QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Chin, L. Articles by Fisher, D. E. Search for Related Content PubMed PubMed Citation Articles by Chin, L. Articles by Fisher, D. E. Related Content Cancer and Disease Models Social Bookmarking                   What's this?

REVIEW

Malignant melanoma: genetics and therapeutics in the genomic era

¹ Melanoma Program, Dana-Farber Cancer Institute, Boston, Massachusetts 02115, USA; ² Department of Dermatology, Harvard Medical School, Boston, Massachusetts 02115, USA; ³ Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts 02115, USA; ⁴ The Broad Institute of Harvard and Massachussetts Institute of Technology, Cambridge, Massachusetts 02142, USA; ⁵ Department of Pediatric Hematology/Oncology, Dana-Farber Cancer Institute and Children's Hospital of Boston, Boston, Massachusetts 02115, USA

	Abstract

Top Abstract The disease The genetics of melanoma The genomics of melanoma The biology of melanoma--lessons... Emerging paradigms for melanoma... Acknowledgments References

 
Cell for cell, probably no human cancer is as aggressive as melanoma. It is among a handful of cancers whose dimensions are reported in millimeters. Tumor thickness approaching 4 mm presents a high risk of metastasis, and a diagnosis of metastatic melanoma carries with it an abysmal median survival of 6–9 mo. What features of this malignancy account for such aggressive behavior? Is it the migratory history of its cell of origin or the programmed adaptation of its differentiated progeny to environmental stress, particularly ultraviolet radiation? While the answers to these questions are far from complete, major strides have been made in our understanding of the cellular, molecular, and genetic underpinnings of melanoma. More importantly, these discoveries carry profound implications for the development of therapies focused directly at the molecular engines driving melanoma, suggesting that we may have reached the brink of an unprecedented opportunity to translate basic science into clinical advances. In this review, we attempt to summarize our current understanding of the genetics and biology of this disease, drawing from expanding genomic information and lessons from development and genetically engineered mouse models. In addition, we look forward toward how these new insights will impact on therapeutic options for metastatic melanoma in the near future.

[Keywords: Development; genetics; genomics; melanoma; therapeutics]

	The disease

Top Abstract The disease The genetics of melanoma The genomics of melanoma The biology of melanoma--lessons... Emerging paradigms for melanoma... Acknowledgments References

In view of the epidemiological and emerging experimental evidence linking melanoma incidence to UV exposure and skin phototype, prevention and screening strategies represent key areas for reduction of disease incidence and severity. For most cutaneous melanomas in the so-called radial growth phase (e.g., thin melanomas), surgical removal affords curative treatment. In contrast, a significant fraction of patients diagnosed with intermediate-thickness (2–4 mm) cutaneous melanoma eventually succumb to recurrence at regional or distant sites.

Critical biological questions facing the melanoma research community include: (1) What genetic and environmental factors contribute to and/or modulate risk of melanoma development in man? (2) What biological or molecular features (biomarkers) in early lesions can predict high risk of subsequent metastasis? (3) What genetic events underlie its propensity for metastasis and treatment resistance (phenotype)? (4) Which genetic alterations responsible for development and progression of melanoma are also essential for maintenance of established disease? (5) Finally, what maintenance-essential biological or molecular pathways/networks might prove amenable to preventive and/or therapeutic intervention in man?

	The genetics of melanoma

Top Abstract The disease The genetics of melanoma The genomics of melanoma The biology of melanoma--lessons... Emerging paradigms for melanoma... Acknowledgments References

 
Many studies conducted over several decades on benign and malignant melanocytic lesions as well as melanoma cell lines have implicated numerous genes in melanoma development and progression. In this review, we focus on validated genetic events, which include predisposing or somatic structural alterations in melanoma specimens on the DNA level, such as translocation, amplification/deletion, and point mutations (Futreal et al. 2004).

CDKN2A, the familial melanoma locus

Physical characteristics such as light complexion, an inability to tan, red hair, and blue eyes correlate with increased risk for melanoma development (Gilchrest et al. 1999). Likewise, the presence of multiple pigmented lesions, including freckles and banal or clinically atypical moles, is associated with a quantitatively increased melanoma risk (Gandini et al. 2005a). However, one of the most significant risk factors for melanoma occurs in individuals with a strong family history of melanoma. Older case-control studies of patients with familial atypical mole-melanoma (FAMM) syndrome suggested an elevated risk of 434-to 1000-fold over the general population (Greene et al. 1985). A more recent meta-analysis of family history found that the presence of at least one first-degree relative with melanoma increases the risk by 2.24-fold (Ford et al. 1995; Gandini et al. 2005b). Genetic studies of this heritable trait in large melanoma-prone families ultimately led to the initial identification of CDKN2A as the familial melanoma gene. Located at chromosome 9p21, LOH or mutation at this locus cosegregated with melanoma susceptibility in familial melanoma kindred (Hussussian et al. 1994; Kamb et al. 1994b), and 9p21 homozygous deletions centered on CDKN2A were frequently observed in cancer cell lines of different types (Kamb et al. 1994a; Nobori et al. 1994). The CDKN2A story is one illustrative example of how rare inherited cancer syndromes can provide key insights into the genes and pathways relevant to malignant transformation of specific cell lineages in both familial and sporadic settings.

The genomic complexity of the 9p21 locus fueled considerable debate on the relative importance of the several overlapping proteins encoded therein. For example, a significant proportion of the 9p21 deletions also included the neighboring CDKN2B gene (for review, see Haluska and Hodi 1998; Ruas and Peters 1998). To confirm a causal role of CDKN2A loss in 9p21-mediated melanoma, a gene targeting approach was used to delete exons 2 and 3 of the Cdkn2a gene in the mouse germline (Serrano et al. 1996). These mice succumbed to fibrosarcomas and lymphomas with high frequency (Serrano et al. 1996) and became highly prone to cutaneous melanomas with short latency when combined with an activated H-RAS mutation in their melanocytes (Tyr-RAS ⁺) (Chin et al. 1997). Moreover, 100% of melanomas that arose in Tyr-RAS⁺; Ink4a/Arf ^+/– animals exhibited loss of heterozygosity (LOH) in the remaining wild-type allele (Chin et al. 1997). Specifically, exon 2 sequences were eliminated in all cases, the significance of which relates to the peculiar genomic organization of this gene that is unique in mammalian cells (human, mouse, opossum) (Sharpless et al. 2003; see below).

CDKN2A encodes two distinct proteins, INK4A and ARF, through utilization of alternative promoters and first exons (1 for INK4A and 1 for ARF) (for review, see Chin 2003). The shared second exons of the two transcripts are translated in different reading frames, thus encoding two proteins with no amino acid homology yet each possessing potent anticancer activities. In particular, INK4A (also known as p16^INK4a)—the founding member of the INK4 (Inhibitor of cyclin-dependent kinase 4) family of proteins—inhibits the G1 cyclin-dependent kinases (CDKs) 4/6, which phosphorylate and inactivate the retinoblastoma protein (RB), thereby allowing for S-phase entry (Serrano et al. 1993). Loss of INK4A function therefore promotes RB inactivation through hyperphosphorylation, resulting in unconstrained cell cycle progression. ARF—the alternative reading frame product of the locus (Quelle et al. 1995) (also known as p14^ARF in humans, or p19^ARF in mice)—inhibits MDM2-mediated ubiquitination and subsequent degradation of p53 (Kamijo et al. 1998; Pomerantz et al. 1998; Stott et al. 1998; Zhang et al. 1998); thus, loss of ARF inactivates p53.

INK4A–CDK4/6–RB pathway In humans, germline intragenic mutations have been identified in familial melanoma kindred and melanoma-prone families that specifically inactivate INK4A while preserving ARF (e.g., exon 1 mutations) (Hussussian et al. 1994; Kamb et al. 1994b; FitzGerald et al. 1996). These findings provided the smoking gun fingering this protein as a bona fide melanoma suppressor (for review, see Chin 2003). In addition to the 25%–40% of melanoma-prone families and 0.2%–2% of sporadic melanoma patients harboring INK4A mutations in the coding region (Aitken et al. 1999; Tsao et al. 2000), polymorphism in both 5' and 3' untranslated regions (UTRs) that alter translation or possibly regulate message stability of INK4A as well as promoter and splicing mutations of INK4A have since been identified in association with 9p21-linked melanoma-prone families (Liu et al. 1999; Kumar et al. 2001; for review, see Sharpless 2004). In line with this, mice with specific Ink4a inactivation did show increased susceptibility to carcinogen-induced and spontaneous melanoma (Krimpenfort et al. 2001; Sharpless et al. 2001); however, the phenotype observed was relatively weak. Consistent with this weak melanoma phenotype, it has recently been shown BRAF^E600-induced senescence in nevus melanocytes is in part INK4A-independent, suggesting presence of another melanoma suppressor(s) that must be inactivated for unconstrained melanoma development (Michaloglou et al. 2005).

Germline mutations of CDK4, an Rb-kinase that is inhibited by INK4A, have also been identified in melanoma-prone kindreds (Wolfel et al. 1995; Zuo et al. 1996; Soufir et al. 1998; Tsao et al. 1998; Molven et al. 2005). These mutations, which are a rare cause of familial melanoma, target a conserved arginine residue (Arg24) and render the mutant protein insensitive to inhibition by the INK4 class of cell cycle inhibitors. Melanomas from patients harboring these germline CDK4 mutations do not demonstrate somatic INK4A inactivation, suggesting that INK4A inactivation and CDK4 activation are mutually exclusive. Accordingly, the mutant Cdk4 Arg24Cys (R24C) "knocked-in" mouse mirrored the human situation, in that it developed increased susceptibility to melanomas after carcinogen treatment, while lacking evidence of somatic Ink4a inactivation (Sotillo et al. 2001). Moreover, this R24C knockin mouse readily cooperates with activated HRAS in spontaneous and UV-induced melanoma genesis (Hacker et al. 2006). In line with these data in the mouse, somatic alteration of CDK4, in form of focal amplification, is observed in sporadic melanomas in human (Muthusamy et al. 2006).

Finally, although less commonly associated with familial melanoma than with hereditary retinoblastoma, patients with germline inactivation of the retinoblastoma gene (RB1) are predisposed to melanoma (Draper et al. 1986; Sanders et al. 1989; Eng et al. 1993; Fletcher et al. 2004). In patients cured of bilateral retinoblastoma, the estimates of increased lifetime risk of melanoma range from fourfold to 80-fold (Eng et al. 1993; Fletcher et al. 2004). These melanomas do not necessarily occur in skin associated with radiotherapy treatment ports of the retinoblastoma, and likewise the excess incidence of melanoma in these cohorts has not been reduced by the practice of decreasing radiotherapy for retinoblastoma. In contrast, the lifetime risk of sarcoma in retinoblastoma survivors has markedly decreased in the setting of less primary treatment radiotherapy, suggesting that the bulk of sarcoma risk in these patients is treatment-related. Therefore, the increased risk of melanoma in patients who survive bilateral retinoblastoma appears to result from the stochastic loss of the remaining wild-type RB1 allele in these patients. These data provide further evidence linking the entire p16/CDK/RB axis to melanoma suppression in human.

ARF–MDM2–P53 pathway Inactivation of the p53 pathway appears to be a rite of passage for virtually all tumor cells. While most human solid tumors inactivate this tumor suppression pathway at the level of p53 itself (e.g., within the TP53 coding region), melanoma provides a notable exception to this rule. Numerous surveys have found either rare or absent TP53 point mutations or allelic loss in surgical specimens from primary and meta-static melanomas (for review, see Chin 2003). Moreover, HMD2 amplification (encodes human ortholog of MDM2) occurs in only 3%–5% of human melanomas (Muthusamy et al. 2006), raising questions as to the pathogenetic relevance of the p53 pathway in melanoma. On the other hand, support for a role for p53 has been suggested by Mintz and colleagues, who showed that SV40 T antigen (which inactivates both RB and p53) generates a highly penetrant and aggressive melanoma phenotype (Bradl et al. 1991). More direct evidence for a role of p53 in melanoma suppression derived from Tyr-RAS ⁺ ; Trp53^+/– mutant mice. These mice readily developed cutaneous melanomas characterized by LOH of the wild-type Trp53 allele and retention of Arf (Bardeesy et al. 2001), a mirror image to the molecular profile of the Tyr-RAS⁺; Ink4a/Arf^–/– melanoma model, which retained wild-type p53 in all cases (Chin et al. 1997). These mouse genetic studies highlighted both the relevance of the p53 pathway in melanoma suppression and the reciprocal pattern of Arf–p53 inactivation, thereby suggesting their functional relatedness. Ultimately, elucidation of the role of ARF as a negative regulator of p53 (Kamijo et al. 1998; Pomerantz et al. 1998; Stott et al. 1998; Zhang et al. 1998) offered a mechanistic explanation for the patterns of genetic mutations in human disease, where the lack of p53 mutation in melanoma reflected preferred pathway inactivation at the level of ARF via 9p21 deletion.

A corollary to the above observation is that ARF represents an independent target of genetic inactivation at the 9p21 locus. Although insertions and deletions in ARF exon 1 without affecting INK4A or INK4B expression have been reported in melanoma cell lines (Kumar et al. 1998), the most convincing data on ARF as an independent melanoma suppressor came from two melanoma patients with germline ARF-specific mutations. One patient harbored a 14-kb deletion in exon 1 sparing both INK4A and INK4B genes (Randerson-Moor et al. 2001), while the second patient had a 16-bp insertion in exon 1 generating a frame-shifted ARF mutant defective in cell cycle arrest (Rizos et al. 2001). However, intact INK4A activity was not demonstrated in either case, precluding definitive assignment of INK4A-independent activity to ARF. In addition, in one case of a splice mutation of exon 1 that was detected in a family of melanoma, inactivation of INK4A was acquired somatically in a melanoma from the family, pointing to the requirement for dual inactivation of these two genes (Hewitt et al. 2002). More recently, a cluster of five different germ-line mutations at the exon 1 splice donor site was identified in melanoma pedigrees known to harbor no mutation in INK4A exons or in CDK4 exon 2 (INK4A-binding site) (Harland et al. 2005). These splice site variants were predicted to impair ARF activity, although their causal roles have not been functionally demonstrated. On the other hand, there is some evidence that point mutagenesis of shared exon 2 still preferentially targets the INK4A transcript. Previously, alterations in the shared exon 2 sequences were considered inactivating events for both INK4A and ARF, a notion that is being challenged by a recent survey, which examined reported exon 2 point mutations (both germline and somatic) and found a higher proportion of synonymous changes and a lower percentage of nonsense mutations in the ARF reading frame when compared with the INK4A reading frame (Yang et al. 2005). In silico modeling of sequence variants that yield missense mutations in both reading frames further suggests that the impact of the codon change is more severe for INK4A than ARF (Yang et al. 2005).

Contrary to the equivocal situation in human, Arf is a potent tumor suppressor in vivo in the mouse. While Ink4a deficiency alone resulted in a modest tumor-prone phenotype (Krimpenfort et al. 2001; Sharpless et al. 2001, 2002), this Ink4a-dependent susceptibility was significantly enhanced in the setting of Arf haploinsufficiency (Krimpenfort et al. 2001), arguing for an independent contribution from Arf in Ink4a/Arf-mediated melanoma suppression. Additionally, when crossed onto the melanoma-prone Tyr-RAS transgenic allele, either Ink4a or Arf deficiency facilitated enhanced melanoma formation, although Arf loss was more efficient than Ink4a (Sharpless et al. 2003). Moreover, the resultant melanomas showed reciprocal p53 or RB pathway inactivation. In other words, Rb pathway lesions were observed in melanomas from Tyr-RAS ⁺ ; Arf ^–/– mice, and p53 pathway lesions were detected in Tyr-RAS ⁺ ; Ink4a ^–/– melanomas (Sharpless et al. 2003). This profile provides unequivocal genetic evidence that both products of the Ink4a/Arf locus served prominent and synergistic/independent roles in melanoma suppression in vivo, which corroborated observation in humans (Hewitt et al. 2002). Moreover, this shared function may underlie the need for coordinated regulation imparted in part by their intimate and evolutionarily conserved genomic arrangement (Sharpless and DePinho 2004). This argument also explains another interesting feature of melanoma: the high frequency of 9p21 deletion as opposed to other mechanisms of INK4A loss such as promoter methylation or point mutation. These latter mechanisms only target INK4A, and are more common in tumors such as non-small-cell lung cancer, where direct p53 inactivation occurs. In summary, while INK4A is the preferred target of inactivation in the INK4A–CDK4–RB pathway, extensive biochemical and genetic data suggest an independent role for the ARF–p53 axis in melanoma suppression, with ARF as its preferred target via 9p21 deletion.

Receptor tyrosine kinase (RTKs) activation

A large body of evidence has implicated hyperactive RTK signaling in the development and progression of melanoma. Although most were based on expression alterations, several RTKs map to known regions of recurrent DNA copy number gain or amplification. Moreover, considering the example of c-Kit (see below), it is expected that systematic resequencing efforts will identify activating mutations in these and other RTKs in melanomas.

Late-stage melanomas often exhibit epidermal growth factor receptor (EGFR) overexpression in association with increased copies of chromosome 7 (Koprowski et al. 1985; Bastian et al. 1998; Udart et al. 2001). Enforced activation of EGFR has been associated in metastatic progression in cell-based study (de Wit et al. 1992; Huang et al. 1996b). However, unlike glioblastomas or lung adenocarcinoma (Maher et al. 2001; Sihto et al. 2005), focal amplification and/or mutation of EGFR has not been reported in melanoma. The nonfocal nature of chromosome 7 gains in melanoma renders it impossible to assign EGFR as a target of such genomic alterations. In an inducible HRAS-driven mouse melanoma model (Chin et al. 1999), transcriptome analysis revealed the existence of a RAS-dependent EGFR signaling loop mediated through up-regulation of EGF family ligands (e.g., amphiregulin and epiregulin) (Bardeesy et al. 2005). This EGFR signaling pathway provides important survival signals involving PI3-K-dependent activation of AKT, as sustained EGFR activity is able to prolong viability of established melanoma upon inactivation of RAS. Conversely, inactivation by dominant-negative EGFR abolishes tumorigenicity of RAS-driven melanoma cells, consistent with observations in other cell systems (fibroblasts, keratinocytes, and intestinal epithelial cells) that autocrine EGFR signaling is required for transformation by activated RAS (Dlugosz et al. 1997; Gangarosa et al. 1997; Sibilia et al. 2000). Thus, in addition to providing experimental evidence that EGFR activation is biologically relevant, the abovementioned study in the inducible model also points out the possibility that EGFR or its ligands may constitute alternative point(s) of therapeutic intervention in RAS-activated melanoma. It should be mentioned that the contribution of EGFR signaling to melanoma development and possibly progression is evolutionarily conserved, as activating mutations in the EGFR homolog, Xmrk, increase melanoma susceptibility in Xiphophorus fish (Wittbrodt et al. 1992; Winnemoeller et al. 2005; Meierjohann et al. 2006; for review, see Bardeesy et al. 2000). It therefore remains possible that similar activating mutations exist in human melanoma, although systematic resequencing of large cohorts of melanomas from different ethnic and/or molecular subclasses will be required to uncover such examples.

The RTK c-MET is normally expressed on epithelial cells and melanocytes (Bottaro et al. 1991) and is activated by binding of its ligand, Hepatocyte growth factor/ scatter factor (HGF). Although MET is normally activated in a paracrine manner, autocrine activation of HGF–MET has been described in melanoma progression (Li et al. 2001; for review, see Vande Woude et al. 1997). Accordingly, increased c-MET expression has been observed in metastatic melanoma (Natali et al. 1993), and copy number gain of the c-MET locus at 7q33-qter seems to be a late event in melanoma progression (Wiltshire et al. 1995; Bastian et al. 1998). However, similar to EGFR above, neither focal MET amplifications nor activating MET point mutations have been detected in melanoma, although both have been observed in other human cancers (Jeffers et al. 1997; Schmidt et al. 1997; Kong-Beltran et al. 2006; Smolen et al. 2006). However, several lines of experimental and functional evidence support a causal role for MET signaling in human melanoma. For example, in explant models, it has been shown that elevated c-Met expression or Met RTK activity may correlate with metastasis (Rusciano et al. 1995). In genetically engineered models, constitutive and ubiquitous HGF expression establishes an autocrine loop with c-Met, leading to stepwise development and progression of cutaneous and metastatic melanomas, which cooperates with UVB and Ink4a/Arf deficiency (Otsuka et al. 1998; Noonan et al. 2001; Recio et al. 2002). Correspondingly, while enforced expression of c-Met in melanocytes provides only weak cancer-initiating activity, this mutation drives the development of metastatic disease, and such tumor lesions show concomitant activation of HGF and establishment of HGF-Met signaling loop (L. Chin, unpubl.). Finally, c-MET was recently shown to be a direct transcriptional target of microphthalmia-associated transcription factor (MITF) (McGill et al. 2006), the melanocytic lineage transcription factor that can be activated by focal amplification in melanoma (see below).

The c-Kit gene encodes a RTK that serves as the receptor for Stem Cell Factor (SCF). Numerous immunohistochemical studies have linked progressive loss of c-KIT expression with the transition from benign to primary and metastatic melanomas (Montone et al. 1997; Shen et al. 2003; Isabel Zhu and Fitzpatrick 2006). Reconstitution of c-Kit in metastatic melanoma cells apparently conferred sensitivity to SCF-induced apoptosis in vitro (Huang et al. 1996a). Thus, at first glance, Kit does not fit the profile of a RTK targeted for activation in melanoma. However, a recurrent L576P mutation in c-Kit has recently been reported in melanoma. Among 153 cases examined, Holden and colleagues identified four metastatic melanomas with robust expression of c-KIT on IHC. High-resolution amplicon melting analyses followed by direct DNA sequencing revealed that three of them harbored a L576P mutation with selective loss of the normal allele (Willmore-Payne et al. 2005, 2006). L576P is a known GIST-associated mutation that maps to the 5' juxtamembrane domain where most activating KIT mutations cluster (Nakahara et al. 1998; Fukuda et al. 2001). Although likely representing an uncommon path to melanoma, the example of EGFR mutational status as a predictor for therapeutic responses in NSCLC (Lynch et al. 2004; Paez et al. 2004) suggests the possibility of identifying a melanoma patient subpopulation that will respond to imatinib based on c-Kit mutational status.

RAS and RAF, activators of mitogen-activated protein (MAP) kinase signaling

Considering the prominent roles of RTKs in transmitting extracellular signals to intracellular effectors, and the importance of homotypic and heterotypic cell–cell interactions in cancers, it is not surprising that almost all of the direct signaling components of RTKs have been implicated in human melanoma. One of the major signaling mediators of RTK is the MAP kinase pathway (ERK1/2), which has been most directly linked to its growth-promoting activities. Since self-sufficiency in growth signaling is a requisite capability acquired by all cancer cells (Hanahan and Weinberg 2000), hyperactive ERKs are common in many human cancers, including melanoma (Takata et al. 2005; Zhuang et al. 2005). Such a hyperactive state can theoretically be achieved by activating mutations of any signaling mediators upstream of ERKs, but there are clear tumor-type-specific patterns of mutational activation. RAS is perhaps the most frequently activated component of this signaling cascade in human cancer with a reported incidence of 15%–30% of all cases (Bos 1989), and in some cancer types, such as KRAS in pancreas cancer, its mutation rate approaches 100% (Hezel et al. 2006). BRAF is also commonly targeted in human cancers with an overall occurrence of 7% (Davies et al. 2002), although it is notable that the mutation frequency approaches 70% in metastatic melanoma (Maldonado et al. 2003; Pollock et al. 2003; Uribe et al. 2003; Daniotti et al. 2004; Kumar et al. 2004; Shinozaki et al. 2004; Libra et al. 2005). However, while classically considered primarily mitogenic, ERK activation can also regulate differentiation, senescence, and survival. For instance, activated ERK protein kinases can phosphorylate enzymes that regulate metabolism, or cytoskeletal proteins that regulate cell shape and migration. In the nucleus, ERK can regulate gene expression by phosphorylating transcription factors such as ETS1/2 (Paumelle et al. 2002), or other protein kinases such as S6 Kinase (Crews et al. 1991). Therefore, genotype–pheno-type correlation must take into account consequences other than growth promotion by activating mutations in components of ERK signaling.

The RAS family of proto-oncogenes: H-RAS, N-RAS, and K-RAS In contrast to other solid tumors, activating mutations of RAS proto-oncogenes are not detected with high frequency in melanoma, ranging from low to 10%– 15% incidence, with the highest frequencies detected in the amelanotic nodular subtype (for review, see Chin et al. 1998). N-RAS is the most frequent RAS family member targeted in the melanocyte lineage, with activating mutations in as many as 56% of congenital nevi (Papp et al. 1999), 33% of primary and 26% of metastatic melanoma samples (Demunter et al. 2001). Activating N-RAS mutations have been correlated with nodular lesions and sun exposure (Jafari et al. 1995; van Elsas et al. 1996). Interestingly, N-RAS mutations are rarely found in dys-plastic nevi (Albino et al. 1989; Jafari et al. 1995; Papp et al. 1999), which may imply their distinct evolutionary path to melanoma. H-RAS activation has occasionally been detected in melanoma, albeit more commonly associated with Spitz nevi, based on amplification of its genomic locus on 11p and oncogenic point mutations (Bastian et al. 2000). K-RAS mutations have not been described in human melanocytic lesions. These mutation patterns point to distinct biological activities of the different RAS family members in melanocyte biology. The phenotypic impact of activated H-RAS versus N-RAS transgenic mice has reinforced this view. Specifically, an activated H-RAS transgene, together with inactivating mutations in Ink4a, Arf, and/or p53 promotes development of nonmetastatic melanomas (Chin et al. 1997; Bardeesy et al. 2001; Sharpless et al. 2003). In contrast, when targeted to the melanocytic compartment, an activated N-RAS transgene and Ink4a/Arf deficiency drive cutaneous melanomas with high penetrance and short latency, as well as metastatic spread to lymph nodes and other distal sites (e.g., lung and liver) in a third of the cases (Ackermann et al. 2005).

BRAF, a potent activator of ERK protein kinases Activating BRAF mutations are the most prevalent somatic genetic event in human melanoma. Since its discovery through a genome-wide cancer resequencing effort (Davies et al. 2002), mutations in BRAF have been detected in a variety of tumor types, with the highest incidence in melanoma (ranging from 27% to 70%) (Maldonado et al. 2003; Pollock et al. 2003; Uribe et al. 2003; Daniotti et al. 2004; Kumar et al. 2004; Shinozaki et al. 2004; Libra et al. 2005), followed by papillary thyroid tumors, colorectal cancers, and ovarian cancers (Ciampi and Nikiforov 2005; Young et al. 2005). These point mutations clustered in specific regions of biochemical importance, with the predominant melanoma mutation being a single phosphomimetic substitution in the kinase activation domain (V600E), which confers constitutive activation (Garnett and Marais 2004). The biology and detailed characterization of the RAF family in human cancers and in melanoma have been discussed in detail in several other reviews (Garnett and Marais 2004; Gray-Schopfer et al. 2005).

BRAF does not appear to be an inherited cancer predisposition gene (Laud et al. 2003; Casula et al. 2004; Jackson et al. 2005; Kelemen et al. 2005; Xing 2005), as individuals with germline mutations develop instead cardio-facio-cutaneous syndrome, which is not associated with increased cancer risk (Niihori et al. 2006; Rodriguez-Viciana et al. 2006). The high prevalence of BRAF mutations in cutaneous melanoma and the known epidemiological link between UV and melanoma has prompted speculation that the BRAF^V600E mutation is induced by UV damage. Arguing against this supposition is the fact that the T A transversion that converts the valine to glutamic acid at amino acid 600 (V600E) is not classically associated with UV-induced damage (Daya-Grosjean et al. 1995). However, the cutaneous distribution of melanomas harboring this BRAF^V600E mutation suggests a complex yet undefined relationship with UV exposure. Specifically, BRAF mutations are common (59%) in melanomas arising in skin with intermittent sun exposure, such as trunk and arms, compared with only 23% of the acral melanomas and 11% of mucosal melanomas harboring BRAF mutations (Maldonado et al. 2003; Edwards et al. 2004; Curtin et al. 2005), but are absent in uveal melanoma (Cohen et al. 2003; Cruz et al. 2003; Edmunds et al. 2003; Rimoldi et al. 2003; Weber et al. 2003). A clear understanding of the UV–BRAF link is further complicated by the observation that melanomas from chronically sun-exposed areas (defined by histo-pathological evidence of chronic sun damage) possess only an 11% frequency of BRAF^V600E mutation (Maldonado et al. 2003; Edwards et al. 2004; Curtin et al. 2005). Together, these contrasting observations highlight the uncertainties surrounding the molecular factors driving BRAF mutation, in particular, the role of sun exposure.

B-RAF mutations are also common in benign and dys-plastic nevi (Pollock et al. 2003; Yazdi et al. 2003; Kumar et al. 2004; Saldanha et al. 2004), suggesting a role in the earliest stages of neoplasia. It is notable that nevi are considered growth-arrested and only rarely progress into melanoma, raising the possibility of BRAF^V600E-induced checkpoint mechanisms operating to constrain malignant transformation. Indeed, a recent study showed that human congenital nevi are invariably positive for senescence-associated acidic -galactosidase (SA--Gal), the classical senescence-associated marker (Michaloglou et al. 2005). In vitro, BRAF^V600E expression in human melanocytes has been shown to induce cell cycle arrest, with concomitant induction of both INK4A and SA--Gal, as well as methylation of Lys 9 of histone H3 on senescence-associated heterochromatic foci (SAHF). Thus, this study, along with others, provided evidence that oncogene-induced senescence (OIS) is an in vivo phenomenon functioning to constrain progression of early premalignant lesions (Sharpless and DePinho 2005). Intriguingly, expression of INK4A is not in 100% concordance with SA--Gal positivity, suggesting presence of a non-INK4A-dependent pathway in mediating BRAF-induced OIS (Michaloglou et al. 2005). It seems probable that the identification of this pathway would lead to the discovery of a tumor suppressor whose importance in melanoma rivals that of INK4A.

The notion that BRAF^V600E is not sufficient for transformation of melanocytes has also been demonstrated in other model systems. In zebrafish, it has been shown that BRAF activation leads only to development of benign nevi, while progression to frank melanoma requires cooperation of p53 deficiency (Patton et al. 2005). Similarly, BRAF^V600E mutation alone in TERT-immortalized RB–p53 mutant human melanocytes was found to produce only junctional nevi in the human/mouse skin graft, in contrast to activated NRAS or PI3K p110a mutants, which generated invasive melanoma lesions (Chudnovsky et al. 2005). While these biological outcomes indicate distinct roles for NRAS and BRAF activation in melanoma genesis, some functional overlap is suggested by the predominantly nonconcurrent occurrence of activated NRAS and B-RAF alleles in melanoma and other tumor types (Davies et al. 2002; Rajagopalan et al. 2002; Goel et al. 2006). The single and compound mutant mouse models will prove useful in dissecting the common and distinct roles of activated NRAS and BRAF signaling in vivo in the melanocytic lineage.

PTEN, negative regulator of phosphatidylinositol 3-kinase (PI3K)–AKT pathway

The PI3 kinase–AKT pathway is often hyperactive in melanoma. Integrins and growth factors (such as HGF and IGF-1) promote melanoma cell growth and survival via PI3 kinase/AKT activation in vivo (Meier et al. 2005; Robertson 2005). In addition, elevated phospho-AKT levels appear to correlate adversely with patient survival (Dai et al. 2005). Taken together, these observations suggest the strong biological importance of PI3 kinase signaling in this disease. However, unlike the MAP kinase pathway, genetic alterations specifically targeting components of this signaling cascade do not occur at high frequency in melanoma. Of those that do occur, the best known culprit is the PTEN tumor suppressor.

PTEN encodes a lipid and protein phosphatase that resides on chromosome 10q, a region known to sustain LOH in many human cancers, including melanoma (Bastian 2003; Wu et al. 2003). This protein regulates extracellular growth signals that use the lipid phosphatidylinositol phosphate (PIP₃) as an intracellular second messenger. In the presence of growth factor signaling, the intracellular level of PIP₃ rises, leading to phosphorylation of AKT, which is known to promote cell cycle progression and inhibit apoptosis. PTEN regulates PIP₃ levels, and its inactivation results in accumulation of PIP₃, AKT hyperphosphorylation, and enhanced cell survival/ proliferation. In melanoma, allelic loss or altered expression of PTEN comprises 20% and 40% of melanoma tumors, respectively (Pollock et al. 2002; Mikhail et al. 2005; Slipicevic et al. 2005; Goel et al. 2006), although somatic point mutations and homozygous deletions are rarely observed. Functionally, ectopic expression of PTEN in PTEN-deficient melanoma cells can abolish phospho-AKT activity, induce apoptosis, and suppress growth, tumorigenicity, and metastasis (Robertson et al. 1998; Stewart et al. 2002; Stahl et al. 2003; for review, see Robertson 2005). Correspondingly, germline or somatic inactivation of Pten in the mouse strongly promotes tumor phenotypes in multiple cell lineages (Di Cristofano et al. 1998; Stambolic et al. 1998; Podsypanina et al. 1999; Ma et al. 2005), including melanoma (You et al. 2002).

In line with the experimental evidence supporting a melanoma-suppressive role of PTEN, constitutive activation of AKT has been shown to be a potent oncogenic lesion for melanocyte transformation (Chudnovsky et al. 2005). In addition, DNA copy gain involving the AKT3 locus has recently been described in melanoma, and selective AKT3 activation may characterize 40%–60% of sporadic tumors (Stahl et al. 2004). However, the complexity of this signaling cascade has not been fully understood. Recent data have suggested that activation of different AKT isotypes may elicit distinct effects on cell proliferation and survival. For example, one report found that AKT3 correlated most strongly with melanoma tumor progression among the three AKT isotypes, as described above; and that targeted AKT3 depletion triggered apoptotic signaling (Stahl et al. 2004). On the other hand, AKT1 activation was found to inhibit the migration and invasion of certain cancer cell lines (Yoeli-Lerner et al. 2005), including MDA-MB435, a line previously believed to derive from breast cancer but subsequently shown through transcriptional and SNP array profiling studies to be a melanoma cell line (Ross et al. 2000; Garraway et al. 2005). Thus, although the PI3 kinase/AKT pathway clearly demonstrates enhanced activity in many melanomas, the extent to which this constitutes a critical melanoma dependency remains unresolved.

Lastly, although PTEN is a bona fide tumor suppressor of 10q24, the existence of additional melanoma suppressors has been inferred by the observation that many more human melanomas with 10q LOH exist than do those with PTEN-specific homozygous/hemizygous loss or mutation, and that reintroduction of PTEN into such melanoma cells seems to have no growth-suppressive effect (Robertson et al. 1998). The Myc antagonist, MXI1, is a candidate for this, as Myc is amplified or overexpressed in RAS-induced Trp53-deficient melanomas in mouse (Bardeesy et al. 2001); however, a role for Mxi1 in melanoma genesis in mice and humans has not been rigorously evaluated.

	The genomics of melanoma

Top Abstract The disease The genetics of melanoma The genomics of melanoma The biology of melanoma--lessons... Emerging paradigms for melanoma... Acknowledgments References

 
Beyond these characterized signature pathway alterations, genomic profiling of human melanoma has revealed a highly rearranged melanoma genome, attesting to underlying molecular heterogeneity of this disease. Mining of such heterogeneity represents an active area of investigation, as it is widely assumed that embedded within this complexity are clues to disease pathogenesis, clinical behavior, and possibly response to therapy. In addition, high-resolution characterization has revealed the existence of many recurrent nonrandom amplification and deletions with as yet no obvious or validated targets, potentially representing many yet-to-be-discovered melanoma-relevant genes. That these genomic events serve as productive entry points for discovery of melanoma genes is exemplified by discovery of MITF (see below). However, the high level of complexity inherent in the melanoma genome also brings into focus a critical need for biological systems capable of prioritizing in-depth validation efforts as illustrated in the case of NEDD9 (see below).

Genomic heterogeneity of melanoma

The underlying molecular heterogeneity of human melanoma has long been apparent on clinical grounds. For example, melanomas arising at different sites of the body may exhibit markedly distinct biological and clinical behaviors. Lentigo maligna melanomas are indolent tumors that develop over decades on chronically sun-exposed area such as the face. In contrast, acral lentigenous melanomas, which develop on sun-protected regions, tend to be more aggressive. "Thick" and "thin" primary melanomas provide additional examples: Although thick lesions have a much higher risk for metastasis than do their thinner counterparts, there is a subset of thin cutaneous melanoma that metastasizes early (Gimotty et al. 2005), suggesting inherent differences within this subset.

Recent advances in genome-wide technologies enable systematic documentation of tumor heterogeneity at high resolution, opening doors to better stratification and, ultimately, improved management of melanoma patients. Transcription profiling studies have provided evidence for distinct molecular subclasses of melanoma (Bittner et al. 2000; Segal et al. 2003; Tschentscher et al. 2003; Onken et al. 2004; Haqq et al. 2005). At the genomic DNA level, the nonrandom nature of the chromosomal alterations characteristic of melanoma likely also dictates disease behavior; thus, patterns of alterations detectable at either the DNA or RNA level may segregate melanoma tumors into subtypes with distinct clinical behaviors and possibly therapeutic responses. Indeed, a recent genome-wide CGH profiling and targeted resequencing study on 126 primary melanomas showed that distinctive patterns of genomic alterations, including BRAF and NRAS mutation frequencies, can be identified in melanoma arising in different anatomic sites and with varying UV exposure histories (acral, mucosal, or with and without chronic sun damage) (Curtin et al. 2005). Similar genetic and biological heterogeneity characterized metastatic melanomas. High-resolution CGH studies of >80 metastatic melanomas found that metastatic tumors can be separated into three distinct classes by unsupervised classification. Such subdivision does not correlate with sites of metastasis; rather, these classes are determined primarily by patterns of chromosomal aberrations. When intersected with clinical outcome, one of the three classes showed a more significant survival advantage than the other two (L. Chin, unpubl.), suggesting that these DNA-based classes of metastatic melanomas may prove biologically pertinent.

The melanoma gene atlas

Beyond the characteristics described above, the melanoma genome also possesses numerous recurrent non-random chromosomal rearrangements indicative of the existence of many additional genetic elements governing disease genesis and progression. Whole-genome screening technologies such as spectral karyotype analysis and array-CGH have identified many recurrent nonrandom chromosomal structural alterations, particularly in chromosomes 1, 6, 7, 9, 10, and 11 (Bastian et al. 1998; Curtin et al. 2005; Garraway et al. 2005); however, in most cases, no known or validated targets have been linked to these alterations. Thus, there exists significant opportunity to discover novel melanoma genes and translate such discoveries into meaningful clinical endpoints.

In a systematic high-resolution genomic analysis of melanocytic genomes, array-CGH profiles of 120 melanocytic lesions, including 32 melanoma cell lines, 10 benign melanocytic nevi, and 78 melanomas (primary and metastatic) have revealed a level of genomic complexity not previously appreciated. In total, 435 distinct copy number aberrations (CNAs) were defined among the metastatic lesions, including 163 recurrent, high-amplitude events. These include all previously described large and focal events (e.g., 1q gain, 6p gain/6q loss, 7 gain, 9p loss, and 10 loss). Not surprisingly, compared with the metastatic tumors, the degree of genomic instability (as reflected by number of CNAs) evident in primary melanomas and benign nevi is significantly diminished (Fig. 1). These types of primary versus meta-static comparisons provide an effective biological filter for progression-related events, thereby enriching for genetic events that constitute likely drivers of metastasis. Such comparative approaches should provide robust future grounds for melanoma gene discovery, as primary melanomas are largely curable with surgical resection, while metastatic disease is near universally fatal.

View larger version (46K): [in this window] [in a new window]   Figure 1. The genomes of benign and malignant melanocytic lesions. Genome-wide profiles of representative benign nevus, primary, and metastatic melanoma showing increasing genomic complexity.

The promise of DNA-based structural alterations as the entry point for gene discovery has been illustrated by the recent identification of MITF as a melanoma onco-gene. The discovery of MITF amplification in melanoma derived from an integrated analysis of genomic copy gains and losses, together with sample-matched mRNA expression data (Garraway et al. 2005). When clustering algorithms were applied to SNP array-derived chromosomal copy number data generated for the NCI-60 cancer cell line collection, some of these cell lines aggregated according to tissue of origin, including several melanoma cell lines. The bidimensionality of the hierarchical algorithm also enabled the identification of chromosomal alterations driving these lineage-restricted clustering patterns, and suggested that lineage-specific cancer genes might reside within the genomic regions implicated. For the melanoma cell lines, the common genomic alteration was a region of copy gain at chromosome 3p14-3p13. To facilitate the identification of an oncogene targeted by this amplification event, the NCI-60 collection was partitioned based on the presence or absence of copy gain at the relevant chromosome 3p locus (Garraway et al. 2005). This partitioning served as a two-class distinction that drove a supervised analysis of sample-matched gene expression data. Although the gene expression signature that emerged was dominated by melanocyte lineage genes (as expected given that only melanoma cell lines comprised the 3p-amplified class), MITF was the only gene showing significantly increased expression in association with the 3p-amplified melanoma cell lines that also mapped to the common region of 3p copy gain.

MITF amplification was subsequently detected in 10% of primary cutaneous and 15%–20% of metastatic melanomas. Although the majority of amplifications were low level (e.g., four to six copies per cell), high-level amplicons were also observed, including one sample that exhibited >100 copies per diploid genome. A Kaplan-Meier analysis performed on metastatic melanomas suggested that MITF amplification in this setting correlated with adverse 5-yr patient survival. Finally, ectopic MITF overexpression complemented BRAF^V600E in conferring soft agar colony growth to immortalized melanocytes engineered to express TERT, and to lack the pRB and p53 pathways. These functional studies thereby suggested a genetic context that might characterize a subset of human melanomas whose survival is dependent on MITF (Garraway et al. 2005). MITF also exemplifies a newly recognized "lineage survival" oncogenic mechanism (Garraway and Sellers 2006a, b), wherein tumor genetic alterations may target survival functions also operant in the relevant cellular lineages during development and differentiation. Thus, while the discovery of MITF amplification began as a systematic genomics-based survey of many human cancer types, it provides a striking convergence of melanoma oncogene discovery and melanocyte development.

NEDD9, a melanoma metastasis gene identified by cross-species comparison The power of genetically engineered mouse (GEM) models as biological filters for human cancer gene discovery was illustrated in a recent study that identified NEDD9 as a novel melanoma metastasis gene (Kim et al. 2006). Since evolution preserves key molecular mechanisms governing biological processes, it is hypothesized that overlapping genomic alterations in mouse and human tumors reflect selection driven by similar genetic pressure, and are thus more likely to represent critical events in tumorigenesis. Exploiting the experimental merits of an inducible mouse model of melanoma (Chin et al. 1999), Kim et al. (2006) compared nonmetastatic parental and metastatic variant melanomas in the mouse by genome-wide high-resolution array-CGH and identified a recurrent focal amplification associated with acquisition of metastatic potential. This amplification in mouse is syntenic to human 6p24-25, a region that sustains copy number gain in 36% of human metastatic, but not primary, melanoma (Bastian et al. 1998; Namiki et al. 2005). Although the recurrence of this regional event speaks to its potential pathogenetic and/or prognostic importance, chromosome 6p gain in human tumors (including metastatic melanoma) typically involves an expansive region spanning >35 Mb, rendering identification of target(s) difficult to impossible. The cross-species comparison quickly narrowed one region of interest on 6p to an 850-kb region encompassing only eight annotated genes (Fig. 2). Further integration with expression data pointed to NEDD9 as the potential target. Accordingly, NEDD9 protein expression was shown to be up-regulated in a progression-correlated manner in human melanoma (Kim et al. 2006).

View larger version (43K): [in this window] [in a new window]   Figure 2. Nedd9 amplification in metastatic mouse melanoma maps to region of frequent gain in human metastatic melanoma. (Top) Genome-wide array-CGH profile of one mouse metastatic melanoma with Nedd9 amplification on chromosome 13. (Bottom) One metastatic human melanoma profile showing gain of chromosome 6p, spanning a large region. (Arrows) Mouse chromosome 13 amplicon maps by synteni to human chromosome 6p24-25.

In addition to illustrating how comparative oncogenomics has enabled the identification and facilitated the validation of NEDD9 as a melanoma metastasis gene, this study also demonstrated evolutionarily conserved genomic activation of a component of the focal adhesion complex during melanoma metastasis (Kim et al. 2006), raising the possibility that inhibition of this signaling complex would halt progression of primary melanoma to metastatic disease. Along this line, it is interesting that NEDD9 expression appears to be positively selected for even at the primary melanoma stage, suggesting the potential predictive value of NEDD9 expression for future metastasis in primary melanoma.

Melanoma maintenance genes as rational therapeutic targets

The concept of tumor maintenance stems from the observation that cancer is the phenotypic end-point of numerous genetic and epigenetic alterations; moreover, the genomic complexity of human melanoma suggests that full malignant transformation of melanocytes requires the cooperating effects of multiple genetic lesions. This requirement has been well-demonstrated in genetically engineered models (GEMs), even when transformation has been driven by a potent initiating mutation (such as RAS). In the melanocyte lineage, even if one or more gain-of-function mutations are engineered into GEM models, transformed phenotypes only emerge (1) after a period of latency and (2) often with incomplete penetrance. Each of these characteristics implies a need for acquisition/selection of secondary cooperating events. Consistent with this, candidate or genome-wide molecular analyses of GEM melanoma have often identified spontaneous genetic alterations acquired de novo. For instance, in the HRAS-driven GEM model of melanoma on an Ink4a or Arf-deficient background, reciprocal p53 (e.g., p53 mutation or Arf loss) or Rb (e.g., Cdk6 amplification or Ink4a loss) pathway lesions, respectively, have been detected in the resultant melanomas, such that melanoma from either genetic backgrounds harbored concomitant inactivation of both p53 and Rb pathway (Kannan et al. 2003; O'Hagan et al. 2003; Sharpless et al. 2003). Along the same line, even on an In4a/Arf double-null or p53-deficient background, new and non-random genomic alterations can be detected by array-CGH profiling in HRAS-driven melanoma (Bardeesy et al. 2001; Hochedlinger et al. 2004). In addition to the genetic lesions mentioned here, biological processes that maintain and sustain an established tumor likely involve epigenetic dysregulation and intimate tumor–stromal interactions, the molecular basis of which remains poorly understood. Clearly, efficacious therapeutic targets must play requisite roles in maintenance of established melanomas.

Inducible onco-transgene models have been developed to assess the tumor maintenance role of activated oncogenes in GEM (Chin and DePinho 2000). Using the tetracycline regulatory system, an inducible HRAS-driven melanoma model (Tyr-rtTA ⁺ Tet-RAS ⁺ Ink4a/Arf ^–/–) has been generated whereby expression of the oncogenic RAS can be regulated in a tissue-specific and developmental stage-specific manner (Chin et al. 1999). In this system, despite the presence of irreversible genomic alterations in RAS-initiated melanomas, inactivation of RAS via doxycycline withdrawal in fully established melanomas led to clinical and histopathological regression, a remarkable finding that attests to the essential role of RAS activation in tumor maintenance. In addition to addressing the role of an oncogene in maintenance, inducible GEMs provide an in vivo system to examine the heterotypic interaction of tumor and stromal cells. For example, it has been shown that RAS-mediated melanoma regression is associated with dramatic activation of apoptosis in both tumor and host-derived endothelial cells, a phenotypic outcome not necessarily anticipated based on the intimate link of RAS activation to proliferation via MAP kinase activation in vitro (Chin et al. 1999). Real-time imaging during tumor regression by MR and bioluminescence provided additional evidence that loss of vascular function in regressing melanomas upon RAS inactivation represented an active response to loss of RAS activity upon doxycycline withdrawal, rather than a passive bystander effect of tumor cell deaths (Tang et al. 2005). Together, these experiments underscore the importance of GEMs to interrogate the tumor maintenance requirement of putative melanoma genes prior to their designation as a rational therapeutic target.

	The biology of melanoma—lessons from development

Top Abstract The disease The genetics of melanoma The genomics of melanoma The biology of melanoma--lessons... Emerging paradigms for melanoma... Acknowledgments References

 
It is a commonly accepted premise in cancer biology that signature phenotypic features of a malignancy may be traceable to the cell and developmental biological characteristics of its lineage origins. Melanoma is no exception. Multiple key tumor biological hallmarks mirror the distinctive features of the melanocytic lineage. Indeed, a defining aspect of melanoma is a strong propensity for metastasis, reflecting well the developmentally encoded dispersion of melanocytes from the neural crest to cutaneous sites. Thus, it stands to reason that valuable information may derive from the careful scrutiny of melanoma biology through the lens of melanocyte lineage development.

Melanocyte stem cells

Melanocyte progenitor cells migrate over considerable distances, along a predictable route upon their exit from the neural crest. They traverse discrete stages of differentiation recognizable in terms of marker expression, ultimately populating the basal epidermis as well as the hair follicle in human. In certain furry mammals such as the rodents, the epidermal population may be transient, a feature likely arising from a lack of SCF expression by basal keratinocytes since persistent epidermal melanocytes have been observed in transgenic mice, which ubiquitously express the cytokine HGF (Takayama et al. 1996) or keratinocyte-targeted SCF (Kunisada et al. 1998). Within hair follicles, melanocytes reside in two locations: at the bulb (where they function as differentiated cells to provide pigment to the growing hair matrix) and as stem cells within the bulge region, located at the base of the permanent portion of the follicle (Nishimura et al. 2002).

As demonstrated by Nishimura et al. (2002), follicular bulge melanocytes satisfy the criteria of stemness, including properties of immaturity, slow cycling, self-maintenance, and the capacity to repopulate the hair follicle at the anagen phase. At the beginning of anagen, signals within the stem cell niche trigger an emergence from quiescence that results in two distinct melanocytic progeny. The differentiated population migrates to the base of the follicle (hair bulb), where melanin synthesis and transport to adjacent keratinocytes occur, resulting in melanization of newly formed hairs. Later in the hair follicle cycle, the deep (transient) portion of the follicle undergoes a regressive (catagen) phase characterized by apoptosis of the deep structures. Following a resting (telogen) phase of variable duration, a new anagen phase is triggered.

The existence of melanoma stem cells has also been investigated recently. Herlyn and colleagues identified a CD20-enriched subpopulation of cells from melanoma lines or metastases that grew as nonadherent spheroids when cultured in human ES-like medium. Moreover, these CD20-enriched cells could be induced to differentiate toward multiple lineages, including melanocyte, adipocyte, osteoblast, and chondrocyte, suggesting the presence of such a subpopulation with a stem cell phenotype (Fang et al. 2005). Separately, Frank and colleagues identified a subpopulation of melanoma cells with expression of a drug efflux transporter ABCB5 and showed that this subpopulation was enriched for the CD133+ (stem cell marker-enhanced) fraction (Frank et al. 2005). Interestingly, ABCB5 expression was associated with resistance to adriamycin treatment, and correspondingly, ABCB5 blockade resulted in adriamycin accumulation and enhanced drug sensitivity. While it remains unclear how melanoma stem cells might be related to melanocyte stem cells, these studies suggest novel strategies for identification and targeting of clinically critical tumor subpopulations that exhibit stem-like features in melanoma.

Pigmentation

Considerable information on genetic and signaling pathways important for melanocyte biology has come from analysis of hair pigmentation dynamics. These studies are aided by the accumulation of easily recognizable coat color mutations in multiple animal species, as well as the generally benign nature of altered melanocyte function and therefore survival for the host organism. Examination of hair graying represents a good example of such analysis. The geographic separation between follicular melanocyte stem cells (bulge region) relative to their differentiated progeny (hair bulb) permitted relatively simple quantitative analysis of each population. Using transgenic mice in which the melanocyte lineage was tagged by LacZ (Mackenzie et al. 1997), it was observed that senile graying is associated with progressive depletion of the stem cell pool (Nishimura et al. 2005). Melanocyte stem cell depletion was preceded by the stochastic appearance of pigmented melanocytes within the bulge region—a population that was not maintained, suggesting that it may represent a dead-end route away from the stem cell pool. Similar observations were made in aging human hair follicles (Nishimura et al. 2005). In addition, several murine models of accelerated graying have been analyzed, and demonstrate either an accelerated version of the age-related process (Mitf^vit/vit ) or abrupt and selective stem cell loss (Bcl2 ^–/–). Interestingly, graying in the Bcl2-deficient background is dramatically rescued by simultaneous deficiency of the proapoptotic family member BIM (Bouillet et al. 2001).

The cellular and molecular events regulating apoptosis of the melanocyte population during the regressive (catagen) phase of the hair follicle cycle are being