesis. The patterns and frequencies of somatic variations vary depending on the nature of the surrounding chromatin. On the other hand, epigenetic alterations often induce genomic instability prone to mutation. Therefore, genetic mutations and epigenetic alterations in hepatocellular carcinoma appear to be inseparable factors that accelerate tumorigenesis synergistically. We have summarized recent findings on genetic and epigenetic modifications, their influences on each other’s alterations and putative roles in liver tumorigenesis.

Keywords::

• DNA methylation
• epigenetic modifications
• genetic mutations
• hepatocellular carcinoma
• signaling pathways

Figure 1. Mutational and epimutational pathway representation in hepatocellular carcinoma.

Among genes mentioned in the text, the type of mutations (genetic mutation; purple, aberrant DNA methylation; red and blue) are shown along with their effect on gene expression (red circle; activation, green circle; suppression). Key pathways and cellular processes affected by the alterations are shown. Mutation rate of each gene is shown using the data obtained from exome sequencing analysis of 243 liver tumors [11] and SNP genotyping of 286 liver tumors [55]. We also showed epigenetically altered genes (WNT5A, NFATC and CCND1) identified from DNA methylation analysis of hepatocellular carcinoma patients (n = 379) from TCGA (The Cancer Genome Atlas).

Hepatocellular carcinoma (HCC) is found prevalently in Asia and Africa, where hepatitis viral infection and consumption of Aflatoxin B1-contaminated food are high. Liver cirrhosis, caused by high alcohol consumption or obesity, is another risk factor for HCC [1]. These HCC risk factors usually cause chronic inflammation, which provides a favorable environment for tumor growth by generating diverse cytokines that promote cell growth and check point escape [2]. The chronic inflammation also enables the accumulation of high levels of genetic and epigenetic damages due to continued viral integration and harsh environmental conditions. The recent development of sequencing-based mapping technology has enabled many comprehensive studies for HCC genomes and epigenomes, resulting in the accumulation of substantial evidence elucidating the molecular culprits of hepatocarcinogenesis. Somatic alterations including mutations, copy number alterations and chromosomal rearrangements have been identified as tumor drivers in HCC along with global changes in DNA methylation, histone modification and chromatin structure. In particular, hypermethylation of promoter CpG islands (CGIs) and global hypomethylation of repeats are considered as additional tumor-promoting changes. These genetic and epigenetic alterations accumulate gradually throughout tumor progression and act together to promote HCC development [3]. As more data that simultaneously analyzes somatogenic and epigenetic mutations in cancer cells are gathered, the integrated data begin to elucidate the mechanism of the tumorigenesis. Therefore, we aimed to summarize the interaction between genetic and epigenetic factors in the development of HCC.

Risk factors

Approximately 70–80% of HCC occurs in cirrhotic liver, but diverse spectrums of histological defects have been observed in advanced HCC, reflecting the complexity of the etiological factors. For example, diverse etiological factors can induce HCC, but its incidence rate varies geographically, and is more frequently found in men than in women. Consequently, diverse environmental agents can induce alterations that are commonly required for the development of HCC, but the distinct biological and chemical nature of risk factors may produce unique genetic and epigenetic modifications, resulting in different HCC phenotypes. Therefore, understanding the genetic and epigenetic effects of these risk factors is important for the development of effective therapeutic approaches.

Hepatitis B virus & hepatitis C virus infection

Hepatitis B virus (HBV) infection is the major cause of HCC in east Asia and Africa in general, but hepatitis C virus (HCV) is the main type of infection in Japan, USA and Europe. Although only 10% of the HBV infected people develop chronically active infection, a significant number of HBV-infected patients develop HCC [2]. On the other hand, approximately 60–80% of HCV infection evades host immunity and develops into a chronically active infection, while only 2.5% of HCV-infected patients develop HCC [2]. These observations suggest that some additional factors of HBV infection, other than the chronic inflammation that accompanies both HBV and HCV infection, must play important roles in tumorigenesis. A higher genome integration rate or the oncoviral HBx protein of HBV may be responsible for the increase in tumor development observed in patients chronically infected with HBV relative to HCV.

Several additional differences between HBV- and HCV-related HCC are as follows. TERT is the most frequently mutated gene in HCC and its mutation is associated with poor prognosis. Alterations in TERT can result from promoter mutation or viral DNA integration into the locus. TERT promoter mutation is more frequently found in HCV-related HCC (HCV-HCC). Unlike HCV-HCC, HBV frequently integrates into the TERT locus in HBV-related HCC, which may produce a selective advantage over mutation since HBV integration and promoter mutation are mutually exclusive in HBV-HCC [4]. In addition to these differential effects of the viral risk factors on genetic alterations of HCC, the viral risk factors also affect epigenetic alterations differently. Although both HBV- and HCV-HCC show mutations in chromatin remodeling factors such as ARID-containing proteins (ARID family) and MLL family, chromosomal aberrations and genome instability are more frequent in HBV-HCC [5]. Similarly, CpG hypermethylation is a general feature shared in most HCC, but methylation profiles for HBV- and HCV-HCC also show different features. Hypermethylation of NAT2, CSPG2, DCC, NTKR3, TNFSF10, TNFRSF10C and RASGRF1 was associated with HBV-HCC, while hypermethylation of RIK and CHGA genes was more frequently found in HCV-HCC [6]. How these differences in methylation patterns affect HCC development remained poorly defined.

Alcohol & Aflatoxin B1

The leading nonviral cause of HCC is alcohol abuse in the USA and some European countries, while Aflatoxin B1 exposure is the main cause in China and Africa [1]. Acetaldehyde and free radicals generated by metabolizing alcohol induce DNA damage and oxidative stress, which often accelerate monocyte activation and telomere shortening [7]. The alcohol-derived risk factors (acetaldehyde and free radicals) seem to function mainly as a mutagen affecting DNA integrity; however, its role in epigenetic regulation in liver cells is now being discovered. Intriguingly, alcohol-mediated HCC (alcohol-HCC) showed a higher frequency of epigenetic aberration compared with other HCC types. Mutation in ARID1A, a key component of the switch/sucrosenon-fermentable (SWI–SNF) chromatin remodeling complex, occurs significantly more frequently in alcohol-HCC than in other risk factor-associated HCCs [8]. In addition, aberrant DNA methylation was observed more frequently in alcohol-HCC than in HCV-HCC [9]. Aflatoxin B1 in contaminated foods such as corn and peanuts is metabolized into a highly reactive epoxide to form a stable ring-opened adduct in the DNA [10]. Therefore, aflatoxin B1 can induce genetic mutations that are associated with a high rate of C to A transversion [10,11]. These lines of evidence indicate that different risk factors elicit different genetic and epigenetic alterations in HCC (Table 1).

Genetic alterations in HCC influenced by epigenetic modifications

Diverse epigenetic modification patterns can affect the types and frequencies of genetic alterations at the neighboring chromatin regions. Integrative analysis of whole genome sequencing with epigenome sequencing have identified how the epigenetic landscape influences the accumulation of genetic alterations during hepatocarcinogenesis [11–15]. The analysis also showed that several cancer driver genes are mutated either by genetic or epigenetic alterations. Genetic mutations that are affected or substituted by epigenetic alterations include single nucleotide variant (SNVs) mutations, small insertions and deletions (indels) and DNA copy number variations (CNVs). We have summarized recent findings that demonstrate molecular links between genetic mutations and epigenetic modifications.

Single nucleotide variants

The average number of somatic point mutations in HCC is 2.8 per megabase at the whole-genome level but only 1.3 mutations per megabase in coding sequences, thus highlighting the effect of chromatin organizations associated with coding regions on mutation frequencies [11,15]. Chromatin modifications could be the major factor influencing this somatic substitution density at the neighboring locations. Repressive histone modification, H3K9 methylation and H4K20 trimethylation were positively correlated with somatic substitution density [16]. DNA methylation also seemed to affect somatic substitution density as differentially methylated regions in the liver were specifically enriched with SNPs affecting cancer-related pathways [17]. For example, cancer risk SNPs at the BRD2 and HLA gene loci were significantly tied to differentially methylated DNA regions in HCC [17].

Because diverse mutational mechanisms are at play during tumorigenesis, the SNV patterns of each tumor are different and reflect the nature of the tumor-inducing conditions. HCC shows high frequencies of C to T, C to A and T to C nucleotide substitutions [18]. Among these, C to T transitions at CpG dinucleotide sequences are caused by the relatively elevated rate of spontaneous deamination of 5-methyl-cytosine in tumors. Thus, the DNA demethylation levels of tumor cells could influence the mutation frequencies induced by the deamination process. In addition, the significant upregulation of the APOBEC family in HBV- and HCV-HCC promotes C to T and C to A mutations by deaminating cytidine to uracil (C to U), coupled with the base excision repair and DNA replication processes. Besides APOBEC enzymes, several cellular and viral proteins also affect base substitutions epigenetically. A methyl-CpG-binding protein, MBD4, can affect the C to T transition rate, probably by regulating accessibility of the methylated C for deamination or repair enzymes. HCC with a reduced level of MBD4 showed higher incidence of C to T transitions, which might be responsible for the poor differentiation phenotype observed [19]. Similarly, MBD4 knock out mice displayed an approximately threefold increase in C to T transitions compared with wild-type littermates, demonstrating its protective role in C to T mutation [20]. These lines of evidence suggest that DNA methylation plays an important role in the accumulation of C to T mutation in HCC. In addition, HBx, an HBV oncoprotein, could promote higher accumulation of C to T transitions in HBV-positive HCC patients by competitive inhibition of TDG required for base excision repair [21].

C to A and T to C mutations are often found in the highly transcribed strand due to removal of the bulky DNA adducts by transcription-coupled nuclear excision repair. This indicates a link between epigenetic status of the DNA templates and mutations caused by error-prone repairs. These types of mutations are highly associated with Aflatoxin B1 and tobacco use [11,15]; however, no obvious mutational patterns were found associated with virus infection [13,15].

Chromosomal instability & copy number variants

DNA methylation status also has a strong effect on chromosomal integrity. Global DNA hypomethylation is observed in HCC and can induce activation of transposons and chromosomal instability, thus contributing to the generation of a large number of CNVs during hepatocarcinogenesis. Hypomethylation-associated reactivation of repetitive elements such as LINE-1, ALU and juxtacentromeric SAT2 is frequently observed in HCC along with copy number variations caused by insertions and deletions [22]. Intriguingly, the loss of repetitive DNA appears tightly associated with its hypomethylation [23].

Chromosomal insertions associated with HBV integrations also show a dependence on the surrounding chromatin structures. As expected, HBV virus integrations are detected more frequently in tumors (86.4%) than in adjacent HBV-infected liver tissues (30.7%) [12,15]. Besides, the increased integration in tumors is targeted to the coding regions: HBV breakpoints in the tumors are observed more frequently in exons and promoters of coding genes contrary to nontumor samples, which exhibit major integrations into introns. However, further studies are required to understand whether a competitive enrichment of disruptive integrations or a different epigenetic landscape of exonic areas in cancer cells is responsible for the increase in mutational incidence observed.

Epigenetic alterations influenced by genetic changes in HCC

During the course of HCC development diverse genetic mutations accumulate; however, not all mutations function as ‘drivers’ for cancer progression. Totoki et al. identified 30 candidate driver genes from 503 liver cancer genomes, and Schulze et al. identified 161 putative driver genes from 243 liver cancer genomes through exome sequencing analysis [11,15]. Some of these genes are key components of core pathway modules that control cell type-specific epigenetic patterns. Among these cancer-associated mutations, diverse factors in epigenetic modifications (32% of HCC) and chromatin remodeling (28% of HCC) were significantly enriched in HCC, indicating the importance of epigenetic alterations in driving cancer development [11]. Abnormal histone modifications or chromatin structures in HCC are suspected to drive cancer development by generating abnormal chromatin structures that are prone for mutations. Here, we have summarized alterations in genes that could contribute to cancer-driving epigenetic modifications (Table 2).

Chromatin remodeling factors

Diverse chromatin remodeling factors are often mutated in HCC. ARID1A and ARID2 are components of the SWI/SNF-related chromatin remodeling complexes and are significantly mutated in HCC (ARID1A, 16.8%; ARID2 5.6%) [24]. ARID1A mutations resulting in inactivation of its function are frequently observed in alcohol-related HCC, whereas ARID2- inactivating mutations were frequently observed in individuals with HCV-associated HCC (18.2%) [8]. As key components of the SWI/SNF chromatin remodeling complexes required for high levels of immune cell activations, ARID mutations may contribute to abnormal interferon-α-induced JAK–STAT signaling to permit lifelong persistence of HCV infection [8]. Mutations in other SWI/SNF components, including SMARCA2, SMARCA4, SMARCB1, SMARCC1 and SMARCC2, have also been found frequently in HCCs. Particularly, inactivating SMARCA2 mutations (2.6%) were significantly enriched in alcohol-related HCC [11].

Histone modification factors

The enzymes responsible for active histone H3 lysine 4 methylation (H3K4me) marks, namely MLL1–5, are inactivated by recurrent somatic mutations (approximately 1–6%) in HCC [11,13]. In addition, a high incidence of HBV integration into the MLL2 and MLL4 gene loci was identified in HCC [12]. In contrast, increased H3K4me3 was associated with mutations in SMYD3 methyltransferase and showed a strong correlation with poor prognosis, particularly in the early stages of HCC [25]. How these contrasting mechanisms can trigger development of HCC has yet to be fully elucidated. Long noncoding RNA HOTAIR represses SETD2 which trimethylates H3K36 in human liver cancer tissues. High HOTAIR levels are associated with poor prognosis, a larger primary tumor size and reduced expression of SETD2. This reduction could affect the DNA damage response due to abnormal H3K36me3 patterns [26]. Above all, SETD2 mutations were found frequently (2.6%) in cancer patients [11]. Another histone methyltransferase, SETDB1, for H3K9 was found to be upregulated in HCC and associated with cancer progression and metastasis. The gain of chromosome 1q21, which contains SETDB1, and downregulation of miR-29, which targets SETDB1 RNA, are often found in HCC and are probably linked to higher expression of SETDB1 [27].

Defects in histone acetylation/deacetylation processes are also strongly associated with HCC. BRD7 and BPTF, which contain a bromodomain that recognizes acetylated lysine residues, were frequently mutated, and their loss-of-functions showed a tight association with HCC [13,15]. Nevertheless, the viral oncoprotein HBx can mediate aberrant histone acetylation through direct interaction with CBP/P300 and HDAC1 in HBV-associated HCC [28]. Besides, diverse additional histone modification factors show abnormal expression and appear to play important roles in HCC development; however, genetic defects responsible for these aberrant expressions have yet to be clearly identified for most of them.

DNA methylation

A systematic analysis of genetic mutations affecting the HCC-associated DNA methylation pattern is currently insufficient, but several factors involved in the regulation of DNMTs have been reported. Viral infection, chronic inflammation, oxidative stress and even methyl-deficient diets have been shown to affect DNMT expression in HCC [9]. Viral oncoproteins, such as the HCV core and the HBx protein of HBV, upregulate DNMT1 and DNMT3 [9]. While HBx can directly interact with DNMT3A to induce abnormal hypermethylation at tumor suppressor gene (TSG) promoters [9], HBx also causes hypomethylation of intragenic CGIs indirectly through the epigenetic repression of DNMT3L expression, which induces transcriptional silencing of the associated genes [29]. In addition to transcriptional regulation, DNMT3B4, a splice variant of DNMT3B lacking DNA methyltransferase activity, is highly expressed in HCC, causing heterochromatin instability associated with hypomethylation of pericentromeric satellite regions [30]. Targeting TET1, methylcytosine dioxygenase, by oncogenic miR-494 reduced the level of 5′-hydroxymethylcytosine in the proximal CpGs of TSGs, including multiple invasion-suppressor miRNA genes [31]. The methyl donor S-adenosyl-l-methionine [32] is primarily synthesized and degraded in liver by MAT1. Thus, downregulation of MAT1A in HCC may cause a passive loss of methylated cytosines in HCCs [33]. Identification of genetic mutations affecting these processes will help us to understand the interactions between genetic mutations and DNA methylation leading to HCC development.

DNA methylation changes as cancer drivers in HCC

HCC is a heterogeneous disease affected by various lifestyles and environmental factors. Epigenetic alterations are frequently caused by these factors and contribute to hepatocarcinogenesis. Abnormal DNA methylation in many cancers is the most studied epigenetic mechanism in modulating expression of the associated gene. Global hypomethylation and targeted hypermethylation at CpG sites in the 5′ regions of TSGs in HCC are typical abnormal epigenetic alterations. Aberrant DNA methylation has been detected as early events during hepatocarcinogenesis and gradually increases with cancer progression [34]. In this section, we will describe the influence of aberrant DNA methylations on driving HCC.

DNA methylation changes in cirrhotic liver & HCC

Ammerpohl et al. observed abnormalities of DNA methylation patterns in cirrhotic liver and HCC using the Infinium Human Methylation27 Bead Array [35]. They identified 247 and 1276 differentially methylated CpG loci in cirrhotic livers and HCC, respectively, compared with that of normal livers. Among them, 80 CpG loci were commonly hyper- or hypomethylated in both cirrhotic livers and HCC. The commonly hypermethylated genes were devoid of CGIs and PRC2 target sites. In contrast, the genes specifically hypermethylated in HCC were enriched with CGIs and PRC2 target sites. Shen et al. examined genome-wide DNA methylation profiles in 66 pairs of HCC and adjacent nontumor tissues using the Infinium Human Methylation 450 Bead Array [36]. They found that 130,512 and 17,207 CpG loci exhibited significantly different methylation levels in HCC and cirrhosis, respectively, with only eight CpG loci in common. These results confirm the distinct patterns of aberrant DNA methylation associated with cirrhosis and HCC. In particular, as tumors progress from early, less progressed, to highly progressed HCCs, DNA methylation levels of several TSGs gradually increased, with enrichment of different subsets of hypermethylated TSGs in various stages of HCC development [34]. This result suggests that DNA methylation of TSGs has a strong cancer-driving potential. Indeed, Villanueva et al. found 36 DNA methylation markers that can predict poor survival of HCC, indicating their possible role in promoting cancer [37]. Some of these DNA methylation changes, such as LINE-1 hypomethylation and promoter hypermethylation of STEAP4, RASSF1A, CDKN2A and RUNX3, occur early and are stable in tumorigenesis, as confirmed in the plasma of HCC patients [22,38].

Abnormal regulation of gene expression by DNA hypermethylation & hypomethylation in HCC

Hypermethylated CpG loci in HCC are located close to transcription start sites, resulting in transcriptional inactivation of downstream TSGs [39]. Transcriptional repression associated with hypermethylation can facilitate clonal selection and regulate cell-cycle control, cell proliferation and apoptosis [22]. The hypermethylation of TSG promoters is a more predominant defect in HCV-related cases than in HBV [36,40]. Although promoter hypomethylation in HCC was also reported [41], the majority of hypomethylated CpGs in HCC were located in intergenic regions or gene bodies (approximately 80%), as well as repetitive DNA elements [36,39]. Some tissue-specific genes and imprinted genes have also been reported to undergo hypomethylation in HCC [42]. The biological significance of this hypomethylation in intergenic regions or gene bodies remains largely unknown. However, the genes associated with gene-body hypomethylation in cancerous cells often have functions in cell type specific differentiation [43].

Genetic & epigenetic modulations in core signaling pathways in HCC

Most cancers accumulate at least five genetic mutations [14] and numerous etiologies are associated with the development of HCC. As expected, HCC is heterogeneous and involves alterations in diverse signaling pathways. Viral infection (HBV and HCV) and chronic inflammation are the major risk factors associated with HCC development and contribute to the establishment of the tumor-promoting microenvironment. Although it is challenging to elucidate the chronological order of events leading to hepatocarcinogenesis, chronic inflammation, cell proliferation, resistance to apoptosis and cancer stem cell establishment are the major events of hepatocarcinogenesis. In this environment, both genetic and epigenetic alterations would accumulate in liver cells. Among the genes genetically mutated in HCC, TERT, TP53, CTNNB1 and AXIN are most frequently mutated [14]. Epigenetic modification at specific target genes is mostly observed as hypermethylation. For example, inactivation of the RB pathway through hypermethylation of the CDKN2A promoter was found in 73% of HCC tissues [44], and methylation of RASSF1A, GSTP1 and MGMT was also found in 85, 50–90 and 40% of HCC, respectively [45].

Genes involved in these processes can be categorized by distinct signaling pathways. However, alterations in the expression of certain genes do not always lead to the same biological consequence due to the branching of downstream signaling pathways, crosstalk between different pathways and tissue-specific selection or function in different signaling pathways. Since it is difficult to correspond signaling pathways to their actual role in cancer development, we summarized the signaling pathways altered in HCC according to their expected primary function in HCC development.

Establishing the tumor-promoting microenvironment

Chronic inflammation is one of the major risk factors that plays a critical role in initiating HCC [46]. Free radicals released by activated inflammatory cells can cause DNA damage, which lead to gene mutations or epigenetic modifications in tumor formation [47]. For example, ROS produced after HBV infection leads to hypermethylation of SOCS-3, resulting in its gene silencing [48]. Epigenetic silencing of SOCS-3 results in activation of the IL-6/STAT3 pathway, which promotes proliferation and growth of HCC cells. IL-6 is a pleiotropic cytokine which promotes cell proliferation and plays an essential role in hepatocarcinogenesis by activating STAT-3 signaling and high levels of circulating IL-6 has been reported in more than 40% of HCC [46]. Overexpression of IL-6 can occur as a consequence of deregulation of upstream controlling genes [49] as well as by genetic mutation [50]. However, epigenetic mutation of IL-6 has not been reported.

NF-κB, the master regulator of inflammation, is upregulated in HCC [51], and HBx, HCV core protein and LPS have been suggested to activate NF-κB in liver disease patients [52]. Blocking NF-κB activation significantly inhibits HCC progression, suggesting that NF-κB is necessary in inflammation-induced HCC [51]. Association of NF-κB genetic polymorphisms with immune persistence of HBV mutations leading to HCC has also been reported [53]. Many downstream genes regulated by NF-κB are highly methylated and silenced in HCC [54].

NF-κB can also be activated by increased accumulation of the upstream regulators, RIP1 and TRAF2, which is mediated by overexpression of AEG-1 in HCC [52]. AEG-1 is known to be an essential component involved in the onset and progression of various cancers. AEG-1 is also overexpressed in more than 90% of HCC patients, and a direct relationship between its level of expression and the pathological stages of HCC in patients has been reported [52]. Genomic amplification of the locus is one of the major genetic alterations associated with AEG-1 overexpression [55], while only a few tumor-specific SNPs in the promoter region of AEG-1 have been reported [56]. Studies show that AEG-1 is a scaffold protein that interacts with diverse protein complexes. Although creating an inflammatory microenvironment is one major important role of AEG-1 in promoting HCC onset and progression by activating the NF-κB/IL-6/STAT3 pathway, it can also activate the PI3K/Akt, Wnt/β-catenin and MEK/ERK pathways [52].

Proliferation stimulation pathway

IGF signaling can induce cell proliferation and inhibit cell apoptosis, which would transform normal cells into malignant cells. Alterations in the expression of key molecules in this pathway can be found in HCC. Expression of IGF2, IGF1R, FGF, IRS1 and IRS2 is found to be upregulated in HCC [57]. On the contrary, inhibition of the IGF pathway reduced the growth and invasion of HCC [58]. IGF2 acting through IGF1R might be the main mitogenic IGF signaling pathway in HCC. Overexpression of IGF2 can be achieved by increased transcription due to changes in imprinting in the promoter region and hypomethylation at exons 8–9 [59]. HBV-derived HBx, TP53 mutation and Aflatoxin B1 administration can also increase IGF2 expression. All SNPs found in IGF2, IGF2R and IRS2 were associated with higher risk of HCC [60]. Increased IGF2 could result from decreased expression of IGF2R to some extent as IGF2R functions to sequester IGF2. Alterations in imprinting, loss of heterozygosity and mutations in IGF2R associated with HCC have also been reported [59]. IGFBP is another component that can control the bioavailability of IGF ligands. Expression of antiproliferative IGFBPs, namely IGFBP1, 3 and 4, is downregulated in HCC, allowing IGF to become more available [59]. Last, hypermethylation of the IGFBP3 promoter has been reported [61].

FGF and HGF are also potent mitogens for HCC. FGF1, 2, 3, 4, 8, 19, 21, 23 and FGFR2, 3 and 4, are upregulated in HCC and can stimulate proliferation in most HCC cell lines. In particular, FGF19 signaling through FGFR4 is known to be important in liver carcinogenesis. Amplification of the FGF19 locus through chromosomal translocation results in chimeric receptor formation with constitutive activity of its signaling. Mutations, such as FGFR3 G380R which prevent degradation of FGFR3, have been found to prolong FGF signaling [62]. HGF acts as a potent mitogen in primary hepatocytes but HGF expression is decreased in HCC. Although, its receptor, c-MET, is overexpressed in HCC [63]. Thus, kinase activity of c-MET seems to be more necessary for tumorigenesis than HGF activation. The gene amplification of c-MET has been reported in HCC [55]. Activation of insulin, IGF, FGF and HGF receptors can trigger the PI3K/AKT/mTOR and RAS/MAPK/ERK pathways, thus promoting mitogenesis and antagonizing apoptosis during tumor development.

Evading guardian system: loss of growth suppressor & cell death function

Apoptosis is one way to eliminate cells with DNA damage, and defects in this process would lead to the accumulation of genetic mutations and tumor progression. TP53, which is known to function in growth suppression and apoptosis, is the most frequently mutated gene in human cancer. Disruption of the p53/ARF pathway has also been reported in 50% of HCC cases [64]. Missense or nonsense point mutations in the conserved region of the p53 DNA-binding domain results in a lower affinity for binding p53 target genes. Despite the presence of wild-type copies of TP53, most of the mutations act in a dominant-negative manner and – as a result, tumor suppressive function is lost. However, some mutations, such as TP53 R249S, result in a gain of function. In this case, mutated p53 acts independent of wild-type p53 and contributes to genome instability and tumor invasion [65,66]. Cumulative evidence shows that different types of mutations in TP53 can result in distinct profiles that lead the cancer progression differently [65]. Although some methylation in the TP53 promoter has been reported, significant association with HCC could not be found [67].

Alterations in the regulators of TP53 also contribute to deregulation of the p53 signaling pathway. MDM2 interacts with p53, which allows it to undergo proteasome-mediated degradation. Copy number gain and MDM2 overexpression is frequently observed in HCC [68]. Inactivation of IRF2, a partner of MDM2, has also been reported, and deletion, splice site or missense mutations of IRF2 underlie the same phenotype [24]. Binding of HBx protein to p53 to alter the selectivity of its binding site could be another mechanism that deregulates p53 function [69]. HEYL is frequently hypermethylated in its promoter region and is inactivated in more than 75% of HCC. Because Serine¹⁵ phosphorylation which is important for p53-dependent apoptosis and p53 accumulation is regulated by HEYL, silencing HEYL expression leads to loss of p53 apoptosis function [70]. The CDKN2A locus encodes p16INK4a and p14ARF, both of which are involved in two tumor suppressor pathways, namely TP53/ARF and RB/p16INK4a. This gene is second most commonly inactivated gene in cancer after TP53. Its mutation, recurrent homozygous deletions and hypermethylation in the promoter region are all reported in HCC [11,71].

Promoting angiogenesis

Angiogenesis is a common event that is required for the development, maintenance and metastasis of solid tumors. Hypervascularization is a major characteristic of HCC [72]. As in other cancers, hypoxia is a central stimulus of angiogenesis and VEGF is frequently upregulated in HCC. Its high expression level has some correlation with inferior patient survival. Copy number multiplication along with SNP of VEGF and the VEGF receptor have been reported to be associated with HCC [11,73].

While VEGF is the major driver of tumor angiogenesis, crosstalk between VEGF and FGF, and upregulation of other pro-angiogenic factors, such as PDGF, HGF, FGF2 (bFGF) and ANG2, have also been reported in HCC [72]. Among the FGF family, FGF2 seems to be the major one associated with HCC angiogenesis. However, the association of genetic or epigenetic mutations within PDGF, HGF, FGF2 and ANG2 with HCC has not been extensively studied.

Establishing the stem cell self-renewal environment

Nuclear accumulation of β-catenin has been reported to promote cell proliferation [74]. However, recent studies suggested that the LSD1/Prickle1/APC/β-catenin signaling axis is a novel molecular circuit that regulates stemness and chemoresistance in HCC [75]. Elevated accumulation of β-catenin has shown a correlation with tumor progression and poor prognosis in HCC [74]. Alteration in the activity of this pathway, both in vivo and in vitro, could lead to chemoresistance and loss of stem cell-like properties [75]. CTNNB1 is one of the genes most frequently mutated somatically in HCC [14]. Most of the mutations are reportedly within the region that is responsible for phosphorylation and destruction of β-catenin [5]. In addition, upregulation can be achieved by altered activity of degradation component APC, AXIN1 and AXIN2 and Wnt/Frizzled (FZD) receptor elements. Decreased expression caused by somatic mutations and aberrant epigenetic modifications in the promoter region of AXIN1 or AXIN2 was also found in HCC [76]. Although AXINs are components of the Wnt/β-catenin pathway, mutations in CTNNB1 and AXINs are mutually exclusive in HCC [11]. Hypermethylation was found in the promoter CpG islands of APC extensively.

TERT expression is primarily repressed in somatic cells, except for self-renewing cells; however, it is reactivated in 70–90% of cancer cells [77]. Single nucleotide mutation in the TERT promoter region, which leads to the creation of a potential binding site for ETS/TCF transcription factors, increases promoter activity in 59% of HCC [78]. Hypermethylation of the TERT promoter region was also reported, but unusually, hypermethylation was correlated with high TERT expression [79]. Frequent HBV integration in the TERT locus has also been reported [13].

While significant association of TERT- and CTNNB1-activating mutations has been reported in HCC [78], p53 and CTNNB1 occurrence is mutually exclusive [11]. Although TP53 is mainly known as an anti-oncogenic gene, this protein can also function in a pro-oncogenic role. Recent studies demonstrated that TP53 also plays important roles in stem cell maintenance [80]. Interestingly, TERT, TP53 and CTNNB1, all three major genes that are mutated in most HCC seem to function in stem cell maintenance [80].

We summarized the combined effect of genetically and epigenetically altered genes of HCC on key signaling pathways (Figure 1). Several key components of each signaling pathway were affected either by genetic mutations or epigenetic alterations (DNA methylation), indicating their collaborative effect on the alteration of the signaling events. Besides, there are preferential targets of genetic and epigenetic alterations in HCC, suggesting a combined analysis of HCC for both genetic and epigenetic alternations is necessary to understand the defects deriving HCC development.

Conclusion

The significant changes in genetic sequence and epigenetic modification cooperate to transform the cell by modulating key signaling pathways and chromatin stability. Until now, only approximately 100 somatic mutations have been identified as cancer driver mutations, with five to six driver mutations believed to actually cause cancer in each patient. However, it is not easy to detect cancer-driving mutations among the high numbers of passively accumulating mutations in cancers. However, most cancers usually carry distinct epigenetic abnormalities that appear to crosstalk with the somatogenetic mutations of cancer cells. For example, genomic regions with repressive histone modification patterns show a higher density for somatic mutations and CNVs show patterns that correlate with the distribution of 5-methylcytosines. On the other hand, diverse epigenetic modification patterns are affected by mutations not only in chromatin regulators, but also in diverse signaling molecules. Therefore, mutual cooperation between genetic and epigenetic alterations may facilitate cancer development in HCC.

Future perspective

Through massive analysis of the genome and epigenome of HCC, we are elucidating the major mechanisms leading to tumorigenesis in liver cells. Diverse genetic modifications in HCC can modulate the core signaling pathways, transcription factors and epigenetic modifiers which promote cellular transformation. The effects and patterns of genetic alterations depend on etiological factors, race and gender. However, the epigenetic patterns associated with cancers and their effect on cancer development are not well studied. Several challenges must be solved in order to understand the roles of epigenetic alterations in tumorigenesis. First, a large portion of the epigenetic regulatory mechanism remains poorly understood. Thus far, studies have been focused on the effect of promoter hypermethylation or repeat hypomethylation. The roles of nonpromoter-associated modifications, noncoding RNA and chromatin architecture in the regulation of transcription remain to be studied. Second, in the case of DNA methylation, there are no standard ‘normal’ methylation values for each CpG site. The DNA methylation rate on each CpG is highly variable for each individual cell. Determining this rate would require the development of new techniques that enable the detection and analysis of DNA methylation at a single-cell level.

Table 1. Molecular characteristics of the major hepatocellular carcinoma types.

Characteristics	HBV-HCC	HCV-HCC	Alcohol-HCC
Geographical distribution	East Asia, Africa	Japan, USA, Europe	USA, Europe
Risk factors	Chronic inflammation, viral integration, HBx oncoprotein	Chronic inflammation	Actaldehyde, free radicals
Frequent mutations	TERT inactivation by HBV integration	TERT promoter mutations	ARID1A mutation
	Inactivation of TDG by HBx causing accumulation of C to T transition		Severe DNA hypomethylation due to reduced hepatic methionine adenosytransferase
Hypermethylated genes	NAT2, CSPG2, DCC, NTKR3, TNFSF10, TNFRSF10C and RASGRF1	RIK and CHGA	RASSF1, GSTP1, APC, MGMT, CDKN2A and CHRNA3

HBV: Hepatitis B virus; HBx: HBV X protein; HCC: Hepatocellular carcinoma; HCV: Hepatitis C virus; TDG: Thymine DNA glycosylase.

Table 2. Genetic mutations in epigenetic modifiers.

Category	Gene	Genetic alteration		Expression	Ref.
Chromatin remodeler	ARID1A (SWI/SNF complex)	Substitution	Missense (Val2084Asp)	Inactivated	[24]
			Nonsense (Tyr1281X)
		Indel	Insertion (Met390IlefsX10)
			Deletion (Gly1274GlufsX7)
	ARID2 (SWI/SNF complex)	Substitution	Missense (Arg601Gln)	Inactivated	[8,24]
			Nonsense (Glu393X, Glu712X, Gln983X)
		Indel	Insertion (Leu1330ValfsX3)
		Splice-site	g.chr12: 44410093G>A
	SMARCA2	Substitution	Missense (Ile779Thr, Asn1134Thr, His1151Arg, Gln1155Leu)	Inactivated	[11,24]
		Indel	Deletion (Arg183EfsX50)
	SMARCA4	Substitution	Missense (Asp1177Glu)	Unknown	[24]
Histone methylation	MLL (H3K4me3)	Substitution	Missense (Gly73Glu, Pro179Arg, Val1237Met)	Inactivated	[13]
	MLL2 (H3K4me3)	Substitution	Missense (Ser1744Ile, Arg3582Leu, Cys5133Trp)	Inactivated	[11]
		Indel	Deletion (E905DfsX25, E2643KfsX11)
		Splice-site	R4692_splice
	MLL3 (H3K4me3)	Substitution	Missense (Ser939A)	Inactivated	[11]
		Indel	p.L3212FfsX37
	MLL4 (H3K4me3)	Substitution	Missense (Tyr1540Cys, Tyr2515Cys)	Inactivated	[11]
			Nonsense (E183X, G221X)
		Indel	Insertion (Asp2635fs)
	MLL5 (H3K4me3)	Substitution	Missense (Thr11Ile, Pro1547Leu)	Inactivated	[11,13,18]
	SMYD3 (H3K4me3)	Substitution	Missense (Glu260Val)	Unknown	[11]
	SETD2 (H3K36me3)	Substitution	Missense (Val2345Met)	Unknown	[11]
			Nonsense (Arg1592X, Gln1996X)	Damaging
		Indel	Insertion (g.47058719_47058720insAGGA)	Damaging
			Deletion (g.47098587_47098593del, g.47098717del)	Damaging
Histone acetylation	BRD7 (Bromo domain)	Substitution	Nonsense (Glu36X, Tyr457X)	Damaging	[18]
	BPTF (Bromo domain)	Substitution	Missense (Arg215Gly, Ala796Val)	Damaging	[11]
		Indel	Deletion (g.65905755_65905789del)
DNA methylation	DNMT1	Unknown	Unknown	Upregulation
	DNMT3			Upregulation
	DNMT3L			Downregulation
	TET1			Downregulation
	MAT1A			Downregulation

Executive summary

Genetic alterations influenced by epigenetic modifications in hepatocellular carcinoma

• Nucleotide mutations are affected by deamination of 5-methylcytosines mediated by the APOBEC family, MBD4 and TDG, which are abnormally regulated in hepatocellular carcinoma (HCC).

• Somatic substitution density correlated with repressive histone modification patterns.

• The loss of 6q, 8p,13q and 17p is associated with DNA hypomethylation.

Epigenetic alterations influenced by genetic alterations in HCC

• Liver-specific DMRs are specifically enriched with SNPs linked to high cancer risk.

• Inactivating mutations of epigenetic modifiers are frequently observed in HCC.

Key signaling pathways modulated by genetic & epigenetic alterations in HCC

• IL-6/STAT-3 can be activated in HCC, thereby promoting the microenvironment established by genetic alterations of AEG-1/NF-κB and epigenetic modification of SOCS-3.

• IGF2 and IGF1R are the main mitogenic IGF signaling components in HCC, and expression of IGF2 is altered by modification in imprinting and methylation.

• The three major genes, namely TERT, TP53 and CTNNB1, most frequently mutated in HCC can function in tumor stem cell maintenance.

Financial & competing interest disclosure

This research was supported by the Global Research Laboratory Program of the National Research Foundation (NRF) funded by the Ministry of Science, ICT and Future Planning (grant number NRF-2007-00013) and Collaborative Genome Program for Fostering New Post-Genome industry through the National Research Foundation of Korea (NRF) funded by the Ministry of Science ICT and Future Planning (grant number 2015M3C9A4053251) to Y-J Kim and by the Mid-Career Researcher Program (NRF-2012R1A2A2A01010176) funded by the Ministry of Science, ICT and Future Planning to J Kim-Ha. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.

Additional information

Funding

This research was supported by the Global Research Laboratory Program of the National Research Foundation (NRF) funded by the Ministry of Science, ICT and Future Planning (grant number NRF-2007-00013) and Collaborative Genome Program for Fostering New Post-Genome industry through the National Research Foundation of Korea (NRF) funded by the Ministry of Science ICT and Future Planning (grant number 2015M3C9A4053251) to Y-J Kim and by the Mid-Career Researcher Program (NRF-2012R1A2A2A01010176) funded by the Ministry of Science, ICT and Future Planning to J Kim-Ha. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. No writing assistance was utilized in the production of this manuscript.