Role of nodal signaling and the microenvironment underlying melanoma plasticity


  • Lynne-Marie Postovit,

    Corresponding author
    1. Children’s Memorial Research Center, Cancer Biology and Epigenomics Program, Robert H. Lurie Comprehensive Cancer Center, Northwestern University’s Feinberg School of Medicine, Chicago, IL, USA
    2. Department of Anatomy and Cell Biology, Schulich School of Medicine and Dentistry, University of Western Ontario, London, ON, Canada
    Search for more papers by this author
  • Naira V. Margaryan,

    1. Children’s Memorial Research Center, Cancer Biology and Epigenomics Program, Robert H. Lurie Comprehensive Cancer Center, Northwestern University’s Feinberg School of Medicine, Chicago, IL, USA
    Search for more papers by this author
  • Elisabeth A. Seftor,

    1. Children’s Memorial Research Center, Cancer Biology and Epigenomics Program, Robert H. Lurie Comprehensive Cancer Center, Northwestern University’s Feinberg School of Medicine, Chicago, IL, USA
    Search for more papers by this author
  • Mary J. C. Hendrix

    1. Children’s Memorial Research Center, Cancer Biology and Epigenomics Program, Robert H. Lurie Comprehensive Cancer Center, Northwestern University’s Feinberg School of Medicine, Chicago, IL, USA
    Search for more papers by this author

*Address correspondence to Lynne-Marie Postovit, e-mail: or Mary Hendrix, e-mail:


The incidence of melanoma has increased dramatically over the last 50 yr, and although melanoma accounts for only 10% of all skin cancers, it is responsible for over 80% of skin cancer deaths. Recent studies have uncovered critical molecular events underlying melanocytic transformation and melanomagenesis. Among these noteworthy observations are the acquisition of stem cell-associated proteins, such as the Notch receptors and Nodal, which have also been implicated in melanoma progression. For example, we have demonstrated that Nodal expression is limited to invasive vertical growth phase and metastatic melanoma lesions, and that inhibition of Nodal signaling promotes the reversion of metastatic melanoma cells toward a more differentiated, less invasive non-tumorigenic phenotype. In addition, molecular cross-talk exists between the Notch and Nodal signaling pathways. Interestingly, the acquisition of stem cell-associated plasticity is often acquired via epigenetic mechanisms, and is therefore receptive to reprogramming in response to embryonic microenvironments. Here, we review the concept of melanoma plasticity, with an emphasis on the emerging role of Nodal as a regulator of melanoma tumorigenesis and progression, and present findings related to epigenetic reprogramming.

Melanoma defined

Melanoma arises from the transformation of neural crest derived melanocytes that reside in the basal layer of the epidermis. Normally, melanocytes are evenly dispersed into epidermal-melanin units, consisting of approximately 36 keratinocytes per melanocyte. Each melanocyte transfers pigment-containing melanosomes to the keratinocytes in its unit via dendritic processes. The melanin contained in these melanosomes absorbs and scatters ultraviolet radiation, thereby shielding the nucleic acids in the skin from damage. Interestingly, in vitro skin models have demonstrated that the keratinocytes dynamically regulate this process by controlling dendrite growth, melanocyte proliferation and melanin production (Chin, 2003; Hsu et al., 2002).

During melanoma progression there is a general dysfunction of this complex epidermal-melanin unit. Initially, the melanocyte:keratinocyte ratio increases, resulting in the formation of common nevi (moles) which may lead to dysplastic nevi with structural atypia. Dysplastic nevi can subsequently progress into a radial growth phase (RGP) melanoma, characterized by lateral growth that is largely confined to the epidermis. Radial growth phase tumors may then acquire the ability to invade into the dermis and subcutaneous tissue, to form a vertical growth phase (VGP) melanoma. Histologically, VGP melanomas are best characterized as expansive nodules of malignant cells that have penetrated the epidermal basement membrane. Unlike RGP melanomas, which remain dependent on keratinocyte-derived growth factors and cannot undergo anchorage-independent growth, VGP melanomas have escaped keratinocyte control, can undergo anchorage-independent growth and have acquired metastatic competency (Bennett, 2008; Chin, 2003; Hsu et al., 2002). Metastatic melanoma is characterized by a high mortality rate of over 80% and a median survival of only 7.5 months. In the last 40 yr, the incidence of melanoma in the USA has increased by 15-fold and cutaneous melanoma has become the most common cancer afflicting young adults. When diagnosed prior to the onset of VGP disease, melanoma is generally curable with surgery (Chin et al., 2006; Chudnovsky et al., 2005). However, patients with metastatic melanoma have few clinical options, because of a high resistance to therapy, exacerbated by a very rapid disease progression. It is therefore imperative that the molecular events that characterize the melanocytic neoplasia be determined, so that targets for early detection and intervention can be developed.

Multipotent melanoma cells

There is a notable body of literature detailing the molecular signature of aggressive melanoma cells and embryonic stem cells (ESCs), which reveals an intriguing similarity in the pluripotent gene expression patterns that characterize these cell types (Hendrix et al., 2007). In the field of melanoma research, studies have utilized comparative global gene analyses to decipher some of the major gene expression patterns that arise as a consequence of genomic and epigenetic transforming events and that characterize the transition of melanocytes to poorly and then highly metastatic melanoma cells (Bittner et al., 2000; Hendrix et al., 2003; Hoek, 2007; Smith et al., 2005). With this approach, we and others have shown that aggressive melanoma cells manifest a functional plasticity characterized by the simultaneous expression of genes from a variety of cell types, including stem cells, concomitant with a reduction in the expression of genes specific to their parental cell lineage (Table 1). For example, aggressive melanoma cells aberrantly express genes (and proteins), such as Vascular Endothelial Cadherin (VE-Cadherin), which are normally associated with endothelial cells, and also express Keratins, which are intermediate filaments characteristically associated with epithelial cells (Hendrix et al., 2003). Furthermore, the expression of melanocyte-specific markers is dramatically reduced, and sometimes absent, in aggressive melanoma cells: Melan A is reduced by more than fivefold and Tyrosinase, which catalyses the conversion of tyrosine to the pigment melanin, is reduced by more than 35-fold in aggressive melanomas relative to their poorly aggressive counterparts (Hendrix et al., 2003). In addition, reduced Tyrosinase levels are associated with immune evasion (Takeuchi et al., 2003). Collectively, this gene expression pattern confers a functional plasticity upon aggressive melanoma cells that enables them to thrive and metastasize. For example, VE-Cadherin expression by melanoma cells is essential for the formation of tumor-derived vascular networks, thought to provide the tumor with a paravascular perfusion pathway, while the expression of Keratins is associated with enhanced invasion and metastasis (Hendrix et al., 1992, 2001).

Table 1.   Molecular profile of aggressive melanoma cells expressing multiple cellular phenotypesa
  1. aAltered gene expression in human melanoma cells was identified by cDNA microarray analysis and confirmed by Western blot.

  2. bSelected genes are reported as a ratio for highly aggressive C8161 human cutaneous melanoma cells compared to poorly aggressive, isogenically matched C81-61 cells.

ESM-1Endothelial surface molecule44
VE-CadherinEndothelial adhesion molecule10.7
TIE-1Endothelial protein receptor>100
EphA2 (Eck)Epithelial cell kinase77
Keratins 7,8,18Epithelial intermediate filaments20–80
Mart-1/Melan AMelanocyte antigen5.1↓
LSP1Lymphocyte specific protein4.8
HCLS1Hematopoietic lineage protein10
KITStem cell factor receptor2.5
NodalEmbryonic stem cell marker and morphogen20
NotchStem cell marker and receptor5.3

Aggressive melanomas also express stem cell-associated proteins (including the Notch receptors, CD133, Wnt-5a, and Nodal) which have been shown to play a role in the maintenance of pluripotency (Balint et al., 2005; Frank et al., 2005; Hendrix et al., 2003; Hoek et al., 2004; Weeraratna et al., 2002). These intriguing findings support the premise that aggressive melanoma cells acquire a multipotent, plastic phenotype, a concept that challenges our current thinking of how to target tumor cells with stem cell-like properties. Indeed, while previous therapeutic strategies have focused on eliminating a homogeneous tumor expressing traditional biomarkers, new treatment modalities should attempt to target a heterogeneous population of cancer cells whose stem cell-like phenotype facilitates adaptation and consequently survival (Hendrix et al., 2007). Therefore, pluripotency-promoting pathways, which maintain tumor cell plasticity, would be ideal targets for early diagnosis and therapeutic intervention.

Nodal as a melanoma plasticity biomarker

Recent studies in our laboratory have revealed a new regulator of melanoma plasticity and tumorigenicity, called Nodal (Topczewska et al., 2006). Nodal is a member of the Transforming Growth Factor Beta (TGF-β) superfamily and is a pivotal inhibitor of hESC differentiation (James et al., 2005; Mesnard et al., 2006; Vallier et al., 2005). Indeed, Nodal has been shown to maintain the pluripotency of ESCs and is one of the first genes to be down-regulated as totipotent hESCs differentiate during embryoid body formation. Moreover, inhibition of the Nodal signaling pathway, through pharmacological inhibition of its receptor, results in hESC differentiation (Vallier et al., 2004). We recently discovered that Nodal expression is positively associated with melanoma tumor progression: As indicated by Western blot analysis, tumorigenic melanoma cells lines (C8161, WM793, and 1205Lu) express high levels of Nodal, whereas Nodal is absent in normal melanocytes and in non-tumorigenic melanoma cells (C81-61) (Figure 1A). Nodal expression is also positively correlated with melanoma progression clinically (Figure 1B–E). Indeed, immunohistochemical analysis has shown that Nodal protein is absent in normal skin and rare in poorly invasive RGP melanomas. This is in contrast to invasive VGP melanomas and melanoma metastases where Nodal expression is detectable in up to 60% of cases. We have also demonstrated that Nodal plays an instrumental role in the maintenance of melanoma cell plasticity and tumorigenicity (Topczewska et al., 2006). Metastatic C8161 melanoma cells re-expressed Tyrosinase, a melanocyte marker, and down-regulated VE-Cadherin and Keratin 8/18, markers of endothelial and epithelial lineages respectively, in response to Nodal inhibition with a Nodal specific Morpholino (MONodal) (Table 2). As a complement to these findings, we utilized an orthotopic mouse model to examine the effect of Nodal inhibition on melanoma tumor formation (Figure 2). Palpable subcutaneous tumors arose within 7 days following the injection of only 250 000 control C8161 cells. In contrast, knocking down Nodal expression resulted in a significant reduction in C8161 tumorigenicity when the same number of cells was injected (Figure 2A). Indeed, a 30% diminution of tumor incidence in addition to a decrease in tumor growth occurred when Nodal expression was inhibited (Topczewska et al., 2006). Previous results indicated that down-regulation of Nodal expression using MONodal lasted for approximately 14 days – during which time there was no significant tumor formation. By 17 days, Nodal was re-expressed in the melanoma cells, and tumorigenicity resumed (Topczewska et al., 2006). To establish a mechanism for the reduction in tumorigenicity, we have examined the effects of this treatment on in vivo tumor cell proliferation and apoptosis (Postovit et al., 2008). Using immunohistochemical staining for Ki67 as a measure of proliferation, and terminal deoxynucleotidyl transferase biotin-dUTP nick-end labeling (TUNEL) as a measure of apoptosis, we determined that inhibition of Nodal expression with MONodal decreases proliferation and increases apoptosis in orthotopic melanoma tumors (Figure 2B). These in vivo data support a role for Nodal in the maintenance of melanoma tumorigenicity and implicate the potential involvement of apoptotic pathways. Furthermore, our data suggest that Nodal is a biomarker of melanoma progression – from a treatable RGP disease to a more aggressive VGP disease, to the presence of metastases.

Figure 1.

 Nodal expression correlates with melanoma progression. (A) Western blot analysis of Nodal in: C8161, human metastatic cutaneous melanoma cells; normal human melanocytes; C81-61, non-tumorigenic melanoma cells (isogenically matched to C8161 cells); WM793 human vertical growth phase cutaneous melanoma cells; 1205Lu, melanoma cells derived from an experimental metastasis of WM793; and A375P, a tumorigenic human cutaneous melanoma cell line. Actin is used as a loading control. Nodal expression is exclusive to the tumorigenic melanoma cells lines. (B–E) Immunohistochemical analysis of Nodal staining (red color) in (B,C) primary cutaneous melanomas and (D,E) cutaneous melanoma metastases. Tumor-associated mast cells are also immunoreactive for Nodal protein. Isotype controls are pictured in the insets. Brown areas are melanotic cells, and bars equal 50 μm.

Table 2.   Summary of biomarker expression in metastatic melanoma cells treated with a morpholino designed to inhibit nodal expression (MONodal) or exposed to a hESC-derived microenvironment (hESC CMTX) as compared to control cellsa
 Nodal (stem cell)Tyrosinase (Melanocyte)Melan A (Melanocyte)VE-Cadherin (Endothelial)Keratin 8/18 (Epithelial)
  1. aRelative expression of biomarkers indicative of melanoma plasticity (Nodal, VE-Cadherin and Keratin 8/18) or a differentiated phenotype (Tyrosinase and Melan A) were assessed using RT-PCR analysis and confirmed by Western blot.

MONodal↑↑↑ ↓↓↓↓↓↓
hESC CMTX↓↓↓ ↑↑↑↓↓↓ 
Figure 2.

 Nodal inhibition diminishes tumorigenicity in an orthotopic model. (A) In vivo tumor formation in a nude mouse injected with C8161 cells treated with either MOControl or MONodal. Previous work has shown that Nodal expression is down-regulated for approximately 14 days using MONodal. Values represent the median tumor volume (mm3) ±interquartile range, and the MOControl and MONodal tumor volumes were significantly different at the time points indicated by an asterisk (*) (n = 5, P < 0.05). (B) Immunohistochemical analysis of Ki67 expression (red/brown) and TUNEL (red fluorescence) staining in orthotopic melanoma (C8161) tumors. Prior to injection into a mouse, cells were treated with MONodal or left untreated (control). Proliferation is indicated by Ki67 staining and apoptotic nuclei were detected with confocal microscopy as red fluorescence staining localized to the nuclei of apoptotic C8161 cells. For the TUNEL analyses, cell nuclei are counterstained blue with DAPI. Bar equals 25 μm.

Given the emerging role of Nodal in melanoma progression, we must understand how this gene is regulated, so that its aberrant expression may be prevented and/or reversed. The Nodal signaling pathway is tightly regulated by a complex array of transcriptional regulators, post-translational modifications and extracellular factors (see Figure 3 for summary). The human Nodal gene, containing three exons, is located on Chromosome 10q22.1. In mice, Nodal expression is enhanced by at least three separate transcriptional regulatory regions, the Node Specific Enhancer (NDE), approximately 10 kb upstream of the gene locus, the Left Side specific Enhancer (LSE), approximately 4 kb upstream of the translational start site; and the ASymmetric Enhancer (ASE), located in the first intron. (Norris and Robertson, 1999; Saijoh et al., 2005; Vincent et al., 2004). Studies have determined that the LSE and the ASE are regulated by Nodal via a positive-feedback loop that culminates in the activation of FoxH1. In contrast, the NDE has been shown to induce Nodal expression in response to Notch signaling (Krebs et al., 2003; Raya et al., 2003). Gene alignments indicate that the human Nodal locus contains similar enhancer elements, so it is likely that human Nodal expression is regulated in a similar manner. Indeed, a positive-feedback loop, similar to that described for the LSE and ASE in mice, has been documented to sustain Nodal expression in human ESCs and, most recently, melanoma cell types (Besser, 2004; Hendrix et al., 2007; Topczewska et al., 2006). Moreover, our preliminary studies indicate that like mouse Nodal, human Nodal is up-regulated by Notch signaling in melanoma cells (Postovit et al., 2007b). There are four known mammalian Notch receptors (Notch1-4) and five ligands (Jagged1, Jagged2, Delta1, Delta3, and Delta4) (Bray, 2006; Pinnix and Herlyn, 2007). The Notch receptors are activated by binding ligands expressed on adjacent cells. Upon activation, the Notch ectodomain is cleaved by a metalloproteinase and the Notch IntraCellular Domain (NICD) is subsequently released as a consequence of γ-secretase-mediated cleavage. The NICD translocates to the nucleus where it interacts with CSL, a protein that binds to the DNA consensus sequence CGTGGGAA and normally inhibits transcription by associating with co-repressor proteins. The NICD generated upon ligand binding competes with these repressor proteins to form a NICD-CSL complex which is recognized by Mastermind/Lag (MAML). This complex initiates transcriptional activation of target genes such as c-myc (Bray, 2006; Krebs et al., 2003; Raya et al., 2003). Of note, the NDE of the Nodal gene contains two CSL binding sites, and this region has been shown to respond to Notch signaling (Krebs et al., 2003; Raya et al., 2003). We have recently determined that inhibiting Notch in metastatic melanoma cells with a γ-secretase inhibitor (DAPT) results in decreased Nodal expression. Moreover, using specific siRNAs, we have found that Notch-4 may preferentially regulate Nodal expression in these cells (Postovit et al., 2007b).

Figure 3.

 Regulation of the nodal signaling pathway. Nodal is secreted from the cell where it can act as an autocrine or paracrine factor. The Nodal precursor is cleaved and activated by the pro-protein convertases (SPC), Pace-4, and Furin. Nodal propagates its signal by binding to heterodimeric complexes between type I (ALK 4/7) and type II (ActRIIB) activin-like kinase receptors, resulting in the phosphorylation and activation of ALK 4/7 by ActRIIB, followed by the ALK 4/7-mediated phosphorylation of Smad-2 and possibly Smad-3. The epidermal growth factor-coreceptor (EGF-CFC), Cripto-1 (Cripto), is often a component of this receptor complex and can enhance Nodal signaling. Phosphorylated Smad-2/3 associates with Smad-4 and translocates to the nucleus where it regulates gene expression through the association with transcription factors such as FoxH1. Nodal up-regulates its own expression by stimulating transcriptional activation at the Left Side specific and Asymmetric Enhancers (LSE and ASE). Extracellular Nodal inhibitors, most notably Lefty A and B (Lefty), spatially and temporally restrict Nodal signaling levels through antagonism of Nodal and/or Cripto-1. Notch receptors become activated following binding to ligands, such as Delta and Jagged, expressed on adjacent cells. Once activated, the intracellular component of the Notch receptor (NICD) is released following a γ-secretase-dependent cleavage. The NICD translocates to the nucleus where it interacts with CSL, a protein that binds to the DNA and inhibits transcription by associating with co-repressor proteins. The NICD competes with these repressor proteins to form a NICD-CSL complex enabling the transcriptional activation of target genes, including Nodal, which has CSL binding domains in its Node Specific Enhancer. Nodal levels may also be regulated by miRNA and possibly DNA methylation of a CpG island.

As a complement to canonical regulators of transcription, Nodal expression is also governed by gene methylation and miRNA-directed degradation. For example, we have determined that there is a sizable CpG island (>1300 base pairs) near the transcription start site (TSS) of the Nodal gene, and that this site may regulate Nodal expression (Postovit et al., 2007a). Moreover, a novel miRNA (miR-430) has been shown to block the translation of a Nodal homolog, squint, in zebrafish (Choi et al., 2007). MiR-430 target sites are also present in the mammalian Nodal gene; and so it is likely that Nodal expression is similarly affected by miRNA-mediated degradation in humans.

Nodal is also regulated post-translationally by subtilisin-like pro-protein convertases, including PACE-4 and Furin (Beck et al., 2002), and by glycosylation. In a manner akin to most TGF-β family members, Nodal is synthesized as a pro-protein that is activated following proteolytic processing by covertases at R-X-(K/R)-R and R-X-X-R consensus sequences (Schier, 2003). Removal of the pro-domain reduces Nodal stability and signaling range, thereby promoting autocrine signaling (Le Good et al., 2005). Conversely, glycosylation of mature Nodal affords the protein with increased stability so that it can potentiate paracrine signaling events (Le Good et al., 2005). Hence, post-translational modifications of Nodal are important mediators of Nodal signaling outcomes.

Nodal propagates its signal by binding to heterodimeric complexes between type I (ALK 4/7) and type II (ActRIIB) activin-like kinase receptors. Assembly of this complex results in the phosphorylation and activation of ALK 4/7 by ActRIIB, followed by the ALK 4/7-mediated phosphorylation of Smad-2 and possibly Smad-3 (outlined in Figure 3). Phosphorylated Smad 2/3 subsequently associates with Smad-4 and then translocates to the nucleus where it regulates gene expression through an association with transcription factors such as FoxH1 and Mixer (Schier, 2003). Genetic studies in zebrafish and mice have defined an essential role for Cripto-1, an Epidermal Growth Factor–Cripto-1/FRL1/cryptic (EGF–CFC) family member, in Nodal function. Indeed, embryological studies have determined that Cripto-1 directly associates with ALK 4 (with its CFC domain) and Nodal (with its EGF domain) and that these associations may be required for Nodal to propagate its signal (Bianco et al., 2002; Yeo and Whitman, 2001). This prerequisite is perhaps best exemplified in Cripto-1 null mice which die at day 7.5 as a result of the inability to gastrulate (a Nodal-dependent phenomenon) (Ding et al., 1998; Liguori et al., 1996). Studies have determined that Nodal may also signal in a Cripto-1-independent fashion. For example, Reissmann et al. (2001) revealed that Nodal can bind to and activate ALK 7 in the absence of Cripto-1, but that Cripto-1 markedly enhances this process. Another study determined that the Nodal precursor can bind to ALK 4 in the extraembryonic ectoderm of the developing mouse embryo in a Cripto-1-independent manner, and that this binding results in the expression of Nodal-responsive genes (Ben-Haim et al., 2006). Finally, using murine knock out models, Liguori et al. (2008) recently demonstrated that Nodal can signal extensively and control axis specification in the absence of Cripto, if its inhibitor Cerberus is also knocked out.

Nodal up-regulates its own transcription via a positive-feedback loop. Hence, to control the levels of this potent morphogen, hESCs also secrete Nodal inhibitors such as Lefty A, Lefty B, Cerberus and Tomoregulin-1 (Schier, 2003; Tabibzadeh and Hemmati-Brivanlou, 2006). Of these factors, the Lefty molecules, highly divergent members of the TGFβ superfamily, are expressed to the greatest extent. In fact, studies have demonstrated that in conjunction with Nodal and Oct 3/4, Lefty A and B are among the most enriched genes expressed in hESCs (Sato et al., 2003; Tabibzadeh and Hemmati-Brivanlou, 2006). Extracellular Nodal inhibitors control Nodal signaling by spatially and temporally restricting the Nodal-mediated activation of ALK 4/7. For example, Lefty A and B specifically antagonize the Nodal signaling pathway by binding to and interacting with Nodal and/or with Cripto-1 in a manner that blocks ALK activation (Schier, 2003; Shen, 2007). This restriction of Nodal signaling can occur in the extracellular microenvironment, where Nodal and sometimes Cripto-1 are present, as well as at the cell surface. Of note, the Lefty proteins have not been found to bind ALK4 or ActRIIB; hence these Lefty proteins are not competitive inhibitors of the ALK receptor complex. Furthermore, in embryological systems, the Lefty genes are often downstream targets of Nodal signaling, thereby providing a powerful negative-feedback loop for this pathway (Schier, 2003; Shen, 2007). In contrast, we have determined that Nodal-expressing melanoma cells do not express Lefty, thereby allowing Nodal signaling to go unchecked in this tumor-associated system (Postovit et al., 2007a, 2008).

The myriad of regulatory mechanisms characterizing the Nodal signaling pathway likely underlies its propensity for aberrant expression in melanoma. However, this complexity also affords a number of putative strategies for the inhibition of Nodal signaling and the circumvention of melanoma progression. One such approach involves the epigenetic silencing of Nodal expression.

Epigenetic reprogramming of multipotent melanoma cells

Although the role of genetic mutations in oncogenic transformation is indisputable, a great deal of evidence suggests that the tumorigenic potential of a transformed cell is also attributable to epigenetic modifications. Unlike genetic changes, epigenetic adjustments do not affect the primary DNA sequence. Rather, they involve interactions among cells and cell products, which lead to alterations in reversible phenomena such as cell signaling and DNA modifications (Postovit et al., 2007a; Rothhammer and Bosserhoff, 2007). Exemplifying the importance of epigenetics in melanomagenesis is a recent study by Jaenisch et al. in which nuclear transplantation of a melanoma nucleus into an oocyte gave rise to ESCs with the capacity to differentiate into non-tumorigenic cell types such as melanocytes and fibroblasts (Hochedlinger et al., 2004). Nuclear transplantation into an oocyte induces a dramatic hypomethylation of the donor DNA, exposing promoters and enabling transcription (Lotem and Sachs, 2006). These phenomena confer the broad developmental spectrum observed when normal or neoplastic somatic nuclei are transplanted. As differentiation and cell fate specification ensue, there is a marked increase in DNA methylation leading to the down-regulation of most genes and a consequential specialization in gene expression (Lotem and Sachs, 2006). In Jaenisch’s study, the tumorigenic potential of the melanoma nuclei was temporarily reversed in response to nuclear transplantation, even though genetic mutations persisted (Hochedlinger et al., 2004). Hence, the mutations which characterized the melanoma genome worked in concert with epigenetic factors to sustain transformation in the donor melanoma cells.

Epigenetic modifications, initiated via microenvironmental factors, have also emerged as major players in melanocyte transformation and transdifferentiation (Bedogni et al., 2005; Carreira et al., 2006). For example, Goding et al. recently determined that the microphthalmia-associated transcription factor, Mitf, epigenetically regulates diaphanous-related formin (Dia1) expression. Dia1, which promotes actin polymerization and coordinates the actin cytoskeleton and microtubule networks at the cell periphery, inhibits invasion and induces cell-cycle arrest. Hence, by regulating Dia1 transcription, alterations in Mitf, which occur in response to microenvironmental factors, influence melanoma cell proliferation and invasion (Carreira et al., 2006). In another study, Bedogni et al. (2005) demonstrated that hypoxia, which characterizes the microenvironment of many solid tumors and has been shown to promote melanoma cell invasion and metastasis, also contributes to melanocyte transformation (Bedogni et al., 2005). This study revealed that constitutively active Akt, which is observed in a high percentage of melanomas, can transform melanocytes exclusively when oxygen levels are low (Bedogni et al., 2005). Furthermore, the skin’s distance from superficial blood vessels renders it mildly hypoxic with oxygen levels between 1 and 5%. This microenvironmental milieu permits melanocytes to stabilize the transcriptional co-factor hypoxia inducible factor 1 alpha (HIF-1α), which promotes hypoxia-associated gene expression. It was discovered that this up-regulation of HIF-1α enhances melanocyte transformation by synergizing with constitutively active Akt to promote anchorage-independent growth in vitro, and tumor formation in vivo (Bedogni et al., 2005). This finding exemplifies how the microenvironment can complement aberrant genetic changes to promote melanomagenesis. Moreover, these findings highlight the importance of epigenetic phenomena (and resultant gene expression patterns) in melanocyte transformation and melanoma progression.

Epigenetic phenomena are theoretically reversible. Hence, the plastic, stem cell-like phenotype of aggressive tumor cells should be receptive to reprogramming (i.e. redifferentiation) (Gerschenson et al., 1986). In support of this concept, embryonic microenvironments have been shown to inhibit the tumorigenicity of a variety of cancer cell lines (Gerschenson et al., 1986; Pierce et al., 1982; Podesta et al., 1984). For example, B16 murine melanoma cells were unable to form tumors and appeared to differentiate toward a neuronal phenotype following exposure to microenvironmental factors derived from the embryonic skin of a developing mouse (Gerschenson et al., 1986). In another set of experiments, Bissell and colleagues documented that Rous sarcoma virus, which causes a rapidly growing tumor when injected into hatched chicks, is non-tumorigenic when injected into 4-day-old chick embryos, despite viral replication and v-src oncogene activation (Dolberg and Bissell, 1984).

More recently, we employed an in vitro 3D model to examine whether the microenvironment of human embryonic stem cells (hESCs) could similarly reprogram the metastatic melanoma cell phenotype (Postovit et al., 2006;Postovit et al., 2008). In this model, hESCs were allowed to ‘condition’ a 3D matrix (CMTX), which would subsequently receive multipotent metastatic melanoma cells. Because the hESCs were removed prior to the addition of melanoma cells, the melanoma cells were exposed only to the extracellular microenvironment of the hESCs, thereby removing the complexity of cell–cell interactions from the vast array of mechanisms that may be working to epigenetically modulate cell behavior. Utilizing this approach we determined that, similar to Nodal inhibition, exposure of melanoma cells to a hESC microenvironment results in the re-expression of Melan-A, a melanocyte specific marker, as well as a reduction in the expression of VE-Cadherin (Table 2) (Abbott et al., 2008; Hendrix et al., 2007; Topczewska et al., 2006). Aggressive melanoma cells exposed to hESC microenvironments also experienced an 87% reduction in Nodal expression concomitant to a significant decrease in tumorigenicity (Postovit et al., 2008). Indeed, exposure to the hESC microenvironment significantly diminished the ability of human metastatic melanoma cells to undergo anchorage-independent growth, a phenomenon that was rescued by the inclusion of rNodal (100 ng/ml). Moreover, exposure of these cells to hESC-derived CMTX resulted in an inhibition of tumor growth in an orthotopic mouse model (Hendrix et al., 2007; Postovit et al., 2008). In a manner similar to Nodal inhibition, we found that exposure to hESC CMTX decreased proliferation and increased apoptosis in the orthotopic tumors (Postovit et al., 2008), implicating the potential involvement of apoptotic pathways in the tumor suppressive effects of the hESC microenvironment. Collectively, these findings illuminate the remarkable ability of hESC-derived factors to inhibit melanoma tumorigenicity and suggest that this tumor-suppressive phenomenon is mediated via an inhibition of Nodal expression and signaling.

The ability of hESCs to reprogram aggressive melanoma cells is reversible over time (Abbott et al., 2008; Postovit et al., 2007a, 2008). As such, this phenomenon is likely because of epigenetic alterations, such as DNA methylation. Given the location of a sizable CpG island near the TSS of the Nodal gene, together with the marked down-regulation in Nodal expression observed in melanoma cells exposed to hESC microenvironments, we hypothesized that the Nodal CpG island is differentially methylated in cells exposed to a hESC microenvironment. Using bisufite-sequencing technology, we determined that exposure of aggressive melanoma cells to matrices conditioned by hESCs resulted in a marked increase in site-specific methylation in the Nodal CpG island. Although hESC microenvironments did not drastically affect global methylation, we observed specific areas, in the first half of the CpG island, where a 32% increase in DNA methylation occurred (Postovit et al., 2007a). Sequence analyses determined that these areas contain putative consensus sequences for transcription factors including Sp1, Egr-1, and GATA-4. It is therefore plausible that hESC-derived microenvironments can alter Nodal expression in melanoma cells by epigenetically methylating transcription factor binding sites. These modifications may canonically decrease the accessibility of the Nodal promoter for transcriptional activators, thereby decreasing Nodal expression commensurate with differentiating melanoma cells and abrogating their tumorigenicity. Our ongoing studies will further explore this possibility to decipher some of the epigenetic mechanisms by which embryonic microenvironments reprogram aggressive cancer cells.

Conclusions and future perspectives

There is consensus in the melanoma research community and affiliated patient advocacy groups that new therapeutic strategies are needed to treat advanced stages of this disease. However, one of the greatest obstacles to achieving success has been a lack of understanding of the basic biology underlying the oncogenic transformation of melanocytes, the aggressive plasticity of aggressive melanoma, and the possible epigenetic regulators of the metastatic phenotype. What we have come to appreciate is that aggressive melanoma cells share many characteristics with embryonic progenitor cells. In fact, our work has illuminated a new pathway in melanoma which is directly related to tumor cell plasticity, and represents the convergence of embryonic and tumorigenic signaling pathways – via Nodal, a member of the TGF-β superfamily responsible for the pluripotency of hESCs.

Protein and immunohistochemical analyses of Nodal demonstrate that this embryonic morphogen is aberrantly expressed in melanoma cells with tumorigenic potential and in VGP and metastatic lesions. These observations suggest that Nodal expression may be associated with the acquisition of an aggressive phenotype in melanoma. Additional studies are needed to determine the prognostic value of Nodal as a new biomarker for disease progression. Compelling evidence supporting a direct relationship between Nodal expression, tumorigenicity and plasticity is provided in Morpholino experiments showing that when Nodal was down-regulated in aggressive melanoma cells, tumor formation was significantly diminished and apoptosis was induced. However, tumorigenicity resumed when Nodal was re-expressed in these same melanoma cells. Equally noteworthy is the finding linking down-regulation of Nodal expression in melanoma to loss of plasticity markers, such as VE-Cadherin and Keratin 8/19, and the re-expression of melanocyte differentiation pathway specific genes such as Tyrosinase and Melan-A. Future studies will focus on new strategies to down-regulate Nodal expression for a longer duration. In addition, the molecular cross-talk revealed between Notch-4 and Nodal may provide novel approaches to targeting a broader signaling pathway underlying melanoma aggressiveness and plasticity.

Based on the plastic phenotype of aggressive melanoma cells and their similar characteristics to stem cells, we tested the possibility that the microenvironment of hESCs could reprogram the multipotent phenotype. This unique experimental approach generated important clues related to the epigenetic reprogramming of plastic melanoma cells, including an 87% reduction in Nodal expression and a 32% increase in DNA methylation near the transcriptional start site of the Nodal gene. These exciting results have stimulated additional studies related to the identification of factor(s) in the hESC microenvironment that might contribute to this important reprogramming. As a consequence of exposure to the normal hESC microenvironment, the melanoma cells lost plasticity markers, regained melanoma differentiation markers, and underwent apoptosis; similar to that described for Nodal down-regulation. Collectively, these observations begin to elucidate a new pathway in melanoma progression that deserves additional scientific scrutiny. Targeting Nodal and associated pathways with molecular cross-talk may provide valuable new insights into managing the plastic melanoma phenotype.


This work was supported by grants from the Illinois Regenerative Medicine Institute, U.S. National Institutes of Health (CA50702 and CA121205) and Charlotte Geyer Foundation to M.J.C.H., and a Canadian Institutes of Health Research Post-doctoral Fellowship to L.M.P. The authors wish to thank Drs Brian Nickoloff and Bento Soares for helpful scientific discussions. We apologize to those colleagues whose studies were not cited in this review because of space limitations.