The Molecular Pathology of Cutaneous Melanoma

Cancer Biomarkers 9 (2011) 267–286 267 DOI 10.3233/CBM-2011-0164 IOS Press

The molecular pathology of cutaneous melanoma

Thomas Bogenriedera and Meenhard Herlynb,∗ aBoehringer Ingelheim RCV, Dr. Boehringer Gasse 5-11, 1121 Vienna, Austria bThe Wistar Institute, 3601 Spruce Street, Philadelphia, PA, USA

Abstract. Cutaneous melanoma is a highly aggressive cancer with still limited, but increasingly efficacious, standard treatment options. Recent preclinical and clinical findings support the notion that cutaneous melanoma is not one malignant disorder but rather a family of distinct molecular diseases. Incorporation of genetic signatures into the conventional histopathological classification of melanoma already has great implications for the management of cutaneous melanoma. Herein, we review our rapidly growing understanding of the molecular biology of cutaneous melanoma, including the pathogenic roles of the mitogen- associated protein kinase (MAPK) pathway, the phosphatidylinositol 3 kinase [PI3K]/phosphatase and tensin homologue deleted on chromosome 10 [PTEN]/Akt/mammalian target of rapamycin [mTOR])PTEN (phosphatase and tensin homolog) pathway, MET (hepatocyte growth factor), Notch signaling, and other key molecules regulating cell cycle progression and apoptosis. The mutation Val600Glu in the BRAF oncogene (designated BRAF(V600E)) has been associated with clinical benefit from agents that inhibit BRAF(V600E) or MEK (a kinase in the MAPK pathway). Cutaneous melanomas arising from mucosal, acral, chronically sun-damaged surfaces sometimes have oncogenic mutations in KIT, against which several inhibitors have shown clinical efficacy. These findings suggest that prospective genotyping of patients with melanoma, combined with the growing availability of targeted agents, which can be used to rationally exploit these findings, should be used increasingly as we work to develop new and more effective treatments for this devastating disease.

Keywords: Melanoma, cell cycle, pRb pathway, p16INK4A, p14ARF, cyclins, CDKs, Apaf1, Bcl2, MAPK pathway, Ras, BRAF, PTEN, PI3K/AKTpathway, SCF, ckit, HGF, cmet, cmyc, Notch

1. Introduction mately lead to more successful therapeutic strategies. The goal of this chapter, therefore, is to highlight some As the treatment of metastatic melanoma moves of the issues that have arisen. For the purpose of this rapidly into the era of targeted therapy, there is an chapter we will focus on the molecular events occurring ever growing need to untangle the underlying genet- within the melanoma cell, especially those pathways ic complexity and cellular signalling heterogeneity of which have been the scope of intensive research over the this highly malignant tumor. It is widely hoped that past decade and which are now targets of an increasing an understanding of how genetic and signalling profiles armamentarium of novel therapeutic approaches. Here, dictate pharmacologic response will allow for the se- we discuss recent findings and hypotheses on the role lection of optimal patient sub-populations for clinical of growth factor pathways in the molecular pathology trials. of melanoma. The matter of melanoma-stroma interac- The molecular pathology of melanoma development tions is too involved to thoroughly explore here. As the and progression is now being deciphered (Fig. 1). Al- literature on molecular and cellular events in melanoma though we have an increasingly good understanding of development and progression is now huge and includes the genetics and biochemistry of these events,the mechanisms underlying their combinatorial interactions are comprehensive reviews [1–5], we will take the liberty still poorly defined and need to be fathomed to ulti- of a more subjective view and review molecules and pathways that now take center stage as well as discuss emerging areas and interesting questions in the ∗Corresponding author. E-mail: [email protected]. melanoma field.

Fig. 1. Development of melanoma. This model implies that melanoma commonly develops and progresses in a sequence of steps from nevic lesions, which can be histologically identiﬁed in approximately 35% of cases. The model also acknowledges that melanoma may also develop directly from normal cells. The exact role of tumor-initiating or melanoma stem cells remains to be elucidated. Genetic (such as mutations and deletions of critical genes) and epigenetic (such as DNA-methylation leading to gene silencing) alterations of critical pathways as well as the acquired capability of evading apoptosis, progressively disturb normal skin homeostatis leading to melanoma development. Genetic changes are expected at the transition from common acquired (benign) nevus to dysplastic nevus and primary melanoma allowing cells to persist. During melanoma invasion and metastasis increased growth, invasion and stromal ‘landscaping’ by proteolysis occurs, leading to an increased manipulation of the microenvironment and interactions with stromal skin cells.

2. The unmet clinical need: “Melanoma is the with metastatic melanoma ranges from only 4.7 to 11 tumor that gives cancer a bad name” months, with a median survival of 8.5 months [8].

Melanoma is the deadliest form of skin cancer and 2.1. Melanoma is characterised by the dysfunction of one of the most aggressive of all neoplasms. The in- the epidermal melanin unit cidence of cutaneous melanoma has increased rapidly during the past several decades, and continues to do so Melanoma arises from the transformation of neu- at an alarming rate: The National Institutes of Health ral crest-derived melanocytes, the pigment cells of the predicts 62,480 new cases of melanoma in the year skin, which reside in the basal layer of the epidermis. 2009, with 8,420 of the patients succumbing to their dis- We now view melanoma as a complex tissue resulting ease [3]. The majority of these deaths are due to distant from disrupted skin homeostasis, rather than focusing metastases from the primary site because melanoma is on the melanoma cell, and the genes within it, alone. notorious for its propensity to metastasize. Although Normal skin homeostasis is maintained by dynamic in- localized melanoma is frequently curable by surgical teractions between the melanocytes and their microen- excision, metastatic melanoma is inherently resistant vironment, such as keratinocytes, ﬁbroblasts, endothe- to most systemic treatments, and survival of patients lial and immunocompetent cells, and the extracellular with advanced disease has not improved in more than matrix. Similarly, during transformation and progres- 30 years [6,7]. Melanoma is poorly responsive to cyto- sion of melanocytes to melanoma cells, there are, how- toxic chemotherapy, and thus the survival of patients is ever deregulated, reciprocal and conspirational inter- based on screening, early detection, and wide resection actions between the neoplastic cells and the adjacent of the primary lesion. The overall survival for patients stromal cells [9,10]. T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma 269

Under normal physiological conditions,melanocytes sufficient phosphorylation of Rb, E2F is released and and keratinocytes form the ‘melanin unit’ in the epi- transactivates genes required for S phase entry, includ- dermis. In this unit, melanocytes and keratinocytes are ing cyclins E and A. CDK inhibitors (CKIs) serve as evenly aligned along the basement membrane zone at a negative regulators of the Rb pathway. CKIs are clas- ratio of 1:5–8. Each melanocyte extends with its den- sified into two distinct families on the basis of their drites into the upper layers of the epidermis, transfer- structural and functional characteristics. The mem- ring pigment-containing melanosomes to approximate- bers of the INK4 family of CKIs (p16Ink4a, p15Ink4b, ly 35 keratinocytes in its unit via dendritic process- p18Ink4c, and p19Ink4d) contain multiple ankyrin re- es. The melanin contained in these melanosomes ab- peats and act as negative regulators of CDK4/6 by bind- sorbs and scatters damaging ultraviolet radiation, there- ing to the catalytic subunit and preventing formation of by shielding the nucleic acids on the skin from damage the active cyclin-CDK complex. The Cip/Kip family of (reviewed in [9–12]). CKIs (p21Cip1/W af1, p27Kip1, and p57Kip2)ismore The undifferentiated keratinocytes of the basal lay- broadly acting and regulates both CDK4/6 and CDK2 er regulate melanocyte growth, number of dendrites activity. Each member of the family contains a char- and expression of cell surface molecules, providing ev- acteristic motif within the amino-terminal region that idence that this highly ordered organizational pattern enables them to bind to both cyclin and CDK subunits. serves as the structural basis for intercellular regulation. The stoichiometry between CDKs and CKIs is impor- Information is not yet available on how melanocytes tant and determines the activity of Rb and, ultimately, and keratinocytes can maintain this lifelong balance, the proliferative state of cells. which is only disturbed during transformation into a melanocytic nevus (‘mole’) or a melanoma. It is now 3.1. Alterations of the pRb pathway in melanoma generally accepted that melanoma risk is modulated by skin pigmentation patterns, such as those linked to The p16-cyclin D/Cdk4-pRb pathway is deregulated MC1R polymorphisms, and early exposure to ultravio- in a large majority of human cancers, either through let (UV) light (reviewed in [13]). loss of p16 or pRb function or through deregulated expression of cyclin D or cdk4 (reviewed in [14–16]). Several mechanisms of inactivation of the pRb path- 3. Selected components of the cell cycle machinery way have been noted in various malignant tumors and cell lines thereof. The most common known ways in Recent excellent reviews summarize our current un- which these gene products are dysregulated in neo- derstanding of cell cycle regulation [14–16]. Mitogenic plasms are through gene deletion (homozygousor hem- growth factors promote the entry of quiescent cells into izygous); inactivating mutations; epigenetic changes the first gap phase (G1) and initiation of DNA synthesis such as promoter methylation, transcriptional repres- (S phase) of the cell cycle (Fig. 2). G1 cyclin-dependent sion, protein sequestration and inactivation (e.g., vi- kinases (CDKs) serve as positive regulators. D-type ral oncoproteins); and posttranslational modifications cyclins (D1, D2, D3) complex with CDK4 and CDK6 (e.g., inactivating phosphorylation events). The p16- to stimulate their kinase activities, which in turn cause cyclin D/Cdk4-pRb pathway as a functional unit is fre- the phosphorylation and inactivation of the retinoblas- quently altered in melanoma pathogenesis [14]. In toma (Rb) tumor suppressor protein. By binding to fact, in one study, virtually all (96%) of investigated E2F, Rb recruits histone deacetylases to the promoters melanoma cell lines harboured alterations at the DNA of E2F-responsive genes and represses their transcrip- level within CDKN2A/p16, CDKN2B/p15, CDK4 and tion. Multiple Ras effector pathways (details of which CCND1/cyclin D1 [17]. Inactivation mutations in the forthcoming) appear to play an important role in the pRb ‘pocket’ region, which blocks binding to E2F, regulation of cyclin D1. Cyclin D1, in part, regulates are rare in melanoma. Instead, the function of pRb the kinase activities of both CDK4 and CDK6. These in melanoma is commonly inhibited by hyperphos- complexes are formed in the cytoplasm and are trans- phorylation and, to a lesser extent, by underexpres- ported into the nucleus and undergo stimulatory modi- sion [18]. In a recent study, Curtin and colleagues fications including phosphorylation by CDK-activating conducted a genome-wide analysis of DNA copy num- kinase (CAK) to yield active holoenzymes. Further in- ber and mutational analysis of BRAF and NRAS in 126 to G1, cyclin E complexes with CDK2 and causes ad- melanomas from individuals with varying UV exposure ditional phosphorylation and inactivation of Rb. With histories [19]. They found that amplification of CDK4, 270 T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma

Fig. 2. Cell cycle regulatory proteins. The cell cycle is controlled by a series of complex interactions among multiple proteins that interact in a precisely orchestrated sequence. Illustrated here is a highly schematic summary of the primary protein interactions, with special attention to components involved in the G1 to S transition of the cell cycle. Rb is one of the targets for the enzymatic activity of cyclin-cdk complexes. p21 is transcriptionally regulated by p53 and inactivates various cdks. p27 inactivates the same complexes as p21. p16 and p15 have a more restricted activity and form binary complexes with cdk4 and cdk6, displacing D-type cyclins. TGF-β can downregulate the cyclin E/cdk2 activity by activating the transcription of two inhibitors of cyclin-dependent protein kinases, p21 and p15. A surge in the level of p15 displaces p27 from the complex of cyclin D/cdk4/6 and the resulting p27 binds to the cyclin E/cdk2 complex. This inactivates its kinase activity and blocks cell cycle progression. was commonly seen in acral and mucosal melanomas ic lesions representing all stages in the progression of but not observed in tumors with activating BRAF or melanoma, Reed et al. [20] suggest that loss of p16 NRAS mutations or Cyclin D1 copy number gains. Fur- protein expression is not necessary for melanoma initi- thermore, losses/deletions of the melanoma suppres- ation as all in situ lesions and the majority of invasive sor CDKN2A locus were also more commonly detected melanomas retained expression of this protein. Loss is in mucosal and acral melanomas, but only in samples potentially more related to invasiveness or the ability without CDK4 ampliﬁcation. to metastasize, because 52% of primary invasive tumors and 72% of metastatic lesions showed partial or 3.2. p16INK4A/p14ARF in primary specimen no expression of p16. In thin to intermediate (< 4 mm) sporadic melanomas, however, the frequencies of LOH Since its discovery as a CKI in 1993, the tumor at the INK4A locus, INK4A intragenic mutations and suppressor p16 has received widespread attention in promoter methylation are extremely low (10%) [21]. melanoma research (see [14] and references therein). Loss of p16 expression is associated with progression A plethora of studies show that in sporadic melanomas, of disease [20], suggesting that other regulatory mech- p16 is inactivated through homozygous deletion (6– anisms allow for decreased p16 in melanomas. 25%), mutation (0–26%) and methylation (0-10%) In this regard, the transcriptional regulatory protein in roughly one forth of primary tumors. Based on Id1 (inhibitor of differentiation) has recently been iden- an immunohistochemical analysis of 103 melanocyt- tiﬁed as a repressor of p16/INK4A transcription [21]. T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma 271

In melanomas, Id1 expression is limited to the in situ lar melanoma, whereas cyclin D1 does not have any component of invasive melanomas and perivascular impact on disease-free and overall survival. regions of metastatic tumors. Id1 expression corre- p27Kip1 is strongly expressed by normal melano- lates with decreased p16INK4A expression in RGP cytes and benign nevi, whereas in nodular melanomas, melanomas, suggesting that Id1 transcriptional repres- the level of p27Kip1 was found to correlate significant- sion may represent one of the earliest mechanisms of ly with the thickness of the tumor, with less protein p16 dysregulation in melanoma initiation. Consequent- expressed in thicker lesions [23]. Patients having tu- Kip1 ly, a model for p16 inactivation via multiple steps in mors with fewer than 5% p27 -staining cells have melanomagenesis has been proposed [21]: Reversible a significantly higher risk of early relapse of their dis- transcriptional inactivation in RGP melanomas (via ease compared with those expressing moderate or high Kip1 Id1 or other repressors), subsequent acquired epige- levels. In contrast, the level of p27 does not corre- netic changes (e.g., promoter methylation), and finally late with tumor thickness or disease-free survival in pa- irreversible genetic alterations (e.g., mutations, dele- tients with superficial spreading melanomas. Further- Kip1 tions) associated with melanoma invasion and metas- more, p27 does not appear to have an influence on tasis. Taken together, these findings indeed implicate overall survival for either subgroup. When examined Cip1/W af1 Kip1 deregulation of the p16/pRb pathway as a critical early the combined effect of p21 and p27 on or initiating event in the development of melanoma. clinical outcome, it was found that analysis of these The INK4A/ARF locus on chromosome 9p21 also two CKIs together may have greater prognostic poten- encodes another tumor suppressor gene called p14ARF tial than either alone. These data underscore the value (p19ARF in mice). Exon 1β, located approximately of analyzing multiple cell cycle regulatory proteins to 15kb centromeric to exon 1α of p16, is spliced on- obtain the most reliable indication of prognosis [23]. to exons 2 and 3 of p16 to generate this alternative A significant increase of cyclin E and CDK2 ex- transcript. Alternate promoters regulate the indepen- pression occurs during tumour progression in melan- dent production of these 2 mRNAs and translation of omas [24]. Cyclins B1, D2, and D3 show significant- the p14ARF mRNA occurs in a reading frame alter- ly increased expression in metastases, but normal or nate to p16. p19ARF acts through MDM2 to inhibit even decreased expression in primary melanomas. In p53 degradation and exon 1β is sufficient for p19ARF - contrast, cyclins A and D1, and CDK1 and CDK4 are mediated stabilization of p53 function (see [11,16] for expressed very weakly in situ with no significant differ- review). Thus far, there are no molecular data support- ences between nevi, primary melanomas or metastases, ing a tumor suppressive role for p14ARF in sporadic and histopathological staging. Approximately 30% of primary melanomas and 40% of metastases completely melanoma. Clearly, however, homozygous deletions Cip1Waf1 at the INK4A/ARF locus have the capacity to inacti- lack p21 expression. In superficial spread- vate both p14ARF and p16 function through the shared ing melanomas, a significant correlation between pro- second exon. tein expression and tumor thickness was found, with thin lesions showing low protein levels. By comparing primary and metastatic specimens obtained from the 3.3. Expression and prognostic relevance of same patient, a reduction in p21Cip1Waf1 staining was Cyclins/CDKs/CKIs in melanoma observed in the latter [25].

As in other malignant tumors, certain cell-cycle regulators are differentially expressed during melanoma 4. Selective components of the apoptotic program progression and are independent prognostic markers, albeit with somewhat incongruentresults. Florenes and Acquired resistance toward apoptosis (programmed colleagues [11,16,22] observed only rare cyclin D3- cell death) is a hallmark of cancer [26]. The concept positive cells and no cyclin D1-positive cells in benign that apoptosis might influence the malignant phenotype nevi. In contrast, cyclin D3/cyclin D1 was found to was first raised in 1972 by Kerr, Wyllie and Currie. be expressed by 96%/62% of primary and 97%/29% However, the importance of apoptosis in the pathogen- of metastatic melanomas, respectively. Kaplan-Meier esis of cancer remained under-appreciated for almost analysis revealed that high levels of cyclin D3 is an in- two decades. A comprehensive overview of apopto- dicator of early relapse and decreased overall survival sis is beyond the scope of this chapter (for review, see for patients with superficial spreading, but not nodu- refs [27,28]). As a general summary, most apoptotic 272 T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma pathways involve a sensor that detects a death-inducing tion induces apoptosis, and loss of p53 function leads signal, a signal transduction network and execution to the survival of these damaged cells thereby initiating machinery that actively carries out the process of cell tumor development. Hence, loss of apoptotic function death. DNA damage can induce apoptosis through a can impact tumor initiation, progression and metasta- central sensor, p53, although p53-independent path- sis. ways clearly exist (Fig. 3). Many of the signals that elic- p53 mutations were found to be associated with poor it apoptosis converge on the mitochondria, which re- prognosis, de novo or acquired resistance, and relapse spond to pro-apoptotic signals by releasing cytochrome in a broad field of solid and hematologic malignancies c. Cytochrome c can interact in a multi-protein com- (for review, see refs [31]). Melanoma is a radiation- plex with Apaf1 and pro-caspase-9, leading to caspase- and chemotherapy-refractory neoplasm and there is no 9 activation and the initiation of a protease cascade. standard therapy for patients with disseminated dis- One of the most important modulators is Bcl-2, owing ease: the commonly used anticancer drugs do not al- to its ability to affect cytochrome c release from mi- ter the prognosis, i.e. the fatal outcome [8], and are tochondria and/or modulate the Apaf1/caspase-9 inter- highly heterogeneous in their in vitro effects. However, action. However, neither Bcl-2 nor Bcl-xL has been contrary to other malignancies, p53 mutations are very shown to be completely cytoprotective against Apaf- rare in melanoma, being less than 5%. Tumor-derived 1/caspase-9 mediated cell death. p53 does localize to the nucleus in melanoma, howev- Bcl-2 is the founding member of an expanding gene er, its transcriptional activity is weak and overexpres- family that includes other anti-apoptotic molecules sion of wild-type p53 does not induce immediate cell such as Bcl-xL as well as pro-apoptotic members such death. Another intriguing factor is that cells which car- as Bax. A series of enzymes known as caspases are ry mutated p53 undergo apoptosis when overexpressed considered the engine of apoptotic cell death. Cas- with wild-type p53 [35]. This suggests that p53 is kept pases are cysteine proteases that are expressed as in- in an inactive state, either by other factors (such as active pro-enzymes. The “signaling” caspase-9 asso- Apaf-1) or through post-transcriptional mechanism(s), ciates with Apaf-1 (apoptotic protease activating factor- 1), and oligomerization of this complex in the presence thereby functionally inactivating its pathways. Thus, in of cytochrome c can activate the downstream “effector” melanoma studies fail to correlate p53 mutations with caspase cascade. Thus, essential downstream compo- reduced toxicity to anticancer agents. nents of p53-induced cell death include caspase-9 and its adapter Apaf-1 [29]. It is now becoming increas- 4.2. Apaf-1 ingly clear that mutations in many cancer-related genes can disrupt the apoptotic machinery and compelling Findings by Scott Lowe and colleagues at the begin- evidence indicates that apoptotic activity is important in tumor suppression. ning of the decade [36] provided exciting and promising new insights into mechanisms how melanomas become 4.1. Defects of the apoptotic program in melanoma apoptosis-, and thus chemotherapy-resistant, despite a fully functional p53. They show that melanoma cells p53 was the first tumor suppressor gene linked to of different progression stages can avoid apoptosis by apoptosis. p53 mutations occur in the majority of hu- inactivating a downstream component of the apoptotic man tumors and are often associated with advanced tu- pathway, namely Apaf-1, thus disabling the p53 apop- mor stage and poor patient prognosis. Moreover, sev- totic program. What makes the results of Soengas et al. eral upstream and downstream components of the p53 even more interesting is that they showed a) a deletion pathway (e.g. Mdm-2, p14ARF and Bax) are altered in of one of the alleles in 42% of the specimens tested and human tumors [26,30,31] (Fig. 3). What triggers apop- b) a reversible “switching off” of the remaining allele tosis during tumor development? A variety of signals by methylation, rather than a permanent deactivation appear important [26,31–34]. Extracellular triggers in- by deletion or mutation. The methylation of Apaf-1 clude radiation, growth/survival factor depletion and could experimentally be reversed by a methylation in- loss of cell-matrix interactions and cell-cell adherence- hibitor (5-aza-2dC) and thus its pro-apoptotic function based survival signals. In the skin, excessive exposure restored. Pre-treatment assessment of Apaf-1 status to UV radiation induces apoptosis, which presumably may therefore improve the therapeutic management of serves to eliminate heavily damaged cells. UV irradia- patients with melanoma. T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma 273

Fig. 3. Interrelationship between selected key molecules in the apoptotic program in melanoma. Apoptosis may involve either death- receptor-mediated (extrinsic) or mitochondria-mediated (intrinsic) events. The basic premise is that p53, although wild-type, is functionally inactive to induce apoptosis due to silencing of Apaf-1. The apoptotic p53 pathway includes cytochrome c release from mitochondria, activation of Apaf-1, its complex formation with caspase-9, and the “effector” caspase family. Apaf-1-mediated apoptosis is inhibited by IAPs (inhibitor of apoptosis proteins). Post-transcriptional modiﬁcations such as phosphorylation by Akt/PKB also inﬂuence the apoptotic effects of caspase-9. In addition, cytochrome c release from the mitochondria is inhibited by anti-apoptotic proteins such as Bcl-2/Bcl-xL.

4.3. Bcl-2 and melanoma blocks the release of cytochrome C [28]. Experimen- tal transfection of cells with Bcl-2 confers a multidrug Bcl-2 is a strongly antiapoptotic protein that is ex- resistant phenotype in both hematologic and solid tu- pressed in normal melanocytes, benign nevi, primary mor cells [39]. Conversely, pharmacologic reduction or melanoma, and melanoma metastases. At this time, the targeted inactivation of Bcl-2 amplifies anticancer re- overall importance of Bcl-2 expression by stage and sponses to chemotherapy in multiple in vivo models, in- lesion type remains controversial (reviewed in [37]). cluding melanoma [40]. Hence, several factors support Bcl-2 protein is located within the inner layers of the a role for Bcl-2 antisense therapy to treat melanoma, mitochondrial membrane, where it prevents apoptosis based on evidence suggesting a critical role for Bcl-2 by blocking the release of cytochrome C from mito- in melanoma cell survival. chondria. In this role, Bcl-2 promotes resistance to anticancer therapy and could contribute to enhanced metastatic potential. Moreover, Bcl-2 protein has been 5. Mitogen-activated protein kinase (MAPK) found in normal developing melanocytes, and the loss signal transduction pathways of Bcl-2 function in Bcl-2 knockout mice leads to accelerated disappearance of melanocytes. The mice are Mitogen-activated protein kinase (MAPK) pathways viable but rapidly lose hair color and turn from black are cellular signalling pathways that enable cells to to white [38]. Drug resistance in melanoma has been transduce extracellular signals to an intracellular re- partially attributed to overexpression of Bcl-2, which sponse (Fig. 4) [41,42]. The molecular details of mam- 274 T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma malian MAP kinase signal transduction pathways acti- distinct proteins, H-ras, N-ras, and K-ras (K4A- and vated by stress and inflammation have only begun to be K4B-). Activating single-point mutations of the ras dissected. The impact of these pathways on, amongst gene can result in constitutive activation of Ras pro- others, the pathology of cancer, the side effects of can- tein, thereby continuously stimulating cell prolifera- cer therapy, chronic inflammation, embryonic develop- tion and inhibiting apoptosis. These mutated forms ment, innate and acquired immunity, is profound. Thus of Ras have impaired GTPase activity. Although they it is not surprising that understanding these pathways still bind GTPase activating proteins (GAP), there is no has attracted wide interest, and in the past 10 years, “off” sign, since GTPase is no longer activated. Acti- dramatic progress has been made. Accordingly, it is vating mutations are limited to a small number of sites now becoming possible to envisage the transition of (codons/amino acids 12, 13, 59, and 61), all of which these findings to the development of novel treatment abolish GAP-induced GTP hydrolysis of the Ras pro- strategies. teins. Oncogenic K-ras mutations occur frequently in non-small-cell lung, colorectal, and pancreatic carcino- 5.1. Ras-Raf-MEK-MAP kinase pathway mas; H-ras mutations are common in bladder, kidney, and thyroid carcinomas; and N-ras mutations are found Since the discovery of the role of Ras oncogenes in hepatocellular carcinoma, hematologic malignancies in carcinogenesis in the early 80s, we have witnessed and melanoma [41,42]. an explosion of research in the signal transduction area [41,42]. Likewise, a host of growth signal- 5.2. Ras genes and melanoma ing pathways is now implicated in melanocyte and melanoma biology. They include cell growth and dif- Within the ras gene family, only N-ras has a signif- ferentiation, melanogenesis, cell fate determination, icant association with melanoma progression, whereas survival and apoptosis. The Ras gene product is a H-ras or K-ras are rarely mutated. Mutations in N-ras monomeric, membrane-localized guanosine triphos- have since been identified in 15–20% of all melanomas, phate (GTP)/guanosine diphosphate (GDP)-binding and are most commonly the result of the substitu- (G) protein of 21 kd that functions as a nodal point, tion from leucine to glutamine at position 61 [43,44]. a molecular “switch” converting signals from receptor The presence of N-ras mutations appears to corre- and non-receptor tyrosine kinases on the cell membrane late with the vertical growth phase (VGP) and with to downstream cytoplasmic or nuclear events (Fig. 4). melanoma arising in chronically UV-exposed body There are multiple effects that occur after Ras activa- sites. Moreover, N-ras mutations are associated with tion, initiating a multitude of cytoplasmic signalling nodular melanomas and to a lesser extent with lentigo cascades, leading to protein synthesis and regulation maligna melanoma. Benign and dysplastic nevi did of cell survival, proliferation, and differentiation. The not exhibit ras mutations, however, N-ras mutations in precise functions of Ras proteins continue to be elu- congenital nevi have been observed. N-ras mutations cidated. Ras utilizes a variety of downstream effec- did not correlate with metastasis or survival parameters tors and mediates its actions in mechanistically distinct (reviewed in [3,4,11]). ways, depending on both species and cell-type identity. In turn, these further complicate our ability to define a 5.3. BRAF in benign nevi and melanoma simple relationship between Ras and, for example, cell cycle progression. The MAP kinase (MAPK) pathway When active in its GTP-bound state, RAS activates has emerged as the crucial route between membrane- a number of downstream effectors, one of which is bound Ras and the nucleus (Fig. 4). This pathway en- the RAF family of serine/threonine kinases. There compasses a cascade of phosphorylation events involv- are three isoforms of RAF, namely, A-Raf, BRAF, ing three key kinases, namely Raf, MEK (MAP kinase and CRAF (also called Raf-1) (reviewed in [3,41,42, kinase) and ERK (MAP kinase). The critical nuclear 45]). Once activated, RAF stimulates the MAPK cas- target of the Ras-Raf-MEK-MAP kinase pathway are cade, resulting in the sequential activation of MEK1 the transcription factors. Another well-characterized and MEK2, which in turn activates ERK1 and ERK2 effector of Ras is the phosphoinositide 3-phosphate (Fig. 4). Once activated, the ERKs either activate cy- lipid kinase (PI3K)-AKT/PKB pathway (Fig. 4). toplasmic targets or migrate to the nucleus, where they Each human cell contains at least three distinct ras phosphorylate transcription factors. In melanocytes, proto-oncogenes, which encode four highly related but the MAPK (ERK) pathway is activated by growth fac- T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma 275

Fig. 4. The Ras/RAF/MEK/ERK and the PI3K/AKT signalling pathways in melanoma. Ligand-mediated dimerization of the receptor tyrosine kinases (RTKs) and subsequent autophosphorylation or transphosphorylation leads to their association with a variety of cytoplasmic phosphoty- rosine binding proteins. This results in the initiation of a phosphorylation cascade and activation of several downstream pathways involved in cell growth and survival, including the Ras/Raf/MEK/ERK and PI3K/Akt pathways. Stimulation of these pathways transmits a signal to the nucleus resulting in modification of gene transcription patterns that ultimately affects processes such as cell division, apoptosis, adhesion, migration, and/or differentiation. tors released from the local microenvironment, such as ically sun-damaged skin, but frequent in skin with in- stem cell factor (SCF), α-melanocyte-stimulating hor- termittent sun exposure and often accompanied by am- mone (α-MSH), and hepatocyte growth factor (HGF). plification of the mutant allele [47]. The importance of Under physiological conditions, these growth factors localization is underlined by the observation that B-Raf only induce a weak stimulation of the MAPK (ERK) mutations do not occur in melanomas of the uvea [48]. pathway that is insufficient to induce melanocyte pro- Furthermore, the frequency of BRAF mutations in liferation. In most melanoma cells, the situation is very melanoma is positively linked with genetic variants of different and it has been shown that > 90% of clinical the melanocortin-1 receptor in melanocytes [49]. Ad- melanoma specimens have continuous hyperactivity in ditionally, melanomas with wild-type BRAF or N-RAS the MAPK (ERK) pathway. frequently have increases in the number of copies of Particularly BRAF has recently attracted enormous the genes for cyclin-dependent kinase 4 (CDK4)and biological and therapeutical interest, as the b-raf cyclin D1 (CCND1), downstream components of the gene is found mutated in at least 60% of cutaneous RAS-BRAF pathway [19]. What is more, constitutive melanomas [44], and at a lower frequency in many MAPK (ERK) activity increases cyclin D1 and down- other human malignancies. This discovery has firmly regulates p27 expression in melanoma cells [50]. established the involvement of Raf kinases in cancer. Interestingly, however, a high frequency of BRAF The most common mutation (over 80%) is a V600E V600E mutations (73–82%) has also been reported change in the activation loop that induces the constitu- in common benign naevi, suggesting that BRAF active activation of catalytic activity [46]. Curiously, in tivation might be the critical inducer of these benign melanoma this mutation is rare in unexposed or chron- ‘precursor’ lesions [51–53]. Hence, acquisition of the 276 T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma

BRAF V600E mutation seems to be an early event MAPK signalling from BRAF to CRAF (also called in melanoma development with a high percentage of Raf-1) is suggestive of a mechanism by which resis- melanocytic nevi also found to harbor the mutation. In- tance to BRAF inhibitors may be acquired and could al- terestingly, the presence of the BRAF V600E mutation so potentially explain side effects observed with BRAF alone is not sufficient to oncogenically transform naevi inhibitors [63–65]. cells into melanoma [54]. Instead, the forced expression of BRAF V600E in primary human melanocytes leads to an irreversible growth arrest characteristic of 6. PTEN and the phosphatidylinositol 3-kinase senescence [55]. Pathological studies have confirmed (PI3K)/Akt signaling pathway these in vitro findings, and showed that most nevi are growth arrested and express many markers of senes- PTEN (phosphatase and tensin homolog delet- cence [55]. This process, termed “oncogene-induced ed on chromosome 10), also known as MMAC senescence,” is thought to be an important mechanism 1 (mutated in multiple advanced cancers 1) or in protecting cells from malignant change [3]. TEP1 (TGFβ-regulated and epithelial cell enriched It has been suggested that the presence of a BRAF phosphatase 1) was recently identified as a tu- V600E mutation, rather than an NRAS or low-activity mor suppressor gene located at 10q23.3 and has BRAF mutation, predicts for sensitivity to both MEK been shown to be deleted or mutated in a variety and BRAF inhibitors. [4,56,57]. Recent investigations of advanced malignant tumors (see [3–5,66,67] for have also sought to determine the prognostic relevance more comprehensive reviews). PTEN is a dual- of BRAF mutations in melanoma. However, studies specificity phosphatase capable of dephosphorylating to date have failed to identify a link between BRAF both tyrosine phosphate and serine/threonine phos- mutational status and disease outcome [58]. In another study, patients with BRAF mutation showed longer phate residues on proteins. It also functions as a disease-free survival (median of 12 months) than pa- major lipid phosphatase, counteracting phosphatidyli- tients without mutation (median of 5 months), although nositol 3-OH kinase (PI3 kinase) by dephosphory- this association was not statistically significant [59]. lating the second messengers phosphatidylinositol- From such tiny samplings it would be reckless to draw 3,4,5-triphosphate (PIP3) and phosphatidylinositol- definite conclusions, but it is interesting to note that 3,4-diphosphate (PIP2), which are required for the ac- thus far no molecule has emerged as a bona fide pre- tivation of PKB/Akt, an important mediator of cell sur- dictive or prognostic maker in melanoma for routine vival protecting cells from apoptosis (Fig. 4). PTEN clinical use. has been shown to be involved in cell migration, There are also possibilities that BRAF/MAPK sig- spreading, and focal adhesion formation through di- naling may allow melanoma cells to escape immune rect dephosphorylation and inactivation of focal ad- surveillance through suppression of their highly im- hesion kinase (FAK). In addition, PTEN inhibits munogenic pigmentation antigens [60]. Moreover, Shc phosphorylation, thus preventing association of it has been shown that treating melanoma cells with Shc with Grb2/Sos and subsequent activation of the an RNAI to V600E BRAF mutation reduces the re- Ras/raf/MEK1/MAPK pathway. PTEN suppresses the lease of immunosuppressive cytokines from melanoma stabilization of hypoxia-mediated HIF-1α (hypoxia- cells [61]. inducible favor 1α), which when stabilized through The melanoma MAPK (ERK) signalling pathway is the PI3K/AKT pathway, upregulates VEGF expression. highly robust, with BRAF-V600E mutated melanoma This suggests a possible role for PTEN in angiogene- cells rapidly developing resistance to BRAF inhibition sis. While some of the functions of PTEN have been in vitro [62]. Rather than this resulting from an ac- delineated, the nature of the factors involved in PTEN’s quired genetic change in the melanoma cells, resis- decision towards inducing cell-cycle arrest or apopto- tance instead results from a simple switching within sis are still elusive. Germline mutations of PTEN have the MAPK pathway so that the signal is routed from been found in the autosomal dominant Cowden syn- BRAF to CRAF (Fig. 4). Studies on melanoma cell drome, which is characterized by multiple hamartomas, lines harbouring NRAS mutations have already shown especially of the skin, and an increased risk of breast that melanomas can use CRAF as a mechanism to ac- (25–50% of affected females) and thyroid (10% of af- tivate downstream MAPK signalling [63]. The pres- fected individuals) cancers as well as, to a lesser extent, ence of a network connection that can easily switch malignant melanomas. T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma 277

Loss of chromosome 10q has been detected in 30 equally important (reviewed in [3–5,66,67]). To to 50% of both early- and advanced-stage sporadic date, the most widely studied of these is the phospho- melanomas, and has been associated with poor clini- inositide 3-kinase (PI3K)/protein kinase B (AKT) path- cal outcome. There is a host of studies examining the way [70,71]. Activation of the PI3K/AKT pathway genomic status of PTEN in melanoma cell lines and in melanoma occurs through either paracrine/autocrine non-cultured melanomas of various stages, with incon- growth factors or the loss of expression and/or muta- gruous results (reviewed in [3–5,66,67]). Overall it ap- tion of negative pathway regulators. In particular, the pears that homozygous deletions and intragenic muta- insulin-like growth factor-1 (IGF-1) is known to aid the tions of PTEN occur in the metastatic setting, especial- growth of early-stage melanoma cells, at least in part, ly in cell lines. Typically, PTEN expression is lost in through the activation of PI3K/AKT [72]. up to 30% of melanoma cell lines and 10% of human The AKT family consists of three members, AKT1– tumor material. There is also evidence of inactivating 3 [73], which exhibit different expression patterns PTEN mutations in > 10% of short-term melanoma depending on the cell type. Of these, 43–50% cell cultures [68]. One study examined the mutational of melanomas have selective constitutive activity in status of both PTEN and NRAS in 53 melanoma cell AKT3 [70]. There is evidence that overexpression of lines. The authors found in 30% alterations in PTEN AKT3 may occur as a result of copy number increases and 21% to harbor activating NRAS mutations, with in the long arm of chromosome 1 [74]. Another mech- only 1 cell line having mutations in both, suggesting anism for PI3K/AKT pathway activation in melanoma that PTEN and NRAS may function in the same path- is through the acquisition of activating E17K mutations way. PTEN can also function as a haploinsufﬁcient in AKT3, but this is thought to be relatively rare [75]. tumor suppressor, and it is known that allelic loss of AKT has a critical role in cancer development through PTEN occurs in at least 58% of melanoma metastases. its ability to regulate apoptosis through the direct phos- With the accumulating knowledge of PTEN inactiva- phorylation of BAD as well as it effects many oth- tion, it appears that there are several mechanisms which er pathways, including the inhibition of forkhead sig- lead to PTEN inactivation: Initial work on tumor cell nalling and the inhibition of glycogen synthase kinase- lines almost uniformly suggested that PTEN intragenic 3 [76]. As already mentioned, one of the most critical mutations, homozygous deletions, and the two struc- regulators of AKT is the phosphatase and tensin ho- tural hits (one structural hit – LOH – followed by the molog (PTEN), which degrades the products of PI3K, second epigenetic hit) would be the rule across a large thereby preventing the activation of AKT. In summary, variety of tumors. In this regard, studies on melanomas the mechanism by which the PI3K-AKT pathway is ac- have shown that biallelic structural inactivation occurs tivated in melanoma is not yet fully elucidated, but may by either homozygous deletion at 10q23 or somatic involve the loss of expression or functional inactivation intragenic PTEN mutation plus loss of the remaining of PTEN. wild-type allele. Zhou et al. [69] have proposed that non-cultured melanomas might be somewhat unique in 6.2. Entente fatale: For melanoma progression both PI3K/AKT and MAPK (ERK) signalling are that, although PTEN inactivation is seen to occur by required one structural hit (LOH) followed by the second epigenetic hit, PTEN protein expression in melanomas can Data are now emerging from both pre-clinical and be biallelically silenced with relatively high frequency clinical studies that the MAPK (ERK) and the PI3K- by epigenetic phenomena as well: Examination of 4 AKT pathways have overlapping functions with regard primary and 30 metastatic melanoma showed little to to melanoma biology (Fig. 4). Analysis of melanoma no expression of PTEN in 65% of the cases. These re- lines and human melanoma samples have shown that sults would support PTEN as a major tumor suppressor those mutational proﬁles are favoured which simul- on 10q involved in melanoma tumorigenesis. taneously promote both MAPK and PI3K/AKT signalling [71]. As activating NRAS mutations can stim- 6.1. Activation of the PI3K/AKT pathway ulate both the MAPK (ERK) and PI3K/AKT pathways, they are rarely found in concert with BRAF mutations Activation of the MAPK (ERK) pathway clearly does and these mutations are mutually exclusive [19]. Like- not account for all aspects of melanoma biology and wise, BRAF mutations are frequently found in combi- there is evidence that other signaling pathways may be nation with either PTEN loss/inactivation or activating 278 T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma

AKT3 mutations, again showing a requirement for dual known to recruit and activate a number of intracellu- MAPK (ERK) and PI3K/AKT pathway activity [68,71, lar signalling pathways implicated in tumor progres- 75]. sion, such as phosphoinositide-3 kinase (PI3K)/AKT, AKT3 seems to cooperate with the BRAF V600E Src, mitogen-activated protein kinase (MAPK), janus mutation in melanoma initiation by suppressing MAPK kinase (JAK)/signal transducers and activators of tran- signalling activity through phosphorylation at Ser364/ scription (STAT) and phospholipase-C (PLC)-γ [82]. 428 of BRAF [77]. It is therefore suggested that The role of c-KIT signalling in melanoma has been AKT3’s ability to downregulate MAPK (ERK) path- controversial; although c-KIT activity is critical to way activity may allow early-stage melanoma cells to melanocyte development and migration, its expression escape from oncogeneinduced senescence. To date, tends to be lost in most melanomas [83]. Progres- the strongest experimental proof of concept for the sion of melanoma towards the invasive and metastat- cooperation between the BRAF V600E mutation and ic phenotype is associated with loss of expression of PI3K/AKT signalling in melanoma development is de- c-kit (see [11,82,84] for review). c-Kit plays a piv- rived from the mouse modelling work of Dankort and otal role in normal growth and differentiation of em- co-workers [78]. These authors have elegantly shown bryonic melanoblasts. SCF is highly mitogenic for hu- that the conditional expression of melanocyte-specific man melanoblasts and melanocytes in culture and for mutated BRAF alone led to the development of be- melanocyes in vivo. In mice, c-kit has been mapped nign melanocytic hyperplasia but not melanoma. Full to the dominant white spotting (w) locus; its ligand is transformation to melanoma occurred only when the the product of the steel (sl) locus on chromosome 10. BRAF-V600E mutation was activated in the presence Mutations in the w or sl locus, or injection of neutral- of PI3K/AKT activity after suppression of PTEN ex- izing anti-c-KIT antibodies into pregnant mice, results pression. in the piebald phenotype, characterised by white spot- Hence, the intracellular melanoma signalling net- ting of the fur and attributed to a local reduction of the work is highly interconnected with a great deal of func- number of cutaneous melanocytes. Mutations in the tional redundancy built into it (Fig. 4). Studies have c-kit receptor also have been identified in human pa- already shown that single-agent MEK inhibition is rel- tients with piebaldism, suggesting that normal function atively ineffective at inducing melanoma regression in of c-kit is required for human melanocyte development both pre-clinical models [56] and in early-phase clinical and migration [11,82]. trials (reviewed in [79]. The disappointing clinical ac- Several early pathological studies have demonstrat- tivity seen after single-agent MEK inhibition parallels ed that the progression of melanoma is associated with with that observed with single-agent sorafenib [80],and a loss of expression of c-kit [11,82]. The vast major- thus clearly argue in favour of a combination of several ity of metastatic lesions and cell lines do not express targeted agents. Smalley et al have identified other sites detectable levels of the c-kit receptor. Transfection of redundancy within the melanoma signaling network, of c-kit into c-kit negative, highly metastatic human with cyclin D1 amplifications being found in concert melanoma cells significantly inhibited tumor growth with BRAF V600E mutations in minor sub-groups of and metastasis in nude mice [85]. Exposure of c-kit- melanomas [81]. positive melanoma cells in vitro and in vivo to SCF triggered apoptosis of these cells but not of normal melanocytes. Loss of c-kit receptor may thus allow 7. Stem cell factor (SCF)/c-kit signalling pathway melanoma cells to escape SCF/c-kit-mediated apopto- as the driving oncogenic event in distinct sis [85]. On the basis of the aforementioned results subtypes of melanoma it was assumed that c-KIT was primarily a regulator of melanocyte behavior and therefore dispensable for c-KIT is a transmembrane receptor tyrosine kinase melanoma growth. (RTK) to the PDGF/CSF-1 receptor subfamily, whose However, the recent publication showing the pres- aberrant activation is implicated in the progression ence of activating c-KIT aberrations (either amplifi- of gastrointestinal stromal tumors (GIST) and some cation and/or mutation) in mucosal (39%) and acral acute myeloid leukemias (reviewed in [82]). The c- (36%) melanomas, as well as melanomas arising on KIT receptor ligand is the glycoprotein Stem Cell skin with chronic sun damage (28%), has renewed in- Factor (SCF), which is also known under a variety terest in c-KIT signaling in melanoma [86]. In ad- of other names including steel factor (SF). c-KIT is dition, subsequent studies have shown that c-KIT is T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma 279

Table 1 Selected genetic alterations in cutaneous melanoma Gene Alteration frequency/type(s) References BRAF 50–70% mutated (e.g., V600) [44] NRAS 15–30% mutated [44,71] AKT3 overexpressed; [70] 43–50% selective constitutive activity Cyclin D1 6–44% amplified (depending on subtype: chronic sun exposure/face, [19,118] acral and mucosal melanoma) CDK4 amplification commonly seen in acral and mucosal melanomas [19] MITF 10–16% amplified [119] Notch overexpressed [110,112] c-met overexpression [97,98] c-kit amplification and/or mutation: 22–39% (mucosal), 36% (acral), 28% [86–88] (sun-damaged skin), 15% (anal) c-myc overexpression [100,120,121] CDKN2A (p16INK4A) 30–70% deleted, mutated or silenced [122] PTEN 5–20% deleted or mutated [71,101,123] APAF-1 40% silenced [36] p53 10% lost or mutated [124] expressed in 88% of oral mucosal melanomas, and a multifunctional cytokine acting as a mitogen, moto- that at least 22% of these harbored activating muta- gen, and morphogen for many epithelial cells. HGF is tions [87]. In these instances, most of the c-KIT ex- physiologically secreted by cells of mesenchymal ori- pression was in the in situ component, with strong ex- gin and acts on neighboring epithelial cells through a pression reported in the invasive portion in only 22% of paracrine loop [91]. Hepatocyte growth factor (HGF) is cases. Another recent paper also reported the presence a multifunctional cytokine acting as mitogen, motogen of the activating L576P mutation in c-KIT in 15% of and morphogen for many epithelial cells through its anal melanomas, a mutation that the authors showed tyrosine kinase receptor c-Met, which is present in ep- to be imatinib sensitive in vitro [88]. What’s more, ithelial cells and melanocytes. HGF is physiologically work from our own laboratory has further identified a secreted by cells of mesenchymal origin and acts on subset of melanomas that lack BRAF mutations, with neighboring epithelial cells through a paracrine loop. constitutive c-KIT signaling activity resulting from c- However, coexpression of HGF and c-Met occurs in a KIT/CDK4 co-overexpression [89]. Although the ini- variety of transformed cells and tumors both in vitro tial clinical trials of the c-KIT inhibitor imatinib me- and in vivo and has been shown to be involved in tumor sylate in melanoma were negative, some dramatic re- development and invasion (reviewed in [11]). sponses have been seen in patients with very high c- HGF/c-Met signaling has been implicated in the de- KIT expression and/or documented activating muta- velopment of melanoma [92]. HGF stimulates prolifer- tions, fostering the belief that focused studies in pa- ation and motility of human melanocytes [93]. In trans- tients selected on the basis of c-KIT mutational status genic mice that ubiquitously expressed HGF, ectopic will yield more encouraging results [82]. localization of melanocytes and hyperpigmentation in skin were observed and melanoma arose spontaneous- 8. Hepatocyte growth factor (HGF)/scatter factor ly. In these mice, ultraviolet radiation-induced carcino- (SF)/c-Met tyrosine kinase receptor genesis was accelerated( reviewed in [94]). It has been suggested that c-Met autocrine activation induced de- c-Met is a receptor tyrosine kinase (RTK) that has velopment of malignant melanoma and acquisition of been known to stimulate the invasive growth of can- the metastatic phenotype [95]. cer cells, increase their metastatic potential, and is al- Many missense mutations of c-Met have been report- so known to be expressed and mutated in a variety of ed in a variety of cancers, with the majority of them solid tumors [90]. The structure of c-Met consists of identified in the cytoplasmic activation loop tyrosine adisulfide-linked α − β heterodimer with a molecular kinase domain. c-Met activating point mutations in the weight of 190 kDa (see [11] for review). The physio- kinase domain are implicated as the cause of hereditary logical ligand for c-Met is the hepatocyte growth factor papillary renal carcinoma. In addition, kinase domain (HGF), which is also known as scatter factor. HGF is mutations have been observed in sporadic papillary re- 280 T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma nal carcinoma, ovarian cancer, childhood hepatocellu- c-myc spontaneously formed macroscopic metastases lar carcinoma, and gastric cancer [90]. (lymph nodes and lung) in severe combined immun- In normal skin, c-Met is present on epithelial cells odeficient (SCID) mice, whereas less prominent c-myc and melanocytes, whereas HGF is produced mainly by overexpression caused the development of only lung mesenchymal cells and, consequently, interacts with its micrometastases [103]. These findings suggest that receptor in a paracrine manner [96]. HGF is a mitogen constitutive high c-myc expression in human melanoma of human melanocytes, and overexpression of c-Met results in a more aggressive growth behavior both in correlates with the invasive growth phase of melanoma vitro and in vivo, and favors metastasis. cells [97]. c-Met is overexpression is also associated More recently, it has been suggested that one of the with patient survival [98]. Studies by our group have major functions of c-myc overexpression in melanoma shown that most of melanoma cells, but not normal progression is to continuously suppress BRAFV 600E melanocytes, produce HGF,which can induce sustained or NRASQ61R-dependent oncogene-induced senes- activation of its receptor [91]. Hence, an autocrine cence [104]. HGF/c-Met signaling loop may be involved in the development of melanomas. Consistent with this, pro- longed HGF stimulation induces the down-regulation 10. Notch signaling and melanoma of the intercellular adhesive molecule E-cadherin that is implicated in the control of melanocyte prolifera- The Notch signaling pathway is an evolutionarily tion [91]. Finally, in transgenic mice that ubiquitous- conserved signaling cascade that affects cell fate deci- ly expressed HGF, ectopic localization of melanocytes sions and many differentiation processes during both and hyperpigmentation in skin were observed, and embryonic and postnatal development (see [105] for melanoma arose spontaneously. In these mice, UV review). Notch signaling is critical for developing and radiation-induced carcinogenesis was accelerated [99]. maintaining tissue homeostasis. Its pathway compris- It is suggested that c-Met autocrine activation induced es a family of transmembrane receptors and their lig- the development of malignant melanoma and the acqui- ands, negative and positive modifiers, and transcrip- sition of the metastatic phenotype [95]. More recent- tion factors. To date, 4 mammalian receptors (Notch1 ly, Puri and co-workers showed that a small-molecule through Notch4) and at least 5 ligands (Delta 1, 3, tyorsine kinase inhibitor of c-met inhibits growth of and 4 and Jagged 1, 2) have been identified. Bind- melanoma by causing apoptosis and induces differen- ing of the ligand renders the Notch receptor suscepti- tiation [90]. ble to metalloprotease- and γ-secretase–mediated pro- teolytic cleavage, which in turn results in the release of the Notch intracellular domain (NIC) from the plas- 9. c-myc ma membrane and its subsequent translocation into the nucleus. c-myc is a gene coding for a nuclear phosphopro- Notch signaling has also been implicated in a vari- tein that is involved in cell proliferation and differ- ety of neoplastic malignancies reviewed in [106–108]). entiation. c-myc acts as a central regulator of cellu- Notch can function as a tumor promoter or a suppres- lar proliferation and is sufficient to induce G1 entry sor depending on the cell type and context. In exper- and cell proliferation in the absence of external growth imental mouse models of skin tumorigenesis, Notch1 factors (reviewed in [11]). Moreover, c-myc func- shows antitumor activity. Ablation of Notch1 in ker- tions as an inducer of apoptotic cell death under con- atinocytes leads to spontaneous development of basal ditions of growth factor deprivation or in the presence cell carcinoma–like tumors, likely as a result of reacti- of wild-type p53 protein. In primary melanomas, over- vation of the β-catenin signaling pathway [109]. expression correlates with tumor thickness, the pres- Recent data suggest that Notch activation may al- ence of ulceration and disease-free survival. Overex- so play a role in melanoma progression. Our group pression of c-myc also predicts poor outcome in both has previously shown that activation of Notch signal- primary and metastatic disease [71,100–102]. Trans- ing promotes the progression of early-stage melanoma fection of c-myc into a melanoma cell line led to in- cell lines in a β-catenin–dependent manner both in vit- creased growth rates, anchorage independent growth ro and in vivo [110]. Blocking Notch signaling sup- and directed cell movement in culture. Subcutaneously pressed whereas constitutive activation of the Notch1 implanted IGR39D melanomas highly overexpressing pathway enhanced primary melanoma cell growth both T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma 281 in vitro and in vivo yet had little effect on metastat- melanocytes and define the importance of Notch sig- ic melanoma cells. Activation of Notch1 signaling naling in human melanoma. enabled primary melanoma cells to gain metastat- Gene expression analyses supported these observa- ic capability. Furthermore, the oncogenic effect of tions and aided in the identification of MCAM, an ad- Notch1 on primary melanoma cells was mediated by hesion molecule associated with acquisition of the ma- β -catenin, which was upregulated following Notch1 lignant phenotype, as a direct target of Notch trans- β activation. Inhibiting -catenin expression reversed activation. NIC-positive melanocytes grew at clonal Notch1-enhanced tumor growth and metastasis, there- density, proliferated in limiting media conditions, and fore suggesting a β-catenin–dependent, stage-specific also exhibited anchorage-independent growth. Hence, role for Notch1 signaling in promoting the progression Notch alone is a transforming oncogene in human of primary melanoma. What’s more, microarray profiling comparing the melanocytes, a phenomenon not previously described gene expression of normal human melanocytes to hu- for any melanoma oncogene [112]. man melanomas revealed up-regulation of Notch tar- Our aforementioned microarray data describing get genes in melanoma cells, suggesting activation Notch pathway activity is underscored by a recent of the Notch signaling pathway in melanoma [111]. report by Hoek and colleagues that revealed up- More recently, an analysis of cell lines and patient le- regulation of Notch2 and Hey1 in a separate but dis- sions showed that Notch activity is significantly higher tinct panel of malignant melanoma cell lines, sug- in melanomas than their nontransformed counterparts. gesting a role for Notch activation in the transforma- Notch and Notch target genes are up-regulated in both tion of melanocytes [111]. Previous immunohisto- melanoma lesions and melanoma cell lines. Notch re- chemical studies on early-phase melanoma lesions have ceptors 1, 2, and 4 are overexpressed in melanoma cell shown overexpression of full-length Notch1 protein lines and lesions, particularly when compared against in melanoma tissue compared against benign human primary melanocytes or normal human skin [112]. nevi [110] and normal human skin [113]. Moreover,Notch1 signaling is activated in melanoma What’s more, our current studies strongly suggest cells, but not melanocytes, and constitutive Notch1 that constitutive Notch signaling is associated with activation confers transforming properties to primary melanocytes in vitro. The use of a constitutively active, melanocyte transformation and melanoma tumorigen- γ truncated Notch transgene construct (NIC) was exploit- esis. Of particular significance is the ability of a - ed by Pinnix and co-workers to determine if Notch ac- secretase inhibitor to selectively inhibit the growth of tivation is a “driving” event in melanocytic transfor- melanoma cell lines. These findings are consistent with mation or instead a “passenger” event associated with a recent report identifying a γ-secretase inhibitor that melanoma progression. NIC-infected melanocytesdis- induced effective apoptosis in human melanoma cells played increased proliferative capacity and biologi- while sparing melanocytes [114]. cal features more reminiscent of melanoma, such as These findings were recently confirmed and expand- dysregulated cell adhesion and migration. Specifi- ed by Bedogni and co-workers who found that Notch1 IC cally, ectopic N expression induced gross morpho- signaling is elevated in human melanoma samples and logic changes, increased growth, adhesion, migration, cell lines and is required for Akt and hypoxia to trans- and survival, and resulted in the loss of E-cadherin form melanocytes in vitro [115]. Notch1 facilitated expression and up-regulation of MCAM, two well- melanoma development in a xenograft model by main- characterized events in melanoma development. In ad- taining cell proliferation and by protecting cells from dition, MCAM was identified as a direct Notch target stress-induced cell death. Hyperactivated PI3K/Akt due to the presence of two high-affinity CSL binding IC signaling led to upregulation of Notch1 through NF- sites present in the MCAM promoter. The N onco- κ protein conferred anchorage-independent growth, in- B activity, while the low oxygen content normally creased survival, and loss of contact inhibition; sup- found in skin increased mRNA and protein levels of α pression of Notch signalling decreased the growth of Notch1 via stabilization of HIF-1 . Taken together, melanoma cell lines, whereas primary melanocytes these findings demonstrate that Notch1 is a key effec- were unaffected. Taken together, these data suggest tor of both Akt and hypoxia in melanoma development that deregulation of Notch signaling plays a specific and identify the Notch signaling pathway as a potential role in promoting a transformed phenotype in human therapeutic target in melanoma treatment. 282 T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma

11. Conclusions and outlook: A changing tions of cyclin D1 also occur in distinct histologic sub- treatment paradigm based on molecular types of melanoma [118]. classification As we have seen, multiple studies support the ex- istence of distinct genetic pathways associated with Cutaneous melanoma can be perceived as a disease melanoma development, mirroring different molecu- of communication between and within skin cells. The lar and clinical behaviour. The complexity of the sig- molecular aberrations are pleiotropic, but growth factor nalling observed in melanoma,and outlined in this book receptor and mitogen-activated protein kinase (MAPK) chapter, calls for a network-biology approach and a pathways feature prominently. Following the discov- combination of targeted therapies. The challenge for ery that the overwhelming proportion of cutaneous melanoma investigators is to identify the key signalling melanomas have constitutive activity in the extracellu- “hubs” or relay stations in this melanoma network. Al- lar signal regulated kinase (ERK) pathway, there has though we have made great strides in unravelling the been considerable interest in pharmacologically target- mysteries of melanoma genetics and molecular biology, the important task now facing clinicians and researchers ing this pathway using small molecule inhibitors [3, is to translate these discoveries into novel therapeutic 67]. Although there is evidence to suggest that the strategies that will improve patient outcomes. presence of the BRAF V600E mutation is predictive of During the past three decades the translation of response to BRAF and MEK inhibitors [116], recent molecular and immunological studies in melanoma bi- clinical trials on MEK inhibitors have not led to the ology into useful clinical correlates or novel effica- expected favorable results [3,4]. BRAF/MAPK sig- cious therapies has been overwhelmingly disappoint- naling may be more heterogeneous than first thought ing. Even though molecular markers and novel thera- and locally regulated by the microenvironment [56, peutic compounds look promising in small-scale stud- 104]. It also is possible that other factors, such as en- ies, the vast majority have failed to prove clinically use- hanced phosphoinositide-3-kinase/AKT signaling ac- ful in large-scale (phase III) studies. Today, in 2010, we tivity, may further influence response to BRAF/MEK clearly stand at the crossroads of melanoma therapy. In inhibition [117]. view of the plethora of targeted agents currently under As yet, very little is known about the factors un- clinical investigation and our emerging knowledge of derlying resistance to BRAF inhibition in the BRAF molecular sub-classifications, we may already have the V600E–mutated melanoma population. A greater un- pharmacologic arsenal that could control melanoma if derstanding of the genetic basis of response to BRAF targeted to molecularly defined patient subgroups or and other inhibitors is essential in selecting the most if administered in biologically appropriate combina- appropriate patient sub-population for future clinical tions. Clearly, therefore, pharmacogenomic analysis studies and developing strategies to overcome inherent of melanoma populations may be a suitable strategy resistance. for the further subclassification of melanoma, ultimate- Although many melanomas harbor either activating ly leading to more “personalized” therapy approach- mutations in BRAF or NRAS, there remains a sub- es in the future. Lastly, to improve the outcome of stantial, yet little known, group of tumors without ei- melanoma and to determine the biological mechanisms ther mutation. We have recently suggested that co- of efficacy or failure, future clinical studies with target- overexpression of KIT/CDK4 is a potential mechanism ed agents will also require more stringent monitoring of oncogenic transformation in some BRAF/NRAS of biological endpoints. wild-type melanomas. This group of melanomas may be a subpopulation for which KIT inhibitors may con- stitute optimal therapy [89]. References We now also realize that there are differences in the genetic profiles of melanomas that originate from skin [1] A.M. DeLuca, A. Srinivas and A. Alani, BRAF kinase in melanoma development and progression, Expert Rev Mol that is either chronically sun-damaged (as defined by Med 10 (2008), e6. the appearance of solar elastosis) or skin that lacks [2] K.S. Smalley and M. Herlyn, Integrating tumor-initiating sun-induced damage [4]. Thus, melanomas that arise cells into the paradigm for melanoma targeted therapy, Int J on skin with chronic sun-induced damage have a low Cancer 124 (2009), 1245–1250. [3] K.S. Smalley, Understanding Melanoma Signaling Networks incidence of BRAF mutations and instead showed in- as the Basis for Molecular Targeted Therapy, J Invest Der- creased cyclin D1 copy number. Frequent amplifica- matol (2009). T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma 283

[4] K.S. Smalley, K.L. Nathanson and K.T. Flaherty, Genetic [24] J. Georgieva, P. Sinha and P. Schadendorf, Expression of subgrouping of melanoma reveals new opportunities for tar- cyclins and cyclin dependent kinases in human benign and geted therapy, Cancer Res 69 (2009), 3241–3244. malignant melanocytic lesions, J Clin Pathol 54 (2001), 229– [5] A. Sekulic, P.Jr. Haluska, A.J. Miller et al., Malignant 235. melanoma in the 21st century: the emerging molecular land- [25] G.M. Maelandsmo, R. Holm, O. Fodstad et al., Cyclin kinase scape, Mayo Clin Proc 83 (2008), 825–846. inhibitor p21WAF1/CIP1 in malignant melanoma: reduced [6] E. Bajetta, V.M. Del, C. Bernard-Marty et al., Metastatic expression in metastatic lesions, Am J Pathol 149 (1996), melanoma: chemotherapy, Semin Oncol 29 (2002), 427–445. 1813–1822. [7] L. Serrone, M. Zeuli, F.M. Sega et al., Dacarbazine-based [26] D. Hanahan and R.A. Weinberg, The hallmarks of cancer, chemotherapy for metastatic melanoma: thirty-year experi- Cell 100 (2000), 57–70. ence overview, J Exp Clin Cancer Res 19 (2000), 21–34. [27] R.A. Hussein, A.K. Haemel and G.S. Wood, Apoptosis and [8] W. Sun and L.M. Schuchter, Metastatic melanoma, Curr melanoma: molecular mechanisms, J Pathol 199 (2003), Treat Options Oncol 2 (2001), 193–202. 275–288. [9] L.M. Haass and M. Herlyn, Normal human melanocyte [28] M.S. Soengas and S.W. Lowe, Apoptosis and melanoma homeostasis as a paradigm for understanding melanoma, J chemoresistance, Oncogene 22 (2003), 3138–3151. Investig Dermatol Symp Proc 10 (2005), 153–163. [29] S.W. Soengas, R.M. Alarcon, R.M. Yoshida et al., Apaf-1 and [10] N.K. Haass, K.S. Smalley, L. Li et al., Adhesion, migra- caspase-9 in p53-dependent apoptosis and tumor inhibition, tion and communication in melanocytes and melanoma 56, Science 284 (1999), 156–159. Pigment Cell Res 18 (2005), 150–159. [30] K.S. Smalley, K.S. Contractor, N.K. Haass et al., An [11] T. Bogenrieder, T. Elder and M. Herlyn, Molecular and cel- organometallic protein kinase inhibitor pharmacologically lular Biology, in: Cutaneous Melanoma, C.M. Balch, A.N. activates p53 and induces apoptosis in human melanoma Houghton, A.J. Sober and S.J. Soong, eds, 2003, p. 751. cells, Cancer Res 67 (2007), 209–217. [12] K.S. Smalley, M. Lioni and M. Herlyn, Targeting the stromal [31] M.S. Soengas and S.W. Lowe, p53 and p73: seeing double? fibroblasts: a novel approach to melanoma therapy, Expert Nat Genet 26 (2000), 391–392. Rev Anticancer Ther 5 (2005), 1069–1078. [32] M.R. Hussein, A.K. Haemel and G.S. Wood, Apoptosis and [13] M. Thompson, M. Scolyer and M. Kefford, Cutaneous melanoma: molecular mechanisms, J Pathol 199 (2003), melanoma, Lancet 365 (2005), 687–701. 275–288. [14] M. Li, A. Sanki, A. Karim et al., The role of cell cycle regu- [33] S.W. Lowe and A.W. Lin, Apoptosis in cancer, Carcinogen- latory proteins in the pathogenesis of melanoma, Pathology esis 21 (2000), 485–495. (Phila) 38 (2006), 287–301. [34] K. Satyamoorthy, T. Bogenrieder and M. Herlyn, No longer [15] M. Malumbres, P. Pevarello, M. Barbacid et al., CDK in- a molecular black box–new clues to apoptosis and drug re- hibitors in cancer therapy: what is next? Trends Pharmacol sistance in melanoma, Trends Mol Med 7 (2001), 191–194. Sci 29 (2008), 16–21. [35] K. Satyamoorthy, N.H. Chehab, M.J. Waterman et al., Aber- [16] M. Malumbres and M. Barbacid£‹ Cell cycle, CDKs and rant regulation and function of wild-type p53 in radioresistant cancer: a changing paradigm, Nat Rev Cancer 9 (2009), melanoma cells, Cell Growth Differ 11 (2000), 467–474. 153–166. [36] M.S. Soengas, P. Capodieci, D. Polsky et al., Inactivation of [17] G.J. Walker, J.F. Flores, J.F. Glendening et al., Virtually the apoptosis effector Apaf-1 in malignant melanoma, Nature 100% of melanoma cell lines harbor alterations at the DNA 409 (2001), 207–211. level within CDKN2A, CDKN2B, or one of their downstream [37] J.A. Bush and G. Li, The role of Bcl-2 family members in the targets, Genes Chromosomes Cancer 22 (1998), 157–163. progression of cutaneous melanoma, Clin Exp Metastasis 20 [18] J. Bartkova, J. Lukas, J. Guldberg et al., The p16-cyclin (2003), 531–539. D/Cdk4-pRb pathway as a functional unit frequently altered [38] K. Yamamura, S. Kamada, S. Ito et al., Accelerated disap- in melanoma pathogenesis, Cancer Res 56 (1996), 5475– pearance of melanocytes in bcl-2-deficient mice, Cancer Res 5483. 56 (1996), 3546–3550. [19] J.A. Curtin, J. Fridlyand, T. Kageshita et al., Distinct sets of [39] C.A. Schmitt, C.T. Rosenthal and S.W. Lowe, Genetic anal- genetic alterations in melanoma, N Engl J Med 353 (2005), ysis of chemoresistance in primary murine lymphomas, Nat 2135–2147. Med 6 (2000), 1029–1035. [20] T. Reed, F. Loganzo, Jr., F. Shea, Jr. et al., Loss of expres- [40] B. Jansen, H. Schlagbauer-Wadl, B.D. Brown et al., bcl-2 sion of the p16/cyclin-dependent kinase inhibitor 2 tumor antisense therapy chemosensitizes human melanoma in SCID suppressor gene in melanocytic lesions correlates with in- mice, Nat Med 4 (1998), 232–234. vasive stage of tumor progression, Cancer Res 55 (1995), [41] A.S. Dhillon, S. Hagan, O. Rath et al., MAP kinase signalling 2713–2718. pathways in cancer, Oncogene 26 (2007), 3279–3290. [21] D. Polsky, A.Z. Young, A.Z. Busam et al., The transcrip- [42] P.J. Roberts and C.J. Der, Targeting the Raf-MEK-ERK tional repressor of p16/Ink4a, Id1, is up-regulated in early mitogen-activated protein kinase cascade for the treatment of melanomas, Cancer Res 61 (2001), 6008–6011. cancer, Oncogene 26 (2007), 3291–3310. [22] A.Z. Florenes, A.Z. Faye, G.M. Maelandsmo et al., Levels [43] M.S. Brose, P. Volpe, M. Feldman et al., BRAF and RAS of cyclin D1 and D3 in malignant melanoma: deregulated mutations in human lung cancer and melanoma, Cancer Res cyclin D3 expression is associated with poor clinical outcome 62 (2002), 6997–7000. in superficial melanoma, Clin Cancer Res 6 (2000), 3614– [44] H. Davies, G.R. Bignell, C. Cox et al., Mutations of the 3620. BRAF gene in human cancer, Nature 417 (2002), 949–954. [23] G.M. Florenes, G.M. Maelandsmo, G.M. Kerbel et al., Pro- [45] N. Dhomen and R. Marais, BRAF signaling and targeted tein expression of the cell-cycle inhibitor p27Kip1 in ma- therapies in melanoma, Hematol Oncol Clin North Am 23 lignant melanoma: inverse correlation with disease-free sur- (2009), 529–545. vival, Am J Pathol 153 (1998), 305–312. 284 T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma

[46] P.T. Wan, M.J. Garnett, S.M. Roe et al., Mechanism of ac- [64] J.P. Arnault, J. Wechsler, B. Escudier et al., Keratoacan- tivation of the RAF-ERK signaling pathway by oncogenic thomas and squamous cell carcinomas in patients receiving mutations of B-RAF, Cell 116 (2004), 855–867. sorafenib, J Clin Oncol 27 (2009), e59–e61. [47] J.L. Maldonado, J. Fridlyand, H. Patel et al., Determinants of [65] K. Ehrenreiter, F. Kern, V. Velamoor et al., Raf-1 addiction BRAF mutations in primary melanomas, J Natl Cancer Inst in Ras-induced skin carcinogenesis, Cancer Cell 16 (2009), 95 (2003), 1878–1890. 149–160. [48] A. Weber, U.R. Hengge, D. Urbanik et al., Absence of [66] L.A. Fecher, S.D. Cummings, M.J. Keefe et al., Toward mutations of the BRAF gene and constitutive activation of a molecular classiﬁcation of melanoma, J Clin Oncol 25 extracellular-regulated kinase in malignant melanomas of the (2007), 1606–1620. uvea, Lab Invest 83 (2003), 1771–1776. [67] V. Gray-Schopfer, C. Wellbrock and R. Marais, Melanoma [49] M.T. Landi, J. Bauer, R.M. Pfeiffer et al., MC1R germline biology and new targeted therapy, Nature 445 (2007), 851– variants confer risk for BRAF-mutant melanoma, Science 857. 313 (2006), 521–522. [68] W.M. Lin, A.C. Baker, R. Beroukhim et al., Model- [50] K.V. Bhatt, L.S. Spofford, G. Aram et al., Adhesion control ing genomic diversity and tumor dependency in malignant of cyclin D1 and p27Kip1 levels is deregulated in melanoma melanoma, Cancer Res 68 (2008), 664–673. cells through BRAF-MEK-ERK signaling, Oncogene 24 [69] X.P. Zhou, O. Gimm, H. Hampel et al., Epigenetic PTEN (2005), 3459–3471. silencing in malignant melanomas without PTEN mutation, [51] P.M. Pollock, U.L. Harper, K.S. Hansen et al., High frequen- Am J Pathol 157 (2000), 1123–1128. cy of BRAF mutations in nevi 308, Nat Genet 33 (2003), [70] J.M. Stahl, A. Sharma, M. Cheung et al., Deregulated Akt3 19–20. activity promotes development of malignant melanoma, Can- [52] P. Uribe, I.I. Wistuba and S. Gonzalez, BRAF mutation: a cer Res 64 (2004), 7002–7010. frequent event in benign, atypical, and malignant melanocytic [71] H. Tsao, V. Goel, H. Wu et al., Genetic interaction between lesions of the skin, Am J Dermatopathol 25 (2003), 365–370. NRAS and BRAF mutations and PTEN/MMAC1 inactiva- [53] R. Kumar, S. Angelini, E. Snellman et al., BRAF mutations tion in melanoma, J Invest Dermatol 122 (2004), 337–341. are common somatic events in melanocytic nevi, J Invest [72] K. Satyamoorthy, G. Li, B. Vaidya et al., Insulin-like growth Dermatol 122 (2004), 342–348. factor-1 induces survival and growth of biologically early [54] P. Uribe, L. Andrade and S. Gonzalez, Lack of associa- melanoma cells through both the mitogen-activated protein tion between BRAF mutation and MAPK ERK activation in kinase and beta-catenin pathways, Cancer Res 61 (2001), melanocytic nevi, J Invest Dermatol 126 (2006), 161–166. 7318–7324. [55] C. Michaloglou, L.C. Vredeveld and M.S. Soengas et al., [73] D.P. Brazil, J. Park and B.A. Hemmings, PKB binding pro- BRAFE600-associated senescence-like cell cycle arrest of teins. Getting in on the Akt, Cell 111 (2002), 293–303. human naevi, Nature 436 (2005), 720–724. [74] B.C. Bastian, P.E. LeBoit, H. Hamm et al., Chromosomal [56] N.K. Haass, K. Sproesser, T.K. Nguyen et al., The mitogen- gains and losses in primary cutaneous melanomas detected by activated protein/extracellular signal-regulated kinase kinase comparative genomic hybridization, Cancer Res 58 (1998), inhibitor AZD6244 (ARRY-142886) induces growth arrest in 2170–2175. melanoma cells and tumor regression when combined with [75] M.A. Davies, K. Stemke-Hale, C. Tellez et al., A novel AKT3 docetaxel, Clin Cancer Res 14 (2008), 230–239. mutation in melanoma tumours and cell lines, Br J Cancer [57] K.S. Smalley and M. Herlyn, Targeting intracellular signaling 99 (2008), 1265–1268. pathways as a novel strategy in melanoma therapeutics, Ann [76] G.P. Robertson, Functional and therapeutic signiﬁcance of N Y Acad Sci 1059 (2005), 16–25. Akt deregulation in malignant melanoma, Cancer Metastasis [58] M. Shinozaki, A. Fujimoto, D.L. Morton et al., Incidence of Rev 24 (2005), 273–285. BRAF oncogene mutation and clinical relevance for primary [77] M. Cheung, A. Sharma, S.V.Madhunapantula et al., Akt3 and cutaneous melanomas, Clin Cancer Res 10 (2004), 1753– mutant V600E B-Raf cooperate to promote early melanoma 1757. development, Cancer Res 68 (2008), 3429–3439. [59] R. Kumar, S. Angelini, K. Czene et al., BRAF mutations [78] D. Dankort, D.P. Curley, R.A. Cartlidge et al., Braf(V600E) in metastatic melanoma: a possible association with clinical cooperates with Pten loss to induce metastatic melanoma, outcome, Clin Cancer Res 9 (2003), 3362–3368. Nat Genet 41 (2009), 544–552. [60] M. Kono, I.S. Dunn, P.J. Durda et al., Role of the mitogen- [79] K.S. Smalley and K.T. Flaherty, Integrating BRAF/MEK inactivated protein kinase signaling pathway in the regulation hibitors into combination therapy for melanoma, Br J Cancer of human melanocytic antigen expression, Mol Cancer Res 100 (2009), 431–435. 4 (2006), 779–792. [80] T. Eisen, T. Ahmad, K.T. Flaherty et al., Sorafenib in ad- [61] H. Sumimoto, F. Imabayashi, T. Iwata et al., The BRAF- vanced melanoma: a Phase II randomised discontinuation MAPK signaling pathway is essential for cancer-immune trial analysis, Br J Cancer 95 (2006), 581–586. evasion in human melanoma cells, JExpMed203 (2006), [81] K.S. Smalley, M. Lioni, P.M. Dalla et al., Increased cy- 1651–1656. clin D1 expression can mediate BRAF inhibitor resistance [62] C. Montagut, S.V. Sharma, T. Shioda et al., Elevated CRAF in BRAF V600E-mutated melanomas, Mol Cancer Ther 7 as a potential mechanism of acquired resistance to BRAF (2008), 2876–2883. inhibition in melanoma, Cancer Res 68 (2008), 4853–4861. [82] K.S. Smalley, V.K. Sondak and J.S. Weber, c-KIT signaling [63] N. Dumaz, R. Hayward, J. Martin et al., In melanoma, as the driving oncogenic event in sub-groups of melanomas, RAS mutations are accompanied by switching signaling from Histol Histopathol 24 (2009), 643–650. BRAF to CRAF and disrupted cyclic AMP signaling, Cancer [83] P.G. Natali, M.R. Nicotra, A.B. Winkler et al., Progression Res 66 (2006), 9483–9491. of human cutaneous melanoma is associated with loss of expression of c-kit proto-oncogene receptor, Int J Cancer 52 (1992), 197–201. T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma 285

[84] J.M. Grichnik, Kit and melanocyte migration, J Invest Der- [101] H. Wu, V. Goel and F.G. Haluska, PTEN signaling pathways matol 126 (2006), 945–947. in melanoma, Oncogene 22 (2003), 3113–3122. [85] S. Huang, M. Luca, M. Gutman et al., Enforced c-KIT expres- [102] D.A. Ross, J.H. Laing, R. Sanders et al., Long term follow-up sion renders highly metastatic human melanoma cells suscep- of c-myc, p53 and proliferation measurements in malignant tible to stem cell factor-induced apoptosis and inhibits their melanoma, Eur J Surg Oncol 32 (2006), 80–84. tumorigenic and metastatic potential, Oncogene 13 (1996), [103] H. Schlagbauer-Wadl, M. Griffioen, E.A. van et al., Influence 2339–2347. of increased c-Myc expression on the growth characteristics [86] J.A. Curtin, K. Busam, D. Pinkel et al., Somatic activation of human melanoma, J Invest Dermatol 112 (1999), 332– of KIT in distinct subtypes of melanoma, J Clin Oncol 24 336. (2006), 4340–4346. [104] D. Zhuang, S. Mannava, V. Grachtchouk et al., C-MYC [87] R.S. Rivera, H. Nagatsuka, M. Gunduz et al., C-kit protein overexpression is required for continuous suppression of expression correlated with activating mutations in KIT gene oncogene-induced senescence in melanoma cells, Oncogene in oral mucosal melanoma, Virchows Arch 452 (2008), 27– 27 (2008), 6623–6634. 32. [105] G.D. Hurlbut, M.W. Kankel, R.J. Lake et al., Crossing paths [88] C.R. Antonescu, K.J. Busam, T.D. Francone et al., L576P with Notch in the hyper-network, Curr Opin Cell Biol 19 KIT mutation in anal melanomas correlates with KIT protein (2007), 166–175. expression and is sensitive to specific kinase inhibition, Int J [106] E.J. Allenspach, I. Maillard, J.C. Aster et al., Notch signaling Cancer 121 (2007), 257–264. in cancer, Cancer Biol Ther 1 (2002), 466–476. [89] K.S. Smalley, R. Contractor, T.K. Nguyen et al., Identifi- [107] C.C. Pinnix and M. Herlyn, The many faces of Notch sig- cation of a novel subgroup of melanomas with KIT/cyclin- naling in skin-derived cells, Pigment Cell Res 20 (2007), dependent kinase-4 overexpression, Cancer Res 68 (2008), 458–465. 5743–5752. [108] B.J. Nickoloff, M.J. Hendrix, P.M. Pollock et al., Notch and [90] N. Puri, S. Ahmed, V.Janamanchi et al., c-Met is a potentially NOXA-related pathways in melanoma cells, J Investig Der- new therapeutic target for treatment of human melanoma, matol Symp Proc 10 (2005), 95–104. Clin Cancer Res 13 (2007), 2246–2253. [109] M. Nicolas, A. Wolfer, K. Raj et al., Notch1 functions as [91] G. Li, H. Schaider, K. Satyamoorthy et al., Downregulation a tumor suppressor in mouse skin, Nat Genet 33 (2003), of E-cadherin and Desmoglein 1 by autocrine hepatocyte 416–421. growth factor during melanoma development, Oncogene 20 [110] K. Balint, M. Xiao, C.C. Pinnix et al., Activation of Notch1 (2001), 8125–8135. signaling is required for beta-catenin-mediated human pri- [92] T. Kunisada, H. Yamazaki, T. Hirobe et al., Keratinocyte mary melanoma progression, Clin Invest 115 (2005), 3166– expression of transgenic hepatocyte growth factor affects 3176. melanocyte development, leading to dermal melanocytosis, [111] K. Hoek, D.L. Rimm, K.R. Williams et al., Expression Mech Dev 94 (2000), 67–78. profiling reveals novel pathways in the transformation of [93] R. Halaban, J.S. Rubin, Y. Funasaka et al., Met and hepato- melanocytes to melanomas, Cancer Res 64 (2004), 5270– cyte growth factor/scatter factor signal transduction in nor- 5282. mal melanocytes and melanoma cells, Oncogene 7 (1992), [112] C.C. Pinnix, J.T. Lee, Z.J. Liu et al., Active Notch1 confers 2195–2206. a transformed phenotype to primary human melanocytes, [94] F.P. Noonan, J. Dudek, G. Merlino et al., Animal models of Cancer Res 69 (2009), 5312–5320. melanoma: an HGF/SF transgenic mouse model may facil- [113] C.L. Curry, L.L. Reed, T.E. Golde et al., Gamma secretase itate experimental access to UV initiating events, Pigment inhibitor blocks Notch activation and induces apoptosis in Cell Res 16 (2003), 16–25. Kaposi’s sarcoma tumor cells, Oncogene 24 (2005), 6333– [95] T. Otsuka, H. Takayama, R. Sharp et al., c-Met autocrine acti- 6344. vation induces development of malignant melanoma and ac- [114] J.Z. Qin, L. Stennett, P. Bacon et al., p53-independent quisition of the metastatic phenotype, Cancer Res 58 (1998), NOXA induction overcomes apoptotic resistance of malig- 5157–5167. nant melanomas, Mol Cancer Ther 3 (2004), 895–902. [96] M.Y. Hsu, F. Meier and M. Herlyn, Melanoma development [115] B. Bedogni, J.A. Warneke, B.J. Nickoloff et al., Notch1 is and progression: a conspiracy between tumor and host, Dif- an effector of Akt and hypoxia in melanoma development, J ferentiation 70 (2002), 522–536. Clin Invest 118 (2008), 3660–3670. [97] P.G. Natali, M.R. Nicotra, M.F. Di Renzo et al., Expression of [116] D.B. Solit, L.A. Garraway, C.A. Pratilas et al., BRAF mu- the c-Met/HGF receptor in human melanocytic neoplasms: tation predicts sensitivity to MEK inhibition, Nature 439 demonstration of the relationship to malignant melanoma (2006), 358–362. tumour progression, Br J Cancer 68 (1993), 746–750. [117] K.S. Smalley, N.K. Haass, P.A. Brafford et al., Multiple sig- [98] J. Cruz, J.S. Reis-Filho, P.Silva et al., Expression of c-met ty- naling pathways must be targeted to overcome drug resis- rosine kinase receptor is biologically and prognostically rele- tance in cell lines derived from melanoma metastases, Mol vant for primary cutaneous malignant melanomas, Oncology Cancer Ther 5 (2006), 1136–1144. 65 (2003), 72–82. [118] E.R. Sauter, U.C. Yeo, S.A. von et al., Cyclin D1 is a can- [99] F.P. Noonan, T. Otsuka, S. Bang et al., Accelerated ultravi- didate oncogene in cutaneous melanoma, Cancer Res 62 olet radiation-induced carcinogenesis in hepatocyte growth (2002), 3200–3206. factor/scatter factor transgenic mice, Cancer Res 60 (2000), [119] L.A. Garraway, H.R. Widlund, M.A. Rubin et al., Integrative 3738–3743. genomic analyses identify MITF as a lineage survival onco- [100] A.J. Ramsden, R. Grover, J. Chana et al., A prospective gene amplified in malignant melanoma, Nature 436 (2005), analysis of c-myc oncoprotein levels as a prognostic marker 117–122. in malignant melanoma, J Plast Reconstr Aesthet Surg 60 (2007), 626–630. 286 T. Bogenrieder and M. Herlyn / The molecular pathology of cutaneous melanoma

[120] G.M. Kraehn, J. Utikal, M. Udart et al., Extra c-myc onco- melanoma, Oncogene 22 (2003), 3092–3098. gene copies in high risk cutaneous malignant melanoma and [123] J.M. Stahl, M. Cheung, A. Sharma et al., Loss of PTEN pro- melanoma metastases, Br J Cancer 84 (2001), 72–79. motes tumor development in malignant melanoma, Cancer [121] D.A. Ross and G.D. Wilson, Expression of c-myc onco- Res 63 (2003), 2881–2890. protein represents a new prognostic marker in cutaneous [124] O. Straume and L.A. Akslen, Alterations and prognostic sig- melanoma, Br J Surg 85 (1998), 46–51. niﬁcance of p16 and p53 protein expression in subgroups of [122] E. Sharpless and L. Chin, The INK4a/ARF locus and cutaneous melanoma, Int J Cancer 74 (1997), 535–539.