*4.1. DNA Methylation in CM Development*

Disruption of DNA methylation is a common event in cancer. Both focal hypermethylation at CpG islands and global hypomethylation are constant hallmarks of the cancer genome and often coexist in tumor cells, impacting tumor biology and behavior. As with other cancers, CM initiation and progression have been associated with loss of tumor suppressors and oncogene activation (Figure 2) [86].

Inactivation of tumor suppressor genes (TSGs) due to specific DNA methylation in the promoter regions was the first epigenetic alteration studied in CM more than 10 years ago [87] (Figure 1). So far, dozens of genes are known to be regulated by this mechanism. These genes appear to be involved in various signaling pathways, usually disrupted in CM, such as the phosphatidylinositol 3-kinase/protein kinase B (PI3K/Akt) and MAPK pathways, cell cycle, DNA repair, retinoblastoma (RB) and Wnt signaling [86]. Furthermore, it was also observed that three TSGs are frequently inactivated by methylation: RASSF1A (55%), RAR-β2 (70%), and MGMT (34%) can also be identified in the circulating tumor DNA of CM patients, which makes them useful diagnostic and prognostic biomarkers in the clinical setting [88,89]. In several cancers, but also in a significant proportion of melanomas, a gradual increase in DNA hypermethylation was observed along with tumor aggressiveness; this phenomenon, called CpG methylator phenotype (CIMP), was reported for the first time in colorectal cancers, a finding that highlighted a tight correlation between altered DNA methylation patterns and the clinical outcome of the affected patients. Tanemura at al. demonstrated that during tumor progression, several tumor-related genes and loci, including WIF1, SOCS1, RASSF1A, TFPI2, MINT17, and MINT31, gain methylation with advancing stages. These genes have been suggested to constitute CM's CIMP [90]; however, recent research highlights that CIMP is usually associated with an NRAS-mutant phenotype, which is more aggressive than a non-NRAS-mutant tumor [91]. Therefore, future approaches should correlate CIMP with CM patients' clinical outcomes and mutational profiles. This information would be essential for developing novel tools for prognosis and response to therapy in the clinical cohorts of melanoma patients.

Epigenetic silencing of tumor suppressor genes has been one of the best-studied phenomena in CM; however, although less studied, DNA hypomethylation is equally important in the initiation and development of CM. It was shown that in many cancers, hypomethylation contributes to tumor progression by inducing genome instability via the demethylation of transposons and pericentromeric repeats or the activation of certain oncogenes [92]. Figure 2 depicts hypomethylation mechanisms as a hallmark of melanomagenesis. LINE-1 elements are one of the most abundant classes of mobile DNAs within the human genome [93]. LINE-1 hypomethylation, detected in both tissues and plasma circulating DNA of melanoma patients seems to be a hallmark of the metastatic capacity of primary melanomas [94,95]. Other reports highlighted that LINE-1 hypomethylation may predict the OS in stage III CM patients [96,97]. In parallel, DNA hypomethylation has been described as one of the main mechanisms regulating the expression of cancer-testisantigens (CTAs) in human melanomas [98]. CTAs are a specific group of tumor-associated antigens (TAAs) whose expression in normal tissues is generally restricted to the gametogenic tissues of the testis and fetal ovaries [99]; nonetheless, CTAs were found to be re-expressed via hypomethylation in CM, regulating vital cellular processes such as tumor cell division, differentiation, invasion and drug resistance [98,100–102].

Several studies have revealed that DNA methylome analysis may help discriminate between normal melanocytes, nevi, and melanomas. For instance, Fujiwara et al. identified several novel genes that were hypermethylated in melanomas compared to melanocytes, such as KRTCAP3, AGAP2, ZNF490, and TTC22, in addition to those previously documented, such as COL1A2, GPX3, and NPM2 [103]. Among those genes, they found that NPM2 showed distinct immunohistochemical (IHC) staining in normal melanocytes, whereas its expression was lost in CM samples [103]. Moreover, Gao et al. reported a diagnostic algorithm based on the methylation patterns of CLDN11, CDH11, and PPP1R3C genes that can differentiate between dysplastic nevi and primary melanomas with a specificity of 89% and a sensitivity of 67% [104]. Other reports highlighted that several methylation subgroups might be associated with different clinical characteristics of the disease, potentiating that the evaluation of DNA methylomes may have prognostic applications in CM. A study led by Lauss et al. revealed three methylation clusters: MS1, MS2, and MS3, which differ significantly in terms of promoter methylation, proliferation, and presence in immune cells [33]. The MS1 group has the highest methylation level, especially at CpG islands and poised promoters, enriched in polycomb repressive complex (PRC2) target genes. The MS1 subtype also showed an increased frequency of homologous deletions of CDKN2A and IDH1R132 hotspot mutations. The MS3 group had the lowest methylation levels, similar to peripheral blood leukocytes, and MS2 was intermediate. No correlations were identified between methylation clusters and clinicopathological variables or actionable mutations such as BRAF or NRAS. However, the tumors bearing MS1-signature were associated with the lower patient OS (20 months for MS1 vs. 60 months for MS3). Interestingly, genetic analysis revealed methylation clusters are associated with different biological and clinical behaviors. The MS1 subtype termed "proliferative" was associated with the upregulation of TP53, MDM2, CDK4, CDK6, CCND1, CCNE1, and E2F3, as well as epigenetic modifiers TET1, JARID1B, SWI/SNF chromatin remodelers, and DNMT3A; in contrast, the MS3 subtype harbored an "immune high" signature, possibly explaining the better survival of patients appending to the MS3 cluster [33]. Similarly, Yamamoto et al. stratified 51 CM into two risk groups based on the promoter region's methylation status. The high-methylated subgroup was positively associated with a thicker tumor progression and hence a worse clinical prognosis. Among the 27 genes proposed to distinguish between the two subtypes, TFPI2 was the most frequently hypermethylated gene in the aggressive subtype [34]. In addition, altered methylation patterns of the homeobox D cluster were linked with melanoma metastasizing to the brain [105].

Further complicating this scenario, genome-wide mapping of CM revealed that 5 hmC levels are progressively lost during tumor progression from benign nevus to malignant melanoma, via IDH2 and TET family downregulation [106]. Elevated levels of 5 hmC were subsequently validated by IHC staining as predictors of metastasis-free survival and overall survival in CM patients [107]. Taken together, all this information supports the further development of 5-hmC IHC expression as a prognostic biomarker that can add some precision to the American Joint Committee on Cancer (AJCC) staging system [108].
