Advertisement

High Frequency of p16INK4A Promoter Methylation in NRAS-Mutated Cutaneous Melanoma

  • Author Footnotes
    2 These authors contributed equally to this work.
    Anders Jonsson
    Footnotes
    2 These authors contributed equally to this work.
    Affiliations
    Department of Oncology and Pathology, Cancer Center Karolinska, Karolinska Institutet/Karolinska University Hospital, Stockholm, Sweden
    Search for articles by this author
  • Author Footnotes
    2 These authors contributed equally to this work.
    Rainer Tuominen
    Footnotes
    2 These authors contributed equally to this work.
    Affiliations
    Department of Oncology and Pathology, Cancer Center Karolinska, Karolinska Institutet/Karolinska University Hospital, Stockholm, Sweden
    Search for articles by this author
  • Eva Grafström
    Affiliations
    Department of Oncology and Pathology, Cancer Center Karolinska, Karolinska Institutet/Karolinska University Hospital, Stockholm, Sweden
    Search for articles by this author
  • Johan Hansson
    Affiliations
    Department of Oncology and Pathology, Cancer Center Karolinska, Karolinska Institutet/Karolinska University Hospital, Stockholm, Sweden
    Search for articles by this author
  • Suzanne Egyhazi
    Correspondence
    Department of Oncology and Pathology, Cancer Center Karolinska R8:03, S-171 76 Stockholm, Sweden
    Affiliations
    Department of Oncology and Pathology, Cancer Center Karolinska, Karolinska Institutet/Karolinska University Hospital, Stockholm, Sweden
    Search for articles by this author
  • Author Footnotes
    2 These authors contributed equally to this work.
      The p16INK4A tumor suppressor is often deleted, or otherwise inactivated, in malignant melanoma. To investigate the loss of p16INK4A in greater detail, we analyzed 77 cutaneous melanoma metastases. Of these 56 retained at least one p16INK4A allele, and 21 had biallelic deletions. Using methylation-specific PCR, direct sequencing, and immunohistochemical methods, we analyzed p16INK4A promoter methylation, mutations, and protein expression, respectively. In addition, 14 corresponding primary tumors were analyzed for protein expression. Results were compared to clinicopathological parameters and previously obtained data regarding mutations in proto-oncogenes NRAS and BRAF. Results revealed that p16INK4A promoter methylation was present in 15 of 59 (25%) metastases; nonsynonymous mutations in 9 of 56 (16%) metastases; and protein expression in 12 of 67 (18%) metastases. Protein expression was lost during progression from primary to metastatic tumors, 71% (10 of 14) and 43% (6 of 14) being positive, respectively. However, the genetic and epigenetic alterations of p16INK4A observed could not explain the lack of p16INK4A protein in 27 metastases, indicating the presence of additional inactivating mechanisms for p16INK4A. Interestingly, p16INK4A promoter methylation was significantly overrepresented in NRAS-mutated samples compared to NRAS wild-type samples (P=0.0004), indicating an association between these two events.

      Abbreviations

      CMM
      cutaneous malignant melanoma
      IHC
      immunohistochemistry
      MSP
      methylation-specific PCR
      WT
      wild-type

      Introduction

      Cutaneous malignant melanoma (CMM) is a skin cancer with a poor prognosis at the metastatic stage (stage III–IV), and the disease shows a pronounced increase in incidence throughout Caucasian populations. The genetic and epigenetic aberrations known to be related to the development of CMM include inactivation of the CDKN2A gene by deletion, mutation, or promoter methylation (
      • Serrano M.
      • Lin A.W.
      • McCurrach M.E.
      • et al.
      Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a.
      ;
      • Ruas M.
      • Peters G.
      The p16INK4a/CDKN2A tumor suppressor and its relatives.
      ;
      • Dahl C.
      • Guldberg P.
      The genome and epigenome of malignant melanoma.
      ). The CDKN2A gene has two alternative reading frames resulting in two unrelated tumor suppressor proteins, p16INK4A and p14ARF, which are associated with the tumor suppressive functions of the RB protein and the p53 protein, respectively (
      • Serrano M.
      • Lin A.W.
      • McCurrach M.E.
      • et al.
      Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a.
      ;
      • Lowe S.W.
      • Sherr C.J.
      Tumor suppression by Ink4a-Arf: progress and puzzles.
      ;
      • Ha L.
      • Ichikawa T.
      • Anver M.
      • et al.
      ARF functions as a melanoma tumor suppressor by inducing p53-independent senescence.
      ). Mutations in CDKN2A are found both as somatic events in tumors, and rarely as germ-line alterations in the hereditary situation. Germ-line mutations, occur in approximately 25–40% of melanoma families (
      • Hussussian C.J.
      • Struewing J.P.
      • Goldstein A.M.
      • et al.
      Germline p16 mutations in familial melanoma.
      ;
      • Kamb A.
      • Gruis N.A.
      • Weaver-Feldhaus J.
      • et al.
      A cell cycle regulator potentially involved in genesis of many tumor types.
      ;
      • Curtin J.A.
      • Fridlyand J.
      • Kageshita T.
      • et al.
      Distinct sets of genetic alterations in melanoma.
      ;
      • Chin L.
      • Garraway L.A.
      • Fisher D.E.
      Malignant melanoma: genetics and therapeutics in the genomic era.
      ). In Sweden, <10% of the familial melanoma kindreds are found to carry germ-line mutations of CDKN2A (
      • Platz A.
      • Hansson J.
      • Mansson-Brahme E.
      • et al.
      Screening of germline mutations in the CDKN2A and CDKN2B genes in Swedish families with hereditary cutaneous melanoma.
      ). Somatic genetic aberrations in CDKN2A are dominated by deletions (30–60%) (
      • Flores J.F.
      • Walker G.J.
      • Glendening J.M.
      • et al.
      Loss of the p16INK4a and p15INK4b genes, as well as neighboring 9p21 markers, in sporadic melanoma.
      ;
      • Funk J.O.
      • Schiller P.I.
      • Barrett M.T.
      • et al.
      p16INK4a expression is frequently decreased and associated with 9p21 loss of heterozygosity in sporadic melanoma.
      ).
      In addition, CMM etiology very commonly involves constitutive activation of the RAS–RAF–MEK–ERK signal transduction pathway, leading to enhanced proliferation and survival of tumor cells. In total, 70–90% of CMM metastases carry mutations in one of the proto-oncogenes BRAF or NRAS (
      • Omholt K.
      • Platz A.
      • Kanter L.
      • et al.
      NRAS and BRAF mutations arise early during melanoma pathogenesis and are preserved throughout tumor progression.
      ;
      • Haluska F.G.
      • Tsao H.
      • Wu H.
      • et al.
      Genetic alterations in signaling pathways in melanoma.
      ). However, prolonged oncogenic stimuli may also trigger cellular senescence through induction of p16INK4A and p53 proteins (
      • Serrano M.
      • Lin A.W.
      • McCurrach M.E.
      • et al.
      Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a.
      ;
      • Narita M.
      • Nunez S.
      • Heard E.
      • et al.
      Rb-mediated heterochromatin formation and silencing of E2F target genes during cellular senescence.
      ;
      • Michaloglou C.
      • Vredeveld L.C.
      • Soengas M.S.
      • et al.
      BRAFE600-associated senescence-like cell cycle arrest of human naevi.
      ). Thus, avoiding cell-cycle arrest and senescence during melanoma development possibly necessitates inactivation of CDKN2A, which influences both the p16INK4A and p53 pathways. This is an attractive explanation as to why CDKN2A is so frequently inactivated in melanoma.
      Previously, we have screened for mutations in NRAS (exon 2) and BRAF (exons 11 and 15) in CMM metastases (
      • Omholt K.
      • Platz A.
      • Kanter L.
      • et al.
      NRAS and BRAF mutations arise early during melanoma pathogenesis and are preserved throughout tumor progression.
      ;
      • Edlundh-Rose E.
      • Egyhazi S.
      • Omholt K.
      • et al.
      NRAS and BRAF mutations in melanoma tumours in relation to clinical characteristics: a study based on mutation screening by pyrosequencing.
      ). We have also previously characterized the prevalence of allelic losses of CDKN2A in these tumors (
      • Grafstrom E.
      • Egyhazi S.
      • Ringborg U.
      • et al.
      Biallelic deletions in INK4 in cutaneous melanoma are common and associated with decreased survival.
      ). We found that samples with bi- or monoallelic deletions in the INK4 region (containing the CDKN2A and CDKN2B genes, which code for the tumor suppressor proteins p16INK4A, p14ARF, and p15, respectively) had shorter median overall survival compared to patients without deletions. However, patients with tumors with biallelic deletions had the same overall survival as those with tumors with monoallelic deletions. This could be due to other mechanisms of gene inactivation in tumors that carry monoallelic deletions such as promoter methylation or mutations.
      This study was aimed at investigating the prevalence of sporadic mutations and transcriptional silencing of p16INK4A by promoter methylation, in CMM metastases with monoallelic deletions and metastases without deletions in p16INK4A. Furthermore, we wished to relate inactivation of p16INK4A to clinical and pathological parameters and the previously studied NRAS and BRAF oncogene mutation status in these metastases. To obtain a more complete picture of p16INK4A aberrations in melanoma metastases, we also included tumors with biallelic deletions.

      Results

      p16INK4A promoter methylation

      Promoter methylation of p16INK4A was observed in 15 of 59 (25%) metastases using methylation-specific PCR (MSP) (Figure 1, Table 1). Three of the tumors with biallelic deletions have been analyzed and none of them showed any promoter methylation, as expected because both alleles have been lost. Assuming that none of the tumors with biallelic deletions have any promoter methylation, p16INK4A promoter methylation occurred in 19% (15 of 77) of this cohort. Promoter methylation among tumors without p16INK4 deletions was similar to those with monoallelic deletions, 7 of 21 (33%) and 8 of 35 (23%), respectively (P=0.53).
      Figure thumbnail gr1
      Figure 1Representative gel images from the methylation-specific PCR (MSP) analyses. L=100bp DNA ladder, U/M/Wt=PCR products from the PCR specific for unmethylated, methylated, and non-CT converted-specific MSP reactions, respectively. Samples were M21, M38, and M01, and control reactions for the MSP were the methylation-positive cell line 505, Wt-positive CT-converted control DNA (-CT), and non-template control for PCR (H2O). CT, bisulfite conversion of cytosine to thymidine.
      Table 1Clinical and pathological characterization of the subjects in the study with data from the analyses of p16INK4A
      IDAge at diagnosis (years)SexHistopathological subtypeBreslow thickness (mm)Oncogene mutatedp16INK4A deletionp16INK4A promoter methylationp16INK4A/ p14ARF mutationp16INK4A protein metastasis (nuclear/ cytoplasmic), primary tumor [nuclear/cytoplasmic]
      M0153MSSM1.5BRAFNoMethylp.L63P(0/0)
      M0258FSSM0.9BRAFNoUnmethylp.A67fs145X/ p.A123fs212X(0/0)
      M0335FSSM0.5BRAFNoUnmethylp.Q50X(0/0)
      M0451FSSM6BRAFNoUnmethylp.R80X/p.P135L(0/0)
      M0542FSSM2.2BRAFNoUnmethylNo(0/0)
      M0670MNM8BRAFNoUnmethylNo(0/0)
      M0782MSSM3.1BRAFNoUnmethylNo(0/2)
      M0891MNM13BRAFNoUnmethylNo(0/3)
      M0979MUnknNABRAFNoUnmethylNo(2/3)
      M1071MSSM0.5BRAFMonoMethylNo(0/0)
      M1146MNM3BRAFMonoUnmethylp.E33X(0/0)
      M1236MSSM3.2BRAFMonoUnmethylp.G23S, p.L97P(2/2) [3/3]
      M1369FUC3.5BRAFMonoUnmethylNo(0/0) [0/0]
      M1462MSSM1.6BRAFMonoUnmethylNo(0/0) [3/3]
      M1549MUnknNABRAFMonoUnmethylNo(0/0)
      M1640MSSM1.2BRAFMonoUnmethylNo(0/0)
      M1741FSSM1.9BRAFMonoUnmethylNo(0/0)
      M1850MUnknNABRAFMonoUnmethylNo(0/0)
      M1971MUC7BRAFMonoUnmethylNo(0/0)
      M2068FUC3BRAFMonoUnmethylNo(0/0)
      M2184FSSM1.6BRAFMonoUnmethylNo(0/0)
      M2278FUC0.7BRAFMonoUnmethylNo(0/0)
      M2334FUnknNABRAFMonoUnmethylNo(0/0)
      M2445FNM1.4BRAFMonoUnmethylNo(1/2) [0/0]
      M2557FALM0.8BRAFMonoUnmethylNo(1/2)
      M2640FSSM4.8BRAFBiUnmethylND(0/0)
      M2780MUC3BRAFBiUnmethylND(0/0)
      M2854MSSM2.7BRAFBiNDNDND
      M2972FSSM0.4BRAFBiNDNDND
      M3093MNM7.1BRAFBiNDNDND
      M3124FSSM0.3BRAFBiNDNDND
      M3278MNM1.2BRAFBiNDND(0/0)
      M3359FNM1.1BRAFBiNDNDND
      M3429MSSM1BRAFBiNDNDND
      M3572MSSM2.5NRASNoMethylNo(0/0) [0/1]
      M3658MNM4.3NRASNoMethylNo(0/0) [0/2]
      M3776FUnknNANRASNoMethylNo(0/0)
      M3865MUnknNANRASNoMethylNo(0/0)
      M3972FNM2.4NRASNoMethylNo(0/0)
      M4079MUCNANRASNoMethylNo(0/0)
      M4175FNM1.4NRASNoUnmethylNo(0/0)
      M4277FNM2.5NRASNoUnmethylNo(1/1) [3/3]
      M4324FSSM1.4NRASNoUnmethylNo(1/1)
      M4475MSSM5.3NRASMonoMethylNo(0/0) [0/0]
      M4585MNM9NRASMonoMethylNo(0/0) [0/1]
      M4634FNM3.8NRASMonoMethylNo(0/0) [3/2]
      M4750MNM3.5NRASMonoMethylNo(0/0)
      M4879FNM5.1NRASMonoMethylNo(0/0)
      M4984FNM2.7NRASMonoMethylNo(0/0)
      M5053FSSM1.1NRASMonoUnmethylp.P70L(2/3) [2/2]
      M5186FUC0.9NRASMonoUnmethylp.W110X/p.G166R(0/0)
      M5269MSSM1.6NRASMonoUnmethylNo(0/0) [2/3]
      M5354FUC0.8NRASMonoUnmethylNo(0/0)
      M5471MSSM1.3NRASMonoUnmethylNo(0/0)
      M5569MNM5NRASMonoUnmethylNo(0/0)
      M5668MSSM3NRASMonoUnmethylNo(0/0)
      M5765MUnknNANRASMonoUnmethylNo(0/0)
      M5844MSSM3.5NRASBiNDND(0/0)
      M5981MSSM4.4NRASBiNDNDND
      M6045FSSM1.8NRASBiNDNDND
      M6153MNM5NRASBiNDND(0/0)
      M6273MNM10.5NRASBiNDND(0/0)
      M6350MUC8NRASBiNDNDND
      M6468FUC1NRASBiNDNDND
      M6547FSSM0.7NoNoUnmethylNo(0/0)
      M6681MSSM2.1NoNoUnmethylNo(1/1) [0/0]
      M6765MALM10NoNoUnmethylNo(1/2) [3/3]
      M6870FSSM2.6NoMonoMethylNo(0/0)
      M6958MLMM0.5NoMonoUnmethylp.F90L/p.L145P(0/1)
      M7040FNM7NoMonoUnmethylNo(0/0)
      M7165MNM10NoMonoUnmethylNo(0/0)
      M7265FUC3.5NoMonoUnmethylNo(0/0)
      M7389MSSM8.5NoBiUnmethylND(0/0)
      M7477FALM30NoBiNDND(0/0)
      M7549FSSM0.7NoBiNDND(0/0)
      M7647MNM2.6NoBiNDND(0/0)
      M7768FNM6NoBiNDND(0/0)
      Abbreviations: ALM, acral lentigious melanoma; LMM, lentigo maligna melanoma; NA, not available; ND, not determined; NM, nodular melanoma; No, no mutation/deletion; SSM, superficial spreading melanoma; UC, unclassifiable; Unkn, patient with unknown primary tumor. The boldface entries in the column for p16INK4A/p14ARF mutation are related to P16INK4A, and the non-bold entries next to the bold entries are related to P14ARF.

      p16INK4A mutation screening

      Sequencing of CDKN2A exons 1α, 2, and 3 revealed that 9 of 56 (16%) metastases carried mutations in the coding region of p16INK4A (Table 1). Assuming that none of the tumors with biallelic deletions carry any mutation, 12% (9 of 77) of this cohort carried CDKN2A mutations. All CDKN2A mutations were heterozygous except in the M69 case that was hemi/homozygous. Interestingly, one metastasis (sample M12) had three different p16INK4A mutations (two nonsynonymous and one synonymous). A summary of all observed genetic variants found is shown in Supplementary Table S3 online. The two tandem mutations and four of the single-nucleotide substitutions showed classical UVB mutation signature patterns, i.e. G:C>A:T transitions at dipyrimidine sites, and GG:CC>AA:TT tandem alterations (
      • Hocker T.
      • Tsao H.
      Ultraviolet radiation and melanoma: a systematic review and analysis of reported sequence variants.
      ), indicating that several of the p16INK4A mutations were caused by UV exposure.
      Promoter methylation and p16INK4A mutation rarely coexisted—only sample M01 had both.

      Expression of the p16INK4A protein

      A majority of the metastases lacked p16INK4A protein expression, including all those with p16INK4A promoter methylation (Table 1). In total, only 12 of 67 (18%) metastases were p16INK4A positive. As expected, none of the 11 analyzed tumors with biallelic deletions were p16INK4A positive. Assuming that all the tumors with biallelic deletions lack p16INK4A protein expression, 16% (12 of 77) of our cohort was p16INK4A positive. There were 7 of 21 (33%) p16-positive tumors without p16INK4A deletions, compared to 5 of 35 (14%) tumors with monoallelic deletions (P=0.11). In three cases only cytoplasmic staining was observed. Interestingly, three p16INK4A-positive metastases (samples M12, M50, and M69) were also found to carry both a mutation in the coding region of p16INK4A and a monoallelic deletion. However, all samples with truncating mutations (i.e., p.E33X, p.Q50X, p.R80X, and p.W110X) and the tumor with a microdeletion causing a frameshift reading and truncated translation (p.A67fs145X) were negative for staining with the monoclonal p16INK4A antibody.
      Furthermore, immunohistochemistry (IHC) analysis of the 14 primary tumors revealed that a majority of these expressed p16INK4A (10 of 14 positive, i.e. 71%) and that expression was lost in some of their corresponding metastases (6 of 14 positive, i.e. 43%) (Figure 2, Table 1). The primary tumors showed a more heterogeneous staining pattern for p16INK4A compared to the metastases, with negative areas (nests or clones) within the tumor. Unexpectedly, two cases negative for p16INK4A expression in the primary tumors were positive in the corresponding metastases (samples M24 and M66). An explanation could be that these primary tumors metastasized before they lost p16INK4A expression or that the metastases originated from minor p16INK4A-positive clones within the primary tumors.
      Figure thumbnail gr2
      Figure 2Images of p16INK4A immunohistochemistry (IHC) sections. Panels show the metastases M42 and M46 and their corresponding primary tumors: (a) primary M46, negative control; (b) primary M46, anti-p16INK4A; (c) metastasis M46, negative control; (d) metastasis M46, anti-p16INK4A; (e) primary M42, negative control; (f) primary M42, anti-p16INK4A; (g) metastasis M42, negative control; (h) metastasis M42, anti-p16INK4A. Both primary tumors and metastasis M42 were regarded positive for p16INK4A expression and M46 metastasis was regarded negative. Bar=100μm.
      The number of p16INK4A positive tumors was significantly higher among tumors without any genetic or epigenetic alterations in comparison to tumors with one or two alterations, 7 of 11 (64%) and 5 of 56 (9%), respectively (P=0.0002). Noteworthy, 27 of 55 (49%) p16INK4A protein-expression-negative metastases had at least one intact p16INK4A allele (Table 1).

      Correlation of different p16INK4A aberrations with NRAS and BRAF oncogene mutation status

      Interestingly, p16INK4A promoter methylation predominantly occurred in NRAS-mutated tumors with 12 of 15 (80%) methylated tumors carrying NRAS mutation (Table 1). Promoter methylation has been found in 12 of 23 (52%) NRAS-mutated metastases, 2 of 27 (7%) BRAF-mutated metastases, and 1 of 9 (11%) wild-type (WT) metastases (Table 2). The proportion of samples with p16INK4A promoter methylation among NRAS-mutated samples was significantly higher than the proportion in those with WT NRAS (P=0.0004).
      Table 2Summary of the alterations of p16INK4A in relation to NRAS and BRAF oncogene mutation status
      Total (n=77)BRAF (n=34)NRAS (n=30)WT (n=13)
      p16INK4A deletion
       No21 (27%)9 (26.5%)9 (30%)3 (23%)
       Mono35 (45%)16 (47%)14 (47%)5 (38.5%)
       Bi21 (27%)9 (26.5%)7 (23%)5 (38.5%)
      p16INK4A promoter methylation*
       No44 (75%)25 (93%)11 (48%)8 (89%)
       Yes15 (25%)2 (7%)12 (52%)1 (11%)
       ND18774
      p16INK4A/p14ARF mutation
       No47 (84%)19 (76%)21 (91%)7 (88%)
       Yes9 (16%)6 (24%)2 (9%)1 (12%)
       ND21975
      p16INK4A protein expression
       No55 (82%)22 (79%)23 (88%)10 (77%)
       Yes12 (18%)6 (21%)3 (12%)3 (23%)
       ND12660
      Abbreviation: ND, not determined.
      *P=0.0004.
      This is in contrast to samples with p16INK4A mutations, which seemed to be more frequent in BRAF-mutated metastases (6 of 25, 24%) compared to NRAS-mutated (2 of 23, 9%) and WT metastases (1 of 8, 12%) although this difference was not statistically significant (P=0.27) (Table 2).
      However, the NRAS or BRAF mutation status shows little difference in the degree of inactivation of p16INK4A at the protein level (Table 2).

      Correlation of different p16INK4A aberrations with clinicopathological parameters

      None of the clinicopathological parameters analyzed (gender, age at diagnosis, histopathological type, and Breslow thickness) showed any significant association with p16INK4A biallelic deletion, promoter methylation, mutation, or protein expression (data not shown). However, patients with tumors with CDKN2A mutations tended to have an earlier onset (median age, 53 years) than those lacking CDKN2A mutations (median age, 69 years) (P=0.053).

      Discussion

      In this study, we analyzed p16INK4A promoter methylation, mutation, and protein expression in tumors either lacking p16INK4A deletions or having monoallelic or biallelic deletions. Although the smallest deletions are difficult to detect and to correctly assign the tumors in their respective deletion categories, we consider that the data presented in this work is accurate enough to give a good picture of the total impact of p16INK4A aberrations in melanoma metastases. The major obstacles of performing high-quality analyzes on excised tumor material are the presence of nontumor cells in the samples, degradation of DNA in archival tumor tissues and during the bisulfite treatment, and the cellular heterogeneity within the tumors. For these reasons, the included tumors and sections were selected so as to have high percentages of tumor cells (>70%), the PCR reactions were designed to be compatible with moderately degraded DNA, and evaluation of IHC sections was performed for multiple regions.
      We found that p16INK4A promoter methylation occurred in 19% of the metastases (including the tumors with biallelic deletions), which is similar to previous studies, reporting between 5 and 19% in primary melanomas and 27 and 33% in metastases (
      • Straume O.
      • Smeds J.
      • Kumar R.
      • et al.
      Significant impact of promoter hypermethylation and the 540 C>T polymorphism of CDKN2A in cutaneous melanoma of the vertical growth phase.
      ;
      • Sharpless E.
      • Chin L.
      The INK4a/ARF locus and melanoma.
      ;
      • Freedberg D.E.
      • Rigas S.H.
      • Russak J.
      • et al.
      Frequent p16-independent inactivation of p14ARF in human melanoma.
      ). There was a somewhat higher proportion of methylated samples among tumors without p16INK4A deletions (33% methylated) compared to those with monoallelic deletions (23% methylated), but this difference was not statistically significant. Our results indicate that monoallelic deletions and promoter methylation are not mutually excluding events, and both alterations contribute to p16INK4A inactivation. However, all samples with promoter methylation were negative for p16INK4A protein, suggesting that both alleles are silenced, although this has to be confirmed using allele-specific methylation analyses.
      An interesting finding was that 52% (12 of 23) of the NRAS-mutated tumors had p16INK4A promoter methylation compared to only 7% (2 of 27) of the BRAF-mutated tumors. The preferential coexistence of p16INK4A promoter methylation and NRAS mutation in the same tumors may be related to a cooperation of these genetic changes. Despite that tumors with NRAS or BRAF mutations show little difference in the degree of inactivation of p16INK4A at the protein level (88 vs. 79% inactivation, respectively) and both oncogene mutations are known to induce senescence in melanocytes (
      • Narita M.
      • Nunez S.
      • Heard E.
      • et al.
      Rb-mediated heterochromatin formation and silencing of E2F target genes during cellular senescence.
      ;
      • Michaloglou C.
      • Vredeveld L.C.
      • Soengas M.S.
      • et al.
      BRAFE600-associated senescence-like cell cycle arrest of human naevi.
      ), the sequence of events leading to neoplastic development may be different. The senescence inducing effect of NRAS mutation may be more pronounced and faster, thereby necessitating rapid (or preexisting) inactivation of p16INK4A whereas BRAF mutations might coexist with a functional p16INK4A protein early in the tumor development.
      Another possible explanation for the differential existence of promoter methylation and NRAS mutation compared to methylation and BRAF mutation could be presence of a mechanistic link between establishment of the NRAS mutation and the events leading to aberrant promoter methylation of p16INK4A. Support for such a link is that the transcription factor E2F, downstream of ras, has been shown to influence DNA methyltransferase 1 activity in a mouse model (
      • Kimura H.
      • Nakamura T.
      • Ogawa T.
      • et al.
      Transcription of mouse DNA methyltransferase 1 (Dnmt1) is regulated by both E2F-Rb-HDAC-dependent and -independent pathways.
      ). In addition, H-ras transformation has been shown to be accompanied by increased levels of DNA methyltransferase activity and repression of several genes by methylation. Treatment of 5-aza-deoxycytidine or suppression of oncogenic ras led to reexpression of these genes showing that oncogenic ras activity is linked to epigenetic silencing (
      • Lund P.
      • Weisshaupt K.
      • Mikeska T.
      • et al.
      Oncogenic HRAS suppresses clusterin expression through promoter hypermethylation.
      ).
      Overall, 12% (including the tumors with biallelic deletions) of our samples carried p16INK4A mutations, which is similar to previous reports (around 8%) on cutaneous melanoma (
      • Hocker T.
      • Tsao H.
      Ultraviolet radiation and melanoma: a systematic review and analysis of reported sequence variants.
      ;
      • Freedberg D.E.
      • Rigas S.H.
      • Russak J.
      • et al.
      Frequent p16-independent inactivation of p14ARF in human melanoma.
      ). In malignant melanoma, the majority of somatic mutations in CDKN2A have been shown to exhibit UVR signature (
      • Hocker T.
      • Tsao H.
      Ultraviolet radiation and melanoma: a systematic review and analysis of reported sequence variants.
      ). Several of the p16INK4A mutations observed in this study were C>T or CC>TT transitions, typically caused by UVB exposure.
      In several studies, it has been found that p16INK4A protein expression is gradually lost during melanoma progression: almost all benign nevi are positive for p16INK4A (74–100%), whereas primary melanomas show reduced levels of p16INK4A expression (28–78% p16 positive) and, in metastases, only 6–24% of samples are p16INK4A positive (
      • Ghiorzo P.
      • Mantelli M.
      • Gargiulo S.
      • et al.
      Inverse correlation between p16INK4A expression and NF-kappaB activation in melanoma progression.
      ;
      • Mihic-Probst D.
      • Mnich C.D.
      • Oberholzer P.A.
      • et al.
      p16 expression in primary malignant melanoma is associated with prognosis and lymph node status.
      ;
      • Fearfield L.A.
      • Larkin J.M.
      • Rowe A.
      • et al.
      Expression of p16, CD95, CD95L and Helix pomatia agglutinin in relapsing and nonrelapsing very thin melanoma.
      ;
      • Sanki A.
      • Li W.
      • Colman M.
      • et al.
      Reduced expression of p16 and p27 is correlated with tumour progression in cutaneous melanoma.
      ). Our results are in line with other reports on p16INK4A expression. In our study 16% of the metastases (including tumors with biallelic deletions) expressed p16INK4A protein. Three of these had both a monoallelic deletion and carried a mutation in the coding region. We and others have found that p16INK4A mutation-positive tumors express the protein although it has been shown that some of these point mutations impair the function of p16INK4A protein (
      • Ghiorzo P.
      • Villaggio B.
      • Sementa A.R.
      • et al.
      Expression and localization of mutant p16 proteins in melanocytic lesions from familial melanoma patients.
      ;
      • Scaini M.C.
      • Rossi E.
      • de Siqueira Torres P.L.
      • et al.
      Functional impairment of p16(INK4A) due to CDKN2A p.Gly23Asp missense mutation.
      ). In addition (
      • Yang G.
      • Rajadurai A.
      • Tsao H.
      Recurrent patterns of dual RB and p53 pathway inactivation in melanoma.
      ) predicted the effects of 117 reported CDKN2A point mutations showing large variations regarding the impact on p16INK4A protein.
      However, lack of expression of p16INK4A protein in 49% of the samples with at least one functional gene copy could not be fully explained by the genetic and epigenetic alterations identified in our study. The reasons for this could be misclassification of tumors with biallelic deletions into (primarily) the monoallelic deletion group in particular if tumors with deletions spanning only few kilobases are common. This is unlikely because pattern of deletion mapping suggest that small deletions are uncommon (
      • Gonzalgo M.L.
      • Bender C.M.
      • You E.H.
      • et al.
      Low frequency of p16/CDKN2A methylation in sporadic melanoma: comparative approaches for methylation analysis of primary tumors.
      ;
      • Palmieri G.
      • Cossu A.
      • Ascierto P.A.
      • et al.
      Definition of the role of chromosome 9p21 in sporadic melanoma through genetic analysis of primary tumours and their metastases. The Melanoma Cooperative Group.
      ).
      Other means of inactivation might also take place in these samples—for example, repressors of p16INK4A and other epigenetic events apart from promoter methylation. One such mechanism might be transcriptional repression by β-catenin (
      • Delmas V.
      • Beermann F.
      • Martinozzi S.
      • et al.
      Beta-catenin induces immortalization of melanocytes by suppressing p16INK4a expression and cooperates with N-Ras in melanoma development.
      ). Constitutive activation of the Wnt/β-catenin pathway is common in melanoma, and
      • Delmas V.
      • Beermann F.
      • Martinozzi S.
      • et al.
      Beta-catenin induces immortalization of melanocytes by suppressing p16INK4a expression and cooperates with N-Ras in melanoma development.
      reported that activated β-catenin in cooperation with activated N-ras promotes melanoma development in mice. Posttranscriptional micro-RNA-mediated suppression of p16INK4A translation by miR-24 has also been reported (
      • Lal A.
      • Kim H.H.
      • Abdelmohsen K.
      • et al.
      p16(INK4a) translation suppressed by miR-24.
      ).
      In summary, we have characterized mechanisms and patterns of p16INK4A inactivation in human melanoma. In our study p16INK4A is most commonly inactivated by deletions and/or promoter methylation (Table 2). Although the analysis of CDKN2A is limited by the lack of study of p14ARF, we found that four of nine mutations in exon 2 also affect the p14ARF protein (Table 1). In addition, we have previously shown that p14ARF is lost in 67% of the cases with p16INK4A biallelic deletions and that p14ARF biallelic deletions rarely occur without simultaneous loss of p16INK4A (
      • Grafstrom E.
      • Egyhazi S.
      • Ringborg U.
      • et al.
      Biallelic deletions in INK4 in cutaneous melanoma are common and associated with decreased survival.
      ). We found interesting patterns regarding inactivation of p16INK4A in melanoma metastases, in relation to NRAS and BRAF oncogene mutation status—possibly reflecting genetic and epigenetic alterations that occur during melanoma development and progression. Intriguingly, p16INK4A promoter methylation predominantly occurred in NRAS-mutant cases indicating an association between the two events.

      Materials and Methods

      Patients and tumor samples

      A total of 77 melanoma metastases, including 21 with biallelic deletions, from equal number of melanoma patients who underwent surgery between 1992 and 2002 at the Karolinska University Hospital were included in the study. Both formalin-fixed, paraffin-embedded archival samples and fresh-frozen tissue samples were used for DNA extraction. In addition, 14 corresponding primary tumors were available for analysis of p16INK4A protein expression. Approval has been obtained from the ethics committee of Karolinska Institutet. Patients have given informed consent. The study has been conducted according to the Declaration of Helsinki Principles. Patient characteristics with clinical and pathological parameters are summarized in Supplementary Table S1 online and given in Table 1 for the individual cases.
      The relative allelic concentrations of p16INK4A in intron 1 (immediately 5′ of exon 2) had been analyzed previously by quantitative real-time RT-PCR (
      • Grafstrom E.
      • Egyhazi S.
      • Ringborg U.
      • et al.
      Biallelic deletions in INK4 in cutaneous melanoma are common and associated with decreased survival.
      ).

      DNA extraction

      DNA was extracted from two to four 10-μm-thick slices of formalin-fixed, paraffin-embedded tissue using the QIAamp DNA Mini kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. DNA was also extracted from frozen tissue samples in 10 cases, using the same kit.

      Methylation-specific PCR

      After bisulfite treatment of DNA, promoter methylation of p16INK4A was analyzed by MSP, using primers specific for unmethylated (U) and methylated (M) sequences. Bisulfite conversion of DNA was performed using the EZ DNA methylation Gold Kit (Zymo Research, Orange, CA) following the manufacturer's instructions. DNA (500ng) was used for each bisulfite conversion reaction, which was performed at least in duplicate for each sample. The completeness of the bisulfite conversion reaction was controlled using non-CT converted specific PCR for all samples (WT primers) (
      • Herman J.G.
      • Graff J.R.
      • Myohanen S.
      • et al.
      Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands.
      ). Primer sequences are given in Supplementary Table S2 online.
      Approximately 100ng bisulfite-modified DNA was used in each (U, M, and WT) PCR, which also contained primers (2μM per primer); dNTPs (each 400μM), 1 × PCR buffer, 1.5mM MgCl2; 2U of Platinum Taq polymerase (Invitrogen, Paisley, UK); and 1% BSA (New England Biolabs, Hitchin, UK), in a reaction volume of 20μl. DMSO (5%; Sigma-Aldrich Sweden AB, Stockholm, Sweden), was used in the WT mix. PCR conditions were 3minutes at 95°C, 36 cycles × (20seconds at 94°C, 15seconds at annealing temperature, 20seconds at 72°C), then 5minutes at 72°C, and soak at 10°C.
      In addition, a nested PCR approach was used for samples with low initial DNA concentration. Bisulfite-converted DNA was then amplified in a two-step PCR procedure using previously published outer primers (
      • Holst C.R.
      • Nuovo G.J.
      • Esteller M.
      • et al.
      Methylation of p16(INK4a) promoters occurs in vivo in histologically normal human mammary epithelia.
      ). The first PCR conditions were as described above except reduced number of cycles (20 cycles) (30 cycles in the MSP).
      PCR products were separated on 1.6% agarose gels, stained with ethidium bromide, and visualized under UV light. A sample was regarded positive for methylation when at least two different PCRs from separate bisulfite conversion reactions yielded M products. DNA from cell line A375 was used as an unmethylated control and DNA from cell line 505 was used as a methylated control. Cell line DNA not modified with sodium bisulfite was used as WT control, and water replaced DNA in the contamination control reaction.

      Sequencing of p16INK4A

      Exons 1α, 2, and 3 of CDKN2A (p16INK4A) were PCR amplified (primer sequences and annealing temperatures are found in Supplementary Table S2 online). DNA extracted from frozen tissue was amplified using primers spanning the whole exon 2, whereas paraffin-extracted DNA was amplified in two overlapping fragments. Approximately 100ng DNA was used in each PCR as follows: 3minutes at 95°C, 40 cycles × (20seconds at 94°C, 20seconds at annealing temperature, 20seconds at 72°C), then 5minutes at 72°C, and soak at 12°C. Samples that yielded low amounts of PCR products were amplified further in a nested PCR. PCR products (2μl) from the first reaction were then amplified for additional 24 cycles. All PCR products were enzymatically treated with shrimp alkaline phosphatase and exonuclease I (50minutes at 37°C, followed by 20minutes at 80°C). Bidirectional sequencing was performed using the Big Dye Terminator v1.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA) and same primers as for PCR. Products were sequenced on an automated sequencer (ABI3130XL; Applied Biosystems). To confirm mutations, we reamplified samples and sequenced them again.

      Immunohistochemistry

      IHC was performed on 4μm, formalin-fixed, paraffin-embedded sections. Heat-induced antigen retrieval was performed in Reveal solution in a decloaking chamber (Biocare, Concord, CA) according to the manufacturer's instructions, and thereafter rinsed in hot rinse. Briefly, sections were incubated overnight (4°C) with the monoclonal primary p16INK4A antibody, clone JC8 (Biocare) diluted 1:250 in Tris-buffered saline buffer with 1.5% horse serum. Negative controls were incubated without primary antibody. Secondary antibody incubation using streptavidin/peroxidase complex was according to a kit manual (Vectastain Universal Quick Kit; Vector Laboratories, Burlingame, CA) as was development with 3,3′-diaminobenzidine (Immunkemi, Stockholm, Sweden). Finally, sections were counterstained with Mayer's hematoxylin. Normal breast tissue and a melanoma tumor positive for p16INK4A were included in all IHC batches.
      Independent evaluation of all slides was first performed by three observers (AJ, SE, and EG), thereafter a consensus was reached regarding the p16INK4A protein expression. The intensity was scored (0=absent, 1=weak, 2=moderate, 3=strong staining) for nuclear and cytoplasmic staining. Samples scored 1 or higher were considered positive. Samples with <5% positive cells were regarded negative (score=0).

      Statistical methods

      Statistical tests were two-sided and P-values exceeding 95% confidence level were considered significant (t-test, χ2-test, or Fisher's exact test when appropriate).

      ACKNOWLEDGMENTS

      We thank Diana Lindén for valuable cooperation in providing clinical and pathological data for patients in the study. We also thank Marianne Frostvik-Stolt, Liss Garberg, and Inger Bodin for excellent technical assistance in isolation of DNA from frozen tumor material and for sectioning of the tumor material. This investigation was supported by the Stockholm Cancer Society, King Gustav V's Jubilee Fund, and the Swedish Cancer Society.

      SUPPLEMENTARY MATERIAL

      Supplementary material is linked to the online version of the paper at http://www.nature.com/jid

      REFERENCES

        • Chin L.
        • Garraway L.A.
        • Fisher D.E.
        Malignant melanoma: genetics and therapeutics in the genomic era.
        Genes Dev. 2006; 20: 2149-2182
        • Curtin J.A.
        • Fridlyand J.
        • Kageshita T.
        • et al.
        Distinct sets of genetic alterations in melanoma.
        N Engl J Med. 2005; 353: 2135-2147
        • Dahl C.
        • Guldberg P.
        The genome and epigenome of malignant melanoma.
        Apmis. 2007; 115: 1161-1176
        • Delmas V.
        • Beermann F.
        • Martinozzi S.
        • et al.
        Beta-catenin induces immortalization of melanocytes by suppressing p16INK4a expression and cooperates with N-Ras in melanoma development.
        Genes Dev. 2007; 21: 2923-2935
        • Edlundh-Rose E.
        • Egyhazi S.
        • Omholt K.
        • et al.
        NRAS and BRAF mutations in melanoma tumours in relation to clinical characteristics: a study based on mutation screening by pyrosequencing.
        Melanoma Res. 2006; 16: 471-478
        • Fearfield L.A.
        • Larkin J.M.
        • Rowe A.
        • et al.
        Expression of p16, CD95, CD95L and Helix pomatia agglutinin in relapsing and nonrelapsing very thin melanoma.
        Br J Dermatol. 2007; 156: 440-447
        • Flores J.F.
        • Walker G.J.
        • Glendening J.M.
        • et al.
        Loss of the p16INK4a and p15INK4b genes, as well as neighboring 9p21 markers, in sporadic melanoma.
        Cancer Res. 1996; 56: 5023-5032
        • Freedberg D.E.
        • Rigas S.H.
        • Russak J.
        • et al.
        Frequent p16-independent inactivation of p14ARF in human melanoma.
        J Natl Cancer Inst. 2008; 100: 784-795
        • Funk J.O.
        • Schiller P.I.
        • Barrett M.T.
        • et al.
        p16INK4a expression is frequently decreased and associated with 9p21 loss of heterozygosity in sporadic melanoma.
        J Cutan Pathol. 1998; 25: 291-296
        • Ghiorzo P.
        • Mantelli M.
        • Gargiulo S.
        • et al.
        Inverse correlation between p16INK4A expression and NF-kappaB activation in melanoma progression.
        Hum Pathol. 2004; 35: 1029-1037
        • Ghiorzo P.
        • Villaggio B.
        • Sementa A.R.
        • et al.
        Expression and localization of mutant p16 proteins in melanocytic lesions from familial melanoma patients.
        Hum Pathol. 2004; 35: 25-33
        • Gonzalgo M.L.
        • Bender C.M.
        • You E.H.
        • et al.
        Low frequency of p16/CDKN2A methylation in sporadic melanoma: comparative approaches for methylation analysis of primary tumors.
        Cancer Res. 1997; 57: 5336-5347
        • Grafstrom E.
        • Egyhazi S.
        • Ringborg U.
        • et al.
        Biallelic deletions in INK4 in cutaneous melanoma are common and associated with decreased survival.
        Clin Cancer Res. 2005; 11: 2991-2997
        • Ha L.
        • Ichikawa T.
        • Anver M.
        • et al.
        ARF functions as a melanoma tumor suppressor by inducing p53-independent senescence.
        Proc Natl Acad Sci USA. 2007; 104: 10968-10973
        • Haluska F.G.
        • Tsao H.
        • Wu H.
        • et al.
        Genetic alterations in signaling pathways in melanoma.
        Clin Cancer Res. 2006; 12: 2301s-2307s
        • Herman J.G.
        • Graff J.R.
        • Myohanen S.
        • et al.
        Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands.
        Proc Natl Acad Sci USA. 1996; 93: 9821-9826
        • Hocker T.
        • Tsao H.
        Ultraviolet radiation and melanoma: a systematic review and analysis of reported sequence variants.
        Hum Mutat. 2007; 28: 578-588
        • Holst C.R.
        • Nuovo G.J.
        • Esteller M.
        • et al.
        Methylation of p16(INK4a) promoters occurs in vivo in histologically normal human mammary epithelia.
        Cancer Res. 2003; 63: 1596-1601
        • Hussussian C.J.
        • Struewing J.P.
        • Goldstein A.M.
        • et al.
        Germline p16 mutations in familial melanoma.
        Nat Genet. 1994; 8: 15-21
        • Kamb A.
        • Gruis N.A.
        • Weaver-Feldhaus J.
        • et al.
        A cell cycle regulator potentially involved in genesis of many tumor types.
        Science. 1994; 264: 436-440
        • Kimura H.
        • Nakamura T.
        • Ogawa T.
        • et al.
        Transcription of mouse DNA methyltransferase 1 (Dnmt1) is regulated by both E2F-Rb-HDAC-dependent and -independent pathways.
        Nucleic Acids Res. 2003; 31: 3101-3113
        • Lal A.
        • Kim H.H.
        • Abdelmohsen K.
        • et al.
        p16(INK4a) translation suppressed by miR-24.
        PLoS ONE. 2008; 3: e1864
        • Lowe S.W.
        • Sherr C.J.
        Tumor suppression by Ink4a-Arf: progress and puzzles.
        Curr Opin Genet Dev. 2003; 13: 77-83
        • Lund P.
        • Weisshaupt K.
        • Mikeska T.
        • et al.
        Oncogenic HRAS suppresses clusterin expression through promoter hypermethylation.
        Oncogene. 2006; 25: 4890-4903
        • Michaloglou C.
        • Vredeveld L.C.
        • Soengas M.S.
        • et al.
        BRAFE600-associated senescence-like cell cycle arrest of human naevi.
        Nature. 2005; 436: 720-724
        • Mihic-Probst D.
        • Mnich C.D.
        • Oberholzer P.A.
        • et al.
        p16 expression in primary malignant melanoma is associated with prognosis and lymph node status.
        Int J Cancer. 2006; 118: 2262-2268
        • Narita M.
        • Nunez S.
        • Heard E.
        • et al.
        Rb-mediated heterochromatin formation and silencing of E2F target genes during cellular senescence.
        Cell. 2003; 113: 703-716
        • Omholt K.
        • Platz A.
        • Kanter L.
        • et al.
        NRAS and BRAF mutations arise early during melanoma pathogenesis and are preserved throughout tumor progression.
        Clin Cancer Res. 2003; 9: 6483-6488
        • Palmieri G.
        • Cossu A.
        • Ascierto P.A.
        • et al.
        Definition of the role of chromosome 9p21 in sporadic melanoma through genetic analysis of primary tumours and their metastases. The Melanoma Cooperative Group.
        Br J Cancer. 2000; 83: 1707-1714
        • Platz A.
        • Hansson J.
        • Mansson-Brahme E.
        • et al.
        Screening of germline mutations in the CDKN2A and CDKN2B genes in Swedish families with hereditary cutaneous melanoma.
        J Natl Cancer Inst. 1997; 89: 697-702
        • Ruas M.
        • Peters G.
        The p16INK4a/CDKN2A tumor suppressor and its relatives.
        Biochim Biophys Acta. 1998; 1378: F115-F177
        • Sanki A.
        • Li W.
        • Colman M.
        • et al.
        Reduced expression of p16 and p27 is correlated with tumour progression in cutaneous melanoma.
        Pathology. 2007; 39: 551-557
        • Scaini M.C.
        • Rossi E.
        • de Siqueira Torres P.L.
        • et al.
        Functional impairment of p16(INK4A) due to CDKN2A p.Gly23Asp missense mutation.
        Mutat Res. 2009; 671: 26-32
        • Serrano M.
        • Lin A.W.
        • McCurrach M.E.
        • et al.
        Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a.
        Cell. 1997; 88: 593-602
        • Sharpless E.
        • Chin L.
        The INK4a/ARF locus and melanoma.
        Oncogene. 2003; 22: 3092-3098
        • Straume O.
        • Smeds J.
        • Kumar R.
        • et al.
        Significant impact of promoter hypermethylation and the 540 C>T polymorphism of CDKN2A in cutaneous melanoma of the vertical growth phase.
        Am J Pathol. 2002; 161: 229-237
        • Yang G.
        • Rajadurai A.
        • Tsao H.
        Recurrent patterns of dual RB and p53 pathway inactivation in melanoma.
        J Invest Dermatol. 2005; 125: 1242-1251