Background
Hepatocellular carcinoma (HCC), the most frequent primary liver tumor, is a malignancy affecting around 7 10
5 patients each year, with highest incidences measured in Sub-Saharan Africa and Eastern Asia. HCC is now the fifth most common malignant tumor and the third common cause of cancer-related mortality worldwide [
1]. The Western North-Africa (WNA: Morocco, Algeria and Tunisia), is considered as an area of intermediate endemicities for chronic viral hepatitis B and C, and displays much lower incidences of HCC than both European or African neighboring countries [
2,
3]. It is generally admitted that liver tumorigenesis is a consequence of accumulation of genetic and epigenetic alterations in key genes controlling proapoptotic or prosurvival signals. These changes occur generally in an impaired hepatic tissue undergoing a persistent viral infection and/or a cirrhosis, but it could be promoted also by an exposure to mutagens such as aflatoxin B1 [
4]. Several classifications of HCC were established to differentiate tumors by focusing on histological characteristics, gene mutations in
TP53 and
Wnt pathways and hypermethylation of tumor suppressor genes (TSG) [
5,
6]. In addition, HCC can be genomically characterized from the most instable tumors with frequent
TP53 and
AXIN1 mutations to stable tumors with
β-catenin alterations [
7]. Moreover, transcriptomics revealed an overexpression of imprinted and mitotic cell cycle genes within instable tumors at odds with stable ones that show increased levels of metabolism-controlling genes coupled to an underexpression of stress, immune response and cell adhesion coding genes [
8,
9]. Epigenetic alterations, and particularly DNA methylation, are also suspected to represent a class of decisive events in liver tumorigenesis. Indeed, diverse studies have been carried out on different cohorts of HCC patients affording insight about epigenetic changes controlling liver carcinogenesis [
10]. Hypermethylation of a set of TSG such as
RASSF1, RIZ,
CDK2NA and
GSTP1 is commonly reported in HCC [
11]. Such targeted silencing is generally accompanied by a global hypomethylation affecting repetitive elements covering the genome [
12,
13]. The connections between genetic and epigenetic features mentioned above are still poorly understood. Finally, numerous studies have been performed to explore the association of the genetic background of patients with an eventual individual susceptibility to HCC [
14,
15].
TP53 presents, at codon 72, a functional single nucleotide polymorphism (SNP, R72P, rs1042522) that modulates the susceptibility to several cancers including HCC [
16]. Notably, the presence of an arginine at codon 72 is known to be associated with higher rates of somatic
TP53 mutations in tumors [
17].
Despite this apparent wealth of data, carcinogenesis in specific populations such as WNA inhabitants is still poorly understood. The WNA patients are, actually, characterized by the scarcity of alterations found in HCC [
2]. This situation suggests that epigenetic changes may be the most significant changes in WNA patients. The aim of the study was, thus, to provide an appraisal of the epigenetic changes occurring in HCC from a WNA population. Methylation status at 10 individual loci as well as at repetitive elements (LINE-1) was assessed. Variations in DNA methylation levels were further confronted with genetic data including
β-catenin and
TP53 mutations, chromosomal instability (CIN) and genotypes of selected SNPs. We found that, in HCC from WNA, somatic changes including methylation, mutations, or CIN were primarily conditioned by the genotype at codon 72 of
TP53. The current report represents the first description of the existing correlation between
TP53 R72P and epigenetic changes in tumors.
Discussion
HCC epidemiology is known to follow ethno-geographic variations of incidence. Intriguingly, despite intermediate endemicities for chronic viral hepatitis B and C, low incidences of HCC are registered in the WNA populations (1.2-2.3 cases/10
5 habitants for males, age standardized ratio,
http://globocan.iarc.fr/) [
39,
40]. Remarkably, in WNA, HCC is characterized by a relative paucity of mutations at common genetic targets (
TP53,
CTNNB1,
AXIN1), making of liver tumorigenesis in WNA a rather mysterious process (2). Nevertheless, and despite an apparently mild tumor process, HCC prognosis in WNA is as bad as anywhere else [
41,
42].
To date, no survey characterizing epigenetic features of HCC in WNA patients has been published. Our results were in accordance with those published by Hernandez-Vargas
et al. underlining the importance of aberrant methylation that differentially affects
RASSF1,
RIZ1 and
SOCS1 in tumors and non-tumorous tissues [
43]. A rather high level of methylation was observed in four NT liver DNA (≥4 loci methyl(+), see Figure
2A). We could not find any annotations common to the four patients. However, we observed that their non-tumor liver tissues tended to be more frequently non cirrhotic (3/4 vs 5/41, p = 0 .014) and the FAL in tumors slightly higher (38 vs 28%, ns) than in other samples of the series. These observations might indicate that a stronger carcinogenic process conferring a preneoplastic status to NT liver is at work in these patients. Alternatively, it might be the consequence of a field cancerization process in absence of full-fledged cirrhosis [
44-
46]. Besides, Nishida
et al. have shown that DNA methylation levels are correlated with chromosomal instability,
TP53 and
β-catenin mutations [
47]. Our data broadly corroborated this model, though without integrating
β-catenin, as this alteration is very infrequent in WNA HCC. We hereby presented data refining and strengthening this model by showing a link between polymorphism at codon 72 of
TP53, somatic mutations, chromosomal instability and DNA methylation status in HCC from WNA patients. The preferential association of codon 72 Arg with somatic mutation of
TP53 has been thoroughly reviewed by Soussi and Wiman [
17]. Likewise, the well-known association of chromosomal instability and
TP53 mutation will preferentially occur in an Arg/Arg context [
48]. Finally, it has been shown that DNA methylation is significantly reduced in Li-Fraumeni cell lines [
49]. However, to our best knowledge, a correlation between aberrant methylation and germline background of patients for
TP53 was never reported.
The
TP53 R72P polymorphism is known to be functionally relevant. The Arg variant was shown to be associated with a better apoptotic activity compared to Pro allele [
50]. Furthermore, predisposition studies showed that the presence of the Pro allele is commonly associated with a higher risk of cancer including HCC [
30,
51] and of defects in embryonic implantation [
52]. Our data indicate that codon 72 polymorphism conditions, apparently, genomic and epigenetic alterations encountered in the tumor tissue. In the current series of HCC from WNA patients, the Pro variant is associated with paucimutated and mildly methylated tumors. On the contrary, the presence of Arg allele requires the presence of additional somatic mutations and frequent DNA hypermethylation. It is well known that aging process is accompanied by an increase of genomic methylation that progressively decreases the expression of multiple genes [
53]. Moreover, as shown in mouse models, an excess of p53 activity is characterized by an aging-like syndrome [
54]. Finally, it has been shown in humans that
TP53 codon72 Pro carriers tend to live longer than Arg carriers [
55]. A link between rs1042522 and constitutive DNA methylation was, however, never evoked so far. Our results indicated consistently that
TP53 codon72 Arg/Pro polymorphism may, in the North-African ethno-environmental settings, influence somatic evolution through the modulation of DNA methylation levels. The fact that we were able to reproduce this situation
in vitro reinforces this hypothesis. We did not detect baseline differences of DNA methylation between cell lines but an increased DNA methylation was often detectable in Arg/Arg cells after decitabine-induced demethylation and subsequent doxorubicin-induced DNA re-methylation. Interestingly, three of the genes showing differential methylation between Arg and Pro variants (
GSTP1,
RASSF1, and
SOCS1) are known to be p53 targets [
56-
58]. This situation suggests that p53 may, depending on circumstances activates or methylate its target genes. The system we developed
in vitro is relatively removed from physiological conditions or from a slow and long lasting process as tumorigenesis but it suggests that p53 controls DNA-methylation plasticity in condition of stress. Genome wide DNA methylation studies on normal and pathological conditions are now warranted to confirm our data. Another intriguing feature of DNA methylation in Arg/Arg and Arg/Pro carriers was its bimodal distribution. We were not able to correlate this feature with any of the clinic-biological annotations at our disposal. It is, thus, tempting to hypothesize that p53 impact on DNA methylation could be modulated by another genetic trait. It is, indeed, well known that DNA methylation levels in humans are strongly influenced by polymorphisms affecting genes controlling one-carbon metabolism [
59,
60]. Moreover other polymorphic effectors of DNA methylation, to be found in the oxidative stress pathway or among partners acting directly in DNA methylation (
eg DNMTs), can be suspected to influence methylation levels [
61-
63].
Being aware that we do not provide a mechanistic explanation to our observations, some conspicuous differential expression affecting genes involved in DNA methylation metabolism, particularly those of
DNMT3A (methylating) or
APOBEC3B (demethylating), however, are in keeping with the current model of DNA-methylation metabolism [
64]. It is well known that p53 entertains tight connections with the different cellular DNA-methyltransferases. Indeed, p53 and Dnmt3A are direct interactors as shown by co-immunoprecipitation experiments [
65,
66]. In addition, and in absence of mutagenic stress, p53 is known to directly inhibit
DNMT1 expression by trapping Sp1 and repressive chromatin modifiers on the promoter of the gene [
67,
68]. The links between p53 and Dnmt1 are, however, far more complex. Both proteins have been shown to physically interact on various p53-responsive promoters (
eg survivin/
BIRC5) leading to their inactivation through DNA methylation as well as other chromatin modifications [
38,
69]. These data and the well-described differential activities of p53 Arg/Pro isoforms in other epigenetic-sensitive phenomena such as cancer or ageing, make plausible a differential activity of rs1042522 on DNA methylation levels.
Finally, we showed a mild but significant increased ability to recover from decitabine treatment in Pro-carrying cells than in Arg carriers. This is in line with our hypothesis according to which tumorigenesis is less dependent on DNA methylation in Pro than in Arg carriers. In addition, and despite hitherto disappointing results in solid tumors, our data suggested that decitabine use in a codon 72 Arg/Arg wild-type TP53 context might significantly improve chemotherapeutic treatment [
70,
71]. The hypothesis obviously needs further confirmation but may be considered as a potential novel application in personalized treatment of cancer.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AM and PP designed research. KR and AM performed the molecular genetics study. PP performed cell culture experiments. KR, AM, SE, RA, OB, HT, MK, AE, AD, SB and PP analyzed the data. KR and PP wrote the manuscript. All authors read and approved the final manuscript.