Epigenomics in Malignant Pleural Mesothelioma

1. Introduction

Human malignant pleural mesothelioma (MPM) is an invariably fatal tumor due to its heterogeneity, growing from the serous surfaces of the pleura. Many factors are involved in its occurrence, such as exposure to asbestos fibers and simian virus 40; these factors being those that are strongly associated with the tumorigenesis of this disease. The annual incidence of MPM is relatively low, estimated in a range of 0.6–30/10,00,000, but the global occurrence is expected to increase continuously in future years [1]. MPM is extremely heterogeneous in its morphology and molecular phenotypes. The latency period for MPM development is 10–50 years after asbestos exposure. The prognosis for MPM is generally poor, with a median survival time of 12 months from diagnosis [1].

Intratumor heterogeneity refers to a mixture of phenotypic, functional, and genetic differences within cancer cells with various differentiation or hierarchical statuses within the tumor. It is a common feature in most tumors. This heterogeneity has been considered the greatest obstacle to the effectiveness of most cancer therapies, manifesting itself in its sensitivity to different therapies. Several studies have been focused on genetic alterations as part of the mechanism of tumoral cells for the generation and maintenance of this heterogeneity. In addition, some other studies show the role of epigenetic modifications involved in its heterogeneity. Despite this, there is scarce information about epigenetic modifications that could explain this process [1, 2].

Epigenetic modifications are heritable and stable alterations of genes that do not change the DNA sequence, including DNA methylation, histone modification, and non-coding RNA interference modifications. DNA methylation has been extensively studied in the development of cancer. On the one hand, hypermethylation in cancer-related promoter genes induces the silencing or downregulation of tumor suppressor genes and repair genes. On the other hand, hypomethylation of DNA leads to activation of oncogenes and genomic instability. Several authors suggest that aberration in DNA methylation may play an important role on tumor cells heterogeneity [3, 4, 5].

The exact mechanisms by which asbestos fibers promote the development of cancer are unknown, however, the most accepted theory is the induction of chronic inflammation and signaling pathways in the transformation of reactive oxygen species generated by asbestos fibers. Therefore, this chapter will address an overview of the epigenetic profile of MPM and the mechanisms that promote epigenetic modifications where asbestos fibers might play an important role.

2. Asbestos-induced molecular alterations

As previously mentioned, asbestos exposure is a primary cause of the development of pleural mesothelioma. Molecular analyses show that asbestos-related carcinogenesis is caused by chronic inflammation that both promote the release of oxygen free radicals that alter intracellular components, and DNA mutation and its consequent transformation. Asbestos fibers also contain iron ions and can induce hemolysis by sequestering iron from hemoglobin. This is particularly important since free iron disproportionately releases H₂O₂, which consequently releases hydroxyl radicals (OH) that oxidize DNA, and release nucleic acids, proteins, and lipids. This process is exacerbated by the release of cytokines including tumor necrosis factor-alpha from macrophages and high mobility of box group 1 (HMGB1) proteins from necrotic cells, leading to an amplification of the inflammatory response and an increase in cells that are driven to oxidative damage. Damaged oxidized DNA, if not properly repaired, is highly mutagenic and can lead to genomic instability. A multitude of oxidative DNA lesions includes oxidation of DNA bases, baseless sites, single-strand breaking, double-strand breaking, and interchain breaking, all of which require different pathways for proper repair. These last two chain-breaking types are particularly toxic, since they cause replication collapse, as well as allow chromosome rearrangements, chromosome gains, losses, or fragmentation [2].

There are other mechanisms involved in how asbestos fibers cause MPM (Figure 1). Four proposed models related to asbestos fibers induce genetic and cellular damage to cells, in addition to the previously mentioned chronic inflammation. The different molecular models involved in asbestos exposure are explained below:

1. Reactive oxygen species generated by asbestos fibers with their surface exposed leads to DNA damage and cell membrane rupture. Macrophages that engulf asbestos fibers but cannot digest them also produce abundant reactive oxygen species.

2. Asbestos fibers are also engulfed by mesothelial cells. Asbestos fibers collected in cells can physically interfere with the mitotic process of the cell. The cycle is cut by the interruption of the mitotic spindles. Another important aspect is the entanglement of asbestos fibers with the chromosomes or mitotic spindles that can give rise to structurally damaged chromosomes such as aneuploidies of normal mesothelial cells.

3. Asbestos fibers can absorb a variety of proteins and chemicals on the wide surface of asbestos, which can result in the accumulation of dangerous molecules including carcinogens. The asbestos fibers also bind to important cellular proteins and a deficiency of these proteins can also be detrimental to normal mesothelial cells.

4. Finally, mesothelial cells and macrophages exposed to asbestos release a variety of cytokines and growth factors that induce inflammation and tumor promotion. These include tumor necrosis factor α, interleukin 1β, transforming growth factor β, and platelet-derived growth factor. Tumor necrosis factor-α has been shown to activate nuclear factor-κB, leading to mesothelial cell survival and inhibiting asbestos-induced cytotoxicity. The high mobility group protein box 1 (GAMB1) is released from mesothelial cells, which are exposed to asbestos and then undergo necrotic cell death, promoting an inflammatory response. Thus, aberrantly activated signaling between mesothelial cells, inflammatory cells, fibroblasts, and other stromal cells can create a set of mesothelial cells, which harbor aneuploidy and DNA damage, potentially developing cancer cells and together all these phenomena form a tumoral microenvironment that supports and nurtures them [6, 7, 8].

3. Methylation-mediated suppressor gene silencing

There are several studies that have shown a relationship between silencing suppressor gene by methylation process and the development of MPL. For example, Christensen et al. [23] examined the DNA methylation status of the promoter of six genes that regulate cell cycle progression at 70 MPM. The extent of methylation of these genes was correlated with lung asbestos burden and overall survival. Goto et al. [60] studied methylation process in more than 6000 GpC islands, comparing twenty MPM versus twenty lung adenocarcinomas, using microarray PCR technique. Their results are interesting because they found out that 387 genes (6.3%) were hypermethylated in mesotheliomas, while the number of hypermethylated gene in lung adenocarcinoma were higher with a total amount of 544 genes (8.8%).

MPL patients’ survival is related with DNA methylation levels. In this way, higher levels of DNA methylation correlate with lower patient survival. Three genes; TMEM30B, KAZALD1 and MAPK13, are specifically hypermethylated in MPM. Several reports have documented tumor suppressor gene silencing related to DNA methylation process in MPM (Table 1) and there is evidence of hypermethylation of some of these genes affecting overall survival.

Currently, scientific evidence has shown that recurrent hypermethylation is highly related to tumor suppressor genes in MPM, however, the mechanisms behind this process have been poorly studied. Novel studies have identified TC2N gene as a tumorigenesis promoter by silencing p53. Cytokine signaling participate in modulation process of DNMT expression and mediate hypermethylation of target genes enrolled in some types of cancer such as colorectal carcinoma and erythroleukemia cells [61, 62]. Exposure to asbestos fibers leads to a cytokine cascade induced by high mobility group 1 (HMGB1) or the NLRP3 inflammasome. These cytokines use to change the regulation of the expression of DNMT and other components of the methylation machinery during the process of MPM evolution.

A study of a panel of genes encoding epigenetic regulators in a panel of cultured cell lines derived from asbestos-associated MPM relative to LP-9 (a commercially available normal mesothelial cell line) was recently carried out. Consistent with the study results, TCGA data demonstrate a spectrum of DNMT expression in MPM and suggest that overexpression of DNMT1, DNMT3A, and DNMT3B correlates with decreased survival of pleural mesothelioma patients (Figure 2).

Kim et al. [1] studied gene expression and methylation profiles in pluripotent populations (SP) and non-SP fractions in human MPM samples, using RNA-seq and methylated DNA immunoprecipitation techniques. They found 6400 hypermethylated genes and 3400 hypomethylated genes in SP. Seven hundred and ninety-five genes were upregulated, while three hundred and thirty-five were significantly repressed in SP compared to non-SP fractions. They looked at changes in DNA methylation and expression levels of 122 genes; 118 genes were hypermethylated and downregulated, while 4 were hypomethylated and upregulated. Ten other genes showed hypermethylation and low expression of CpG promoter islands.

4. Loss of imprinting (LI) and de-repression of cancer-germline (CG) genes

The loss of the imprinting process is largely due to DNA hypomethylation. Repression of endogenous retroviral sequences and activation of GC genes can promote malignant transformation by increasing proliferation, genomic instability, and resistance to apoptosis [63, 64].

A fascinating phenomenon can occur during the malignant transformation of somatic cells. The development of highly limited tumor antigens that induce serological and cell-mediated immune responses in cancer patients can be caused by abnormal activation of GC genes [also known as testicular cancer (TC) genes]. As a result, testicular cancer antigens (ATCs) have become popular targets for cancer immunotherapy in recent years [65]. More than 270 GC genes have been registered in the international TC database thus far. Seventy-five percent of these genes are only expressed in normal testes and malignant neoplasms, while the rest have high levels of expression in testes and varying levels of expression in other normal tissues and malignancies. On the X chromosome, around half of the GC genes are encoded. Families of cancer-testis-X (CT-X) chromosomal genes with inverted DNA repeats are common. On the other hand, inverted repetitive DNA sequences or extended families or are not linked to non-X CT genes [66, 67]. Furthermore, CT-X genes are frequently active in malignancies, and genes from families are increased in a tumor-specific manner, implying that the CT-X genes have a transcriptional coregulation and functional link.

In human malignancies, the stage at which the disease is discovered at a specific time corresponds to the degree of CG gene repression. Malignant and aggressive phenotype of cancer cells is promoted activations of this genes. BORIS/CTCFL, for example, upregulates h-TERT and suppresses apoptosis in cancer cells via processes that are still unknown [68, 69]. MAGE-A11 regulates the activity of the tumor suppressor gene RBL1/p107 [63]. MAGE-A11 inhibits the tumor suppressor gene RBL1/p107, while MAGE-B2 promotes cell cycle advancement by increasing E2F activity. MAGE-A2 and MAGE-C2 prevent p53 from binding to target promoters, changing its activities and leading to p53 deacetylation (inactivation) or enhanced ubiquitin-mediated degradation. The absence of CG gene regulation does not appear to be just a symptom of pluripotency, as it is accompanied with chromosomal hypomethylation. In human ESC, mesenchymal stem cells, and adipose-derived stem cells, Loriot et al. [70] found no overexpression of 18 different CG genes. In induced pluripotent stem cells (iPSCs) produced from normal small airway epithelial cells, transcriptional repression of CG genes such as NY-ESO-1, MAGE-A1, and MAGE-A3, which are generally located upward in thoracic malignant tumors, has been detected, which is consistent with these findings. Although these findings imply that iPSC reprogramming is partial, induction of CG genes in cancer cells may necessitate more extensive DNA hypomethylation as well as activation of tissue-specific transcription factors.

There is currently such little information on the expression of CG genes in MPM. MAGE1--4, NY-ESO-1, GAGE1-2, GAGE1-6, SSX2, SSX1-6, and RAGE-1 expression in five MPM lines was compared to normal mesothelial cells employing RT-PCR techniques, according to Sigalotti et al. [71]. In these MPM lines, diverse expressions of the CG gene were identified, with each line exhibiting a unique profile, as previously reported for lung malignancies [72]. None of these genes were found in normal mesothelial cells [71, 73].

5. Polycom complex-mediated epigenetic silencing

Polycomb group proteins (PcG) play an important role as regulators of stem cell pluripotency and differentiation [74], as well as inappropriate gene expression during cancer transformation [75, 76]. In mammals, two main Polycomb repressor complexes (PRCs) have been discovered. PRC-2 is an initiating complex that causes trimethylation of histone 3 lysine and contains the subunits EZH1/EZH2, SUZ12, EED, and RBAP46/48 (H3K27Me3).

PCAF, PHC, RING1, CBX, and BMI1 are components of the housekeeping complex PRC-1, which mediates the ubiquitination of H2AK119 (H2AK119Ub). CRC recruitment and heterochromatin growth are aided by these histone marks, which are frequently detected in the context of DNA hypermethylation and gene suppression [75, 76]. Several proteins, such as including JARID2 and members of the sex comb-like family (ASXL), interact with EZH2 and SUZ12 to lead PRC-2 to polykyl response elements (PRE) throughout the genome [77, 78]. Goto et al. [79] investigated gene repression in MPM and observed that a subset of genes repressed in MPM had H3K27Me3 without DNA hypermethylation, implying that disruptions in polycomb gene expression may play a role in MPM etiology [80, 81].

Several immunoblotting investigations studies revealed that MPM cells overexpress EZH2 with associated increases in H3K27Me3 levels when compared to normal mesothelial cells. Another set of tests, which included QRT-PCR, immunoblotting, and IHC, revealed that EZH2 was overexpressed in almost 80% of primary MPMs (most of which were epithelioid histology). As a result of these findings, it was identified that EZH2 is overexpressed in MPM and that PRC-2 could be considered as a potential therapeutic target in these cancers. The overexpression of EZH2 in MPM was verified by TCGA analysis, as was a strong link between EZH2 upregulation and lower MPM patient survival (Figure 2A). Further TCGA analysis reveals that SUZ12 overexpression is associated with poor survival in MPM patients (Figure 2B). On the other hand, there is no evidence about MPM patients’ survival related to EED expression (Figure 2C).

The foregoing findings are especially important in light of recent findings that inactivating mutations in BRCA-associated protein 1 (BAP1), which encodes a nuclear ubiquitin hydrolase with several functions, are found in uncommon familial MPMs as well as almost 60% of sporadic MPMs. H2AK119Ub is ubiquitinated, for example.

LaFave et al. [82] discovered that BAP1 mutations, which are linked to protein expression loss, enhanced the expression of EZH2 and SUZ12 in MPM cells in a series of experiments. Likewise, overexpression of EZH2 was related to lower levels of H4K2Me1 and less occupancy of L3MBTL2 (an unusual polycomb protein that identifies this repressive histone mark) inside the EZH2 promoter in BAP1 mutant cells. Despite the strong connection between BAP1 mutations and repression of Polycomb stem cell targets, no specific clinical manifestation of BAP1 mutant MPM has been identified. Somatic mutations in BAP1 appear to be more common in current or past smokers with MPM [83].

6. SWI/SNF

SWI/SNF are mammalian homologs of yeast trithorax complexes. Their major purpose is to antagonize PRC-2’s repressive effects by destroying DNA-nucleosome connections allowing movement and ejection, or by switching nucleosomes to increase factor accessibility transcription to DNA [84, 85]. In human malignancies, the genes encoding the SWI/SNF complexes are commonly altered, with various subunit mutations related to specific cancer histologies.

Yoshikawa et al. [86] used whole exome sequencing to identify a substantial number of mutations in genes involved in the SWI/SNF pathways, including homozygous SMARCA4, ARID2, and PBRM1 mutations in short-term established MPM lines [86, 87]. They also evaluated at the loss of somatic copies in the 3p21 region (which is roughly 10.7 Mb in size and contains 251 genes) in 33 MPM samples, using techniques including comparative genomic matrix high-density hybridization (a-CGH) and next-generation targeted sequencing (NGS). Bi-allelic deletions (3 Kb) were observed in 46 genes, four of which have been associated to malignant tumors, including two SWI/SNF-related genes [PBRM1 (15%) and SMARCC1 (6%)], BAP1 (48%) and SETD2 (27%). More than 200 MPM were studied in a recent thorough genomic investigation.

Bueno et al. [88] described mutations in genes encoding SWI/SNF components in 8% of the samples, as well as mutations in two histone methyltransferases (SETDB1 and SETD5) in about 3% of the samples. The discrepancies between the results reported by Yoshikawa et al. [87] and Bueno et al. [88] may be attributable to the identification of minuscule deletions by high-density, a-CGH, and specific NGS that are not detectable by conventional NGS techniques. To establish final conclusions, more studies are needed to determine the frequency and clinical significance of SWI/SNF mutations in mesothelioma.

7. Epigenetics in the treatment of mesothelioma

MPM inhibits tumor suppressor genes by promoting LOI and repression of CG genes by site-specific hypermethylation of DNA and/or polycomb repressor complexes in the context of hypomethylation of the genome. This “DNA methylation paradox” mimics epigenomic conditions in normal germ cells and lays the groundwork for epigenetic regimens that restore tumor suppressor gene expression and trigger growth arrest/apoptosis. Upregulation of CTAs, development of viral mimicry by derepression of endogenous retroviruses, and control of the tumor microenvironment all help to boost antitumor immunity [89, 90].

DNMTs are potential targets for MPM treatment because of their direct functions in suppressing tumor suppressor genes and maintaining pluripotency [91, 92]. Previous clinical attempts to inhibit DNMT activity in MPM, however, have failed miserably. Yogelzang et al. [93] showed a 17% objective response rate in 41 MPM patients who received 120 h of continuous dihydro-5-azacytidine infusions. Amazingly, 6 years following treatment, the single responder was disease-free. The lack of efficacy of DNA hypomethylating drugs in solid tumors could be due to their usage at maximum tolerated doses, resulting in myelosuppression, rather than prolonged use at lower doses to obtain pharmacodynamic effects without systemic toxicity. The Phase I decitabine trial (DAC) clearly demonstrates that chronic exposures are required to achieve maximum gene induction effects in cancer tissues [94].

Furthermore, 5-AZA and DAC administered IV, SQ , or PO have short half-lives (less than 5 min) and poor biodistribution, limiting their potential utility in patients with solid tumors. Cytidine deaminase (CDA), which is found in practically all organs but mainly the gastrointestinal system, quickly inactivates these molecules [95, 96]. Documented toxicity increases Cmax and t1/2 (>50 nM and 4 h, respectively) as well as biodistribution of oral decitabine, decreasing inter-patient variability in drug levels significantly [95, 96, 97, 98]. Significant increases in fetal hemoglobin, without neutropenia, thrombocytopenia, or lymphopenia, are indicative of hypomethylation of systemic DNA caused by oral DAC-THU. A phase II trial (NCT02664181) is currently underway at the Cleveland Clinic and NCI to examine whether DAC/THU can improve responses to nivolumab when given as second-line therapy to patients with non-small cell lung cancer. Despite encouraging preclinical data [26], efforts to target HDAC on MPM have also been disappointing.

As second- or third-line therapy, Krug et al. [99] randomized 661 MPM patients to receive the HDAC inhibitor vorinostat or placebo. Overall survival, as well as the drug’s safety and tolerability, were the key outcomes. Vorinostat-treated patients had a median OS of 30.7 weeks (95% CI 26.7–36.1) compared to 27 weeks (95% CI 23.1–31.9) for placebo-treated patients. Given the absence of evidence for HDAC upregulation in MPM and the limited antitumor effects of HDAC inhibitors alone in preclinical tests, the lack of efficacy of the single-agent vorinostat in patients with MPM is not surprising. Combinated techniques, such as using HDAC inhibitors to sensitize cells to TRAIL-mediated apoptosis or flavopiridol to boost romidepsin-mediated growth arrest and death, might be helpful for future clinical trials. Hypomethylating drugs, on the other hand, do not appear to lessen the incidence of mesothelioma after asbestos exposure. In fact, non-solid cancers such leukemias, lymphomas, and other myelodysplastic syndromes show the best benefits with this medicine.

It is feasible that BAP1 mutations could be used for MPM therapy in the future. BAP1 promotes the recruitment of the polychial deubiquitinase PR-DUB complex to DNA damage sites by stabilizing BRCA-1 and promoting poly (ADP-Ribose) dependent recruitment of the polychial deubiquitinase PR-DUB complex to DNA damage sites. This activity is dependent on deubiquitinase activity and BAP1 phosphorylation. BAP1 mutations, which invariably show as a loss of function, cause BRCA-1 levels to drop and double-stranded DNA repair to be inhibited [100, 101, 102]. A BAP1 isoform including part of the catalytic domain sensitized MPM cells to the PARP1 inhibitor, according to Parotta et al. [102]. (Olaparib). Concomitant treatment with GDC0980, a dual PI3K-mTOR inhibitor that is downregulated by BRCA-1, could improve this sensitivity. These strategies could improve responses to cisplatin/pemetrexed in patients with BAP1 mutant MPM and should be evaluated in future clinical trials.

There is considerable interest in chromatographic remodeling agents with adoptive cell transfer or immune checkpoint inhibitors for cancer therapy, given the extensive preclinical studies showing DNA demethylating agents, HDAC inhibitors, and KMT inhibitors in the immunomodulatory effects of potentials [103]. In a syngeneic mouse tumor model, cytolytic T lymphocytes target testicular cancer antigen in vivo using decitabine to destroy metastatic cancer. The preclinical basis for combining gene induction regimens with cancer adoptive immunotherapy was established in these studies. Furthermore, novel microenvironmental data are likely to have a substantial impact on the outcomes of clinical trials for epigenetic treatments and immunotherapies [89].

8. Conclusions

While malignant pleural mesothelioma is a disease with a low incidence worldwide, with aggressive behavior, its survival does not go beyond 12 months once the diagnosis is made [1, 2]. Its origin has been related to the chronic exposure of asbestos as the main factor. Also, asbestos fibers have been an essential component in structural changes at the molecular level, with much evidence about its genetic behavior and to a lesser extent, its epigenetic behavior. All of this gives it a fairly heterogeneous behavior [13, 14, 15]. New molecular techniques allow a broader understanding of the carcinogenesis of this tumor and an approach to new diagnostic tools. Epigenetic dysregulations require active maintenance and are potentially reversible, making them a therapeutic target [7, 23, 30].

The study of methylome has made it possible to carry out differential diagnoses thanks to the methylation of some specific loci, such as TMEM30B, KAZAZD1, MAPK13 and to demonstrate greater survival rates in patients with low frequencies of methylations [16, 17, 28].

It is important to mention the exposure to asbestos fibers as the main resistance factor associated with the methylation of tumor suppressor genes seen in pleural mesothelial cells such as APC and RASSF1. Additionally, there are direct cellular effects such as chronic inflammation measured by free radicals leading to DNA oxidation, hemolysis with the release of hydroxyl ions, intrachain breakdown plus subsequent chromosomal fragmentation, and production of pro-inflammatory cytokines with higher expression of angiogenic growth factors, another aspect that can be considered a potential therapeutic objective. Genomic responses related to methylation conclude in a gene silencing, most likely in tumor suppressor genes such as SFRP4, FHIT, SLCA20 [69, 71, 80]. Another diagnostic approach that can be observed by methylation is the overexpression of DNMT in patients with MPM and consequently could be an attractive therapeutic target, however, clinical efforts for its inhibition have been disappointing and future studies should focus on the therapeutic approach to the inhibition of DNMT.

A greater association of methylation has been seen in advanced ages and ethnic groups such as the Japanese population. However, the greater association related to histological changes in proliferation, differentiation, invasion, and reduction of apoptosis has been seen with the increased methylation of CpG islands in genes such as CCND2, CDKN2A, and associated with asbestos bodies with RASSF1.

Although methylation is the most studied epigenetic mechanism, there are other modifications that lead to the silencing of tumor suppressor genes, such as the activation of the Polycomb complex and the mutation of the SWI/SNF pathway [82, 83]. Deacetylation mediated by HDAC has been seen in the p53 gene and other aspects such as HAT-mediated acetylation or demethylation by KDMs.

The modification in histone features such as stability in chromatin has a great relationship with HDCAs, thus making them a potential therapeutic target. There are few studies with inhibitors such as vorinostat, however, where there are no positive results due to the low expression in MPM.

Finally, it is clear that there is much to know about the modifications and/or epigenetic changes in MPM. The current evidence of the molecular mechanisms opens up another panorama for us to adjust personalized therapeutic strategies aimed at reversing normal changes and thus be able to identify in a timely manner those patients who are susceptible to such treatments. Therefore, clinical trials should focus on those epigenetic markers that at some point in their disease are overexpressed or silenced.