Frequent alterations of Ras signaling pathway genes in sporadic malignant melanomas



Ras signaling is important for the intracellular transduction of mitogenic stimuli from activated growth factor receptors. We have investigated 37 sporadic malignant melanomas (15 primary cutaneous melanomas and 22 melanoma metastases) and 6 melanoma cell lines for mutations in the 3 Ras genes NRAS, KRAS and HRAS. All tumors and cell lines were additionally analyzed for mutation and expression of BRAF, which encodes a Ras-regulated serine/threonine kinase with oncogenic properties, as well as for expression of RASSF1A, which encodes a Ras-binding protein with tumor suppressor properties. Mutational analyses identified somatic NRAS mutations in 2 primary melanomas, 4 melanoma metastases and 2 cell lines. One melanoma metastasis showed a somatic KRAS mutation whereas HRAS mutations were not detected. Eight primary melanomas, 6 melanoma metastases and 4 melanoma cell lines carried BRAF mutations affecting the known hot-spot codon 599. None of the tumors or cell lines with BRAF mutation demonstrated NRAS or KRAS mutations. Real-time reverse transcription-PCR showed that 8 melanomas (3 primary tumors, 5 melanoma metastases) had reduced RASSF1A transcript levels of ≤50% relative to benign melanocytic nevi and normal skin. Three melanoma cell lines lacked detectable RASSF1A transcripts. The RASSF1A gene promoter was hypermethylated in these 3 cell lines as well as in 6 of 8 melanomas with reduced RASSF1A mRNA levels. Treatment of the cell lines with 5-aza-2′-deoxycytidine and trichostatin A resulted in demethylation of the RASSF1A promoter and re-expression of RASSF1A transcripts. Most tumors and all cell lines with RASSF1A promoter methylation additionally carried BRAF or NRAS mutations, suggesting a synergistic effect of these aberrations on melanoma growth. Taken together, 57% of the investigated melanomas and 100% of the melanoma cell lines carried mutations in either NRAS, KRAS or BRAF. In addition, 22% of the melanomas and 50% of the cell lines showed reduced RASSF1A transcript levels. Thus, alterations of Ras pathway genes are of paramount importance in the pathogenesis of sporadic melanomas. © 2004 Wiley-Liss, Inc.

Malignant melanoma is a highly aggressive form of skin cancer that may progress to a fatal metastatic disease. Unfortunately, both incidence and mortality of melanomas have markedly increased over the past decades.1, 2 Because metastatic melanomas are commonly resistant to available treatment regimens, long-term survival has not significantly improved since the 1970s.3 To develop novel therapeutic strategies, it is important to elucidate the yet poorly characterized molecular mechanisms that lead to melanoma initiation and progression. Molecular genetic studies have identified several genes that are aberrant in variable fractions of sporadic or familial melanomas. These include the tumor suppressor genes CDKN2A and PTEN, as well as the proto-oncogenes CDK4, CTNNB1, NRAS and MYCC.4, 5, 6 Genetic alterations in these genes result in aberrations of different cellular pathways that are crucially involved in the regulation of signal transduction, cell cycle progression and apoptosis.6

We have focussed on the molecular analysis of genetic and epigenetic changes in a set of genes that are important for the intracellular transduction of mitogenic signals from the cell membrane to the cell nucleus, namely the Ras genes NRAS, KRAS and HRAS, as well as the Ras-related genes BRAF and RASSF1A. Previous studies have demonstrated somatic NRAS mutations in between 10–37% of sporadic melanomas and up to 95% of hereditary melanomas from patients carrying germline CDKN2A mutations.7, 8, 9, 10 In addition, BRAF mutations have been detected as a common somatic aberration in both melanomas and melanocytic nevi.11, 12, 13 The Ras proteins are highly homologous small G-proteins with GTPase activity that mediate the cellular response to growth stimuli by signaling via different effector cascades.14, 15, 16 A major mechanism of Ras-induced oncogenic transformation is related to an enhanced mitogen-activated kinase (MAPK) pathway signaling caused by Ras-dependent activation of Raf serine/threonine-specific kinases. Effects on other pathways, however, such as the phosphatidylinositol 3-kinase (Pi3-kinase) and the Ral guanine nucleotide exchange factors (Ral-GEF) signaling cascades, are also important for Ras-induced tumorigenesis.14, 15, 16

The Ras association domain family protein 1-gene (RASSF1), which maps to the short arm of chromosome 3 (3p21.3), has been identified as a novel Ras-binding protein with tumor suppressor properties.17, 18 Several isoforms are generated from RASSF1 due to alternative splicing and differential promoter usage. The longest isoform, RASSF1A, is able to reduce tumorigenity both in vivo and in vitro.17, 19 In addition, RASSF1A has been shown to inhibit cell cycle progression by inhibiting cyclin D1 accumulation.20 Its role as a tumor suppressor is supported strongly by numerous studies showing that RASSF1A either is homozygously deleted or, more commonly, transcriptionally downregulated by promoter methylation in different types of human cancer.21

To better define the role of genetic aberrations in Ras signaling pathway genes in sporadic melanomas, we carried out a systematic mutation analysis of NRAS, KRAS, HRAS and BRAF in a series of 37 melanoma tumors and 6 melanoma cell lines. The tumors and cell lines were additionally investigated for the expression of RASSF1A at the mRNA level, and cases with markedly decreased expression levels were subsequently subjected to methylation analyses of the RASSF1A promoter. Furthermore, we compared the aberrations detected in these 5 genes with previous findings on PTEN mutations in the same tumor series.22


Patient samples and cell lines

We investigated 37 sporadic malignant melanomas from 12 male and 25 female adult patients (mean age = 67 years, range = 29–88 years). The tumor series included 15 primary cutaneous melanomas (8 nodular melanomas, 4 acral lentiginous melanomas, 2 superficially spreading melanomas and 1 polypoid melanoma) and 22 melanoma metastasis (15 cutaneous melanoma metastases, 1 regional lymph node metastasis, 1 spinal metastasis and 5 intracerebral metastases). The bias in our tumor series toward advanced melanomas is related to the fact that it is often impossible to freeze sufficient amounts of tumor tissue for research purposes from early melanoma stages, such as melanoma in situ or SSM. For diagnostic reasons, these early melanomas usually have to be entirely subjected to histological analysis. All tumors of the present series had been studied before for mutation in the PTEN (phosphatase and tensin homolog deleted on chromosome 10) tumor suppressor gene.22 The Clark-level and tumor thickness of each primary melanoma are listed in Table I. Parts of each tumor were frozen immediately after operation and stored at −80°C. The tumor cell content of each specimen subjected to molecular genetic analysis was histologically determined as reported previously.22, 23 Peripheral blood samples for the extraction of constitutive DNA were available from 34 patients. In addition to the melanoma tumors, we investigated 6 established melanoma cell lines (MEL-JUSO, SK-MEL-30, COLO-849, SK-MEL-3, IGR-37 and SK-MEL-1). All cell lines were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany) and grown in RPMI1640 medium (Biochrome, Berlin, Germany), supplemented with 10% FCS (PAA Laboratories, Cölbe, Germany) and 1% penicillin/streptomycin (PAA Laboratories).

Table I. Molecular Genetic Findings in the Investigated Melanomas and Melanoma Cell Lines
Tumor number/ cell lineTumor type1Clark level/ thickness (mm)AgeGender2LocalizationNRAS mutation3KRAS mutation3HRAS mutation3BRAF mutation3BRAF expression3PTEN mutation3RASSF1A expression4RASSF1A methylation5
  • 1

    ALM, acral lentiginous melanoma; CMM, cutaneous melanoma metastasis; l.n. MM; melanoma metastasis in a regional lymph node; i.c. MM; intracerebral melanoma metastasis; NM, nodular melanoma; PM, polypoid melanoma; spinal MM; spinal melanoma metastasis; SSM, superficial spreading melanoma.

  • 2

    F, female; M, male.

  • 3

    no mutation detected.

  • 4

    The BRAF and RASSFIA expression values refer to the densitometrically determined transcript level in each tumor relative to the mean expression level determined for 2 normal skin samples and 4 benign melanocytic nevi. NA, not analyzed.

  • 5

    no RASSFIA promoter methylation detected; +, RASSFIA promoter methylated; NA, not analyzed.

M5ALMIV/4.068FPlantarc.181C>A: Q61K0.51.9NA
M13ALMIII/1.467FPlantarc.1796T>A: V599ENANANA
M15ALMIV/5.580FPlantarc.1796-1798del TGA: V599E, K600del1.0c.164ins45 (dupl. exon 3)0.3+
M12NMIV/3.974FShoulderc.1796T>A: V599E1.00.9
M24NMIV/4.957FFootc.1796T>A: V599E0.6c.1002–1003CC>TT: R335X0.4
M29NMIV/2.929FAbdomenc.1796T>A: V599E0.71.1
M39NMIV/6.068MAbdomenc.1796T>A: V599E0.60.2+
M47NMIV/2.272FNeckc.1796T>A: V599E0.70.6NA
M54NMIV/1.577MShoulderc.1796T>A: V599E0.9c.730C>T: P244S1.8NA
M55PMV/3564FAbdomenc.181C>A: Q61K1.11.0NA
M4CMM72FLegc.145G>A: E49K0.51.1NA
M14CMM66MArmc.34G>T: G12C0.41.4NA
M18CMM59FAbdomenc.236ins27(209–235): p.78ins70–780.40.2
M23CMM86FLabia majora1.00.8NA
M38CMM83FLegc.1796T>A: V599E1.31.9NA
M46CMM63FAbdomenc.1796T>A: V599E0.50.8
M11l.n. MM43MAxilla0.81.5NA
M32spinal MM76FSpinalc.1796T>A: V599E0.51.3NA
M33i.c. MM69MCerebralc.1795-1996GT>AA: V599K0.50.3+
M34i.c. MM53FCerebralc.1796T>A: V599E0.80.9NA
M35i.c. MM76FCerebralc.1796T>A: V599E0.60.7NA
M36i.c. MM60MCerebralc.181C>A: Q61K0.5c.843-850delAGG- ACCAG (frameshift)0.2+
M44i.c. MM72MCerebralc.35G>A: G12D0.70.7NA
MEL-JUSO     c.182A>T: Q61L0.40.0+
SK-MEL 30     c.180–181AC>TA: Q61K0.31.0
COLO-849     c.1796T>A: V599E0.4c.493-634del141 (exon 6)1.0
SK-MEL 3     c.1796T>A: V599E0.31.0
IGR-37     c.1796T>A: V599E0.20.0+
SK-MEL 1     c.1796T>A: V599E1.10.0+

DNA and RNA extraction

Extraction of high molecular weight DNA and RNA from frozen tumor tissue samples and cell lines was carried out by ultracentrifugation over cesium chloride as described in detail elsewhere.24, 25 DNA extraction from peripheral blood leukocytes was carried out according to a standard protocol.26

Mutation analysis

Mutation analyses of NRAS, HRAS and KRAS were carried out at the transcript level. From each tumor and cell line, 3 μg of total RNA were reverse-transcribed into cDNA in a volume of 50 μl using random hexanucleotide primers and Superscript reverse transcriptase (Invitrogen, Karlsruhe, Germany). PCR fragments covering the known mutation hot-spot sites in NRAS, HRAS and KRAS were then amplified by PCR from cDNA. For mutation analysis of BRAF, we amplified exons 11 and 15, which carry the hot-spot mutation sites, by PCR from genomic DNA using intronic primer pairs flanking each exon. The respective primer sequences are listed in Table II. The 6 melanoma cell lines were analyzed for PTEN mutations as reported in detail elsewhere.22 Each PCR product was screened for mutations by single-strand conformation polymorphism (SSCP)/heteroduplex analysis as described before.27, 28 In brief, the PCR products were heat-denatured and then subjected to electrophoresis on non-denaturing polyacrylamide gels. Each fragment was evaluated under at least two different conditions with variations in temperature and polyacrylamide concentration. The SSCP/heteroduplex band patterns were visualized by silver staining of the gels. PCR products showing aberrant band patterns were sequenced in both directions using cycle-sequencing (BigDye cycle sequencing kit, Applied Biosystems, Foster City, CA) and an ABI PRISM 377 semi-automated DNA sequencer (Applied Biosystems). Somatic origin of a detected mutation was confirmed by SSCP/heteroduplex analysis or sequencing of constitutive (leukocyte) DNA of the respective patient.

Table II. Summary of Oligonucleotide Primers Used for SSCP/Heteroduplex Analyses (Application 1), Screening for Gene Amplification (Application 2), Expression Analyses by Real-Time Reverse Transcription PCR (Application 3), and Analysis of Promoter Methylation by Sequencing of Sodium Bisulfite Treated DNA (Application 4)
GenePrimer namePrimer sequenceFragment lengthApplication
BRAFBRAF Exon 11 F5′-cctctcaggcataaggtaatg-3′244 bp1
 BRAF Exon 11 R5′-gacttgtcacaatgtcaccac-3′  
 BRAF Exon 15 F5′-tcataatgcttgctctgatagg-3′244 bp1, 2
 BRAF Exon 15 R5′-gtaactcagcagcatctcag-3′  
 BRAF-TAQ-F15′-ctcttgtggcggaggtag-3′120 bp3
RASSFIARASSFIA-TAQ-F5′-cctctgtggcgacttcatctg-3′108 bp3
 MU 3795′-gttttggtagtttaatgagtttaggtttttt-3′381 bp4
 ML 7305′-accctcttcctctaacacaataaaactaac-3′  
 MU 3795′-gttttggtagtttaatgagtttaggtttttt-3′205 bp4
 ML 5615′-ccccacaatccctacacccaaat-3′  
NRASNRAS-F15′-agcttgaggttcttgctggt-3′230 bp1
 NRAS-F25′-ggacatactggatacagctg-3′253 bp1
 NRAS-F35′-gccaacaaggacagttgatac-3′279 bp1
 NRAS-F45′-gtactgtagatgtggctcgc-3′183 bp1
KRASKRAS-F15′-cgggagagagcctgctg-3′229 bp1
 KRAS-F25′-tcgacacagcaggtcaagag-3′238 bp1
 KRAS-F35′-ccttctagaacagtagacac-3′250 bp1
HRASHRAS-F15′-gagaccctgtaggaggacc-3′233 bp1
ARF1ARF1-F15′-gaccacgatcctctacaagc-3′111 bp3
APRTAPRT-1F5′-cagggaacacattcctttgc-3′136 bp2

Determination of BRAF gene dosage

All tumors were analyzed for BRAF amplification using duplex-PCR with primers for exon 15 of BRAF together with primers for the APRT gene on 16q24.3 (Table II). PCR was carried out with 10 ng of genomic DNA as template and HotStar Taq DNA-polymerase (Qiagen, Hilden, Germany). An initial denaturation of 15 min at 95°C was followed by 30 cycles of 30 sec at 95°C, 30 sec at 58°C and 30 sec at 72°C. The final extension reaction was carried out for 5 min at 72°C. PCR-products were separated on 2% agarose gels and ethidium bromide-stained bands were recorded with the Gel-Doc 1000 system (Bio-Rad, Hercules, CA). Quantitative analysis of the signal intensities obtained for the target gene and the reference gene relative to constitutional (leukocyte) DNA was carried out with the Molecular Analyst software (version 2.1, Bio-Rad). Increases in the target/reference gene ratio of ≥3-fold relative to the target/reference ratio obtained for constitutional DNA were considered as gene amplification. Control experiments with different relative concentrations of PCR-amplified BRAF and APRT fragments mixed together as templates (1:50, 1:30, 1:10, 1:5, 1:4, 1:3, 1:2, 1:1, 2:1, 3:1, 4:1, 5:1, 10:1, 30:1, 50:1) confirmed that the duplex-PCR assay was able to detect these BRAF copy number changes relative to APRT, with the selected BRAF/APRT threshold ratio of 3 corresponding to an approximately 5-fold higher BRAF copy number.

Expression analyses

The expression of BRAF and RASSF1A transcripts was determined by real-time reverse transcription-PCR using the ABI PRISM 5700 sequence detection system (Applied Biosystems), which allows continuous measurement of the PCR product amount by means of SybrGreen fluorescent dye. BRAF and RASSF1A mRNA levels were normalized to the transcript level of the ADP-ribosylation factor gene ARF1 (Table II). As reference tissues for the melanomas, we used normal skin samples from 2 different individuals, as well as benign melanocytic nevi from 4 different patients. Cultured normal melanocytes (PromoCell, Heidelberg, Germany) were used as reference cells for the melanoma cell lines. In addition to real-time reverse transcription-PCR, we carried out simple reverse transcription-PCR as shown in Figure 2b to demonstrate the induction of RASSF1A mRNA after treatment of melanoma cell lines with 5-aza-2′-deoxycytidine and trichostatin A.

Methylation analysis of the RASSF1A promoter

The 8 melanomas and the 3 melanoma cell lines with normalized RASSF1A mRNA levels of ≤0.5 were subjected to RASSF1A promoter methylation analysis using sequencing of sodium bisulfite modified DNA. In brief, DNA from these cases was treated with sodium bisulfite as described elsewhere.29 Then, semi-nested PCR was carried out using primers (Table II) for bisulfite-modified DNA of the RASSF1A promoter region.17 The PCR products were purified with the Jetquick gel extraction spin kit (Genomed, Bad Oeynhausen, Germany) and cloned into pCR 2.1 vectors (Invitrogen, Carlsbad, CA), followed by cycle-sequencing of at least 5 clones from each tumor and cell line. As negative controls, we sequenced bisulfite-modified DNA from 4 benign nevi, as well as 3 selected melanomas (M29, M37, M46) and the 3 melanoma cell lines with normal RASSF1A transcript levels. As a positive control, we methylated DNA from 4 benign nevi in vitro using SssI (CpG) methylase (New England Biolabs, Beverly, MA) followed by sodium bisulfite treatment of native and treated DNA samples, PCR amplification of the RASSF1A promoter, cloning and DNA sequencing. In all 4 benign nevi, native DNA was unmethylated whereas SssI treatment resulted in CpG methylation of the RASSF1A promoter (data not shown).

Treatment of melanoma cell lines with 5-aza-2′-deoxycytidine and trichostatin A

Melanoma cells (MEL-JUSO, SK-MEL1, IGR-37) were seeded with a density of 1–1.6 × 106 cells/75 cm2 and grown for 24 hr in RPMI1640 medium containing 10% FCS and 1% penicillin/streptomycin. After 24 hr, the cultures were treated with 1 μM 5-aza-2′-deoxycytidine (Sigma-Aldrich, Munich, Germany) in the same medium for 72 hr, with daily changes of the medium. For the last 24 hr, 300 nM trichostatin A (Calbiochem, San Diego, CA) was added. Cells were then lysed in 4 M guanidinium isothiocyanate and subjected to DNA and RNA extraction by ultracentrifugation over cesium chloride.24, 25


SSCP/heteroduplex-analysis followed by DNA sequencing identified NRAS mutations in 6 of 37 melanomas (2 primary melanomas and 4 melanoma metastases). Two primary melanomas (M5, M55) and one intracerebral melanoma metastasis (M36) carried identical missense mutations affecting the known hot-spot codon 61 (c.181C>A: Q61K) (Table I). Tumor M14, a cutaneous melanoma metastasis, showed a missense mutation at codon 12 (c.34G>T: G12C), which is a further known mutation hot-spot (Fig. 1a,b). The remaining 2 tumors (M4, M18), both cutaneous melanoma metastases, exhibited NRAS mutations outside the hot-spot regions (Table I). NRAS mutations affecting codon 61 were detected in 2 of 6 investigated melanoma cell lines, namely MEL-JUSO (c.182A>T: Q61L) and SK-MEL-30 (c.180-181AC>TA: Q61K) (Table I). One intracerebral melanoma metastasis (M44) carried a KRAS mutation causing a missense mutation at codon 12 (c.35G>A: G12D) (Table I, Fig. 1c, shown is the sequence of the non-coding strand). None of the investigated tumors or cell lines showed HRAS mutations.

Figure 1.

Demonstration of NRAS, KRAS, and BRAF mutations in malignant melanomas. (a,b) The cutaneous melanoma metastasis M14 showed a point mutation at nucleotide 34 of NRAS (a, arrow) that results in the substitution of the amino acid glycine by cysteine at codon 12 (c.34G>T: G12C). Sequencing of this patient's constitutional (blood) DNA showed the wild-type sequence (b). M14T, M14 tumor DNA; M14B, M14 blood DNA. (c) A heterozygous missense mutation affecting codon 12 of KRAS (c.35G>A: G12D) was detected in the intracerebral melanoma metastasis M44 (shown is the sequence of the non-coding strand, the arrow points to the mutation site). (d) SSCP/heteroduplex analysis of 4 malignant melanomas for mutations in BRAF. Note identical aberrations of the SSCP band patterns in the tumor DNA samples (T) of M12, M24, and M29 as compared to the corresponding blood DNA samples (B) (arrows). Tumor M15 showed a different SSCP band pattern as well as 2 aberrant heteroduplex bands (arrowheads). (e) DNA sequencing of the tumor DNA of M12 confirmed a heterozygous point mutation at nucleotide 1796 (arrow) resulting in the activating V599E missense mutation of BRAF. (f) Tumor M15 carried a deletion of 3 nucleotides (c.1796-1798delTGA), which at the protein level results in the V599E missense mutation accompanied a loss of codon 600. The arrow points to the first deleted nucleotide in the sequence.

Mutational analysis of the BRAF gene showed somatic mutations in 14 of 37 melanomas (38%) and 4 of 6 melanoma cell lines (COLO-849, SK-MEL-3, IGR-37, SK-MEL-1) (Table I, Fig. 1d–f). All but one of the mutations resulted in the V599E missense mutation. This particular mutation has been shown to cause an elevated kinase activity of the protein.11BRAF mutations were exclusively found in tumors and cell lines without a demonstrated NRAS or KRAS mutation (Table I). Analysis of BRAF mRNA expression using real-time reverse transcription-PCR did not show any overexpression, except for one cutaneous melanoma metastasis (M45) that showed a transcript level of 2.5 relative to normal skin and benign nevi (Table I). In line with these expression data, determination of BRAF gene dosage by duplex-PCR did not show any evidence for BRAF gene amplification in our tumors and cell lines.

Analysis of RASSF1A mRNA expression by real-time reverse transcription-PCR demonstrated reduced transcript levels of ≤0.5 relative to the reference tissues (normal skin and benign nevi) in 8 of 37 (22%) melanomas (3 primary tumors, 5 metastases) (Table I, Fig. 2a). Sequencing of bisulfite treated DNA showed hypermethylation of the CpG island within the RASSF1A promoter in 6 of 8 melanomas with reduced RASSF1A mRNA levels (Table I, Fig. 2c). No RASSF1A promoter methylation was detected in 7 control tumors including 4 benign nevi and 3 melanomas (M29, M37, M46) without reduced RASSF1A mRNA levels. Three of the 6 investigated melanoma cell lines lacked detectable RASSF1A transcripts (MEL-JUSO, IGR-37, SK-MEL-1), whereas RASSF1A mRNA was expressed at levels similar to cultured normal melanocytes in the other 3 melanoma cell lines (SK-MEL-3, SK-MEL-30, COLO-849) (Table I). Sequencing of the RASSF1A promoter after bisulfite treatment of the DNA was carried out for all 6 cell lines. The 3 cell lines with RASSF1A expression demonstrated no RASSF1A promoter methylation. In contrast, we found complete methylation of all investigated CpG sites within the RASSF1A promoter in the MEL-JUSO, IGR-37, and SK-MEL-1 lines (Fig. 2d, Table I). Treatment of the latter cell lines with 5-aza-2′-deoxycytidine and trichostatin A resulted in the demethylation of the RASSF1A promoter (Fig. 2e) and a re-expression of RASSF1A transcripts (Fig. 2b).

Figure 2.

Analysis of melanomas for RASSF1A mRNA expression and promoter methylation. (a) Demonstration of markedly reduced RASSF1A mRNA expression in melanoma M39 relative to the benign melanocytic nevus N11 using real-time reverse transcription-PCR. Abscissa, cycle number; Ordinate, amount of PCR product. Although the curves for the reference mRNA (ARF1, right panel) pass the threshold (Ct) value at an approximately equal cycle number in both tumors, the RASSF1A mRNA curve (left panel) obtained for M39 is shifted to the right relative to the RASSF1A mRNA curve obtained for N11. The calculated RASSF1A mRNA expression level in M39 was 0.2 relative to N11 (whose expression value was representative for the mean value determined for normal skin and 4 melanocytic nevi). (b) Induction of RASSF1A mRNA expression in 3 melanoma cell lines by treatment with 5-aza-2′-deoxycytidine and trichostatin A. The individual lanes correspond to: 1, MEL-JUSO; 2, IGR-37; 3, SK-MEL-1; C, water control. All 3 cell lines lacked detectable RASSF1A transcripts before treatment (−) and re-expressed RASSF1A transcripts after 72 hr of treatment (+). (c) Sequencing of the RASSF1A promoter after bisulfite modification of DNA shows methylation of all CpG sites (arrows) in melanoma M39. (d) Similarly, the cell line SK-MEL-1 had completely methylated all investigated CpG sites in the RASSF1A promoter (arrows). (e) Treatment of SK-MEL-1 cells with 5-aza-2′-deoxycytidine and trichostatin A for 72 hr resulted in demethylation of the RASSF1A promoter. The sequences shown in (c–e) are derived from sequencing of the non-coding strand. Depicted is the sequence from nucleotide −30 to −89 of the RASSF1A promoter (GenBank accession number NM_007182).

Six of 8 melanomas and all melanoma cell lines with decreased RASSF1A expression simultaneously carried either BRAF or NRAS mutations (Table I).PTEN gene mutations were present in melanomas from 4 patients and the COLO-849 cell line (Table I). One intracerebral melanoma metastasis (M36) carried mutations in PTEN and NRAS, whereas melanomas from 3 other patients (M15, M24, M54), as well as the COLO-849 cell line showed mutations in both PTEN and BRAF (Table I). Reduced RASSF1A transcript levels were detected in 2 of 4 PTEN mutant melanomas (M15, M24), including one (M15) with RASSF1A promoter methylation (Table I).


Oncogenic mutations in RAS genes have been found in various human malignancies, with carcinomas of the pancreas, colon, thyroid and lung showing the highest incidence.30 Mutations in RAS genes, in particular in NRAS, have also been identified in 10–37% of sporadic melanomas.7, 8, 9, 31, 32, 33 A recent study reported that NRAS mutations are significantly more frequent in familial melanomas than in sporadic melanomas, i.e., NRAS mutations were detected in 95% of melanomas from patients with a CDKN2A germ line mutation but only 10% of sporadic melanomas.10 In contrast to NRAS, mutations in HRAS or KRAS are rare in melanomas,8, 31, 32 although one study suggested KRAS mutation as an early and common event in melanoma development.34 Although some authors claimed an association of NRAS mutations with the progression of melanomas to metastatic disease,7, 35 others considered this aberration as an early event in melanoma development because oncogenic mutations were detected already in congenital melanocytic nevi, melanomas in situ, and the early horizontal growth phase of primary melanomas.36, 37 Comparisons of early lesions with advanced lesions, as well as primary with metastatic melanomas from the same patient showed identical NRAS mutations in the majority of cases.8, 9 In our tumor series, melanomas from 6 of 37 patients (16%), including 2 primary melanomas and 4 melanoma metastases, harbored somatic NRAS mutations. An additional intracerebral melanoma metastasis carried a KRAS codon 12 mutation. Both NRAS mutant primary melanomas (M5 and M55) progressed to metastatic disease later on. None of our melanomas showed the codon 18 mutation in NRAS that has been associated with good prognosis.38 In 4 cases (2 primary melanomas and 2 melanoma metastases) the mutational hot-spot codons 12 and 61 of NRAS were affected by missense mutations known to result in constitutive activation of the protein.14 Several studies have reported that NRAS codon 61 mutations occur predominantly in melanomas located at sun-exposed sites, thus arguing for a role of ultraviolet (UV) exposure.10, 31, 32 This particular codon (CAA) contains a TT pyrimidine doublet in the non-coding strand, which is a prime site for UV-induced DNA damage.39 Most UV-induced mutations do not occur at TT sites but at cytosines in TC or CC sites, however, resulting in C to T or CC to TT transitions.39 In fact, the majority of NRAS mutations in melanomas, including those detected in our tumor series, are not UV-signature transitions.40 In addition, one of 2 NRAS mutant primary melanomas of our series (M5) was located at a sun-protected site (sole of the foot), whereas the other one (M55) was resected from the abdomen, which may be considered as an intermittently sun-exposed site. Taken together, the relationship between UV exposure and NRAS mutation in melanomas may be more complicated as previously suggested, and possibly involves additional molecular aberrations, such as impaired repair of UV-damaged DNA.40

The significance of aberrant Ras-signaling in melanoma pathogenesis has been substantiated by the demonstration of frequent somatic mutations in the BRAF gene in melanomas and melanoma cell lines.11 In most cases, the hot-spot codon 599 within the kinase domain was affected by a missense mutation (V599E) leading to an elevated kinase activity of the protein and the acquisition of transforming properties.11 Mutational analysis of our tumor series identified somatic BRAF mutations in 14 of 37 melanomas (38%) and 4 of 6 melanoma cell lines (66%) investigated. All detected mutations affected the hot-spot codon 599. The incidence of BRAF mutations obtained in our melanoma series is in line with the recent study of Gorden et al.41 who found BRAF mutations in 31 of 77 of melanomas investigated (40%). Other authors detected BRAF mutations in even higher fractions of invasive and metastatic melanomas, ranging between 63–68% of the cases.11, 12, 13, 42 In addition, BRAF mutations were identified at high frequency (71–82%) in melanocytic nevi.13, 42 The latter finding suggests BRAF mutation as a critical step in the initiation of melanocytic tumorigenesis.13 Dong et al.42 found only 2 tumors with BRAF mutation among 20 early (radial growth phase) melanomas investigated. Therefore, these authors regarded BRAF mutation as a progression-associated event rather than an initiating event in the majority of melanomas. We identified BRAF mutations in 2 of 4 ALM, and 5 of 8 NM, but none of 2 SSM included in our primary melanoma series. The low number of primary melanomas in our series, however, does not allow any conclusion about a different frequency of BRAF mutations in SSM as compared to other melanoma subtypes.

We found BRAF mutations exclusively in melanomas and melanoma cell lines without a demonstrated mutation in NRAS or KRAS. A mutually exclusive occurrence of BRAF and RAS gene mutations has also been observed in previous studies on melanomas,11, 12, 41, 42 as well as various other cancers.12, 43, 44, 45 These findings clearly indicate that BRAF and RAS mutations may be regarded as alternative genetic changes that result in the activation of the same signaling pathway, namely the Raf/ERK/MAPK cascade.46 Interestingly, activation of this pathway has been found to be sufficient for the in vitro transformation of rodent cells but not of human cells.47 Thus, it seems likely that aberrant activation of additional Ras-dependent signaling cascades, such as the RalGEF and Pi3-kinase controlled pathways, is necessary for the development of melanomas. The tumor suppressor protein Pten functions as a lipid phosphatase that inhibits Pi3-kinase signaling through dephosphorylation of phosphatidylinositol-(3,4,5)-triphosphate. A previous study reported on the mutually exclusive presence of NRAS and PTEN mutations in melanoma cell lines.33 Therefore, we wondered how the previously reported PTEN mutations22 were related to the RAS and BRAF gene alterations in our melanoma series. Three of the 4 tumors and the COLO-849 cell line with demonstrated PTEN mutations additionally carried BRAF mutations, whereas one tumor had mutations in PTEN and NRAS.

In addition to the mutational analysis of RAS genes and BRAF, we investigated our tumor series for aberrations of RASSF1A mRNA expression. RASSF1A encodes a Ras-effector protein with putative tumor suppressor function, which has been shown to be inactivated by homozygous deletion or promoter hypermethylation in various cancers.17, 21 We found a reduced expression of RASSF1A transcripts in 22% of the melanomas and 50% of the melanoma cell lines investigated. Decreased RASSF1A transcript levels were detected with approximately equal frequency in primary (20%) and metastatic tumors (23%), indicating that this aberration is not restricted to metastatic disease. Hypermethylation of the RASSF1A promoter could be demonstrated in 75% of the tumors with normalized mRNA levels of ≤0.5 relative to normal skin and benign nevi, as well as all 3 cell lines without detectable RASSF1A mRNA. Furthermore, we showed that RASSF1A transcripts were re-expressed after treatment of these melanoma cell lines with 5-aza-2′-deoxycytidine and trichostatin A. In line with our results, a study published during the preparation of our article reported on RASSF1A hypermethylation detected by methylation-specific PCR (MSP) in 55% of the melanomas and 82% of the melanoma cell lines investigated.48 The higher frequency of RASSF1A hypermethylation reported by these authors may in part be due to the different experimental strategy. Spugnardi et al.48 carried out MSP for all tumors without analyzing RASSF1A expression, while we primarily screened for reduced RASSF1A mRNA expression, followed by sequencing of bisulfite treated DNA from the melanomas with mRNA levels of ≤0.5 relative to the normal skin and benign nevi. Thus, we may have missed some tumors with partial methylation of the RASSF1A promotor, which did not result in a reduction of the transcript levels below this arbitrary threshold. In addition, the DNA segment that we analyzed by sequencing did not entirely cover the 2 regions studied with MSP by Spugnardi et al.48 In particular, their most commonly methylated MSP fragment 2, which maps to exon 1α, was not investigated by us. The significance of methylation at this particular segment is questionable because its methylation did not correlate with transcriptional silencing.48 The other fragment (fragment 1) analyzed by Spugnardi et al.48 was located within the RASSF1A promoter and overlapped with the sequence that we studied. Methylation of this particular region was found in 41% of melanomas and 64% of the melanoma cell lines.48 These data and our results strongly support a role of epigenetic downregulation of RASSF1A transcription in a subset of melanomas, both in vitro and in vivo.

With respect to the relationship between RASSF1A and RAS alterations, studies on pancreatic and colon carcinomas have reported on a mutually exclusive occurrence of RASSF1A promoter methylation and KRAS gene mutations, suggesting that these alterations function as alternative mechanisms.49, 50 Our study showed that the majority of melanomas and all cell lines with decreased RASSF1A transcript levels additionally carried either BRAF or NRAS mutations. Therefore, our results suggest that transcriptional downregulation of RASSF1A does not function as an alternative mechanism to oncogenic BRAF or RAS mutation in melanomas. Interestingly, the RASSF1A protein has been found to heterodimerize with a homologous protein called Nore1.51 The corresponding NORE1A gene has been shown recently to be hypermethylated and transcriptionally downregulated in various human cancers.52 Thus, it will be of interest to investigate whether aberrations in NORE1A or yet other genes involved in Ras signaling are also involved in melanoma pathogenesis.

In summary, our results confirm the presence of either BRAF or NRAS gene mutations in the majority of melanomas in vivo and in vitro. We additionally demonstrate that a subset of melanomas shows epigenetic inactivation of RASSF1A, which may be found in tumors with or without NRAS or BRAF mutation. Because we did not study all the genes known to be involved in Ras signaling, it seems likely that melanomas carry aberrations in yet other genes related to the various Ras-regulated intracellular signaling pathways.