Skip to main content

Crosstalk between microRNAs and epigenetics in melanocyte differentiation: A new approach to unravel the complexity of melanoma progression and metastasis

Final Report Summary - EPIMIRMEL (Crosstalk between microRNAs and epigenetics in melanocyte differentiation: A new approach to unravel the complexity of melanoma progression and metastasis.)

Melanoma is the most aggressive form of skin cancer and one of the most invasive tumor types. Melanoma incidence keeps increasing worldwide and is expected to double every 20 years in the absence of lifestyle changes. Despite remarkable advances that have led to recent FDA approval of new compounds, outcomes for metastatic melanoma patients remain poor. Unfortunately, relapse with treatment-refractory disease (MAPK inhibitors) and adverse side effects and eventual emergence of resistance upon immunotherapy are common occurrences (Lo and Fisher 2014; Macdonald et al. 2015; Dranoff 2013). The lack of treatments with durable responses may be due, at least in part, to an incomplete understanding of the molecular mechanisms that regulate tumor initiation and/or progression to metastasis. Therefore, there is an urgent need to find new strategies to expand our knowledge of melanomagenesis.
Strong evidence supports the notion that melanoma cells behave like stem cells (reviewed in (Zabierowski and Herlyn 2008)): they frequently exhibit characteristics of neural crest or embryonic stem cells (Hendrix et al. 2003), retain their morphologic and biological plasticity despite repeated cloning (Kath et al. 1991), often express developmental genes, and they can differentiate into a wide range of cell lineages, including neural, mesenchymal, and endothelial cells. Complementing these findings, gene expression profiling of melanoma has shown that tumor progression and increased invasive behavior correlate with loss of differentiation features (Ryu et al. 2007) and expression of neural crest markers (Uong and Zon 2010; White et al. 2011).

We reasoned that studying a model of differentiation of melanocytes from neural crest stem cells could be useful to identify the mechanisms that modulate melanoma stem-like properties, probably among the most significant contributing features to the tumor aggressiveness. The plasticity and reversibility of the epigenetic mechanisms make them ideal orchestrators of these dynamic processes. Then, we hypothesize that epigenetically controlled miRNAs could be involved in normal differentiation and in tumorigenesis when altered. Thus, aberrant epigenetic modifications could be responsible for reactivation (or retention) of stem-like properties in melanoma cells.
Our initial strategy to study the epigenetic changes occurring during differentiation was to develop a melanocyte differentiation model. We tried different protocols, but they failed to render melanocytes. As alternative strategy, we use DNA from stem cell lines, neural crest cell lines and normal melanocytes to generate the DNA methylation profiles to perform the analyses proposed in our project. As planned, we identified and validated differentially methylated regions comprising not only miRNAs but also protein-coding genes. We integrated those results with DNA methylation profiles from human melanoma patients to identified the best candidates with a potential role during melanoma progression and metastasis. Finally, we perform functional validation both in vitro and in vivo, using melanoma cell lines and mouse metastasis models, respectively, as planned in our project.
Comparing the DNA methylation profiles from stem cells, neural crest cells (NCCs) and melanocytes, we identified several miRNAs epigenetically regulated during differentiation. Next, we analyzed DNA methylation profiles in 109 primary and 364 metastasis melanoma samples available in The Cancer Genome Atlas (TCGA), in the hope of identifying mirroring epigenetic changes during metastatic progression and melanocytic differentiation. Intriguingly, metastatic melanoma samples recapitulate the methylation profile that we found in neural crest cell lines. The most relevant findings were miR-30d, miR-24-1 and miR199a. miR-30d and miR-24-1 were hypermethylated in NCCs (vs. melanocytes) and metastatic melanoma (vs. primary). In both cases, miRNA methylation was accompanied by decreased expression. Moreover, we confirmed these findings in short term cultures (STCs) generated from melanoma metastatic samples at the NYU Melanoma Program. Next, focused in strongly validated miRNA-target interactions using reporter assays, one of the most striking result was the increased expression of SNAI1/Snail (known miR-30d target (Ye et al. 2015; Cai et al. 2016)) in metastatic melanoma. Extensive studies have shown the crucial role of Snail family triggering epithelial–mesenchymal transition (EMT) and conferring migratory and invasive abilities to both neural crest cells during development (Nieto 2002; Bronner 2012; Sauka-Spengler and Bronner-Fraser 2008) and cancer cells during tumor progression (Nieto 2002; Baulida and Garcia de Herreros 2015; Poser and Bosserhoff 2004; Hemavathy, Ashraf, and Ip 2000). Thus, our results revealed that CpG hypermethylation-associated silencing of miR-30d is co-opted for metastatic melanoma cells to enhance migratory and invasive properties in part through increased SNAI1 expression, a mechanism that could inherently occur in neural crest. Regarding miR-24-1, it is located in a cluster that also includes miR-23b and miR-27b, and expression of miR-27b was also significantly decreased in metastatic samples. miR-24-1 and other members of the cluster have been recently described as tumor suppressors in bladder (Inoguchi et al. 2014) and prostate cancer (Goto et al. 2014; Ishteiwy et al. 2012). However, to our knowledge, epigenetic regulation and functional role of miR-24-1 in melanoma progression have not been previously reported. Concerning the identified hypomethylated CpGs in NCCs vs. melanocytes, six of them resided in miR-199a loci, four in miR-199a1 and two in miR-199a2, located in chromosomes 19 and 1, respectively. The fact that epigenetic regulation mechanisms are occurring simultaneously in two distant genomic locations to ensure miR-199a expression in neural crest and metastasis, suggests a critical role of miR-199a in these settings. Interestingly, a recent report has identified miR-199a as endogenous promoter of metastatic invasion, angiogenesis, and colonization in melanoma, being ApoE one of the most relevant targets (Pencheva et al. 2012). When we add miR-199a methylation in the scenario, we found that miR-199a1 and miR-199a2 methylation status were significantly correlated with ApoE expression in melanoma metastasis. Thus, we propose epigenetic regulation of miR-199a1 and miR-199a2 as a mechanism to block the metastasis-suppressive role of ApoE in melanoma metastasis. The invasive features conferred by miR-199a expression are advantageous for both, neural crest and melanoma metastatic cells. The link between epigenetic regulation of the described miRNAs in stemness-differentiation and melanoma progression has not been previously reported for any of them. Thus, although we had to develop a parallel strategy to the one initially proposed, we have identified epigenetically regulated miRNAs with a dual role in differentiation and melanoma progression, in agreement with our working plan. Epigenetic mechanisms regulating miRNA expression during development co-opted during metastasis confer neural crest stem-like features that drive invasiveness, aggressiveness and metastatic potential of melanoma cells.
Moreover, although our project was initially focused in miRNAs, considering the interesting findings we decided to broad our study to protein-coding genes. Analogous to the miRNA strategy, we overlap comparative DNA methylation profiles in neural crest cells vs. melanocytes and metastatic vs. primary melanoma samples to identify developmental programs reinstated during melanoma metastasis, beyond miRNAs. Interestingly, the five most significantly hypomethylated CpGs in NCCs and melanoma metastasis all reside in a nuclear receptor essential for embryonic development and transcriptional regulator of neural crest cells.

Unsupervised clustering analysis revealed a complete lack of methylation in stem and NCCs in contrast to complete methylation in melanocytes, associated to loss of expression. A significant CpG hypomethylation was also detected in metastasis vs. primary melanoma samples in the TCGA cohort (n=473) and validated in an independent cohort (n=76). Integration of CpG methylation and mRNA expression TCGA datasets revealed a significant inverse correlation suggesting that DNA methylation regulates the expression in these tumors. Epigenetic regulation was then confirmed in MeWo methylated melanoma cell line, by restored expression upon treatment with DNA demethylating agent. Collectively, these data demonstrated that this gene is epigenetically regulated during melanocyte differentiation and melanoma progression. The reciprocal patterns of methylation and expression in NCCs/melanocytes versus primary/metastatic melanoma, support the concept of acquisition or reactivation of neural crest stem cell features by metastatic melanoma cells and suggest the possibility that its hypomethylation and expression contributes to such ‘reprogramming’ effect. Thus, we hypothesized that epigenetic activation could contribute to melanoma progression and metastasis. To test this hypothesis, we performed loss-of-function (LOF) and gain-of-function (GOF) in vitro and in vivo experiments. In melanoma cells carrying hypomethylation and expression, shRNA-mediated silencing had no effect on proliferation, but significantly suppressed the ability to form colonies in soft agar (a measure of anchorage-independent growth) and to form spheres from single cell suspensions in low attachment plates (a property of stem cells). Conversely, ectopic overexpression in melanoma cells harboring locus hypermethylation (MeWo) resulted in enhanced anchorage-independent growth and sphere formation.
In in vivo assays in mouse, gene silencing had no effect in tumor growth upon subcutaneous injection. However, we observed a significant reduction in the metastatic spread of sh-knockdown melanoma cells instilled by ultrasound-guided intra-cardiac injection. Conversely, the overexpression promoted melanoma metastasis. Overall, our results indicate a critical role of this factor in melanoma progression and metastasis, strongly supported by clinic-pathological data in melanoma patients and in vivo mouse models.

Next, in order to dissect the underlying mechanism of its pro-metastatic behavior, we performed and integrated RNAseq and ChIPseq experiments in our LOF and GOF cellular models. Our data support a model in which highly localized DNA methylation exquisitely controls the expression during development of a truncated isoform of the nuclear receptor, which lacks a DNA binding domain. We proposed that this mechanism acts as an on/off switch that tightly regulates the transcriptional activity of the full length nuclear receptor, likely by shifting the balance from homodimers to heterodimers, which may harbor reduced DNA binding affinity. In addition to broad our knowledge in melanomagenesis, characterization of epigenetically-mediated alterations in melanoma promises to be a good strategy to identify prognosis biomarkers, but also may provide new targets for therapeutic intervention. Nuclear receptors are among the top four families of drug targets and account for approximately 13% of all Food and Drug Administration (FDA)-approved drugs (Overington, Al-Lazikani, and Hopkins 2006). The nuclear receptor function could be potentially modulated by small molecules, which compete with endogenous ligands to act as either antagonists or agonists. Thus, this project not only will increase our scientific knowledge, but also could have significant clinical and socio-economic contribution in melanoma management.
In summary, our unbiased approach by comparing DNA methylation changes occurring in the differentiation of neural crest cells to melanocytes and those occurring from primary to melanoma metastasis, has revealed developmental mechanisms which are co-opted by melanoma cells to promote tumor aggressiveness. We have successfully identified epigenetically regulated miRNAs, including miR-24-1, miR-30d, miR-199a1 and miR-199a2 that target genes controlling migration and invasive abilities during both development and tumorigenesis, as SNAI1 and ApoE. Moreover, we have identified an epigenetically regulated developmental program lead by a neural crest master regulator that is reinstated during melanoma metastasis. We have described a novel mechanism where a truncated isoform whose expression is tightly controlled by a developmental epigenetically-regulated switch, is in turn able to regulate DNA binding of the corresponding full-length isoform and its downstream gene expression program. This elegant, multilayered mechanism involves also miRNAs as the nuclear receptor is able to regulate metastasis-suppressor miRNAs such as miR-181a2.