Polycomb group proteins – epigenetic repressors with emerging roles in melanocytes and melanoma

Summary

Melanocytes undergo rapid and significant changes in their gene expression programs at regular intervals during development and the hair follicle cycle. In melanoma, the gene expression pattern found in normal melanocytes is disrupted. These gene expression patterns are regulated in part by post-translational histone modifications catalyzed by Polycomb group (PcG) proteins, which play a major role in many developmental processes and are often altered in cancer. In this review, we discuss the role of the PcG proteins in stem cell and cancer biology, in general, as well as in melanocyte development and melanomagenesis. Highlights include discussion of newly identified treatments that target the activity of PcG proteins as well as new developments in the understanding of the role that these proteins play in melanocyte biology.

Both during development and during initiation of the hair follicle (HF) cycle, melanocytes undergo drastic changes in their gene expression programs on a relatively short timescale. They cycle from a state of relative quiescence in telogen to rapid proliferation and migration in anagen. A small number of melanocyte stem cells (MeSCs) must be activated to regenerate the melanocyte population at the next anagen or after wounding (Chou et al., 2013; Nishimura et al., 2002). Dynamic transitions between cell states of non-transformed cells, that is, cells that are not undergoing or have not undergone malignant transformation, require epigenetic regulation of gene expression, principally involving either changes in DNA methylation (Reik et al., 2001), histone modification (Li, 2002), or both in chromatin. One of the major mechanisms regulating histone-dependent changes in transcriptional regulation involves the Polycomb group (PcG) proteins. In this review, we discuss the emerging evidence for roles of PcG and their associated proteins in both melanocyte development and melanoma.

Introduction to Polycomb proteins

PcG proteins were first identified in Drosophila melanogaster, due to the fact that mutations in the genes coding for these proteins cause developmental defects resulting from dysregulation of Hox genes (Frei et al., 1985; Jürgens, 1985). Homologs of these proteins were soon identified in mammals as well, including humans, and were found to play an important role in both development and cancer. PcG proteins function within two distinct complexes: Polycomb Repressive Complex 1 (PRC1) and Polycomb Repressive Complex 2 (PRC2), which generally cooperate to repress gene expression. First, PRC2 is recruited to chromatin and trimethylates lysine 27 on the tail of histone H3, and PRC1 then recognizes this mark and causes further heterochromatization (Figure 1).

Figure 1.

A) One half of a histone core is depicted, with DNA encircling it and several relevant residues of the histone tails are indicated: lysine 27 of histone H3 (H3K27) and lysine 119 of histone H2A (H2AK119).

B) In the classical model of Polycomb regulation, PRC2 is targeted to a locus and catalyzes the trimethylation of H3K27. The core proteins comprising the PRC2 complex, EED, EZH2, and SUZ12 are required for this reaction. In addition, several other proteins, including AEBP2, RBAP46, and JARID2, have been identified as part of the PRC2 complex in various tissues and developmental stages.

C) The core members of PRC1 are a CBX protein, an HPH protein, a PCGF protein, and a RING family member. The CBX family of proteins interacts with H3K27Me3 through its chromodomain and localizes the PRC1 complex at the site that was modified by PRC1. The RING1 family of proteins has ubiquitin ligase function and catalyzes the ubiquitination of lysine 119 of histone H2A (H2AK119), which can enhance chromatin compaction.

D) When reactivation of a locus is required, the SWI/SNF chromatin remodeling complex facilitates the removal of the PRC2 and PRC 1 complexes from the chromatin. In addition, H2A deubiquitinases (H2A-DUBs) catalyze the removal of ubiquitin from H2AK119 and the H3K27Me3 demethylases UTX, KDM6A, and KDM6B catalyze the removal of the methyl groups from H3K27. With these chromatin marks removed, the locus is available for the application of active chromatin marks.

Polycomb Repressive Complex 2 (PRC2)

PRC2 is a multiprotein complex consisting of several core proteins, along with several others that are present in specific contexts. The core members of this complex are EED, SUZ12, and, in mammals, either EZH1 or EZH2, homologues of the Drosophila Enhancer of Zeste (E(z)). In addition, AEBP2, a zinc-finger protein; JARID2, a Jumonji C (JmjC) domain-containing protein that lacks histone demethylase activity; and the histone-binding proteins RBAP46 and RBAP48 are also present in various forms of the PRC2 complex (Kim et al., 2009; Li et al., 2010; Margueron and Reinberg, 2011; Shen et al., 2008). PRC2 possesses a histone methyltransferase activity, specific for lysine 27 of histone H3, associated with the catalytic subunit EZH1 or EZH2. The apparent redundancy in mammalian E(z) forms is at least partially explained by relative differences in the activities and expression of EZH2 and EZH1. EZH2 expression is restricted, and associated with proliferating cells, whereas EZH1 is expressed more ubiquitously in mammalian tissues (Bracken et al., 2003; Margueron et al., 2008). EZH2-containing PRCs have a greater H3K27 di- and trimethylation catalytic activity than EZH1-containing PRCs (Margueron et al., 2008). However, PRC2 complexes containing EZH1 can repress transcription in a methylation-independent manner. Histone deacetylase 1 (HDAC1) has also been shown to associate transiently with this complex (Kuzmichev et al., 2002), which could further aid in chromatin condensation or heterochromatization, given the association between histone deacetylation and transcriptional silencing (Rice and Allis, 2001). EZH1 complexes maintain slow cycling of adult hematopoietic stem cells and inhibit their differentiation and senescence (Hidalgo et al., 2012). Senescence, a nearly irreversible withdrawal from the cell cycle that also includes widespread changes in chromatin configuration and acquisition of a secretory phenotype, is a cellular response to stress, which can include oncogenic stress, representing a powerful inhibitory barrier to malignant transformation (Chinta et al., 2013). In addition to the classically described repressive function of this complex, EZH1-containing PRC2 complexes have also been described as possessing the ability to activate transcription by promoting RNA Polymerase II association with target genes (Mousavi et al., 2012). EZH2 has the ability to methylate non-histone proteins in a PcG-dependent manner (Su et al., 2005; Xu et al., 2012).

Investigations about the roles of EZH1 and EZH2 in epidermal development and homeostasis may portend their roles in melanocyte development and differentiation. Consistent with prior associations of expression with proliferating cells (Bracken et al., 2003), Ezh2 was expressed strongly from E14 in murine proliferating epidermal progenitor cells and at E16-E18 was confined to the proliferative basal layer (Ezhkova et al., 2009). Expression in the epidermis was nearly extinguished shortly after birth at P9. Mice selectively deficient for Ezh2 in epidermal basal cells exhibited an accelerated expression of terminal epidermal differentiation genes and premature development of the epidermal barrier. H3K27me3 marks were present in high concentration at terminal epidermal differentiation genes in wild type but not Ezh2-deficient mice. Neither wild type nor EZH2-deficient mice showed high levels of enrichment at the Krt14, Krt1, or Krt10 loci, which are expressed in epidermal progenitor cells within or immediately outside of the proliferating basal compartment. Accelerated epidermal differentiation in Ezh2-deficient mice was associated with reduced basal cell proliferation and lower levels of H3K27me3 marks at epidermal differentiation genes and the Ink4A-Arf-Ink4b locus, with correspondingly higher expression of those genes detected in the knockout.Although neither loss of Ezh2 or Ezh1 alone in basal epidermal progenitors caused phenotypic deficiencies in epidermal function or HF cycling, loss of both resulted in rapid neonatal demise. Grafting analysis of Ezh2- and Ezh-1-deficient skin showed that HF and sebaceous gland development was markedly impaired. Although stem cells were specified in the deficient grafts, subsequent cell proliferation was markedly inhibited, a property likely attributable to release of PRC2-mediated repression of Ink4b (Ezhkova et al., 2011).

During vertebrate embryonic development, neural crest cells gives rise to melanoblasts, progenitor cells of melanocytes. These melanoblasts migrate outward from the neural crest through underlying mesenchymal tissue into the nascent follicles in the epidermis (Mayer, 1973). In adult skin, follicular melanocyte proliferation is coordinated with the hair follicle cycle. During anagen, melanocytes proliferate and migrate down the lengthening follicle. Many of these melanocytes undergo apoptosis along with the rest of the follicle during catagen, until a small pool of MeSCs is left to generate the cells that will repopulate the follicle in the following anagen (Botchkareva et al., 2003; Nishimura et al., 2002; Sharov et al., 2005). These transitions are relatively rapid, on a time scale of days to weeks in mice, and require precise coordination of multiple gene expression changes. Melanoblasts proliferate rapidly, similar to and coinciding with the proliferation of epidermal progenitor cells described previously, during the embryonic migratory phase (Hornyak et al., 2001; Mackenzie et al., 1997), which is followed by melanocyte stem cell specification as melanoblasts enter the follicular epithelium in the immediate post-natal period (Osawa et al., 2005). It is conceivable that PRC2-mediated control of the timing of proliferation and specification is critical for the formation of a sufficient number of melanoblasts to colonize mammalian HFs during follicular morphogenesis. Release of PRC2-initiated or –mediated repression of cell cycle inhibitors may also be necessary to allow paracrine triggers (Nishimura et al., 2010; Rabbani et al., 2011) to stimulate MeSC proliferation at the onset of follicular anagen, leading to proper pigmentation of the growing hair shaft. Definitive exploration of the expression of both EZH2 and EZH1 in melanoblasts, MeSCs, and adult differentiated melanocytes will be required to infer whether a parallel mechanism for PRC2 is operative during melanocyte development and differentiation.

The phenotype of mice deficient in a PRC2 complex member provides an initial example of how PcG proteins may regulate aspects of melanocyte development. Deletion of the PRC2 component Aebp2, which is expressed in many neural-crest derived tissues in the embryo, led to a partially-penetrant phenotype affecting neural crest cell derivatives in heterozygotes. One-quarter to one-third of the mice exhibited a megacolon phenotype, with a combination of difficult defecation, megacolon, and reduced colonic ganglion cells. The majority of mice had white distal tail bands and white hindlimb digits (Kim et al. 2011). This phenotype resembles both the Dominant megacolon (Dom) (Southard-Smith et al., 1998) and Sox 10-deficient mouse phenotypes (Britsch et al., 2001), as well as the human phenotype of Waardenburg syndrome Type 4 (WS4) (Pingault et al., 2010) that also results from haploinsufficiency for the Sox10 transcription factor. It is intriguing that loss of Aebp2, thought to participate in gene regulation through chromatin-mediated repression, results in a similar phenotype as loss of Sox10 which has been shown previously to function as a direct transcriptional activator of genes in both the melanocytic (Jiao et al., 2004; Potterf et al., 2000; Potterf et al., 2001) and glial lineages (Bondurand et al., 2001; Peirano et al., 2000; Peirano and Wegner, 2000) of the neural crest. As a potential explanation for this paradox, Sox10 may conceivably function as well as a transcriptional activator of genes regulating neural crest cell multipotency, another key function of Sox10 in development (Kelsh, 2006). Inappropriate loss of gene repression in a multipotent neural cell precursor might result in either accelerated lineage differentiation or multiple lineage skewing, causing a phenotype consistent with defects in the migration of multiple neural crest cell types as in WS4 and analogous murine mutants.

Components of both PRC2 and PRC1 are required in mouse development as well as in the maintenance of embryonic stem (ES) cells, as deletion of many of these proteins (e.g. SUZ12, RING1A/B, BMI1, EZH2, AEBP2) cause either developmental defects or the failure of ES cell self-renewal (Endoh et al., 2008; Kim et al., 2011; Leeb et al., 2010; Pasini et al., 2004). However EED has been shown to be dispensable for ES cell pluripotency (Chamberlain et al., 2008; Leeb et al., 2010) even though it was found to be required for repression of PcG targets in these cells (Boyer et al., 2006). One potential explanation for this paradox is that ES cell pluripotency gene expression is largely independent of PcG protein repression, but proper differentiation from the ES cell state requires precise PcG-mediated epigenetic control. SUZ12 is also not required for ES cell maintenance and proliferation, but its deletion creates ES cells that cannot properly differentiate. In addition, regulated expression of many PRC components is required for several lineage-specific progenitor cell types (Ezhkova et al., 2009; Hirabayashi et al., 2009; Su et al., 2003).

Polycomb repressive complex 1 (PRC1)

PRC1 can be regarded as the effector arm of the PcG protein complexes, with canonical PRC1 complexes binding the H3K27me3 modification generated by PRC2 to result in chromatin compaction and transcriptional silencing (Di Croce and Helin, 2013). PRC1 complexes can be quite heterogeneous depending upon which member of each family of core proteins is incorporated. Core PRC1 complex members include CBX (chromobox domain) proteins, a PCGF family member (PCGF1-PCGF6), a RING1 protein (RING1a or RING1b), and an HPH family member (HPH1-HPH3). The core CBX subunit of PRC1 is important for the interaction of PRC1 with H3K27me3 binding sites generated by PRC2. The CBX family of proteins has been shown to be involved in the repression of senescence via repression of the Ink4a/Arf locus (Dietrich et al., 2007; Gil and Peters, 2006). CBX4 possesses a SUMO E3 ligase domain in addition to the chromodomain that is common to this family. While this SUMO E3 ligase domain has been found to be essential for the control of differentiation and proliferation in epidermal stem cells, the chromodomain of this protein has also been shown to play an additional role beyond interaction with methylated histone tails (Luis et al., 2011). The chromodomain is required to repress senescence in these cells, which may contribute to the protection of this stem cell population from exhaustion. Since the chromodomain is shared among the PRC1-associated CBX proteins, this mechanism for senescence suppression may also be common to this group of proteins. It would be interesting to understand how this is modulated in tissues where senescence is important for normal development as well as in cancers that show deregulated senescence.

Non-canonical, or CBX-independent PRC1 complexes lack a CBX protein constituent and are not dependent upon pre-existing H3K27me3 sites for their recruitment to chromatin. Instead, their interactions with DNA are mediated through their binding of sequence-specific DNA binding proteins and transcription factors such as Fbx110, E2F6, REST, and RUNX1. Alternative protein complexes containing PcG proteins have also been described. These include the dRAF and PR-DUB complexes in Drosophila, which ubiquitinate and deubiquitinate H2AK118, respectively (Simon and Kingston, 2013). dRAF also demethylates H3K36me2, normally associated with transcriptional elongation. In mammalian cells, the E2F6.com complex, which methylates H3K9, and the BCOR complex, which like Drosophila may engage in coupled H2AK119 deubiquitination and H3K36 demethylation, contains PcG proteins.

PCGF family members include Bmi-1 (PCGF4) and Mel-18 (PCGF2). Bmi-1/PCGF4 has been implicated independently in the promotion of normal and cancer stem cell renewal and proliferation (Lessard and Sauvageau, 2003; Molofsky et al., 2003; Park et al., 2003) as well as promoting cell proliferation and tumorigenesis, and inhibiting senescence, through epigenetic effects on the Ink4a/Arf locus (Jacobs et al., 1999a; Jacobs et al., 1999b). Mel-18/PCGF2 has been shown to inhibit the expression of Bmi-1/PCGF4 and thus inhibit the proliferation and senescence repression effects of Bmi-1/PCGF4 (Guo et al., 2007a; Guo et al., 2007b). The RING1 proteins are E3 ubiquitin ligases that monoubiquitinate K119 of histone 2A (H2A) with Ring1b the major determinant of that activity (Wang et al., 2004). H2AK119 ubiquitination is important for PcG protein-mediated gene silencing, although not essential for PRC1 to interact with H3K27me3 sites or to stimulate chromatin compaction (Endoh et al., 2012). The recent discovery of interactions between PRC2 members Aebp2 and Jarid2 with ubiquitinated H2AK119 (Kalb et al., 2014) suggests that recruitment of these components following PRC1 activity may create a positive feedback loop favoring H3K27 trimethylation. PRC1 may also inhibit gene repression separately through blocking transcriptional initiation, not by inhibiting TFIID binding but by blocking assembly of the Mediator transcriptional co-activator complex (Lehmann et al., 2012).

Although there is no specific role defined as of yet for individual PRC1 complex members in embryonic melanocyte development, BMI1 is required for self-renewal in both central nervous system-derived neural stem cells and gut-derived neural crest stem cells. Here it functions in a p16^Ink4a and p19^Arf-dependent manner, although it is dispensable for restricted neural progenitors (Molofsky et al., 2005; Molofsky et al., 2003). Taken together with the importance of Aebp2 function in melanocyte development, the results of these studies show that members of both the PRC1 and PRC2 complexes play important roles in the development of cells derived from the neural crest lineage.

Other macromolecules mediating PcG function

Non-PcG proteins have been found to mediate PcG protein function, especially the SWI-SNF ATP-dependent chromatin remodeling complexes. Two members of these complexes are especially important in modulating PcG function: SNF5 and BRG1. SNF5 has been shown to act as a tumor suppressor, and loss of this protein results in an increase in EZH2 expression as well as H3K27Me3 and repression of PcG target genes (Wilson et al., 2010). SNF5 mediates BRG1 occupancy at its target loci. The chromatin remodeling carried out by the SWI/SNF complex mediates the removal of both the PRC1 and PRC2 complexes from and the recruitment of the H3K4 methyltransferase mixed-lineage leukemia 4 (MLL4) to repressed loci, with a net effect of reduced H3K27Me3 and increased H3K4Me3 and increased gene expression (Kia et al., 2008).

Members of the Trithorax group (trxG) of proteins function both as a counterpoise and as cooperators with PcG proteins in epigenetic gene repression. TrxG proteins, which include SET1A, SET1B, and the MLL proteins 1–4, trimethylate a distinct lysine residue, lysine 4, on the H3 tail to generate the H3K4me3 mark which is the epigenetic hallmark of active gene transcription. Consistent with their association with active gene expression, TrxG proteins can not only modify H3K4 but also interact with CBP to acetylate H3K27 and block PcG protein silencing (Tie et al., 2014). Yet initial genome-wide studies of murine embryonic stem cells (mESCs) revealed many gene loci paradoxically expressing both the active H3K4me3 mark along with the repressive H3K27me3 mark generated by PcG proteins (Bernstein et al., 2006). These regions of chromatin, termed bivalent chromatin, are characteristic of developmental regulator genes expressed at low levels in mESCs poised for activation upon initial differentiation.

In addition to the protein modifiers of histone modification, there are several non-coding RNAs (ncRNAs) that are important in the establishment and maintenance of PRC-mediated gene silencing. HOTAIR is a long non-coding RNA (lncRNA) that originates from the HOXC genomic locus and is required for PRC occupancy and H3K27 trimethylation of the HOXD locus (Rinn et al., 2007). This transcript physically interacts with the PRC2 complex in a EZH2-EED heterodimer-dependent manner, and its interaction is enhanced by phosphorylation of EZH2 by CDK1 (Kaneko et al., 2010; Wu et al., 2013). In contrast to HOTAIR, many microRNAs play a role in down-regulating EZH2 expression. miR-101, miR-31, miR-137, and miR-138 all target EZH2 and overexpression of these microRNAs results in downregulation of EZH2 (Asangani et al., 2012; Luo et al., 2013a; Luo et al., 2013b; Zhang et al., 2013). miRNA regulation is not exclusive to the PRC2 complex, as BMI-1 expression is regulated by miR-200c (Liu et al., 2012).

Polycomb proteins in melanoma: current findings and their potential as therapeutic targets

The process of melanomagenesis has not been completely elucidated, but several factors have been found to be important in the transition of normal melanocytes to invasive melanoma. Some general factors parallel the classical hallmarks of cancer, including changes in cell cycle progression, the enhancement of the metastatic process, and alterations in the interactions between melanoma cells and the surrounding stroma (Crowson et al., 2007; Hanahan and Weinberg, 2011). However, others such as inactivation of the CDKN2A locus encoding the p16 and p14 tumor suppressor proteins (Walker et al., 1998) and abrogation of cellular senescence described in melanocytic nevi (Michaloglou et al., 2005), are more specific to melanocyte oncogenic transformation. Although many melanoma cells alterations have occurred because of either DNA mutation events or the acquisition of copy number variations in key genes, some are directly caused by epigenetic dysregulation of gene expression.

Both EZH2 and BMI1 have been shown to be altered in metastatic melanoma as compared to melanocytic nevi. BMI1 is expressed at lower levels in melanoma than nevi, and in melanomas, lower expression of this protein is associated with increased proliferation and decreased survival time (Bachmann et al., 2008). However, EZH2 is expressed at higher levels in metastatic melanoma than in nevi, and this higher expression is associated with increased tumor cell proliferation and decreased survival time (Bachmann et al., 2006; McHugh et al., 2007). This is somewhat paradoxical, since the PRC1 and PRC2 complexes are thought to cooperate in the repression of important tumor suppressor genes (Wang et al., 2012). However, this could simply imply that either different subsets of melanoma are sensitive to either EZH2 or BMI1 alterations, or that there are additional cofactors involved in regulating genes involved in melanomagenesis. In cultured melanoma cell lines, EZH2 depletion reduced proliferation and induced a senescent phenotype through induction of p21 in a p53-independent manner (Fan et al., 2011). PcG-mediated repression of key tumor suppressor genes mediated by EZH2 in early melanoma cells may enable these cells to bypass senescence and acquire aberrant cell growth characteristics, thus promoting tumorigenesis. A similar mechanism was proposed to explain how the H3K9 histone methyltransferase SETD1B, also mediating epigenetic gene repression and often amplified in melanoma, promotes melanoma development in a p53-deficient zebrafish model (Ceol et al., 2011).

Alterations in Polycomb group proteins are common in many cancers, and have been studied most extensively in hematopoietic malignancies. Several point mutations of EZH2 have been identified that alter its specificity, shifting the substrate preference from unmethylated H3K27 to dimethylated H3K27. This mutation can synergize with a wild-type allele to lead to hypertrimethylation of H3K27 (McCabe et al., 2012a; Sneeringer et al., 2010). Similar mutations have begun to be described in melanoma (Harms et al.). Deletions of PRC2 complex genes are frequent in the transition from myelodysplastic syndrome to acute myeloid leukemia. A small deletion of chromosome 6 frequently leads to JARID2 loss, and AEBP2, EZH2, and SUZ12 are also commonly deleted (Puda et al., 2012). In addition, point mutations of SUZ12 and EED that were identified in myelodysplastic syndrome were sufficient to reduce H3K27Me3 activity of the PRC2 complex in vitro (Score et al., 2012). BMI1 was found to be the target of frequent genomic rearrangements in several subtypes of leukemia (Rouhigharabaei et al., 2013). In melanoma, JARID1B, an H3K4 lysine demethylase which also targets many Polycomb-marked genes in ES cells, has been shown to be required for continued tumor growth, but not tumorigenesis (Roesch et al., 2010; Schmitz et al., 2011).

Due to the role of EZH2/PRC2 in promoting proliferation in many types of cancers, there is much interest in the development of inhibitors of H3K27Me3. One of these, DZNep, a S-adenosylhomocysteine hydrolase inhibitor, was initially believed to be a selective inhibitor of this modification, but further study revealed that it was a general inhibitor of histone methylation (Miranda et al., 2009). In addition, inhibition of EZH2 by DZNep results in the loss of expression of both SUZ12 and EED in a proteasome-dependent manner (Tan et al., 2007). Recently, a new inhibitor has been reported that shows greater specificity for inhibition of the methyltransferase activity of EZH2. GSK126 is a small molecule that competes with S-adenosyl methionine at the active site of EZH2 to decrease H3K27Me3 levels genome-wide and reactivate previously silenced Polycomb target genes (McCabe et al., 2012b). Interestingly, this inhibitor seems more effective in PRC2 complexes that were already activated by EED interaction with H3K27Me3 (Van Aller et al., 2013). An inhibitor has been developed that targets PRC1 as well. PRT4165 inhibits PRC1-mediated ubiquitination of histone H2A by blocking the ubiquitin ligase activity of both RNF2 and RING1A (Ismail et al., 2013).

SWI/SNF chromatin remodeling complexes important in the switch between repressed and active chromatin play important roles in melanoma formation. The catalytic subunits BRG1 and BRM are required for tumorigenicity (Keenen et al., 2010). BRG1 also promotes melanoma cell survival after UV radiation exposure by activating expression of ML-IAP (melanoma inhibitor of apoptosis) by promoting a decrease in H3K27Me3 and an increase of histone acetylation at the ML-IAP promoter (Saladi et al., 2013).

Polycomb group proteins have been shown to cooperate with DNA methylation to effect changes in gene expression. While melanomas share a similar genome-wide methylation profile with benign nevi, signatures of hypermethylation at individual genes have been found to differ between these tissues. Although it has not yet been determined if these individual genes show bivalent histone modifications in melanocytes or their precursors, identifying these loci should yield important information on those that are epigenetically labile and potentially more amenable to alteration by pharmaceutical treatments. In embryonic stem cells, these bivalent promoters are frequently bound by the TET1 protein, which catalyzes the formation of hydroxymethylcytosine from 5-methylcytosine (Branco et al., 2012; Neri et al., 2013). The amount of this modification present in a genome has been found to be correlated with Polycomb-mediated repression in many normal tissues and tumor samples (Haffner et al., 2013). Hydroxymethylation has been shown to be significantly decreased in melanoma on a genome-wide scale, when compared with benign nevi (Lian et al., 2012). This is unexpected, since EZH2 is expressed at high levels in melanomas, but it could represent uncoupling of the Polycomb-TET1 regulatory machinery due to depletion of IDH2, which is responsible for the formation of α-ketoglutarate, a necessary cofactor for hydroxymethylcytosine formation (Xu et al., 2011). Since reintroduction of either TET1 or IDH2 into models of melanoma results in slowed tumor growth, pharmaceutical agents that can modulate these pathways could be successful in melanoma therapy.

The interaction between polycomb group proteins and noncoding RNAs has been shown to play a role in melanoma. For example, the lncRNA HOTAIR is expressed at high levels in metastatic melanoma, while its depletion reduced the invasive potential of melanoma cells (Tang et al., 2013). In contrast, several miRNAs that are instrumental in decreasing the expression of EZH2 are found at very low levels in melanoma as compared to melanocytic nevi, and overexpression of these transcripts decreases melanoma cell invasion, migration, and proliferation (Asangani et al., 2012; Luo et al., 2013a; Luo et al., 2013b). The expression of miR-200c downregulates BMI-1 and inhibits both the proliferation and migration of melanoma cells (Liu et al. 2012).

Summary and future research directions

Advances in the understanding of melanocyte development from the neural crest have previously emphasized the discovery and characterization of transcription factors, receptors, and their ligands whose activities in model organisms are connected in networks to drive the specification, differentiation, and migration of pigment cells. However, why certain transcription factors activate gene expression at critical loci in melanoblasts but not in other cell types is largely not understood. Exploring how PcG proteins and other factors modifying chromatin function to render important regulatory sequences accessible to gene activation represents the next level of complexity in the investigation of melanocyte development.

Recent exciting advances in melanoma therapeutics have leveraged discoveries in kinase pathway activation in melanoma cells and their susceptibility to immune checkpoint blockade. In our opinion, it is unlikely that inhibitors of EZH2 or PcG protein function more broadly alone will surpass the efficacy of highly-selective inhibitors of BRAF or antibodies blocking CTLA-4 or PD-1 interactions on T cells. Nevertheless, it is quite possible that epigenetic inhibitors could fulfill important adjunct roles in melanoma therapeutics. Their potential to reactivate expression of important tumor suppressor genes or promote genome stability will be important to evaluate as additional strategies are sought to stave off resistance to highly-selective BRAF inhibitors in tumors harboring activating BRAF mutations, or for driving the re-expression of potential tumor antigens rendering tumors more susceptible to cancer immunotherapy.