Current understanding of epigenetics role in melanoma treatment and resistance

Genetic alterations in melanoma patients were seen to activate the RAS/RAF/MEK/ERK (MAPK) and the PI3K/PTEN/AKT (AKT) signaling pathways, so that it has been found that the growth of melanoma cells can be blocked as a result of inhibiting both ERK and PI3K signaling [11‐13]. The MAPK (Mitogen-activated protein kinase) pathway can affect downstream pathways of some receptors, such as cytokine, heterotrimeric G-protein, and tyrosine kinase receptors. The small G protein Ras, which belongs to a family of hydrolase enzymes, is an anchored protein in the inner leaflet of the membrane bilayer [14‐17].

So, MAPK is an essential pathway in most cases of melanoma. Several mutations, such as NRAS and BRAF can activate this pathway [18]. The most critical downstream molecules of the MAPK pathway are BRAF and CRAF serine-threonine kinases [18]. Both BRAF and CRAF were shown to have a regulatory domain (CRD), a RAS-binding domain (RBD), and a kinase domain that inhibits RAF function [18]. Contemporary, it has been reported that mutation in both NRAS and BRAF is linked with poor prognosis in stage IV of melanoma cases [18]. Such mutations are ascertained in the benign proliferation of melanocytes, metastatic melanoma, and invasive melanoma [18]. Both vemurafenib and dabrafenib, which are approved by FDA, have shown a noticeable efficacy in BRAF inhibition, thus suppressing the tumor cells [18]. Mutations of NRAS (in 15–30 percent of melanoma tumor samples) have been recognized as a vital driver of oncogenesis in melanoma patients [18]. It should be pinpointed that growth factor receptors (namely epidermal growth factor receptor, c-Kit, and c-Met) can be activated by RAS promotion in melanoma cases [18, 19].

On the other hand, it is of note that rampant genetic changes in melanoma are capable of reducing apoptosis through the overexpression of B-cell lymphoma 2 (Bcl-2), loss of both Phosphatase and tensin homolog (PTEN) and nuclear factor-κB (NF-κB), and mutation of Akt3, NRAS, and BRAF [11, 20]. Most importantly, lymphocytes and pigmented melanophages (possessing ingested melanin) are found close to the dermal-epidermal p53-positive cells, suggesting cell death among the melanocytes [21, 22]. Accordingly, after p16INK4A-dependent senescence, melanocytes can be provoked by a p53-dependent ‘back-up’ cellular senility checkpoint, thereby mediating the transformation of NRAS or BRAF [23].

BRN2 is a member of the POU domain family of transcription factors with a crucial function in the progression and metastasis of melanoma [24, 25]. A high level of BRN2 expression could lead to elevated invasiveness as well as suppression of DNA repair and apoptosis in melanoma cell lines [26]. This fact corroborates the notion that BRN2 is involved in high somatic mutations in melanoma cases [24, 27]. Significantly, BRN2 contributes to the melanocytic-lineage oncogenic factor (MITF)-mediated progress of melanoma. MITF is the master regulator in the transcription of melanocytes [24]. An in vivo imaging study on melanoma cell lines has indicated that the BRN2 expression is increased in invasive cells of the primary tumor, while MITF expression is lost [28]. Most biopsy samples of melanoma patients and drug-sensitive melanoma cell lines have shown increased expression of MITF [29]. High expression of MITF is distinguished as a fundamental mechanism of resistance to MAPK pathway suppression [30]. Since overexpression of BRN2 and reduced expression of MITF are directly linked to activation of MAPK pathway, they are significantly associated with early resistance to targeted therapy [31]. Taken together, BRN2 is suggested as a critical regulator that is involved in drug resistance and invasion during melanoma treatment as a counterbalance to the MITF [24]. Even with well-known functions, the enormous scope for uncovering its tumor progression effects, the tumor microenvironment on BRN2, and the involved epigenetic switching mechanisms are still needed.

Hypoxia is the most commonly accruing condition among all solid tumors. It can lead to poor prognosis in cancer patients, irrespective of the kind. Hypoxia can increase the progression of tumor cells via activation of HIF-1α. This protein is responsible for regulating essential genes, which interfere in cell proliferation, angiogenesis, metabolism, and metastasis [32‐36]. Hypoxia-Inducible Factor1 (HIF-1) is a transcription activator, which is sensitive to oxygen and encompasses the HIF-1β and HIF-1α subunits. HIF-1α is activated by post-translation modifications such as phosphorylation, acetylation, hydroxylation, and ubiquitination [32, 37]. HIF-1α also has a prominent role in angiogenesis by affecting cellular metabolism, tumor invasion, vascular endothelial growth factor (VEGF), cell survival, and metastasis [32, 33, 38‐40]. HIF-1α expression is related to aggressive characteristics of melanoma. Hence, the incorporation of HIF-1α as a promising prognostic indicator in melanoma may add growing value to current staging procedures [41]. Accordingly, along with post-translation modifications, some signaling pathways can activate HIF-1α, including the RAS/RAF/MEK/ERK and MAPK/ERK signaling pathways [42, 43]. The MAPK pathway was shown to induce the formation of the HIF-p300/CREB-binding protein (CBP) complex and modulate the transactivation of p300/CBP [44]. The RAS/RAF/MEK/ERK pathway can be stimulated via mutations that occur in membrane receptors and oncogenic genes, namely KIT and both N-RAS and B-RAF, respectively. Hypoxia condition could also be involved in activating the JAK⁄ STAT (signal transducer and transcription activator and Janus kinases) pathway in response to cytokines and growth factors [32, 45‐47]. This phenomenon can be triggered by HIF-1α in multiple cancer cell lines and animal models [47‐49].

Obliterating the STAT3 pathway is demonstrated to block oncogenesis. It has been found that STAT3 signaling disruption could lead to inducing both apoptosis and cell cycle arrest in sundry cancer cell lines. This condition occurs in the cases of prostate cancer, multiple myeloma, and breast cancer [50]. Cao et al. explored the involvement of STAT3 signaling in melanoma occurrence/development by evaluating the anti-melanoma activities of shikonin in cell and zebrafish xenograft models. It has been demonstrated that shikonin (a naphthoquinone pigment extracted from the dried root of Zicao (Lithospermum erythrorhizon, Onosma paniculata, or Arnebia euchroma, as a traditional Chinese herbal medicine) could block the phosphorylation of STAT3, decrease the levels of STAT3-targeted genes involved in melanoma survival and migration (Mcl-1, Bcl-2, MMP-2), and finally suppress melanoma growth [51, 52]. In a different study, the inhibition of the viability and proliferation of A375 and A2058 melanoma cells was shown by the dauricine via blocking the phosphorylation-mediated activation of STAT3 and Src in a dose-dependent manner [34, 53‐55]. Recently, Meng et al. have also evaluated the potent anti-melanoma activity of podocarpusflavone A (PCFA). This compound has been described to inhibit melanoma growth via inhibition of the JAK2/STAT3 pathway [56]. These findings indicate that the JAK2/STAT3 pathway plays a significant role in melanoma occurrence/development.

Ambra1 is identified as a multifunctional scaffolding protein and pro-autophagy protein. It has been reported that functional deficiency in Ambra1 could induce hyperplasia and impaired autophagy in neuro-epithelial cells of mouse embryos [57, 58]. Ambra1 can initiate autophagy via regulation of Unc-51 like autophagy activating kinase (ULK1) and Beclin1 [57, 59]. Autophagy can lead to drug resistance and progression in multiple cancers (e.g. melanoma, acute myeloid leukemia (AML), etc.) [60‐62].

Ambra1 has also been found to have a key function in cell cycle progress and proliferation by affecting the stability of c-Myc, interaction with the protein phosphatase 2A (PP2A) and stability of Cyclin D1 (CCND1) through interaction with the E3 ligase DDB1/Cullin4 [63, 64]. It was shown that concomitant loss of Ambra1 and expression of Loricrin in the peritumoral epidermis could be used as prognostic biomarkers for high-risk tumor subsets and early stages of melanoma [65, 66]. A list of onco-suppressor and oncogenic factors involved in melanoma is presented in Table 1.

The methylation commonly occurs at the fifth carbon atom Cs in cytosine phosphate-guanine (CpG) dinucleotides [5]. DNA methylation is an essential epigenetic modification in many cancers [81]. The CpG dinucleotides are distributed in the human genome [82] that can be either as a dinucleotide or clusters as CpG islands. The CpG islands have some unique properties due to their localization in the promoter region of genes [83] and methylation in neoplastic conditions [84]. It is worth noticing that the hypermethylation of CpG islands has been mostly spotted in the promoter zones of the specific genes inducing the tumor suppressor genes silencing, chromatin remodeling and influencing the transcription, DNA repair, cell signaling, apoptosis, and cell cycle regulation of melanoma thus contributing to melanoma tumorigenesis [85]. In this line, DNA methyltransferases are the enzyme that adds methyl groups to the carbon atom of cytosine, culminating in the methylation of DNA [86]. In mammalian systems, DNA methylation is performed by DNMT1 and DNMT3s (DNMT3A and 3B). DNMT1 is predominantly involved in the maintenance of DNA methylation during cell division, while DNMT3s are involved in establishing de novo cytosine methylation and maintenance in both embryonic and somatic cells [87]. Thus, melanoma pathogenesis is developed owing to epigenetic alterations. Infinium methylation technology identified some CpG sites, associated with more than 14,495 cancer-related genes with significant methylation differences (44 hypomethylated and 106 hyper-methylated CpG islands) [88]. Some of the biomarker genes being modified through methylation are mentioned in Table 2.

The CDKN2A encoding for p16 tumor suppressor is either mutated or omitted in most melanoma cell lines. This gene could be transcribed in alternative reading frames, resulting in two separate gene products, p16 and ARF that both of which can negatively regulate cell cycle progression [52, 89‐91]. The p16 exerts its effects by competitive inhibition of cyclin-dependent kinase 4 (CDK4) [92, 93]. So, p16 mutations increase the possibility of repair failure of DNA before cell division [94]. ARF is the second protein product of the CDKN2A locus, which regulates cell growth by affecting the p53 pathway [95]. The p16 mutation disables two separate pathways of cell growth control, which could indicate ARF role in cell growth [96, 97]. CDKN2A locus also encodes for the p14 protein, which binds to MDM2 and inhibits p53 ubiquitination and proteasomal degradation [97, 98]. Hyper-methylation of p14 has been shown in approximately 57% of human melanocytic nevi samples, while CDKN2A methylation has not been reported [99, 100]. On the other hand, hypermethylation of CpG islands could lead to modification of some genes such as the telomerase reverse transcriptase (TERT) gene, BRCA1-associated protein-1 (BAP1), microphthalmia-associated transcription factor (MITF), and Ras association domain family 1 isoform A (RASSF1A) [101‐104]. Hypomethylation of specific CpG sites being close to the PRAME promoter and the deleted split hand/split foot 1 (DSS1) gene are other genes being regulated through DNA methyltransferases (DNMTs) [105, 106]. Overall, the identification of aberrant DNA methylation modifications in melanoma is considered a critical step toward comprehension and utilization of the methylation landscape in melanoma therapy [107].

Histone modifications are critical epigenetic drivers causing post-transcriptional modifications (PTMs) or altering the chromatin state proper for the cancer progression [108, 109]. Histones are characterized by positively charged and lysine-rich N tail regions [110]. Epigenetic modifications that happened mainly in tail domains can alter transcription and replication, or cause malignant transformations [5]. Multiple histone modifiers have been introduced and some of the histone modifiers are mentioned in Table 2. Ribosylation, phosphorylation, or histone ubiquitination has been involved in the regulation of various pathways in the development of melanoma.

The chromatin compaction and initiation of transcription are influenced by the phosphorylation of histones in mitosis and meiosis [111]. Additionally, histone phosphorylation of H1, H2B, and H3 can have a remarkable impact on DNA repair and gene regulation [112]. Both cancer development and dysregulation of oncogenic kinases are shown to result from an aberrant function of histone phosphorylation. However, modification of histone acetyl groups (via histone deacetylases (HDACs), histone acetyltransferases (HATs), methyl groups (via histone lysine methyltransferase (HKMTase), and histone demethylases (HMDs)) are of great importance and frequency [113].

MicroRNAs (miRNAs), single-stranded short noncoding RNAs, play a vital role in expressing nearly 60 percent of human protein-coding genes that their dysregulations lead to several disorders such as cancer [55, 154]. In Melanoma cancer, miRNAs are involved in numerous cellular events, including melanoma genesis, cell cycle regulation, tumor growth and proliferation, cell migration and invasion, drug resistance, and apoptotic induction. Accordingly, the biological processes of melanoma cancer cells are potentially affected by downregulated miRNAs, including miR-211, miR-196a, miR-21, miR-124, miR-29c, and miR-210 [155, 156]. Interestingly, miR-211 has been identified as differentially expressed in the melanoma cell lines among various types of miRNAs affecting numerous targets like TGFBR2 (transforming growth factor beta receptor 2), RUNX2, IGF2R (insulin like growth factor 2 receptor), and NFAT5 (nuclear factor of activated T cells 5). Moreover, the ectopic expression of miR-211 involves the inhibition of migration and invasion in melanoma cells. This property suggests the tumor suppressor activities of miR-211 [157, 158]. Notably, the expression of miR-196a, miR-200c, and miR-205 could lead to extensive down-regulation of malignant melanoma cell lines and act as the tumor suppressors [159]. In contrast, numerous miRNAs such as miR-210, miR30b, and miR-30 are overexpressed in melanoma and associated with up-regulation of cancer cells leading to melanoma metastasis through promoting invasion and immunosuppression induction [160]. MiR-149 is another overexpressed miRNA in melanoma targeting GSK3a leading to apoptosis resistance in melanoma cells. So, non-coding RNAs (mainly miRNA) deletion can/state control both normal and melanoma cells generally affecting cell cycle regulation [161]. For instance, let-7b is a type of miRNA that inhibits the cell cycle progression by decreasing the CCND1, D3, and CDK4 expressions and it functions as a cancer cell growth [162]. Moreover, miRNA-193b downregulates CCND1 and CCND2 genes, which results in the promotion of melanoma cell proliferation and invasion [163]. Downregulation of miR-206, miR-143, or miR-106b inhibits CCND1 via affecting G1 cell cycle causing decrease in melanoma cells invasion or migration. Several other miRNAs have been documented as the significant cell cycle regulators in a cyclin-independent manner, including the miR21, miR203, miR205, miR18b, miR149, and miR26a [164]. Apoptosis induction by some microRNAs, including miR-155, miR-205, miR-21, miR-26a, miR-15b, and miR-149 have been highlighted in various reports[165]. Alteration in methylation of CpG islands regulate the expression of miR-375, miR-34b, miR-182, miR-148a, miR-203, miR-29, and miR-26 demonstrating the epigenetic regulation of miRNAs in melanoma [166]. As aforementioned, some miRNAs function as anti-melanoma in combination with HDACIs and immune checkpoint inhibitors. For example, downregulated miR-589 promotes melanoma malignancy through accelerating PD-L1 expression level [167]. Further, combinatory inhibition of a miR-146a and PD-L1 promotes the survival in a melanoma mouse model [168]. Although the down/upregulation of miRNAs in melanoma cells and the consequences of such dysregulations have been vastly reported. Although, the related mechanisms of these dysregulations are not entirely understood. Some of the miRNAs affecting apoptosis, migration, and proliferation of melanoma cells through different mechanisms are presented in Fig. 3. Overall, miRNAs can be beneficial in scientific research, the diagnosis of melanoma, and also it can be used to predict the patient’s reactions to treatments. Therefore, the development of miRNAs is considered as critical epigenetic factor in melanoma which could greatly increase the clinical management of melanoma.

Based on reports, cancer patients treated with either HDACIs, DNMT, or PD1/PD-L1 immune checkpoint inhibitors experienced potent treatment responses. These studies suggest that these epigenetic inhibitors may escalate the efficacy of immunotherapy via [1] enhancing the antigenicity, [2] counteracting immunosuppressive mechanisms by the tumor microenvironment, and [3] reversing cytotoxic T cell exhaustion [198]. It is reported that patients suffering from melanoma with the expression of PD-L1 are divided into four groups based on the number of tumor-infiltrating lymphocytes (TILs). Group 1 patients respond to treatment, which is responsive against their tumor cells. Group 2 patients have a low number of TILs and negligible or no PD-L1 expression, thereby not responding to PD1 monotherapy. Group 3 patients have TILs, albeit low or no PD-L1 expression. Eventually, Group 4 patients who have few or no TILs, albeit having PD-L1 expression [138, 139]. It is of note that PD-L1 expression can be increased in epithelial cancer cell lines, which are under treatment with DNMT inhibitors (DNMTis). Moreover, Illumina 450 K arrays revealed that low or no PD-L1 expression is strongly linked with high DNA methylation. This property reveals the role of chromatin methylation to suppress PD-L1 expression. Transcription factors or epigenetic regulators including the EZH2 and SUV39H1 methyltransferases are capable of direct interaction with binding domains (ATRX-DNMT3-DNMT3L (ADD)) of DNMT3A. Therefore, DNA hypomethylation regulates the PD-L1 expression, and it plays a pivotal role in the modulation of responsiveness to PD1 inhibitors [199]. DNA methyltransferase 1 (DNMT1) locates on the daughter strand cytosine at the complementary CpG. This process enforces gene silencing simultaneously to the mammalian cell division [200, 201]. DNA methyltransferase inhibitors (DNMTis), namely 5-azacytidine, decitabine, and guadecitabine, are approved for DNA histone hypo-methylation in patients suffering from the myelodysplastic syndrome or leukemia to reactivate tumor suppressor genes. These Aza nucleosides irreversibly bind to DNMT1 and degrade them by substituting nitrogen with carbon at the C-5 position of the pyrimidine ring. This substitution results in the loss of DNA methylation, expression of genes pertaining to immunomodulatory pathways, and induction of tumor antigen presentation [202]. For instance, NSCLC cell lines under treatment with 5-azacytidine can activate the JAK/STAT signaling pathway and provoke the expression of genes with a role in antigen presentation. This event per se culminates in the expression of PD-L1, considered a vital ligand-mediator of immune tolerance. Moreover, DNMTi has been shown to be linked with the activating of expression for hyper-methylated endogenous retroviral double-stranded RNAs (EVs). This property of DNMTi could lead to the induction of type I interferon response and MHC I expression [203]. Researches in tumor-bearing mice models indicated that previous treatment with decitabine (as a DNMTi) affects either the tumor cells or antigen-specific CD8 + T cells. This treatment accompanied by anti-PD-L1 agents can prevent the acquisition of exhaustion-associated methylation programs. This event makes the T cells become more potent for expansion after immune checkpoint blockade [204, 205]. In a murine ovarian cancer model, anti-CTLA-4 treatment became more potent via its combination with decitabine. This combination increased the differentiation of naive T cells into effector T cells [205].

Aside from the aforementioned, non-coding RNA molecules (ncRNAs), miR-125a, miR-28, miR-125b, miR-100, miR-200c, miR-211, MELOE, SAMMSON, and HOTAIR have been found to play a crucial role in treatment resistance [206]. miRNAs promote the melanoma to secondary sites via several mechanisms, including (A) regulation of MITF-M expression, (B) alteration of the extracellular matrix (ECM), (C) enhancement of reciprocal epithelial-to-mesenchymal transition (EMT), mesenchymal-to-epithelial transition (MET), and (D) preparation of pre-metastatic niche formation [207]. However, some mRNAs such as miR-182, miR-137, miR-211, and miR-107 reduce the MITF-M expression in melanoma cells, leading to an invasive phenotype [208]. For instance, melanoma cell invasion and migration are provoked as a result of miR-182 upregulation which is triggered by the downregulation of both expressions of MITF and FOXO3 [158]. On the other hand, miR-211 can block the invasion and migration of melanoma cells [156], and repress POU3F2 (POU-domain class 3 transcription factor 2, also known as brain-specific homeobox 2 (BRN2)) which acts as a MITF suppressor. Zhao et al. have indicated that downregulation of miR-107 (a tumor suppressor) represses melanoma cell invasion through POU3F2 targeting [209].

miRNAs have conflicting functions, such as enhancing either tumor migration or suppression. For example, the miR-224/miR-452 cluster is directly activated by E2F1, which facilitates the cytoskeletal rearrangement of less aggressive cells and thereby increases the migration and invasion of melanoma cells. Notably, the miR-200 family (miR-200a, miR-200b, miR-200c, and miR-141) induces EMT-like processes via upregulation of Bmi-1 oncogene expression. Thus, the PI3K/AKT and MAPK pathways can be activated. Activation of these pathways negatively impresses the expression of ZEB1 (zinc finger E-box-binding homeobox 1) and E-cadherin, which provokes the expression of vimentin and N-cadherin [209]. Pertaining to IncRNA effects on melanoma cells, it is of note that SPRY4-IT1 was the first lncRNA, which was characterized to be originated from an intron of the SPRY4 gene. Recent studies have found that the expression of SPRY4-IT1 is escalated in cases of melanoma [210]. Siena et al. have identified a remarkable upregulation of ZEB1 antisense RNA 1 (ZEB1-AS1) in metastatic melanoma linked with hotspot mutation in both BRAF and RAS family genes [211]. Their analysis showed that ZEB1-AS1 could function by activation of expression for zinc finger E-box binding homeobox 1 (ZEB1). Activation of EB1 could influence the invasiveness and phenotype switching melanoma cases [211]. GAS5 is a lncRNAs, which diminishes the expression of MMP2. This protein is involved in ECM degradation and can reduce the migration and invasion of human MM cells [212, 213]. Nonetheless, deregulation of the expression for some miRNAs could give rise to drug resistance, particularly in BRAFi or MAPKi-based melanoma therapies. As a good example, miR-31a, miR-100, and miR-125b are shown to stimulate tumor cell proliferation, apoptosis escape, and decline in drug sensitivity among patients who took vemurafenib. Additionally, inhibition of miR-125a causes drug re-sensitization in a subset of BRAFi-resistant cell lines of melanoma. miR-204 and miR-211 could also lead to resistance against vemurafenib in melanoma cells [214‐218].