Alterations in Gene Expression and Signal Transductions in Human Melanocytes and Melanoma Cells

.


I. INTRODUCTION
unknown but epidemiological evidence suggests that excess intermittent exposure to ultraviolet Cutaneous malignant melanoma (CMM) has light (UV) radiation may be partly responsible plagued mankind since recorded history, and the (Longstreth [1988]).first mention of this cancer was presented by Within the past decade, it has become possible Hippocrates in the fifth century (Urteaga and Pack to routinely culture human melanocytes in vitro.[1966]).The examination of pre-Colombian Inca Melanocytes from epidermis can be cultured for a mummies (approximately 2400 years old) dem limited time and require media containing serum, onstrated metastases to the bones (Selby et al. phorbol esters (e.g., 12-0-tetradecanoylphorbol [1956]).Melanoma was described as the "fatal 13-acetate, TPA), agents that elevate cAMP levels, black tumor with metastases and black fluid in the and bovine pituitary extracts (BPE) or basic fibro body" as encountered by Highmore (Urteaga and blast growth factor (bFGF) (Eisinger and Marko Pack [1966]).Despite the recognition and long [1982]; Halaban et al. [1987]; and Kath et al. documented history of this disease, the etiology [1989]).Melanocytes grown in these conditions of melanoma and the mechanisms leading to trans have a diploid karyotype, are nontumorigenic, formation remain largely unknown (Hecht [1989]).
and will not proliferate in anchorage-indepen The incidence of CMM has been increasing at an dent conditions.average rate of about 4% per year in the U.S.
The characterization of normal human pri (Longstreth [1988]).Those lesions that are not mary melanocytes and cells from various stages surgically cured are often fatal.The exact reason of human melanocyte transformation (dysplastic for the increase in the incidence of melanoma is nevi, primary and metastastic melanoma) has al lowed researchers to obtain information concern ing genotypic and phenotypic events associated with melanocyte transformation.These studies have observed differences in human melanocytes and melanoma cells, including alterations in the expression levels of certain genes and their ex pressed proteins, growth factor requirements, and growth characteristics in vitro and in vivo.Previ ous reviews have described cellular (growth re quirements), genetic (gene expression levels), phenotypic (protein expression levels), and clini cal alterations in human melanocytes and mela noma cells (Herlyn [1993]; Albino and Fountain [1993]; Bennett [1993], and Clark [1991]).In this review we summarize recent studies on alterations in gene expression and signal transduction dis covered in human melanocytes and melanoma cells and highlight those factors that may playa role in human melanocyte transformation and how these changes may interact with each other to produce the malignant phenotype.

II. CHROMOSOMAL CHANGES AND A GENETIC MODEL OF TRANSFORMATION
Chromosome alterations have been described in detail in several reviews of human melanoma (for reviews see Albino and Fountain [1993]; Bennett [1993]; Herlyn [1993]; and references therein).Cytogenetic studies have identified chro mosomes 1, 6, 7, 9, 10, and 11 as frequently involved in human melanoma.Based on this in formation, a presumptive working model of ac quired genetic change and phenotypic alterations as a melanocyte progresses to transformation has been proposed by a number of investigators, and this model is updated and presented in Figure 1 (Albino and Fountain [1993]).
Alterations have been observed in chromo some 1p in human melanoma tissue biopsies and cell lines.Bale et al. (1989) have noted either a loss of heterozygosity or cytogenetic alterations for chromosome 1p in human melanoma tissues and cell lines.This loss appears to occur late in human melanocyte transformation.There was a frequent loss of heterozygosity in 43% of mela noma tissue biopsies and 52% of melanoma cell lines on chromosome 1p36, a region that contains genes encoding c-jun, RAB3B, VCAMl, TCL5, and p58 clk -1 • Alterations in the expression of c-jun and p58 clk -1 gene are discussed in more detail in this article.
The next most common alteration observed in human melanoma was in chromosome 6q 12-25 (Trent [1983]; and Parmiter and Nowell [1993]).Several genes have been mapped to this region, including c-myb, e-ros, and others.The most de finitive study to determine the gene involved in this region was initiated by Trent et al. (1990).Using microcell hybrids, the introduction of a normal chromosome 6 into melanoma cells re sulted in the ability of the hybrids to form tumors in vivo.A possible gene located in this region of chromosome 6 is discussed later in this review.Welch et al. (1994) have determined that intro duction of a normal chromosome 6 suppressed the metastatic but not the tumorigenic phenotype of human metastatic melanoma cells.An inverse correlation was also observed with nm23-Hl RNA transcript expression levels and the metastatic ability of the hybrid cell clones.
An increase in the ploidy of chromosome 7 (7pll-l3) has also been observed in human mela noma and correlates with an increase in the copy number of the epidermal growth factor receptor (EGF-R).Alteration in a gene similar to EGF-R has been observed in a swordfish melanoma model.An experimental model for the formation of ultra violet radiation induced melanoma using Fl pro geny from a mating with swordtail (Xiphophorus belleri) and platyfish (Xiphophorus maculatus) has determined that there are two genes involved in the formation of tumors (Ahuja and Anders [976]).One of the genes responsible was a tumor suppressor gene, Tu, which encodes Xmrk, a membrane receptor tyrosine kinase similar to the human EGF-R (Anders [1991]).
An alteration that may occur early in human melanocyte transformation is located on chromo some 9. Cytogenetic studies from several groups suggest the involvement of chromosome 9p21 p22 as a locus for familial melanoma susceptibil ity gene (Cannon-Albright et al. [1992]; Fountain et al. [1992]; Holland et al. [1994]; Isshiki et al. [1994]; and Coleman et al. [1994]).Fountain et al. (1992) determined that 85% of melanoma tis sue biopsies and cell lines had alterations in this region.p16 has been proposed to be the gene altered in this region and is discussed later in this review.
Changes on chromosome 10 have also been associated with human melanocyte transforma tion as well.Parmiter and Nowell (1993) have observed alterations involving chromosome 10 in 10% (1/10) of dysplastic nevi, 67% (2/3) of pri mary melanomas, and 37% (19/51) of advanced melanomas.Using either loss of heterozygosity or cytogenetic studies, other chromosomes that show alterations include 11, 2, and 3, but these changes have been less frequent.The genes in volved at the identified cytogenetic sites of abnor mality are currently unknown, and studies are underway to define these regions.

III. ALTERATIONS IN GENE EXPRESSION
Alteration in the RNA transcript expression levels of various genes has been investigated using Northern blot hybridization analysis and reverse transcriptase in combination with polymerase chain reaction (RT-PCR).Researchers have observed changes in the relative expression levels of specific genes during human melanocyte transformation (Chenevix-Trench et al. [1990]; Herlyn [1993]; and Albino and Fountain [1993].Some of these alterations include: The phorbol ester, TPA, affects the expression of many genes (for review see Karin and Herrlich [1989]).One of the transcription factors involved in modulating the expression levels of TPA-inducible genes is the c-jun protein, which in cellular form is one of several polypeptides that form a complex called AP-l (a combination of the jun and fos fami lies).The c-jun gene belongs to a gene family; the other members beingjun-B andjun-D.Jun-B inhib its the transforming and transactivating activities of the c-jun protooncogene (Schutte et al. [1989]).This complex recognizes a specific DNA sequence that mediates the transcription response to phorbol es 431 ters.Some of the TPA-inducible genes are proto oncogenes (c-fos and c-myc), proteases (collage nase and stromelysin),and cytokines (interleukin 1~ and 2) (for review see Karin and Herrlich [1989]).Thus, change in the expression level of the AP-l complex may induce the expression of genes that assist in melanocyte transformation.
Transcriptional activation of the c-fos gene by various agents has been investigated by many researchers.c-fos belongs to a family of genes, including fos-b, fra-l, and fra-2.c-fos whose mRNA levels increase during stimulation of cells with growth factors or mitogens and after DNA damage (Verma [1986]; Sassone-Corsi et al. [1988]).c-fos is a protooncogene whose elevated expression level can induce tumors in vivo (Jenumein et al. [1985]) and transformation of cells in vitro (Miller et al. [1984]).c-fos expres sion has been investigated in an immortalized, nontumorigenic, murine melanocyte cell line.Hart et al. (1989) observed an increase in c-fos mRNA following treatment of murine melanocytes with dibutryl cyclic AMP (dbcAMP).Investigators have also studied the inhibition of c-fos RNA transcript expression by antisense fos expression vectors.Both the proliferation of dividing (Holt et al. [1986]) and quiescent cells stimulated with mitogens (Nishikura and Murray [1987]) was blocked by antisense fos.
We have described alterations in the expres sion ofjun andfos families in human melanocytes and melanoma cells (Yamanishi et al. [1991a]; Jiang et al. [1993]).We observed an increase in c fos and a decrease in c-jun RNA transcript ex pression levels as melanocytes went from quiescent to proliferative growth states.A similar change in expression of c-jun (decreased), c-fos (increased), andjun-B (decreased) was observed in human melanoma cells compared with melano cytes.Changes in the expression levels of these genes may alter the ability of melanoma cells to undergo differentiation.The shift in the expres sion of the c-jun and c-fos RNA transcripts should shift AP-l activity, resulting in the expression of other genes that are involved in cell proliferation rather than differentiation.
Other genes that have AP-l binding sites lo cated in their promoter are growth factors (e.g., bFGF), cell division cycle genes (e.g., p58), and cytokines.Jiang et al. have determined that mela noma cells induced to differentiate had an el evated c-jun and jun-B RNA transcript expres sion level (Jiang et al. [1993]).Alterations in the expression of these gene families may determine whether melanocytic cells proliferate or undergo differentiation.Future studies using expression vectors will determine what role these genes play in cell growth and differentiation and which genes are altered by overexpression of these transcrip tional regulators.

B. Retinoic Acid Receptor (RAR) Isolypes
The nuclear RAR family has provided a mechanistic basis by which gene expression may be modulated by retinoids.Three RAR isotypes a, ~, and "( have been identified in mammalian cells and are conserved between species (for re view see Chambon et al. [1991] and Mangelsdorf and Evans [1992].Variations in the expression of the RAR isotypes have been found in many tis sues (Elder et al. [1991]; Chambon et al. [1991]; and Mangelsdorf and Evans [1992]).RAR a RNA transcripts are expressed in most tissues, while expression ofRAR ~ RNA transcripts in tissues is variable and RAR "(RNA transcripts are expressed at high levels in the skin.A retinoic acid-respon sive element has been found in the promoter of the RAR ~ isotype and is a direct repeat of the sequence AG(Gff)TCA separated by five nucle otides (De Luca [1991]).A target for retinoid induced transcription activation is the RAR ~ isotype.While the RAR a and "( isotypes are constitutively expressed, the RAR ~ isotype is markedly inducible by all-trans retinoic acid (tRA).
We have determined the basal RNA tran script expression levels of RAR isotypes in hu man melanocytes and melanoma cells using Northern blot hybridization analyses.RAR a (2.8 and 3.6 kb) and RAR ~ (2.8 and 3.1 kb) RNA transcripts were detected in melanocytes.Expres sion of the RAR a, ~, and "( RNA transcripts in melanocytes was only slightly affected by the addition of growth factors in the melanocyte medium.RAR a RNA transcript expression lev els were decreased in melanocytes cultivated in complete medium and RAR ~ RNA transcript expression levels increased as the melanocytes were cultivated in more complete medium.RAR yRNA transcripts (2.8 kb) were detected in mel anocytes cultivated in the various growth me dium.Expression of the RAR isotypes in melanoma cells was different compared with mel anocytes.RAR ex and y RNA transcripts were detected in all four melanoma cell strains, while RAR ~ RNA transcripts were detected only in the melanoma cell strains c8l-46a and c8l-46c.Expression of the RAR isotypes RNA transcripts in human melanocytes and melanoma cells as well as their cell response to 13-cis retinoic acid (13-cisRA) and tRA is summarized in Table 1.
Although tRA was previously shown to in duce the expression of RAR ~ isotype in other cell types, we investigated if the undetectable RAR ~ RNA transcript expression levels in two of the melanoma cell strains was due to an altered cell responsiveness.In those melanoma cell lines that had basal expression of RAR ~, tRA induced RAR ~ RNA transcripts (3-to 8-fold).RAR ex and yRNA transcripts expression levels were also induced following tRA treatment.However, in melanoma cell strains c8l-6l and c83-2c that lacked onstitutive expression of the RAR ~, nei ther of the RAR ~ RNA transcripts were induced following tRA treatment.Loss in the expression of the RAR ~ isotype may be one mechanism by which melanoma cells become resistant to tRA treatment.Future studies using RAR ~ expression vectors may allow us to determine the role of the RAR ~ isotype and provide insight into the mecha nisms involved in determining melanocyte re sponse to tRA.

C. p58 c1k -1
The protein complex that controls the onset of mitosis in normal cell cycle contains at least four members: p3¥dcZ, cyclin B, p13 sucl, and cdc25 (for review see Maller at al. [1991]).The mitotic protein kinase p34 cdcZ is required for both the G zto M-phase transitionand the G 1 -to S-phase transitionsin the cell cycle.p58 c1k -1 was shown to be structurallyand func tionally related to p34 cdcZ (Bunnell et al. [1990]) and has been localized to chromosome 1p36 (Bunnell et al. [1990]; Eipers et al. [1991Eipers et al. [ , 1992]]).Its abnormal expression alters the ability of eukaryotic cells to progress through the cell cycle and changes cell phenotype.p58 c1k -1 may function to negatively regu late normal cell cycle progression.Due to the chro of the p58 cl k mosomal location -1 gene, we investigated its expression in four melanoma cell lines and were unable to detect its expression (Table 2), although the gene was expressed in normal melanocytes.We also observed a decrease in the expression level of the p58 c1k -1 in melanoma tissue biopsies.We have been unable to detect any gross rearrangements in the p5SC 1k -1 gene using Southern blot hybridization analysis, which would suggest an alteration in the activity of the promoter.

D. myc Family
Alteration in the expression of the myc family (c-myc [chromosome 8q], N-myc [chromosome 2p], and L-myc [chromosome Ip]) have been ob served in human tumor cell lines.c-myc RNA transcript expression levels have been shown to be tightly linked to the proliferate state of the cell, and the myc protein may serve to induce GoIG I transition genes.
While Linnenbach et al. (1988) were unable to observe any gross alterations in the myc family from 28 melanoma cell lines, Bauer et al. (1990) observed an amplification of the N-myc gene in biopsies from two melanoma patients.The effect of the overexpression of the c-myc gene on hu man melanoma cells has been investigated by several groups.Versteeg et al. (1988) determined the expression levels of the c-myc gene and HLA class I antigens in 11 melanoma cell lines.A high level of c-myc expression was observed to corre late inversely with the expression of the HLA class I antigen (Versteeg, et al. [1989]; Schrier and Peltenburg [1993]).Transfection of mela noma cells (IGR 39 cells with a low c-myc and high HLA class I antigen expression level) with a c-myc expression vector resulted in a decrease in the HLA class I antigens and ~2-microglobulin expression (Versteeg et al. [1988]).Lenardo et al. (1989) determined that N-myc may work by alter ing the factor binding to the HLA class I gene enhancers.The effect of myc expression on HLA class I antigens was determined to be through enhancer inactivation.There are two distinct ele ments in the MHC class I promoter that are sus ceptible to N-myc suppression (Lenardo et al. [1989]).However, a consistent overexpression of the myc genes has not been observed in mela noma cell lines compared with melanocytes (Chenevix-Trench et al. [1990]).
bFGF protein has a basic pI (>9.0) and a high affinity for heparin and glycosaminoglycans.Us ing heparin-affinity chromatography and bFGF antibodies, the translation of bFGF has been inves tigated.Four bFGF polypeptides (18, 21, 22.5, and 24 kDa) have been detected in human cells.Utili zationof non-AUGcodons producedthe three larger molecular weight bFGF isoforms (Florkiewicz and Sommer [1989]).
The role of bFGF in melanocyte cell growth has been studied widely, and this growth factor is a mitogen in vitro (Halaban et al. [1988]).How ever, the expression levels of bFGF in human melanocytesand melanomacells varies.While there is some disagreement on whether human melano cytes express bFGF RNA transcripts, bFGF pro tein was not detected (Yarnanishi et al. [1992]; Rodeck et al. [1991]; Halaban et al. [1988]).Our data on the absence of bFGF RNA transcript ex pression in melanocytes are supported by in situ hybridization studies.Nevertheless, bFGF RNA transcripts are expressed at various stages of mel anocyte progression and were detected in tissue from nevi, primary melanomas, and metastatic melanomas (Scott et al. [1991]).Interestingly, the expression level ofbFGF RNA transcripts detected decreased with increasing progression toward ma lignant tissue (nevi (+2 to +3), primary (+2), and metastatic melanoma [+1]).On the other hand, our studies and those of others indicate that melanoma cells expressed very low levels of bFGF RNA transcripts and proteins (Yamanishi et al. [1992]; Halaban et al. [1988]).While three of four cell strains expressed bFGF RNA transcripts and pro tein isoforms, the cell strain c83-2c only expressed the 1.2-kb RNA transcript and did not synthesize any detectable levels of bFGF protein as measured by Western blot analysis.
The effect of bFGF neutralizing antibodies on human metastatic melanoma cell growth has been investigated also (Becker et al. [1992]).A decrease in melanoma proliferation was observed following incubation of cells with antibodies to bFGF pro tein.Inhibition of cell proliferation by antisense oligonucleotides to the bFGF gene also has been studied and using antisense oligodeoxynucleotides targeted against the bFGF gene, Becker et al. (1989) demonstrated a slight inhibition of cell prolifera tion in melanoma cells as well as inhibition of anchorage-independent growth.The use of either bFGF antisense oligonucleotides or bFGF antibod ies had only a limited effect on cell growth, sug gesting that there may be more than one fibroblast growth factor or bFGF-R involved in melanoma cell growth or that bFGF is of secondary impor tance in contributing to cellular proliferation of melanoma cells.These data suggest that although exogenous bFGF protein may be required for the proliferation of human primary melanocytes and the molecule is expressed frequently by melanoma cell strains, expression of the bFGF protein does not appear to be a consistent or necessary alteration in melanocyte transformation.

F. PKC Isotypes
PKC is a serine/threonine protein kinase that interacts with calcium ions, phospholipids, and diglycerides to form a complex associated with a cellular membrane structure (Nishizuka et al. [1988]).PKC represents a multigene family and 10 different cDNA clones have been isolated to date (Nishizuka [1988]; Ohno [1991]).Three dis tinct groups of PKC genes have been isolated to date.The four conventional PKC (cPKC) cDNAs (a, ~l' ~ll' and y) are encoded by three distinct genes (a, ~, and y) with the ~l and ~ll RNA transcripts formed by utilizing different 3' splice acceptor sites.In addition to the cPKC genes, four novel PKC (nPKC) cDNAs (0, E, Tl/L, and 8) and two atypical (aPKC) cDNAs (~ and A) have been identified and have been shown to have regions of high homology (conserved regions, C) and weak homology (variable regions, V) at the protein level.cPKC isotypes have four highly conserved (C\ C 4 ) regions surrounded by five variable (V,-V s ) regions, whereas the four nPKC and two aPKC isotypes lack the C 2 region.The aPKC isotypes are also missing one of the two cysteine-rich do mains located in the C\ region.
PKC has been implicated in the regulation of many cellular processes, including growth, dif ferentiation, neuronal function, and gene expres sion (Nishizuka [1988]; Nairn et al. [1985]).Although the role of each PKC isotype in cellular processes is unknown, investigations in the overexpression of PKC a gene (Housey et al. [1988]) or expression of a mutated PKC isotype in mouse fibroblasts (Megidish and Mazurek [1989]) have demonstrated either altered growth regulation or complete transformation of the trans fected cells.In addition, PKC has been implicated in the in vitro transformation processes induced by the oncogenes ras, sis, fins, srs, fps, and fes (Jackowski et al. [1986]; Preiss et al. [1986]).Researchers have observed either elevated levels of sn-I ,2-diacylglycerol or phosphorylation of a transformation-related protein and a PKC sub strate in cells transformed by these oncogenes.
We have measured the expression of PKC isotypes in human melanocytes and melanoma cells (Yamanishi et al. [1991b]; unpublished data), due to their involvement in the TPA response (its membrane receptor being the PKC).The expres sion levels are summarized in Table 3.We and others have observed that melanocytes expressed PKC a, ~, and E isotypes, and that the PKC isotypes could be downregulated following TPA treatment (Yamanishi et al. [199lb]; Arita et al. [1992]; Powell et al. [1992]).Melanoma cells expressed PKC a and PKC E RNA transcripts at various levels.Alterations in the expression of PKC a (increased), PKC ~II (undetectable), and PKC E (increased) were common to the mela noma cells (compared with melanocytes) screened.The most striking finding was that PKC ~II RNA transcripts were detected in only primary melano cytes or benign moles (Table 3).
Eighteen metastatic melanoma cell lines (cell lines from ATCC and those derived by Dr. Meyskens' laboratory) have been screened for PKC ~II RNA transcripts using Northern blot hybridization analysis and expression was unde tectable in all of the cell lines (Table 3).Dysplas tic nevi and metastatic melanoma biopsy tissue samples were also screened and were shown to have either decreased or undetectable PKC ~II RNA transcript expression levels in nevi and in melanoma cells and biopsies.We have also inves tigated the expression levels of the PKC isotypes in other human tumors.There was a consistent loss in the expression of the PKC ~II isotype in malignant tissue biopsies and cell lines derived from human brain, head and neck regions, breast, and hemapoietic tissues (unpublished data).
As measured by Southern blot hybridization, no gross alterations were detected in the PKC ~ gene in four melanoma cell strains using the re striction enzymes, BamHI, EcoRI, and Pvull.Our data clearly demonstrate a loss in the expression of the PKC ~II RNA transcripts early in human melanocyte transformation; we speculate this was possibly due to an alteration in the transcriptional activity of the PKC ~ promoter.
Inorder to determine the functional role for the PKC ~II isotype in human melanocyte transforma tion, we have transfected human melanoma cells with an expression vector containing the PKC ~II isotype, pSRa-PKC~.Colonies were isolated in cells transfected with either the selection vector, pSV2-neo, alone or co-transfected with an expres sion vector containing the PKC a isotype (Table 4).However, a decrease in the number of colonies was observed with transfection of the PKC ~I or ~II isotypes.Melanoma cell growth was suppressed after 1 week following transfection with the ex pression vector containing the PKC ~I or ~II isotype, suggesting that the PKC ~ isotype can suppress tumor growth.Choi et al. (1990) also observed similar results when PKC ~I was transfected in colon cancer cells.The cells had reduced tumorigenicity and ability to grow in anchorage independent conditions.Thus, PKC ~II may playa common role early in human transformation and alterations may lead to abnormal (i.e., tumor) cell growth.Although the gene for PKC ~II does not undergo genomic rearrangement, the expression of the PKC ~II isotype is downregulated.This may be another mechanism by which a cell can regulate cell growth and undergo cell transformation.We hypothesize that there is a common alteration dur ing human melanocyte transformation, resulting in the loss of a transcription factor that induces the expression of the PKC ~II isotype.6-9

RNA Transcript Expression
N.T.a 0-2 0-3 Note: Cells were cultivated until 50 to 70% confluent and then transfected with plasmids.Cells were rinsed and prewarmed fresh medium was added to the plates.Cells were allowed to recover for 2 d and then fresh medium containing 250 Ilg/ ml of G418 was added to the plates.Cells were fed twice per week with fresh medium containing G418.
The effect of tRA on the expression of PKC isotypes in murine melanoma cells, B 16, has been investigated by many groups.tRA-induced dif ferentiation of B 16 induces cyclic AMP-depen dent protein kinase and PKC (Ludwig et al. [1980]; Rogelj et al. [1984]).Selective increase in the expression levels of PKC ex isotype with no de tectable alteration in the expression of the PKC <5, E, and ~ was observed in tRA-treated B16 cells (Oka, M. et al. [1993]).Transfection of the PKC ex isotype induced longer doubling times, reduced anchorage-independent growth, longer tumor for mation in vivo, and increased melanin production (Gruber, J. R. et al. [1992]).
Interestingly, Park et al. (1993) has pursued the role of the PKC ~II isotype in pigmentation in human melanoma cells and determined that hu man melanoma cells were depigmented due to a decrease in the phosphorylation of the tyrosi nase protein by PKC ~.Pigmentation in this cell line could be recovered by transfecting the PKC ~ isotype into the melanoma cells.We have also observed that transfection of melanoma cells with a constitutive PKC ~ expression vector induced pigmentation in two amelanotic melanoma cell lines, and a decrease in cell growth [unpublished data].However, we also have a melanoma cell line that does not express PKC ~ RNA tran scripts and is pigmented.It may be that other PKC isotypes can substitute for the PKC ~ and activate the tyrosinase protein.Thus, alteration in the expression of the PKC isotypes may play a role in melanocyte transformation through a number of mechanisms.

G. Epidermal Growth Factor Receptor (EGF-R)
EGF-R is a member of a famiy of growth factor receptor tyrosine kinases (Carpenter [1987]).These receptors have an extracellular domain that contains a site to which growth factor or ligand binds, an intracellular domain encoding a tyrosine kinase and connected by a transmembrane do main of hydrophobic amino acids.Intracellular portion of the EGF-R has three tyrosinase resi dues that are autophosphorylated after binding of EGF to the receptor.Activation of PKC will re sult in the phosphorylation of the threonine 654 residue that contributes to the conversion of EGF-R from the high-to low-affinity state (called transmodulation), which results in a form of de sensitization of the receptor.Ligands for the EGF-R include TGF-ex and EGF.Increased levels of the EGF-R have been found in breast cancer, gliomas, and carcinomas.
An elevated expression level ofthe EGF-R in melanoma cells has been correlated to an increase in the copy number of human chromosome 7.An increase in the expression level also has been used as a marker of tumor progression by in situ studies.The EGF-R is detected in vertical growth phase primary (>80%) and metastatic (>80%) melanomas compared with normal melanocytes and common nevi (0%), and dysplastic nevi and radial growth phase primary melanomas (20%) (De Wit et al. [1992]).

H. Ras Family
There are three members of the ras family: Ha-ras (chromosome IIp), Ki-ras (chromosome l2p), and N-ras (chromosome lp).These genes code for 21-kDa protein (p2l), which functions as a GTP/GDP binding protein with a GTPase activ ity and plays a role in signal transduction, cell proliferation, and differentiation.Mutations in the first, second, and third codons activate the ras genes with transformation potential.van t'Veer et al. ( 1989) used PCR and oligo nucleotide hybridization to detect point mutations in the N-ras gene in 7/37 cutaneous melanomas (primary, metastatic, and cell lines).Interestingly, the tumors with an activated N-ras oncogene were found in sites with continuous sunlight exposure.The site of mutations (codons 12, 13, and 61) were not always at a thymidine dimer but mostly at a thymidine-cytidine dimer site.In a more de tailed study by Shukla et al. (1989) using PCR and oligonucleotide hybridization, activated N-, Ki-, and H-ras oncogenes were detected.In be nign nevi samples, 2/4 samples had a codon 12 Ki-ras mutations.In primary melanoma, 2/22 had a codon 12 ki-ras mutation, 1/22 had a codon 61 N-ras mutation, and 1/22 had a codon 12 Ha-ras and a codon 12 Ki-ras mutations.In lymph node metastases, 2/12 had a codon 12 Ki-ras mutation, 1/12 had two different codon 12 Ki-ras muta tions, and 1/12 had both a codon 12 Ki-ras and a codon 61 N-ras mutation.In systemic metastases, 0/2 had no activated ras gene mutations.Albino et al. (1989) have detected mutations in the ras genes in melanoma using PCR and oligonucleotide hybridization.They were unable to detect an activated ras gene in biopsies from nevi tissue.In primary melanomas, 5% of the samples had an activated N-ras gene at the codon 61st.In metastatic melanoma biopsies, 6% of the samples had a mutation in the 61st codon of the N ras gene, while cultured metastatic melanomas had mutations in the 61st codon (22% or the codon 13 th of the N-ras gene and in the 61st (2% ) codon of the H-ras gene.Interestingly, the melanomas with an activated ras gene were representative of early or intermediate stages of differentiation as determined by the expression of a large number of EGF receptors, class II histocompatibility an tigens (IA), nonpigmented, and with a morphol ogy that was epithelioid/spindle type.
The expression of the ras gene was investi gated using a monoclonal antibody against Ha-and Ki-ras protein (Yasuda et al [1989]).A highly positive reaction was observed in melanocytic nevi located in the dermal region, while low reactivity or no detectable response was detected in com pound nevi and junctional nevi.High reactivity was also detected in nodular melanoma and meta static melanoma.Medium reactivity was observed in acral lentiginous melanoma and superficial spreading melanoma.The authors comment that "the different p2l expression levels among the type of tumors may represent the state of tumor cell differentiation with greater p21 expression with the more immaturity in the melanocyte lineage".Ball et al. (1994) also have observed ras mutations in a subset of melanomas from sun exposed skin.Ras mutations were observed in 56% of tumors from continuously exposed skin sites compared with 21% of tumors from inter mittent or non-sun-exposed skin sites.Most of the mutations in the ras genes occurred as primary melanomas progressed from Clark's level II to III; their results suggest that the ras mutations occur during the phase that melanomas acquire a more aggressive phenotypic behavior.They pro pose that "activated ras contributes to the growth advantage of melanomas in the vertical growth phase of dermal invasion" and may playa role in tumor progression in a subset of melanoma.

I. p53
Mutationsin thep53 gene are the most common alterations in human cancers (Harris [1994], and references therein).Loss of the wild-type p53 gene has been observed in a high proportion of lung, brain, breast, colon, ovary, and bladder tumors.
Alterations in the p53 gene have been ob served in human skin cancer also.An elevated expression level of the p53 protein was detected in malignant melanomas compared with primary melanocytes and dysplastic nevi using immun ofluorescence and flow cytometry (Stretch et al. [1991]; Akslen and Merkve [1992]; Lassam et al. [1993]).However, Volkenandt et al. (1991) were able to detect a mutated p53 gene in only one out of ten melanoma cell lines.Montano et al. (1994) observed a differential increase in nuclear expression of p53 protein in melanoma cells but were not able to detect a mutation in the p53 gene.Our studies on the expression of the p53 gene were similar (Table 5).Although we were able to detect an alteration in the RNA transcript expression level of the p53 gene in four human melanoma cell lines and two meta static melanoma tissue biopsies, we were unable to detect any point mutations in the entire coding regions of the p53 gene in these tissue biopsies (unpublished data).

IV. PROPOSED MOLECULAR MODEL OF PROLIFERATION AND DIFFERENTIATION PATHWAYS
Notwithstanding the large number ofmolecu lar changes that have been detected in human melanomas, we proposed that a few alterations are key and at the heart of regulation of melano cyte growth and maturation (Figure 2).These hypotheses include 1.The interaction of the cAMP transduction pathway with bFGF through the c-fos/AP-l mechanism 2. The effect of TPA (may act as a surrogate for UV radiation) on the PKC pathway and its interaction with c-jun/AP-l mechanism 3. The modulation of the AP-l mechanism via interaction with the RAR family of tran scription factors Several studies have investigated the role of the transcription factor AP-l and the steroid re ceptors in cell pathway programming (ponta et al. [1993] and references therein).The steroid hor mone receptor family includes the retinoic acid, thyroid hormone, and vitamin D receptors.Al though there is crosstalk between the various sig nal transduction pathways, a reductionist model would suggest that AP-l is involved in the prolif eration program, while the steroid hormone re ceptors are involved in the differentiation program (Figure 2).Several studies have investigated how these two pathways interfer and interface with each other, and they probably modulate the growth of human melanocytes in a critical manner.
In one aspect, AP-l may be a key pathway producing a proliferative response in melanocytes.Quiescent melanocytes have a low level of c-fos with a high level of c-jun protein.Stimulation of human melanocytes to proliferate with bFGF, serum, and cAMP induces the expression of c-fos protein levels.An increase in c-fos protein levels will stimulate the formation of AP-l heterodimers and result in a subsequent decrease in c-jun RNA transcript expression levels (induction of other genes rather than inducing its own expression).The activation of AP-l heterodimers may induce  the expression of specific genes involved in cell proliferation through TPA-responsive elements (TRE).Also, TPA or UV affects PKC with down stream modulation of c-jun/AP-l.
On the other hand, RARs largely act on the differentiation pathway.Treatment of cells with RA or other differentiation hormones may trans mit signals to the nucleus to induce cell differen tiation.Alterations in the expression of the RARs may change the ability of the melanocytes to undergo differentiation.A decrease in RAR RNA transcript expression levels and an inability to induce the expression of the RAR ~ isotype was observed in tRA-resistant human melanoma cells.This situation may also occur in human squamous cell carcinomas and other tumor types: a loss in the expression of the RARs may allow a cell to escape from normal cell programming.
In this model, we have proposed how two major signal transduction pathways may modu late the ability of a cell to undergo a proliferation or differentiation response to external stimuli.In normal melanocytes, there exists a homeostatic interaction in the proliferative and differentiation signals.Alterations in this balance (treatment with growth factors or retinoids) may shift the focus of the cell.Transformation of the human melano cyte would likewise result in inhibiting or altering the ability of the cell to undergo differentiation.There may be other "partners" in the scheme of human melanocyte proliferation/differentiation that are not addressed in this model; the role of the AP-1 regulator, IP-1, and the retinoic acid related receptors are reasonable candidates.With future studies, the interaction of these genes and others will be investigated and may allow a clearer picture to develop at the molecular level.

V. INVOLVEMENT OF UV RADIATION IN TRANSFORMATION
The epidemiological evidence that suggests a causative role of nonionizing radiation in the for mation of melanoma include (Longstreth [1988]): 1.The incidence of CMM is higher in lighter pigmented people 2. Freckles and nevi are induced by sunlight exposure 3. Epidemiological correlation between decreasing latitude and increasing sunlight exposure and higher CMM rates 4.
The higher incidence of CMM in patients who cannot repair UV radiation DNA damage 5.
The indication of excess sunlight exposure at an early age and higher CMM incidence Experimentally, UV radiation has been shown to induce cellular and molecular alterations in vivo and in vitro (Kripke [1990]; Ambach and Blumthaler [1993]; Kainaetal. [1989]).Besides its well-known mutagenic response in all cellular sys tems (Hanawalt [1991]), UV radiation, particularly at low doses, induces a transient mitogenic re sponse and alters the expression of genes.The mitogenic and mutagenic responses have been demonstrated in many cells and are either immedi ate or early alterations induced in cells following UV radiation exposure.In contrast, the late events following UV irradiation, particularly at high doses, are cell death, cell mutagenesis, and cell transfor mation (Longstreth [1988]; Ananthaswamy and Piercell [1990]).
In general, the ozone layer in the upper strato sphere blocks out shorter wavelengths of UV ra diation (less than 290 nm) so that UV-C probably does not play a role in melanocyte transforma tion.Whether this is true with a depleted ozone layer is less certain (Jones et al. [1987]; Van der Lubbe et al. [1988]).An increase in the incidence of basal and squamous cell carcinomas (utilizing data from dose-response models) suggest that for each 1% increase in UV-B (290 to 320 nm), a 1.0 to 2.8% long-term increase in skin cancer is pre dicted (Fears et al. [1987]).To date, no dose response model for malignant melanoma has been determined and thus the estimates of the increase in the incidence of cutaneous cancers are on the conservative side (Rogers and Gilchrest [1990]).
Exposure to UV radiation also has the poten tial to interfere with the immunological response to tumors.Two immune responses impaired by high doses of UV radiation are contact hypersen sitivity to skin sensitizers and delayed hypersen sitivity to complex antigens that may be regulated by antigen-specific suppressor T lymphocytes (Kripke [l990]).Kripke et al. have observed an UV-induced immunosuppressive effect on mice and the inability of UV-irradiated mice to reject transplanted tumor cells (melanoma and other skin cancer cells) as well as the development of pri mary skin cancer.Thus, UV radiation may not only directly induce skin cancer but also inter feres with the body's immune responses to the altered phenotype.
What role UV radiation plays in melanoma causation is still not totally understood at the basic mechanistic level.However, UV radiation has been shown to induce several alterations in vitro and in vivo, including I.
Induced DNA damage 2. Altered gene expression 3. Changed cell membrane components 4. Inhibition of the antioxidant system The mutagenic effect ofUV radiation has been well characterized.The two types of DNA damage that are considered to be primarily responsible for the lethal and mutagenic effects of UV radiation 442 are the formation of cyclobutane pyrimidine dimer and (6-4) pyrimidine-pyrimidine photoproducts.Transition and transversion mutations have been observed following in vitro UV irradiation (Vrieling et al [1989]).An assay to measure the repair of UV-induced photoproducts in genes using the en zyme T4 endonuclease V has been developed by Hanawalt (Bohr et aI. [1989]).Investigators have shown that the repair of photoproduct adducts is dependent on gene transcription and DNA methy lation levels (Bohr et al. [1989]).While cell sur vival of UV mammalian cells does not correlate with overall genomic DNA repair, it does correlate with repair of essential genes (Bohr et aI. [1987]).
Decreased repair of UV radiation-induced pyrimidine dimers has been observed in patients with basal cell carcinomas (Alcalay et al. [1990]).Cancer patients and healthy volunteers were treated with a single dose of solar-simulated ra diation (dose equal to one minimal erythema dose), and DNA repair was measured using the T4 endo nuclease V assay.There was a decrease in the DNA repair ofUV-induced photoproducts in can cer patients compared with healthy volunters (22 to 33%, respectively).Cell lines derived from patients with hereditary dysplastic nevus syndrome (DNS) were also shown to be hypennutable fol lowing exposure to UV radiation (Perera et al. [1986]).These DNS cell lines had similar cell survival values following exposure to UV-C but had a two-to threefold increase in frequency of induced mutants (6-thioguanine resistance assay) compared with cell lines derived from control individuals.To examine the mechanism involved in this UV radiation hypennutability, Seetharam et al. (1989) used a transient shuttle mutagenesis assay.There was an increase in frequency of single base mutations in UV-irradiated plasmids iso lated from DNA cell line compared with a normal cell line.Conceivably, a similar situation may occur with melanomas because dysplastic nevi are a precursor to this cancer.Schothorst et al. (1991) have investigated the induction of pyrimidine dimers and DNA repair in cultured human keratinocytes and melanocytes following UV irradiation.Using monochromatic UV radiation of 254, 297, 302, and a light source emitting predominantly 312 nm, the number of T4 endonuclease V-sensitive sites (ESS) in ge nomic DNA was determined.The action spectra for dimer induction in keratinocytes and melano cytes was similar for other cultured mammalian cells.The kinetics and overall genomic DNA re pair was found to be similar for both cell types.Nine hours after UV irradiation, 55% of ESS were removed and after 24 h 70% of ESS were removed.Melanocytes were irradiated with a higher dose (250 J/m 2 ) compared with keratino cytes (200 J/m 2 ) to obtain a similar amount of ESS in the cells, so melanin may have some UV radiation protective effect in melanocytes.
There is also evidence that UV radiation may induce photoproducts in the ras gene, which is mutated in 10 to 25% of human melanomas.Activated ras oncogenes have been found in hu man squamous cell carcinomas, basal cell carci nomas, and metastatic melanomas.Activated c-Ha-ras oncogenes were found in four out of eight human basal and squamous cell carcino mas that occurred on sun-exposed body sites (Anathaswany et al. [1988]; Anathaswany and Pierceall [1990]).Activated ras (N-, Ki-, and Ha-) oncogenes have also been demonstrated in benign atypical nevi, primary, and metastatic melanomas ( Van'T Veer et al. [1989]; Shukla et al. [1989]).UV-induced photoproducts may be involved in the activation of these oncogenes.Interestingly, the sites where the point mutations were located in the ras genes were at potential pyrimidine dimer sites.
UV irradiation at low doses produces a prolif erative response in human melanocytes.Three studies have reported the effect of UV-B on mel anogenesis and proliferation (Friedmann and Gilchrest [1987]; Libow et al. [1988]; Schothorst et al. [1991]) .We also have investigated the role of UV-B on cell growth and gene expression levels in human melanocytes and have observed an increase in DNA synthesis following UV irra diation with low to moderate doses ofUV-B.We also observed an increase in the expression of the c-jun RNA transcripts that returned to baseline after 2 h.A slight delay in the induction of c-fos RNA transcripts was observed that peaked 2 h after UV irradiation.We are also investigating the ability of chemopreventive agents to act as modu lators of UV-induced alterations.

VI. A SYNTHETIC MODEL OF HUMAN MELANOCYTE TRANSFORMATION
We have summarized recent studies on the molecular events that accompany human melano cyte transformation.The use of molecular, cyto genetic, immunological, and biochemical analyses has led to the detection of genetic alterations that may playa role in the transformation of human melanocytes.A probable etiological agent in volved in human melanocyte transformation is UV radiation.Based on this information, we pro pose the following model as one heuristic vehicle by which to investigate melanocyte transforma tion.We recognize that several aspects of this model are speculative, but the approaches allow experimental solutions.
The first question we address is, Is there a specific group of alterations that are required to obtain the transformed melanocyte?Cytogenetists and molecular biologists have observed specific alterations that occur during human melanocyte transformation.The current model involves alter ations on chromosomes 9p, lOq, 6q, 11, Ip, 2, 3, and 7 as well as altered expression levels of the MGSA, EGF-R, Ha-ras, N-ras, bFGF, c-kit, TGF a, TGF-~2, PDGF-A, and c-fos (Albino and Foun tain [1993]).We have added our data involving c-jun, PKC-~ll' p5SC 1k -1 , and RAR-~ to this model and expanded the involvement of certain genes (Figure 1).
One of the first required alterations may be a deletion of a gene(s) to reduce or inhibit the abil ity of the primary melanocyte to undergo terminal differentiation.As human melanocytes undergo changes to become dysplastic nevi, a loss in the ability of these cells to undergo normal pro grammed cell differentiation has been proposed (Clark [1991]).Cytogenetic and molecular re search have pointed to a region of chromosome 9p as one that may contain a tumor suppressor gene involved in the earliest stage of human mel anocyte transformation (Albino and Fountain [1993]) and p16 has been proposed to be the candidate gene (Kamb et al. [1994a]; and Nobori et al. [1994]).p16(MTS/CDKN2) binds to cyclin dependent kinase 4 (cdk4) and inhibits the ability of cdk4 to interact with cyclin D, which has been shown to be involved in the cell cycle and to stimulate the passage of cells through the cell cycle (Musgrove et al. [1994]; Serrano et al. [1993]).The gene encoding p16 is deleted in many tumor cell lines (brain, breast, renal, bone, and melanoma) and contain point mutations in a few melanoma-prone families (Kamb et al. [1994a(Kamb et al. [ , 1994b]]).As the p16 protein functions as a nega tive regulator of cell cycle progression, its gene has the potential to be the melanoma susceptibil ity locus.However, another group has determined that the p16 gene was neither deleted nor mutated in a high percentage of their melanoma-prone families (Hussussian et al. [1994]).Further stud ies will be required to address whether p16 is the melanoma susceptibility locus on chromosome 9 involved in a significant subset of melanoma or involved in the growth of human tumor cell lines in vitro.
A second required alteration may induce the activity of protein kinases resulting in the abnor mal phosphorylation of various proteins (e.g., tran scription factors, membrane receptors, etc.).Loganzo et al. (1993) found an elevated expres yes sion levels of the p62 c -protein and tyrosine kinase activity in melanoma cells (18/20 mela noma cell lines).Easty et al. (1993) also observed that melanoma cells preferentially expressed two families of tyrosine kinases: Eph (Tyro-6 and ECK) and FGF-R (FGF-R4 and Tyro-9).A de crease or loss in the expression of c-kit and c-met has been observed in melanoma cells also (Natali et al. [1992]; Lassam and Bickford [1992]; Halaban et al. [1992]).c-kit has been thought to play an important role in melanocyte differentia tion and growth.Although lost late in melanocyte transformation, loss in the response to the c-kit ligand may allow a melanoma cell to undergo a proliferative response rather than a differentiative response.The expression of bFGF and EGF-R protein levels, which are frequently abnormally expressed, would activate bFGF-R and EGF-R (including other protein kinases), resulting in a proliferate response.
Alterations in protein kinase activities may lead to an increase in cell proliferation and an inability to undergo differentiation.This may re sult via the activation of growth signals from exogenous and endogenous factors.The protein kinases probably activate transcription factors (e.g., AP-l and SP-l), which induce the expres sion of various protooncogenes (e.g., c-fos and c myc), growth factors (e.g., bFGF and cytokines), and immunosuppressive factors (e.g., TGF-a).

Melanocyte
A third required alteration may be the loss in the ability of human melanocytes to protect itself from oxygen-free radicals.Melanocytes are highly susceptible to damage induced by hydrogen perox ide and endogenous reactive oxygen species com pared with human fibroblasts and keratinocytes (Yohn et al. [1991]).Melanocytes have decreased peroxidase, catalase, glutathione peroxidase, and superoxide dismutase enzyme activities compared with keratinocytes and fibroblasts.One of the re gions involved in melanocyte transformation is chromosome 6q, a region that has been shown to include the gene manganese superoxide dismutase (MnSOD) (Church et al. [1993]).Loss ofthis gene may result in nonspecific damage to the cell mem brane and genomic DNA, which could result in alterations leading to a transformed phenotype.
We propose that one pathway to melanoma is vai multiple exposures to UV radiation.We pro vide the following model in Figure 3 as a heuristic vehicle from which to explore this hypothesis.Figure 3 displays some of the changes that UV radiation may induce in irradiated melanocytes and the surrounding cells.UV irradiation can af fect the target cell, the primary melanocyte, as well as the surrounding cells.Following UV irra diation, the target melanocyte may be altered by UV-induced DNA damage and UV-induced gene expression, resulting in various phenotypic ef fects, including production of growth factors.UV irradiation of the surrounding cells will induce host cell death, resulting in the release of growth factors and a localized area of concentrated growth factors.In some cases, due to the protective effect of melanin as well as being quiescent, the mel anocyte will be induced to proliferate rather than cell death being the result.These UV-induced changes may parallel the proliferative response observed in melanocytes in vitro (Friedmann and Gilchrest [1987]; Libow et al. [1988]).This pro liferative response would "fix" the damaged DNA and allow pyrimidine dimers and single strand breaks to result in point mutations, deletions, and rearrangements.With multiple UV irradiations (e.g., intermittent severe sunburns), several muta tions may occur in the target melanocyte and allow receptivity to endogenous and exogenous signals that lead to abnormal cell growth.Also, UV irradiation will induce a regional immunosuppressive response resulting in loss in ability of the host to detect the mutated melano cyte.As the altered melanocyte is exposed to multiple UV irradiations, the cell will become progressively more abnormal and eventually de velop to a fully transformed cell that will have lost the ability to response to growth control sig nals.It will also become less immunologically detectable with the self-production of cytokins (to decrease local immunosurveillance from the im mune system) and a decrease in the expression levels of histocompatibility antigens.
This model is based on several changes occur ring in the same cell involving gene mutations, the abnormal production of growth factors in situ by damaged host cells and a decrease in immuno surveillance.Together these effects allow the growth of an altered subpopulation of melanocytes.With multiple UV exposures, this subpopulation of cells may undergo unscheduled DNA replication result ing in the formation of a melanocyte with cumula tive molecular abnormalities.Several alterations would have had to occur in the same cell; however, a parallel model may already exist to study this phenomenon.Albino et al. (1993) have observed multiple changes in melanocytes infected with an activated ras and cultured for several passages.Growth of these transfected melanocytes led to the formation of a select cell population that had lost specific chromosomes (after 6 months) and be came fully transformed.

VII. SUMMARY
This review has provided information on al terations in gene expression and the signal trans duction pathways in human melanocytes and melanoma cells.The development of an in vitro cell culture system was the first step in unraveling the genetic alterations that occur as a human melanocyte transforms and progresses to malig nant melanoma.Using biochemical, cytogenetic, and molecular techniques to develop an under standing of the changes that occur, we may yet explain the relevant etiological genetic events in human melanocyte transformation.
Signal transduction pathways do not function as independent units but rather form an interac tive network of pathways.This interaction may be synergistic as observed with activators of PKC and PKA in melanocyte proliferation or antago nistic as observed with activators of RAR and other members of the steroid family.It is this combination of proliferation and differentiation signals (Figure 2) that regulates human melano cyte cell growth.Ongoing studies will increase our knowledge of this balance and allow us to develop a further understanding of the key link ages in the different pathways involved as well as the control oftheir interactive nature.Understand ing human melanocyte transformation will allow the accumulation and assimilation of information that will assist clinicians with new diagnostic tools and development of innovative therapies.

FIGURE 1 .
FIGURE 1. Genetic model of human melanocyte transformation.Proposed critical genetic alterations are listed in the upper part of the model.Those genetic alterations listed below the arrow are considered secondary changes.Features in ( ) are supported by relatively less data.

FIGURE 2 .
FIGURE 2. Schematic illustration of a molecular model of proliferation and differentiation path ways in human primary melanocytes. (+) phosphorylated; (-) dephosphorylated.

FIGURE 3 .
FIGURE 3. A synthetic model of human melanocyte transformation.

TABLE 1 Retinoid
Response and RNA Transcript Expression Levels of the RAR Isotypes in Human Melanocytes and Melanoma Cells Note: 13-cisRA and tRA IC50 is in micromolar concentration.RAR ~ induction was determined in cells treated with tRA.RNA transcript expression levels of RAR a, ~, and yare indicated as: present (+), or not detected (-).

Levels of the PKC Isotypes in Human Primary Melanocytes, Tissue Biopsies, and Metastatic Melanoma Cell Lines
Note: N.T. -not tested.