Chapter 10 / Genetics of Melanoma Initiation

10

183

Familial Melanoma Genes, Melanocyte Immortalization, and Melanoma Initiation Dorothy C. Bennett CONTENTS INTRODUCTION THE CLARK MODEL FOR PRIMARY MELANOMA DEVELOPMENT CELL SENESCENCE: TYPES AND MOLECULAR MECHANISMS THE CDKN2A LOCUS, ITS TWO GENE PRODUCTS, AND MELANOMA EFFECTS OF CDKN2A DEFICIENCIES IN CULTURED MOUSE AND HUMAN MELANOCYTES POSSIBLE GENETIC BASIS FOR THE CLARK MODEL AND SUPPORTING EVIDENCE SUPPRESSION OF APOPTOSIS IN VGP MELANOMA CONCLUSIONS AND PERSPECTIVES REFERENCES

Summary The most common types of pigmented lesions have been classified by Clark and colleagues into a series of increasingly malignant types. We have proposed a general model for the genetic events underlying this series of lesions and progression from one lesion type to another. The current form of this model, the evidence that gave rise to it, and the areas of uncertainty that remain are presented here. In particular, evidence for the following is discussed: 1. Involvement of the process of cell senescence in the proliferative arrest seen in nevi. 2. A role for apoptosis and keratinocyte-dependence in the flat growth pattern of radial growth-phase melanomas. 3. The genetic suppression of both senescence and apoptosis in more advanced melanomas. The protein p16, encoded by the familial melanoma locus cyclin-dependent kinase inhibitor 2A (CDKN2A), appears to play a central role in the senescence of nevi and probably also in the keratinocytedependence of thin melanomas, in interaction with two other key tumor suppressors, retinoblastoma (RB)-1 and p53. Key Words: Melanoma; nevus; cell senescence; p16; CDKN2A; RB1; telomere; immortalization; apoptosis.

From: From Melanocytes to Melanoma: The Progression to Malignancy Edited by: V. J. Hearing and S. P. L. Leong © Humana Press Inc., Totowa, NJ

183

184

From Melanocytes to Melanoma

Fig. 1. Clark et al.’s classification of pigmented lesions. The melanocyte is not to scale. RGP, radial growth phase; VGP, vertical growth phase. A VGP melanoma may not necessarily have a visible RGP area. See pages 184–185 for further details.

INTRODUCTION There is, as yet, no type of solid cancer for which the minimal genetic events required to produce that lesion are definitely known and proven. Genetic models have been produced for some tumor types, such as the widely cited model of Vogelstein and colleagues for colorectal cancer and its variants (1), but so far without a conclusive demonstration that the proposed events are necessary or sufficient to produce the cancer. It is possible that cutaneous melanoma may be among the first solid cancers to be genetically “solved” in this way, because an early epidemiological analysis suggested that melanoma might be genetically relatively simple, compared with other solid cancers, in terms of the number of cellular events needed to produce it. This was based on age–incidence curves; the number of independent causal events (interpreted as genetic changes) required to produce each type of cancer could be estimated from a best fit between theoretical and observed curves. Among the cancer types analyzed, melanoma was estimated to require the fewest events for its development (2). The estimated number of events was only one, although it now seems likely that at least three genetic changes are needed to produce even an early melanoma, as discussed from page 189. The fact that n = 1 gives the best mathematical fit suggests that one of the necessary events is substantially more rare than the others, thus being rate-limiting. This chapter will review our current knowledge of the genetic basis of human melanoma and the common benign pigmented lesions. A relatively simple genetic model for sporadic melanoma will be discussed, which brings these data together and relates them to the widely-used clinical classification of pigmented lesions by Clark et al. (3–5), supporting the idea that few genetic events are required to produce a metastatic melanoma. I will begin with some sections of background information needed to understand the biological and genetic events to be discussed.

THE CLARK MODEL FOR PRIMARY MELANOMA DEVELOPMENT Figure 1 shows a version of the series of primary pigmented lesions described by Clark et al. (3–5). These are sometimes taken to represent a linear progression, in which each lesion is the direct precursor of the next, but it seems best simply to regard them as clinically distinct types of lesion, each of which may apparently develop directly from normal skin as well as from one of the more benign types. Various evidence, including clinical and molecular (as will be discussed), indicates that the pathways by which different cases of advanced melanoma develop may vary.

Chapter 10 / Genetics of Melanoma Initiation

185

The first lesion type (Fig. 1) is the benign nevus or mole, small and well-delineated, of which, most of us have a good many on our skin. The clinical definition of the dysplastic nevus is somewhat disputed, but this is broadly agreed to mean a nevus that has one or more features more typical of a melanoma, such as being unusually large, having an irregular or indistinct border and/or uneven pigmentation. Atypical mole has also been suggested as a better clinical term for such lesions, which does not imply a particular histopathological pattern (dysplasia). Next is the radial growth-phase (RGP) melanoma. This lesion grows progressively, but only “radially,” forming a flat plaque within the epidermis or with small dermal nests close to the epidermis. Lastly, a vertical growth-phase (VGP) melanoma forms large nodule(s) in the dermis; this key feature is predictive of the potential to metastasize (4). Interestingly, melanoma differs from most other solid cancers in that acquiring the ability to metastasize does not appear to be a separate step from becoming able to invade neighboring tissues (4).

CELL SENESCENCE: TYPES AND MOLECULAR MECHANISMS I will now introduce the main concepts of cellular senescence and immortalization, as needed to understand the apparent connection with melanoma progression. These concepts have been reviewed more fully by a number of authors, both generally (6–8) and as applied specifically to melanocytes (9,10). Normal somatic mammalian cells (as studied in cell culture) can undergo only a limited number of cell divisions, called their proliferative life-span, although this number varies between animal species, between cell types, and with culture conditions (9). When cells reach the end of this life-span, they irreversibly stop growing, and this arrest is called cellular senescence, as first described by Hayflick (11). We now know that cellular senescence can be affected by more than one molecular pathway (6,7,12). For example, the mechanisms are radically different between mouse cells and human cells; telomere shortening (see next paragraph) is important in the senescence of most normal human cells, but not that of normal mouse cells, which have extremely long telomeres (7). This review will concentrate on human cell senescence, in relation to human melanoma. Even among human cells of different types, different kinds of senescence can be distinguished (Fig. 2). M1 senescence, typical of human fibroblasts, is the best understood. The key initiating signal for M1 senescence appears to be critical shortening of telomeres (the multiple DNA hexamer repeats at the ends of the chromosomes), which become shorter with each division in somatic cells. In germline cells and a few others, together with most cancer cells, telomere length is maintained by the reverse transcriptase, telomerase (13). However, most human somatic cells lack the catalytic subunit, human telomerase reverse transcriptase (hTERT). If these cells undergo numerous divisions, the telomeres shorten, as does a single-stranded loop at the very end (14), resulting in unfolding of the loop, recognition of the single-stranded DNA, and activation of DNA damage-signaling mechanisms, which activate p53 through a pathway involving checkpoint kinase 2 (CHK2) (15). Cellular growth arrest (M1 senescence) results partly from this active p53, which transcriptionally upregulates the growthinhibitor protein, p21, and partly from the parallel upregulation of another important growth inhibitor, p16INK4A or p16, by a less-understood pathway. p16 will be discussed further on page 187, as the product of a melanoma susceptibility gene. At present, we will just note that p16 activates the tumor suppressor and growth inhibitor, RB1 (retinoblastoma), and the two other members of the RB protein family, RBL1/p107 (retinoblas-

186

From Melanocytes to Melanoma

Fig. 2. Types of cellular senescence. See pages 185–187 for explanation.

toma-like 1) and RBL2/p130. p16 does this by inhibiting cyclin-dependent kinase 4 (CDK4), which can inhibit RB activity through phosphorylation. Either the RB proteins (activated by p16) or activated p53 are independently capable of arresting human cell growth during M1 senescence; in other words, cell senescence will still occur through RB if p53 signaling is disrupted, and vice versa (6,12,16). p16 is not present in most normal cells (17), but its synthesis is activated at cellular senescence (8,18). It is not clear how this synthesis occurs in response to extended cell proliferation (8), except that, in senescence of human fibroblasts, an increase and a decrease, respectively, in the amounts of transcription factors ETS1 and ID1 have been implicated in the upregulation of p16 (19). There seems to be no information, however, on how these changes in ETS1 and ID1 levels would result from extended cell proliferation. The second form of cell senescence in Fig. 2, M0 senescence, is precipitated by p16 alone, without involvement of p53. This type of senescence has been reported in various human epithelial cells under stressful culture conditions (20,21), in human fibroblasts also under stressful conditions (21), and in human melanocytes under the most favorable culture conditions known (including growth with “feeder cells”) (10,22,23). Again, we know little of how p16 expression becomes induced in M0 senescence or “stress,”

Chapter 10 / Genetics of Melanoma Initiation

187

although there is some evidence for roles for transcription factors ETS2, ID1 (19), and BMI1 (24). At any rate, the outcome is activation of RB proteins and proliferative arrest. As Fig. 2B shows, expression of exogenous hTERT allows cells to overcome or circumvent M1 senescence, after which they can divide apparently without limit (immortalization). However, this is not true of cells in M0 senescence. These require inactivation of the p16/RB pathway as well as expression of hTERT before they can be immortalized (20,23,25).

THE CDKN2A LOCUS, ITS TWO GENE PRODUCTS, AND MELANOMA Approximately 10% of cutaneous melanoma cases worldwide occur in a loosely familial context, that is, one or more relatives also have melanoma, suggesting inheritance of a susceptibility gene (26,27). There are several detailed recent reviews on melanoma genetics (26–28). To date, mutations of four different genes at three different loci are known to confer susceptibility to melanoma. These are the INK4A (p16) and ARF sequences at the CDKN2A locus (chromosome 9p21); CDK4 (chromosome 12q14), encoding the principal kinase that p16 inhibits; and the recently identified and uncloned locus, CMM4 (cutaneous malignant melanoma 4; chromosome 1p22) (29). Melanomaassociated mutations of INK4A and CDK4 seem to be functionally equivalent, because both block the association of p16 with CDK4, and hence p16’s ability to activate the RB family (30,31). Accordingly, these mutations could interfere with cellular senescence, especially M0 senescence, as observed in melanocytes (see pages 186–187). It is interesting that both INK4A and CDK4 mutations are associated relatively specifically with melanoma; the only other cancer that clearly occurs at a higher than normal rate in families carrying INK4A mutations appears to be pancreatic cancer (32,33). The CDKN2A locus is highly unusual, possibly unique in vertebrates, in that the same DNA sequence is used to encode two different, unrelated proteins in different reading frames, although the two gene products do have separate sequences as their first exons (34,35). One of the products is p16, the other (as mentioned above) is called ARF (alternative reading frame), also known as p14ARF in humans and p19Arf in the mouse (reflecting the different protein sizes in the two species). ARF also appears to have a role in human melanoma. Nearly all tested melanoma-associated mutations in CDKN2A functionally affect p16 protein, whereas many of them do not affect ARF function (34,36). Nonetheless, a few of these mutations apparently have no effect on p16 and do disrupt ARF function. Such mutations have been reported both in melanoma families and in sporadic melanomas (37–39). ARF is thus included here among the melanoma susceptibility genes. Like p16, ARF is a powerful inhibitor of cell proliferation. Its best-known function is stabilization of p53, through inhibition of the ubiquitin ligase MDM2 (mouse double minute 2 homolog, also sometimes—less correctly—called HDM2 in the human). MDM2 can ubiquitinylate p53, leading to its degradation (40,41). In the mouse, Arf is an essential player in cellular senescence; fibroblasts from mice with normal p16 but deleted Arf fail to senesce, and the mice are susceptible to a variety of tumor types, although melanoma is not observed (42). However, there is good evidence that human ARF, surprisingly, plays no part in senescence of normal fibroblasts (43). This may be connected with the role of telomere shortening in activating p53 in human but not mouse cells. In the mouse, a pathway involving Arf may substitute for the short telomere/CHK2

188

From Melanocytes to Melanoma

pathway in activating p53 after a certain amount of cell division. Returning to humans, ARF does appear to have some tumor suppressor function, because individuals heterozygous for a germline ARF mutation appear susceptible to melanoma and probably some types of neural tumors (26,38). It is not, however, clear how ARF acts as a tumor suppressor in humans. One possibility is through RB proteins, because MDM2 can interact in an inhibitory fashion with RB1 (44), thus, ARF may also be able to prevent this. Another possibility is that ARF may be involved in a kind of premature M1-like senescence induced by activation of single mitogenic oncogenes, such as HRAS or MYC, because ARF expression and upregulation of p53 as well as p16 are reported in human fibroblasts under these conditions (45–47).

EFFECTS OF CDKN2A DEFICIENCIES IN CULTURED MOUSE AND HUMAN MELANOCYTES Cell culture studies have yielded some interesting and unexpected information on the biology of CDKN2A deficiency in melanocytes, which will be reviewed here for comparison with the biology of early melanoma. Mouse melanocytes with a Cdkn2a deletion, which abolishes the synthesis of both p16 and Arf, have been cultured and compared with wild-type melanocytes from the same inbred mouse strain. The wild-type melanocytes senesced within 5–6 wk. However, like fibroblasts from these p16 and Arf-null mice (48), the melanocytes grew very well in culture and showed almost no cell senescence, showing that p16, Arf, or both are required for mouse melanocyte senescence (49). Fibroblasts with a single copy of the deletion were reported to senesce similarly to wild-type mouse fibroblasts (48), but melanocyte cultures with only a single copy showed only partial senescence followed by immortalization of all cultures (49), indicating that two intact copies of this locus are needed for normal cell senescence in melanocytes (but not fibroblasts). This suggests that particularly high protein levels of p16, Arf, or both are needed for normal mouse melanocyte senescence, which may be relevant to the melanoma susceptibility of humans who carry a single copy of a mutation in p16. Melanocyte cultures were also prepared from two human familial melanoma patients, both of whom (very unusually) had mutations in both copies of the p16 coding sequence but normal ARF function. The patients are described by Gruis et al. (50) and Huot et al. (51), and the melanocytes by Sviderskaya et al. (23). These cultures, unexpectedly, showed very little net growth in a medium in which normal human melanocytes grow well, although mitotic cells could be observed. The explanation proved to be a high rate of apoptosis in these cells. Further experimentation revealed that the excessive apoptosis was suppressed by coculture with keratinocytes, or with two peptides produced by keratinocytes, stem cell factor, and endothelin 1 (23). Under these conditions, the p16deficient melanocytes divided for far longer than normal adult human melanocytes. They did eventually senesce, with upregulation of p53 and p21 (23). This is reminiscent of the epithelial cell types that undergo M0 senescence under some conditions; if RB function is blocked, then these cells bypass M0, proliferate extensively, and then enter a new senescent phase that is M1-like, because it involves p53 and p21. This second senescent phase can be circumvented by expression of hTERT, leading to cell immortalization (20). The p16-deficient melanocytes could likewise be immortalized using hTERT alone (23). Recent unpublished work indicates similar biology between these human p16-deficient melanocytes and melanocytes from mice that are likewise deficient for p16 but

Chapter 10 / Genetics of Melanoma Initiation

189

Fig. 3. Genetic model for melanoma progression, correlated with the clinical model of Clark et al. Adapted and updated from Bennett (10). Note that not all the stages shown may necessarily be observed in a given case; for example, the genetic changes may happen in a different order.

have normal Arf (52). These mouse melanocytes grow normally in the beginning, then slow down at the expected time of senescence, with apparent cell death. However, the slowing-down can be reversed by either keratinocyte feeder cells or stem cell factor with endothelin 1, in the presence of which, the cells grow well and do not senesce (53, abstract only). This finding preliminarily implicates Arf in the excessive cell death and keratinocyte-dependence of p16-deficient melanocytes, at least in the mouse, because mouse melanocytes null for both p16 and Arf did not show this cell death (49).

POSSIBLE GENETIC BASIS FOR THE CLARK MODEL AND SUPPORTING EVIDENCE These observations of excessive apoptosis (suppressed by keratinocytes) and deficient cell senescence in p16-deficient melanocytes, combined with emerging evidence on genetic alterations and abnormal gene expression in pigmented lesions, led to a hypothesis for the genetic basis for the Clark series of lesions (10). The current version of this model is presented in Fig. 3, and the evidence for each proposed step is now discussed. It has also been reviewed in detail elsewhere (28).

Proliferative Mutation in the Nevus First, it is suggested that a benign nevus is a clone of melanocytes that has undergone a single mutation that overcomes the mechanism that normally limits the local population density of normal epidermal melanocytes, allowing the clone to proliferate into a colony of densely packed melanocytes. The most likely causal mutations are activating mutations of BRAF and NRAS, because approx 90% of benign nevi are reported to carry one or other of these two mutations, usually BRAF (approx 80% of nevi, whether benign, congenital, or dysplastic), and, in general, the two are mutually exclusive (54–56). As inferred from other cell types, these mutations are functionally very similar and would activate the MAPK intracellular signaling pathway, promoting cell proliferation and inhibiting apoptosis; moreover, expression of an activated BRAF sequence has been shown to be both mitogenic and oncogenic in mouse melanocytes (57).

190

From Melanocytes to Melanoma

M0 Senescence in the Benign Nevus It is proposed that the reason that benign nevi stop growing and can remain static for many years is that their cells undergo cell senescence. This idea has already been proposed by several groups (9,58,59). There was circumstantial support for the idea, in that nevi grow initially and then stop growing, showing virtual absence of the proliferation marker Ki67 (60); moreover, nevus cells appear to show poor ability to divide when put into cell culture, compared with normal melanocytes, and the nevus cultures are reported to include giant, highly dendritic cells that are sometimes multinucleate, all properties observed in senescent cells, including melanocytes (reviewed in ref. 9). Descriptions of nevus cell cultures and their proliferation are somewhat variable, which may be attributable to varying proportions of normal melanocytes present in the cultures; nonetheless, they are generally reported to have low proliferative potential (9). The idea that nevi are senescent was recently tested further by examination of additional cellular properties in nevi and other pigmented lesions (25). It was found that benign nevi nearly all expressed substantial levels of p16, and all showed acidic E-galactosidase, a marker with some specificity for senescent cells (61), neither of which was present in normal melanocytes in adjacent skin (25). p16 has been reported in nevi before, but it does not seem to have been realized that this is abnormal as compared with normal melanocytes (62,63). GraySchopfer et al. also reported that benign nevi rarely expressed detectable p53 or p21, indicating that their type of senescence was M0, just as seen in cultured melanocytes (25). This provides an attractive explanation for the observation that families with p16 mutations and susceptibility to melanoma often show dysplastic nevus syndrome, a tendency to have large numbers of nevi, many of which may be unusually large and/or classifiable as dysplastic (27). If the growth arrest of nevi is p16-dependent cell senescence, then it makes sense for a germline defect in p16 to be associated with larger nevi through deficient (delayed) cell senescence. This theory does not explain why only some individuals with p16 mutations have dysplastic nevus syndrome (27).

Escape From M0 Senescence in Dysplastic Nevi The model in Fig. 3 proposes that dysplastic nevi contain or consist of cells that have bypassed M0 senescence, which would necessarily involve a defect in the p16/RB pathway. Alterations and deletions of p16 are reported in some dysplastic nevi, although not in benign nevi (64,65) (it remains unclear how p16 loss may happen in pigmented lesions; Gray-Schopfer et al. reported a tendency for reduced or more patchy expression of p16 in dysplastic nevi (25), consistent with the previous studies). It was predicted that dysplastic nevi might reach a state of p53-dependent senescence, as seen with p16deficient human melanocytes in culture, but immunohistochemical data did not support the presence of p21 in most of these lesions (25). It is therefore possible, instead, that cells in dysplastic nevi are generally not senescent but still proliferating.

Cell Immortalization and Telomere Maintenance in Melanoma; Apoptosis in RGP It has been reported in general that more than 90% of human cancers express hTERT, and that the rest can extend their telomeres by an abnormal mechanism called ALT (alternative lengthening of telomeres), which appears to involve recombination (66). Melanomas appear to conform to the generality concerning telomere maintenance; whereas benign and dysplastic nevi have little or no telomerase, the majority of melanomas are reported to have substantial levels of telomerase activity (67–69). Lower

Chapter 10 / Genetics of Melanoma Initiation

191

telomerase activity in thinner melanomas was provisionally concluded to be an artefact caused by contamination by normal cells, when all melanomas but not host cells were found to express the telomerase template RNA subunit hTR by in situ hybridization (67). However, further evidence is needed on the question of telomerase activity in RGP melanomas, because there does not appear to be clear evidence regarding whether these lesions also express hTERT, also needed for activity of telomerase. Although this gap in the data should be noted, our model (Fig. 3) proposes that the critical difference between melanomas and nevi is cellular immortalization. We postulate that RGP melanomas will often have the minimal genetic changes that would achieve both growth and immortalization, namely: 1. A mutation leading to cell proliferation. 2. A deficit in the p16/RB pathway. 3. The activation of telomerase (or possibly sometimes ALT).

A mutation leading to cell proliferation would usually be attributable to the BRAF and NRAS mutations already mentioned, which are also found in approx 70% of melanomas (55,56,70), and the activation of telomerase (or possibly sometimes ALT) has just been discussed. Actual mutations of p16 are not very common in sporadic melanoma (34), but expression of p16 appears to be very commonly reduced in RGP as well as in advanced melanomas (25,63,71,72), and other RB pathway lesions are also reported in melanoma (73; reviewed in ref. 28). Accordingly, RGP melanoma cells are expected to resemble p16-deficient melanocytes, including the high rate of apoptosis in the absence of keratinocytes or keratinocyte-derived growth factors (23). This fits with the growth pattern of RGP melanomas, namely, growth only in or near the epidermis (4)— it would be expected that the RGP cells would undergo apoptosis if they move beyond the diffusion range of keratinocyte-derived growth factors. Incidentally, benign nevus cells, which are not p16-deficient, would not be expected to be so keratinocyte-dependent, which accords with their ability to grow in the dermis.

SUPPRESSION OF APOPTOSIS IN VGP MELANOMA If RGP melanoma cells grow near the epidermis because they are keratinocyte-dependent, one would expect VGP cells, which grow deeper in the dermis, not to be keratinocyte-dependent. This fits with the known ability of advanced primary and metastatic melanoma cells to grow well in cell culture without any other cells, and indeed with very few growth factors (74,75). There are thousands of cell lines derived from such melanomas. We therefore postulate that VGP melanoma cells possess all the same genetic changes as listed for RGP, but in addition have one or more alterations that inhibit apoptosis. This fits well with our knowledge of common genetic alterations in VGP and metastatic melanoma cells. As shown (Fig. 3), many of these common alterations do reduce apoptosis, as follows. Oncogenic activation of p53 is reported in approx 10% of advanced melanomas (76,77). Loss or silencing (by methylation) of apoptotic protease activating factor 1 (APAF1), an effector of apoptosis that is transcriptionally activated by p53, is also common in melanoma (78,79). A variety of protein tyrosine kinases become overexpressed in melanoma cells, at least as measured in cultured melanoma lines (reviewed in ref. 80); signaling from these kinases would be expected to suppress apoptosis through phosphoinositide signaling, activating the antiapoptotic protein kinase, AKT (81). Likewise, approx 15% of uncultured melanomas have inactivating

192

From Melanocytes to Melanoma

mutations in phosphatase and tensin homolog (PTEN) (82,83). PTEN is a phosphatase that degrades the phosphoinositide and AKT activator, PIP3, thus, PTEN inactivation is another route to excessive AKT activity and reduced apoptosis. Other modulators of apoptosis are members of the B-cell lymphoma 2 (BCL2) protein family, found in mitochondrial membranes, and incidentally, modulated in activity by AKT. Normal melanocytes already have high levels of the suppressor of apoptosis, BCL2 (84,85), and apparently require this for survival when synthesizing pigment (85). Bush and Li (86) recently reviewed alterations in the BCL2 protein family, which include both proapoptotic and antiapoptotic factors. They concluded that, although most family members show no consistent trend in their expression related to melanoma progression, there are consistent reports of overexpression of antiapoptotic factor BCL2L1 (BCL-XL) in melanomas. A last major antiapoptotic pathway to be mentioned is that of E-catenin activation. Activating mutations of E-catenin itself can be found in melanoma but are rare (87). It is more common to find defects or transcriptional silencing of adenomatous polyposis coli protein, which normally facilitates destruction of E-catenin (88). Also fairly common is overexpression of the transcription factor and proto-oncoprotein, SKI, which can also activate E-catenin (89). In summary, there are many genetic and epigenetic routes by which apoptosis becomes inhibited in melanoma, and the collation given here is probably far from exhaustive. Many of the same pathways also promote cell proliferation, thus, they would confer a double advantage for tumor growth.

CONCLUSIONS AND PERSPECTIVES The model presented here seems consistent with our knowledge to date about the biology and genetics of melanoma progression. It can explain some puzzling aspects of the biology of pigmented lesions, such as why nevus cells should be able to grow relatively deep in the dermis, whereas the apparently more malignant RGP melanoma cells do not. At present, however, various aspects of the model are unproven and speculative, and some details may well change as more information emerges. What does seem very likely, in a nutshell, is that cell senescence is important in nevi and escape from senescence is important in melanoma. A number of interesting questions are raised by these considerations. One key remaining question is that of how p16 is activated during M0 senescence. We now know in some detail how shortening of telomeres can activate p53, but we seem to have little idea of which other mechanism for counting cell divisions might result in the increasing expression of p16. It does not appear to be connected with telomere shortening, because telomere maintenance by the forced expression of hTERT does not prevent this type of senescence. Transcription factors ETS1 and ETS2 were reported to be involved in this activation of p16, respectively, in M1 senescence and in the type of accelerated senescence that follows oncogene activation, together with falling expression of ID1 (19) and BMI1 (24). These findings were in human fibroblasts. If they are applicable to other cell types, then the question becomes rephrased as: what determines this changing expression of these transcription factors in dividing normal cells? Another obvious question is whether cell senescence does indeed act as a barrier to other types of cancer, proposed as probable by those who study it. If so, then, in organs in which cancers can develop, there should commonly be benign, static lesions composed of senescent cells following a mitogenic mutation. This can be tested in future by

Chapter 10 / Genetics of Melanoma Initiation

193

looking for the expression of senescence markers in the known types of static lesions, such as epithelial cysts and polyps. A related question is that of why INK4A should be a susceptibility gene only for melanoma and, less commonly, pancreatic cancer. This seems to suggest that escape from M0/p16-mediated senescence may not be rate limiting for the initiation of most cancer types, other than melanoma and, presumably, pancreatic cancer. If not, then what is more usually rate-limiting? It is tempting to suggest M1/p53-mediated senescence. Mutations of the p53 coding sequence TP53 are notoriously common in cancer; germline TP53 mutations predispose to a wide variety of cancers, as do mutations in CHK2, which activates p53 during senescence (90). This is the set of cancers in which we might predict expression of normal p53 and p21, and probably p16 also, in the commonest types of static benign lesions. Lastly, returning to melanoma, it is of special interest to determine whether cell senescence markers may have any value for diagnosis or prognosis. Initial studies indicate a connection with progression, but it remains to be seen whether, using any of these markers, exact criteria may be developed that have better prognostic value than (for example) those offered by Clark et al. over a decade ago. It may at least be hoped that further clarification of our biological understanding of the processes that restrict the growth of benign pigmented lesions may be one route to the identification of therapeutic targets for better treatment of metastatic melanoma.

ACKNOWLEDGMENTS Work in the author’s laboratory reviewed here was supported by European Commission Contract QLK4-1999-01084 and by Wellcome Trust Grants 046038 and 064583. I am grateful to many colleagues for stimulating discussions about this topic.

REFERENCES 1. Arends JW. Molecular interactions in the Vogelstein model of colorectal carcinoma. J Pathol 2000;190:412–416. 2. Cook PJ, Doll R, Fellingham SA. A mathematical model for the age distribution of cancer in man. Int J Cancer 1969;4:93–112. 3. Clark WH, Elder DE, Guerry D, Epstein MN, Greeve MH, Van Horn M. A study of tumor progression: the precursor lesions of superficial spreading and nodular melanoma. Hum Pathol 1984;15: 1147–1165. 4. Clark WH, Elder DE, Guerry D, et al. Model predicting survival in stage I melanoma based on tumor progression. J Natl Cancer Inst 1989;81:1893–1904. 5. Crotty K, McCarthy S, Mihm MC. The histological diagnosis and classification of melanoma. In: Thompson JF, Morton DL, Kroon BBR, eds. Textbook of melanoma. Martin Dunitz, New York, NY: 2004, pp. 115–121. 6. Wynford-Thomas D. Cellular senescence and cancer. J Pathol 1999;187:100–111. 7. Newbold RF. The significance of telomerase activation and cellular immortalization in human cancer. Mutagenesis 2002;17:539–550. 8. Sharpless NE, DePinho RA. Telomeres, stem cells, senescence, and cancer. J Clin Invest 2004;113:160–168. 9. Bennett DC, Medrano EE. Molecular regulation of melanocyte senescence. Pigment Cell Res 2002;15:242–250. 10. Bennett DC. Human melanocyte senescence and melanoma susceptibility genes. Oncogene 2003;22:3063–3069. 11. Hayflick L. The limited in vitro lifetime of human diploid cell strains. Exp Cell Res 1965;37:614–636.

194

From Melanocytes to Melanoma

12. Bond JA, Haughton MF, Rowson JM, et al. Control of replicative life span in human cells: barriers to clonal expansion intermediate between M1 senescence and M2 crisis. Mol Cell Biol 1999;19: 3103–3114. 13. Blackburn EH. Telomere states and cell fates. Nature 2000;408:53–56. 14. Stewart SA, Ben Porath I, Carey VJ, O’Connor BF, Hahn WC, Weinberg RA. Erosion of the telomeric single-strand overhang at replicative senescence. Nat Genet 2003;33:492–496. 15. d’Adda di Fagagna F, Reaper PM, Clay-Farrace L, et al. A DNA damage checkpoint response in telomere-initiated senescence. Nature 2003;426:194–198. 16. Ramirez RD, Herbert BS, Vaughan MB, et al. Bypass of telomere-dependent replicative senescence (M1). upon overexpression of Cdk4 in normal human epithelial cells. Oncogene 2003;22:433–444. 17. Zindy F, Quelle DE, Roussel MF, Sherr CJ. Expression of the p16INK4a tumor suppressor versus other INK4 family members during mouse development and aging. Oncogene 1997;15:203–211. 18. Hara E, Smith R, Parry D, Tahara H, Stone S, Peters G. Regulation of p16CDKN2 expression and its implications for cell immortalization and senescence. Mol Cell Biol 1996;16:859–867. 19. Ohtani N, Zebedee Z, Huot TJG, et al. Opposing effects of Ets and Id proteins on p16INK4a expression during cellular senescence. Nature 2001;409:1067–1070. 20. Kiyono T, Foster SA, Koop JI, McDougall JK, Galloway DA, Klingelhutz AJ. Both RB/p16INK4a inactivation and telomerase activity are required to immortalize human epithelial cells. Nature 1998;396:84–88. 21. Ramirez RD, Morales CP, Herbert BS, et al. Putative telomere-independent mechanisms of replicative aging reflect inadequate growth conditions. Genes Dev 2001;15:398–403. 22. Bandyopadhyay D, Medrano EE. Melanin accumulation accelerates melanocyte senescence by a mechanism involving p16{+INK4a}/CDK4/pRB and E2F1. Ann N Y Acad Sci 2000;908:71–84. 23. Sviderskaya EV, Gray-Schopfer VC, Hill SP, et al. p16/cyclin-dependent kinase inhibitor 2A deficiency in human melanocyte senescence, apoptosis, and immortalization: possible implications for melanoma progression. J Natl Cancer Inst 2003;95:723–732. 24. Itahana K, Zou Y, Itahana Y, et al. Control of the replicative life span of human fibroblasts by p16 and the polycomb protein Bmi-1. Mol Cell Biol 2003;23:389–401. 25. Gray-Schopfer VC, Cheong SC, Chow J, et al. Cellular senescence in melanocytic nevi: a mechanism for melanoma suppression by p16. 2005, submitted. 26. Hayward NK. Genetics of melanoma predisposition. Oncogene 2003;22:3053–3062. 27. Kefford RF, Mann GJ, Newton Bishop J. Genetic predisposition to melanoma. In: Thompson JF, Morton DL, Kroon BBR, eds. Textbook of Melanoma. Martin Dunitz, New York, NY: 2004, pp. 56–64. 28. Gray-Schopfer VC, Bennett DC. The genetics of melanoma. In: Nordlund JJ, Boissy RE, Hearing VJ, King RA, Oetting WS, Ortonne JP, eds. The Pigmentary System. Physiology and Pathophysiology. Oxford University Press, New York, NY: 2005, in press. 29. Gillanders E, Juo SH, Holland EA, et al. Localization of a novel melanoma susceptibility locus to 1p22. Am J Hum Genet 2003;73:301–313. 30. Zuo L, Weger J, Yang Q, et al. Germline mutations in the p16INK4a binding domain of CDK4 in familial melanoma. Nat Genet 1996;12:97–99. 31. Soufir N, Avril MF, Chompret A, et al. Prevalence of p16 and CDK4 germline mutations in 48 melanoma-prone families in France. The French Familial Melanoma Study Group. Hum Mol Genet 1998;7:209–216. 32. Hayward N. New developments in melanoma genetics. Curr Oncol Rep 2000;2:300–306. 33. Tucker MA, Goldstein AM. Melanoma etiology: where are we? Oncogene 2003;22:3042–3052. 34. Ruas M, Peters G. The p16INK4a/CDKN2A tumor suppressor and its relatives. Biochim Biophys Acta 1998;1378:F115–F177. 35. Sherr CJ. The INK4a/ARF network in tumour suppression. Nature Rev Mol Cell Biol 2001;2:731–737. 36. Sharpless NE, DePinho RA. The INK4A/ARF locus and its two gene products. Curr Opinion Genet Dev 1999;9:22–30. 37. Rizos H, Puig S, Badenas C, et al. A melanoma-associated germline mutation in exon 1beta inactivates p14ARF. Oncogene 2001;20:5543–5547. 38. Randerson-Moor JA, Harland M, Williams S, et al. A germline deletion of p14(ARF) but not CDKN2A in a melanoma-neural system tumour syndrome family. Hum Mol Genet 2001;10:55–62. 39. Hewitt C, Lee WC, Evans G, et al. Germline mutation of ARF in a melanoma kindred. Hum Mol Genet 2002;11(11):1273–1279. 40. Pomerantz J, Schreiber-Agus N, Liegeois NJ, et al. The INK4a tumor suppressor gene product, p19ARF, interacts with MDM2 and neutralizes MDM2’s inhibition of p53. Cell 1998;92:713–723.

Chapter 10 / Genetics of Melanoma Initiation

195

41. Momand J, Wu HH, Dasgupta G. MDM2—master regulator of the p53 tumor suppressor protein. Gene 2000;242:15–29. 42. Kamijo T, Zindy F, Roussel MF, et al. Tumor suppression at the mouse INK4a locus mediated by the alternative reading frame product p19ARF. Cell 1997;91:649–659. 43. Wei W, Hemmer RM, Sedivy JM. Role of p14ARF in replicative and induced senescence of human fibroblasts. Mol Cell Biol 2001;21:6748–6757. 44. Hsieh JK, Chan FSG, O’Connor DJ, Mittnacht S, Zhong S, Lu X. RB regulates the stability and the apoptotic function of p53 via MDM2. Mol Cell 1999;3:181–193. 45. Serrano M, Lin AW, McCurrach ME, Beach D, Lowe SW. Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a. Cell 1997;88:593–602. 46. Lin AW, Barradas M, Stone JC, van Aelst L, Serrano M, Lowe SW. Premature senescence involving p53 and p16 is activated in response to constitutive MEK/MAPK mitogenic signaling. Genes Dev 1998;12:3008–3019. 47. Drayton S, Rowe J, Jones R, et al. Tumor suppressor p16INK4a determines sensitivity of human cells to transformation by cooperating cellular oncogenes. Cancer Cell 2003;4:301–310. 48. Serrano M, Lee HW, Chin L, Cordon-Cardo C, Beach D, DePinho RA. Role of the INK4a locus in tumor suppression and cell mortality. Cell 1996;85:27–37. 49. Sviderskaya EV, Hill SP, Evans-Whipp TJ, et al. p16Ink4a in melanocyte senescence and differentiation. J Natl Cancer Inst 2002;94:446–454. 50. Gruis NA, van der Velden PA, Sandkuijl LA, et al. Homozygotes for CDKN2 (p1) germline mutation in Dutch familial melanoma kindreds. Nat Genet 1995;10:351–353. 51. Huot TJ, Rowe J, Harland M, et al. Biallelic mutations in p16INK4a confer resistance to Ras- and Etsinduced senescence in human diploid fibroblasts. Mol Cell Biol 2002;22:8135–8143. 52. Krimpenfort P, Quon KC, Mooi WJ, Loonstra A, Berns A. Loss of p16Ink4a confers susceptibility to metastatic melanoma in mice. Nature 2001;413:83–86. 53. Sviderskaya EV, Hill SP, Krimpenfort P, Bennett DC. p16Ink4a in mouse melanocyte senescence. Pigment Cell Res 2004;17:568. 54. Pollock PM, Harper UL, Hansen KS, et al. High frequency of BRAF mutations in nevi. Nat Genet 2003;33:19–20. 55. Omholt K, Platz A, Kanter L, Ringborg U, Hansson J. NRAS and BRAF mutations arise early during melanoma pathogenesis and are preserved throughout tumor progression. Clin Cancer Res 2003;9:6483–6488. 56. Yazdi AS, Palmedo G, Flaig MJ, et al. Mutations of the BRAF gene in benign and malignant melanocytic lesions. J Invest Dermatol 2003;121:1160–1162. 57. Wellbrock C, Ogilvie L, Hedley D, et al. V599EB-RAF is an oncogene in melanocytes. Cancer Res 2004;64(7):2338–2342. 58. Mooi WJ, Peeper DS. Pathogenesis of melanocytic naevi: growth arrest linked with cellular senescence? Histopathol 2002;41:120–146. 59. Bastian BC. The longer your telomeres, the larger your nevus? Am J Dermatopathol 2003;25:83–84. 60. Healy E, Belgaid C, Takata M, et al. Prognostic significance of allelic losses in primary melanoma. Oncogene 1998;16:2213–2218. 61. Dimri GP, Lee X, Basile G, et al. A biomarker that identifies senescent human cells in culture and in aging skin in vivo. Proc Natl Acad Sci USA 1995;92:9363–9367. 62. Talve L, Sauroja I, Collan Y, Punnonen K, Ekfors T. Loss of expression of the p16INK4/CDKN2 gene in cutaneous malignant melanoma correlates with tumor cell proliferation and invasive stage. Int J Cancer 1997;74:255–259. 63. Keller-Melchior R, Schmidt R, Piepkorn M. Expression of the tumor suppressor gene product p16INK4 in benign and malignant melanocytic lesions. J Invest Dermatol 1998;110:932–938. 64. Park WS, Vortmeyer AO, Pack S, et al. Allelic deletion at chromosome 9p21(p16) and 17p13(p53) in microdissected sporadic dysplastic nevus. Hum Pathol 1998;29(2):127–130. 65. Papp T, Pemsel H, Rollwitz I, et al. Mutational analysis of N-ras, p53, CDKN2A (p16(INK4a)), p14(ARF), CDK4, and MC1R genes in human dysplastic melanocytic naevi. J Med Genet 2003;40:E14. 66. Londono-Vallejo JA, Der-Sarkissian H, Cazes L, Bacchetti S, Reddel RR. Alternative lengthening of telomeres is characterized by high rates of telomeric exchange. Cancer Res 2004;64:2324–2327. 67. Ramirez RD, D’Atri S, Pagani E, et al. Progressive increase in telomerase activity from benign melanocytic conditions to malignant melanoma. Neoplasia 1999;1:42–49. 68. Glaessl A, Bosserhoff AK, Buettner R, Hohenleutner U, Landthaler M, Stolz W. Increase in telomerase activity during progression of melanocytic cells from melanocytic naevi to malignant melanomas. Arch Dermatol Res 1999;291:81–87.

196

From Melanocytes to Melanoma

69. Rudolph P, Schubert C, Tamm S, et al. Telomerase activity in melanocytic lesions: a potential marker of tumor biology. Am J Pathol 2000;156:1425–1432. 70. Davies H, Bignell GR, Cox C, et al. Mutations of the BRAF gene in human cancer. Nature 2002;417:949–954. 71. Straume O, Sviland L, Akslen LA. Loss of nuclear p16 protein expression correlates with increased tumor cell proliferation (Ki-67) and poor prognosis in patients with vertical growth phase melanoma. Clin Cancer Res 2000;6:1845–1853. 72. Pavey SJ, Cummings MC, Whiteman DC, et al. Loss of p16 expression is associated with histological features of melanoma invasion. Melanoma Res 2002;12:539–547. 73. Bartkova J, Lukas J, Guldberg P, et al. The p16-cyclin D/Cdk4-pRb pathway as a functional unit frequently altered in melanoma pathogenesis. Cancer Res 1996;56:5475–5483. 74. Herlyn M, Thurin J, Balaban G, et al. Characteristics of cultured human melanocytes isolated from different stages of tumor progression. Cancer Res 1985;45:5670–5676. 75. Herlyn M, Balaban G, Bennicelli J, et al. Primary melanoma cells of the vertical growth phase: similarities to metastatic cells. J Natl Cancer Inst 1985;74:283–289. 76. Albino AP, Vidal MJ, McNutt NS, et al. Mutation and expression of the p53 gene in human malignant melanoma. Melanoma Res 1994;4:35–45. 77. Castellano M, Parmiani G. Genes involved in melanoma: an overview of INK4a and other loci. Melanoma Res 1999;9:421–432. 78. Soengas MS, Capodieci P, Polsky D, et al. Inactivation of the apoptosis effector Apaf-1 in malignant melanoma. Nature 2001;409:207–211. 79. Baldi A, Santini D, Russo P, et al. Analysis of APAF-1 expression in human cutaneous melanoma progression. Exp Dermatol 2004;13:93–97. 80. Easty DJ, Bennett DC. Protein tyrosine kinases in malignant melanoma. Melanoma Res 2000; 10:401–411. 81. Blume-Jensen P, Hunter T. Oncogenic kinase signalling. Nature 2001;411:355–365. 82. Guldberg P, Thor SP, Birck A, Ahrenkiel V, Kirkin AF, Zeuthen J. Disruption of the MMAC1/PTEN gene by deletion or mutation is a frequent event in malignant melanoma. Cancer Res 1997;57: 3660–3663. 83. Pollock PM, Walker GJ, Glendening JM, et al. PTEN inactivation is rare in melanoma tumours but occurs frequently in melanoma cell lines. Melanoma Res 2002;12:565–575. 84. Plettenberg A, Ballaun C, Pammer J, et al. Human melanocytes and melanoma cells constitutively express the Bcl-2 proto-oncogene in situ and in cell culture. Am J Pathol 1995;146:651–659. 85. Yamamura K, Kamada S, Ito S, Nakagawa K, Ichihashi M, Tsujimoto Y. Accelerated disappearance of melanocytes in bcl-2–deficient mice. Cancer Res 1996;56:3546–3550. 86. Bush JA, Li G. The role of Bcl-2 family members in the progression of cutaneous melanoma. Clin Exp Metastasis 2003;20:531–539. 87. Omholt K, Platz A, Ringborg U, Hansson J. Cytoplasmic and nuclear accumulation of beta-catenin is rarely caused by CTNNB1 exon 3 mutations in cutaneous malignant melanoma. Int J Cancer 2001;92:839–842. 88. Worm J, Christensen C, Gronbaek K, Tulchinsky E, Guldberg P. Genetic and epigenetic alterations of the APC gene in malignant melanoma. Oncogene 2004;23:5215–5226. 89. Chen D, Xu W, Bales E, et al. SKI activates Wnt/beta-catenin signaling in human melanoma. Cancer Res 2003;63:6626–6634. 90. Bell DW, Varley JM, Szydlo TE, et al. Heterozygous germ line hCHK2 mutations in Li-Fraumeni syndrome. Science 1999;286:2528–2531.