Cutaneous melanoma (CM) is the most lethal form of skin cancer and incidence in the Western world is rising. Recently, it was found that 66% of CMs and 82% of benign melanocytic tumors (comprising congenital, common acquired and dysplastic naevi) showed BRAF mutations,1, 2 while NRAS mutations are common in the remainder.2 Both genes encode proteins of the MAPK pathway, which is therefore deregulated in the vast majority of melanocytic tumors. Congenital, common acquired and dysplastic naevi appear to be melanoma precursors based on histologic similarities and apparent development of melanomas from contiguous naevi.3, 4 This suggests that BRAF and NRAS mutations may be early events in melanoma progression, although this issue remains uncertain. A key question now is whether the very high combined frequency of these mutations is common to all melanocytic neoplasia. The answer is important because the relationship of Spitz and blue naevi to melanoma is poorly documented. This may be partly because these tumors occur less frequently than common acquired naevi (CANs) and would therefore be expected to have a less frequent association with melanoma, although large congenital naevi are rarer than Spitz naevi and yet have a significant association with melanoma. In addition, Spitz naevus and melanoma have many histologic similarities, so contiguous Spitz and melanoma would be hard to identify.5 Malignant blue naevi have been reported,6 but their relationship to benign blue naevus is unknown. To address whether these mutations are common in all types of melanocytic tumor, we screened for mutation hot-spot alterations in BRAF and examined a subset of mutant and wild-type cases for NRAS mutation. Because BRAF and NRAS mutations activate the MAPK pathway, we also looked for evidence of this using immunohistochemistry for active dually phosphorylated ERK1/2 protein, a downstream pathway effector.
Cutaneous melanoma (CM) is the most lethal form of skin cancer. Along with some benign melanocytic tumors, the majority shows BRAF or NRAS mutation, but it is not known whether these are essential to all forms of melanocytic neoplasia. We screened 79 melanocytic tumors of different types for BRAF and NRAS mutations and looked at MAPK pathway activity using immunohistochemistry in a subset. Significant differences in BRAF exon 15 mutation frequency were found: 14/16 (87.5%) in common acquired naevi (CANs), 9/12 (75%) in CMs, 0/26 in Spitz naevi and 3/25 (12%) in blue naevi (p < 0.01). We looked at whether Spitz and blue naevi showed a compensatory increase in BRAF exon 11 and/or NRAS exons 1 and 2 mutations to account for the low BRAF exon 15 mutation frequency. NRAS mutations were found in only 1/16 (6.3%) Spitz naevi and 0/15 blue naevi. In addition, NRAS mutations were found in 2/11 (18.2%) CANs and 3/12 (25%) CMs. None of the tumors showed BRAF exon 11 mutations. Despite their low combined BRAF and NRAS mutation frequency, Spitz naevi showed strong MAPK pathway activation as measured by cytoplasmic expression of dually phosphorylated ERK1/2, while blue naevi had weak pathway activation. We conclude that BRAF and NRAS mutations are not necessary for melanocytic tumor development and that some types of tumor must arise by alternative mechanisms. © 2004 Wiley-Liss, Inc.
MATERIAL AND METHODS
Seventy-nine paraffin-embedded tissue samples from different patients were selected from the University Hospitals of Leicester NHS Trust histopathology archive, with local research ethics committee approval. Because microdissection was not routinely performed to separate tumor from surrounding normal skin, only cases where tumor cells occupied at least 25% of the biopsy were used. In our hands, single-strand conformation polymorphism (SSCP) is able to detect mutations where tumor cells comprise as little as 10% of the total sample without false negatives, as confirmed by subsequent laser capture microdissection and of pure tumor cells and by analysis of mutant DNA diluted at varying ratios in wild-type DNA. The tumors comprised 16 CANs, 25 blue naevi, 26 Spitz naevi and 12 melanomas. These were selected randomly apart from 2 melanomas that were specifically included because 1 had initially been interpreted as blue naevus and the other a Spitz naevus. Mutations in NRAS and BRAF genes were detected using SSCP analysis of PCR-amplified DNA from 1–5 × 10 μm tissue sections per sample. The sections were rehydrated, digested in 400 μl rapid extraction buffer (50 mM KCl, 100 mM Tris-HCl, pH 8.3, 2.5 mM MgCl2, 0.45% v/v Tween 20, 0.45% v/v NP40 tergitol) containing 0.5 mg/ml proteinase K (Boehringer Mannheim, Brussels, Belgium) overnight at 58°C, followed by boiling for 1 min and storage at 4°C. Thermal cycling was carried out with the GeneAmp PCR system 9600 (Perkin Elmer, Oak Brook, IL) in final volumes of 50 μl containing 45 mM Tris-HCl (pH 8.8), 11 mM (NH4)2SO4, 4.5 mM MgCl2, 200 μM each dNTP, 6.7 mM β mercatoethanol, 4.4 μM EDTA, 0.2 μM of each primer (for NRAS exon 1, forward 5′-CTCGCCAATTAACCCTGATT-3′, reverse 5′-CCGACAAGTGAGAGACAGGA-3′; for NRAS exon 2, forward 5′-TTCTTACAGAAAACAAGTGGTTATAGA-3′, reverse 5′-GAGGTTAATATCCGCAAATGAC-3′; for BRAF exon 11, forward 5′-TTTCTTTTTCTGTTTGGCTTGA-3′, reverse 5′-TGTCACAATGTCACCACATTACA-3′; for BRAF exon 15, forward 5′-CATAATGCTTGCTCTGATAGGAAA-3′; reverse 5′-GCTTGCTCTGATAGGAAAATGAG-3′), 5 μl of DNA extract and 2.5 U of Taq polymerase (Promega, Madison, WI). The amplification protocol consisted of 40 cycles with denaturation at 94°C, annealing at 60°C for 30 sec and extension at 72°C for 30 sec. An initial denaturation step of 94°C for 10 min and a final incubation at 72°C for 2 min were included. Appropriate SSCP conditions were determined using DNA extracted from control cell lines with known mutations (MOLT-4 for NRAS codon 12, HL60 and HT1080 for NRAS codon 61, SKMel 5 and SKMel 28 for BRAF codon 599). A 1% v/v MDE polyacrylamide gel (BioWhittaker Molecular Applications, Rockland, ME) was sufficient to detect all of the cell line mutations. Gels were loaded with 3 μl of PCR product and run at 300 V overnight in 1 × TBE buffer in a cold room at 4°C. DNA was visualized by silver-staining the gel and recorded using a digital scanner. Repeat PCR of mutant samples was performed to confirm the result. Aberrantly migrating bands were excised with a needle from the gel, reamplified by PCR with the same primers and purified with QIAEX II (Qiagen, Chatsworth, CA). Automated DNA sequencing was then performed using Big Dye terminators and cycle sequencing to confirm and identify the point mutations.
Two cases had weak aberrant bands where mutation was not confirmed by sequencing. DNA was extracted from laser capture microdissected tumor cells from these cases followed by restriction fragment length polymorphism–polymerase chain reaction (RFLP-PCR) as previously described7 using an NRAS codon 61 forward primer 5′-GGACATACTGGATACAGCTGTA-3′ and the NRAS exon 2 reverse primer above, followed by restriction enzyme digestion with Bsp1407I (Helena Biosciences, Sunderland, U.K.).
We identified MAPK pathway activation using an antibody to dually phosphorylated ERK1/2 protein on paraffin-embedded tissue sections. Briefly, 4 μm formalin-fixed paraffin-embedded tissue sections were rehydrated, pretreated with 5 μg/ml Proteinase K for 1 hr at 37°C and blocked in normal rabbit serum 1:5. Primary rabbit antihuman active ERK1/2 Ig (V803A, Promega) 1:1,000 was incubated overnight at 4°C and detected using biotinylated swine antirabbit Ig (E0431, Dako, Carpinteria, CA) 1:800 followed by streptavidin-biotin complex-alkaline phosphatase (K0391, Dako) and visualization in naphthol AS BI phosphate and Fast Red TR as the substrate in the presence of Levamisole to block endogenous alkaline phosphatase activity. Active ERK1/2 localizes to the nucleus but also has cytoplasmic targets, therefore 2 observers (G.S., J.H.P.) scored immunostaining based on overall intensity and extent as absent, weak (< 50% cells, low intensity) or strong (> 50% cells, high intensity) for both nuclear and cytoplasmic localization. The 50% cutoff was chosen on an arbitrary basis. For each case, a paired section where primary antibody was omitted served as a negative control. The endothelium of peritumoral vessels, which had a high intensity of phospho-ERK1/2 staining, served as an internal positive control, while fibroblasts were negative. We used formalin-fixed, paraffin-embedded cytoblock preparations of the melanoma cell line SK-MEL-28 as an external positive control. The specificity of phospho-ERK1/2 antibody was confirmed by immunostaining the same cells after prior PD98059 treatment. The latter specifically inhibits phosphorylation of ERK1/2 by upstream components in the MAPK pathway.8
Probabilities were calculated using a binomial distribution. Formal statistical comparisons employed were 2-tailed chi-squared and Fisher's exact tests; p < 0.05 was considered significant.
Mutations in exon 15 account for the vast majority of BRAF gene alterations in melanomas and naevi analyzed to date. We therefore looked for this in a range of melanocytic tumors, including Spitz and blue naevi (which have not been previously analyzed for this alteration). We first determined the sensitivity of the SSCP assay using dilutions of BRAF exon 15 mutant DNA in BRAF exon 15 wild-type DNA, showing that a mutation could be detected when it comprised as little as 10% of the sample (Fig. 1a). BRAF exon 15 mutation results are demonstrated in Figure 1(b). Mutations were found in 14/16 (87.5%) CANs and 9/12 (75%) melanomas, frequencies comparable to other data.1, 2, 9, 10 One of the BRAF mutant melanomas was originally diagnosed as a blue naevus, but subsequently metastasized to regional lymph nodes and was redesignated as malignant blue naevus. The aberrant bands were all identical and sequencing confirmed that these were BRAF V599E mutations. In contrast, we found 0/26 BRAF exon 15 mutation in Spitz naevi and 3/25 (12%) in blue naevi. The BRAF mutation frequency in these tumors showed significant differences (p < 0.01).
In melanocytic tumors, NRAS and BRAF exon 11 mutations tend to be seen in BRAF exon 15 wild-type cases,1, 2 presumably because the encoded proteins all act in the MAPK pathway and generate overlapping phenotypes. Therefore, the lack of BRAF exon 15 mutations in Spitz and blue naevi could be because the majority harbor these alternative MAPK pathway mutations. To address this, we selected a subset of Spitz and blue naevi from our original sample comprising 16 Spitz naevi with wild-type BRAF exon 15 and 15 blue naevi, 12 with wild-type and 3 with mutant BRAF exon 15. We then looked for NRAS exon 1 and 2 and BRAF exon 11 mutations. The results are shown in Table I. Of the Spitz naevi, only 1 of 16 cases showed further mutations: a G12C in NRAS exon 1. None of the blue naevi showed further mutations. If, as proposed above, the majority of Spitz and blue naevi harbored NRAS or BRAF exon 11 mutations, then the probability of finding only one or less mutations in the Spitz naevi and none in the blue naevi would be less than 0.001. This probability is so low that it would be reasonable to assume that no reciprocal increase in mutation frequency of the magnitude postulated exists. No further cases of Spitz or blue naevi were therefore analyzed.
|Case||BRAF exon 15 mutation1||BRAF exon 11 mutation||NRAS exon 1 mutation||NRAS exon 2 mutation||Histology|
We also performed a similar analysis on a subset of CANs, which included 2 BRAF exon 15 wild-type and 9 BRAF exon 15 mutant cases and all 12 melanomas. NRAS mutations were found in 2/11 (18.2%) CANs. In one, a G12D mutation was present alongside a BRAF exon 15 mutation. In the other, a Q61R mutation was found alongside wild-type BRAF. In the melanomas, no NRAS exon 1 mutations were found, but 3/12 (25%) had NRAS exon 2 aberrant SSCP bands. One of these was a Q61L mutation. The other 2 sequences could not be confirmed because the aberrant bands were weak (Fig. 1c, lanes 2 and 4), suggesting that, while the total number of tumor cells comprised 25% of the sample, the tumor cells actually harboring mutations may have been in the minority. However, RFLP-PCR of microdissected tumor cells confirmed the presence of codon 61 mutations, as shown in Figure 2. Two of the NRAS mutations were present alongside a BRAF exon 15 mutation, and the remaining tumor had wild-type BRAF. The latter tumor was initially diagnosed as a Spitz naevus, but recurred twice over a 7-year period with the frankly malignant histology of a Spitzoid melanoma.
We wanted to determine whether there was evidence of MAPK pathway activation in Spitz and blue naevi, despite their low frequency of BRAF and NRAS mutations. Immunohistochemistry for active dually phosphorylated ERK1/2 protein was performed on 35 cases comprising 10 Spitz naevi, 6 blue naevi, 8 CANs, and 11 melanomas, chosen because they had complete BRAF and NRAS mutation analysis data. The results are shown in Table II and illustrated in Figure 3. We documented whether dually phosphorylated ERK1/2 was present in the nucleus and/or cytoplasm because the relative staining in these 2 compartments would provide important clues about ERK1/2's targets in melanoma.
|Case||Cytoplasmic staining1||Nuclear staining1||BRAF mutation2||NRAS mutation2||Histology|
Nuclear staining was absent or weak in 8/8 CANs, 11/11 melanomas, 6/6 blue naevi and 9/10 Spitz naevi. Strong expression was seen in only 1 Spitz naevus. Cytoplasmic staining showed marked differences between tumor types (p < 0.001). Specifically, 10/10 (100%) Spitz naevi showed strong expression compared to 6/11 (54.5%) melanomas, 1/8 (12.5%) CANs and 0/6 blue naevi. Of the NRAS mutants, 5/6 (83.3%) had strong cytoplasmic phospho-ERK1/2 expression. This was regardless of histology (1/2 CANs, 1/1 Spitz naevus, 3/3 melanomas) or BRAF mutation status (3 with, 3 without). Despite this, the relationship between NRAS mutation and strong cytoplasmic phospho-ERK1/2 staining was not statistically significant (p = 0.08) because it was heavily biased by the 10 Spitz naevi that also showed strong cytoplasmic staining but rarely harbored NRAS mutation. BRAF mutation was strongly associated with absent/weak cytoplasmic phospho-ERK1/2 (p < 0.01), regardless of histology, but not NRAS status, as noted above. Overall, these data show that Spitz naevi had high levels of cytoplasmic phospho-ERK1/2 akin to those seen in NRAS mutant cases, while blue naevi tended to have lower levels akin to those seen with BRAF mutations.
These data show clear differences in the frequency of BRAF mutations and indicate that these do not characterize all types of melanocytic tumor.
The differences in overall BRAF/NRAS mutation frequencies provide insights into possible tumor relationships. For example, BRAF, or more rarely NRAS, mutation must be a common route to CAN, as virtually all have it. Clinical observation indicates that most CANs remain stable, but a small minority progress to melanoma, presumably through acquisition of additional alterations. Because most melanomas also have BRAF or NRAS mutation, it thus seems likely that either the majority arises via a CAN, or there must be an alternative BRAF/NRAS-mediated tumorigenic route to melanoma without a benign intermediary. The findings of Dong et al.11 support the second scenario because they show that BRAF mutations occur relatively late in progression, as melanomas change from radial to vertical growth phase. This would imply that BRAF mutations occur in 2 distinct settings, first in some benign naevi such as CAN, and then during progression of melanoma. This could explain the fact that in most studies the majority of melanomas do not arise next to naevi. In contrast, Spitz and blue naevi had infrequent BRAF and NRAS mutations; therefore, if melanoma is capable of arising from a Spitz or blue naevus, it is likely to be either from those infrequent mutant cases or from wild-type cases that subsequently acquired such mutations. We analyzed 2 melanomas, 1 of which had originally been diagnosed as Spitz naevus and the other as blue naevus, with this issue in mind. We found that they harbored NRAS and BRAF mutation, respectively, thus supporting either of the above proposals. However, we included only 2 such cases in our data and would require more samples to confirm whether this was a consistent finding, although they do provide an indication of the potential differences that may exist between Spitzoid melanoma and Spitz naevus on the one hand and malignant blue naevus and blue naevus on the other. Alternatively, these tumors never were Spitz or blue naevi, but merely showed morphologic resemblance to them leading to diagnostic confusion.
The results of this study pose questions regarding what alternative cellular events give rise to Spitz and blue naevi. To address this, we assessed whether there was evidence of MAPK pathway activation despite a low frequency of BRAF and NRAS mutations in these tumors. We looked for activation of ERK1/2, which has several nuclear and cytoplasmic targets, and found that it was mainly localized in the cytoplasm of melanocytic tumors. Cytoplasmic targets include ribosomal proteins and the cytoskeleton,12 but the precise targets in melanocytic tumors are not known. However, our data suggest that nuclear targets such as transcription factors may be less important than cytoplasmic targets in melanocytic tumors. Blue naevi had weak cytoplasmic phospho-ERK1/2 whether or not they were BRAF mutant. This level of expression was equivalent to that seen in other types of melanocytic tumor with BRAF mutation. Spitz naevi showed strong cytoplasmic phospho-ERK1/2 that was commensurate with that in cases harboring NRAS mutation, suggesting a similar underlying mechanism. Indeed, HRAS amplification and/or mutation are seen in approximately 10% of Spitz naevi.13 Whether other MAPK pathway genes are altered in Spitz and blue naevi remains to be established. Determining these abnormalities is important, especially for Spitz naevi, because their distinction from melanoma on morphologic criteria is a recurring diagnostic dilemma that often results in inappropriate patient management and litigation.14 Development of a molecular diagnostic test to separate these lesions would represent an advance of immense clinical value. The use of BRAF mutations as a diagnostic test would have a high specificity for melanoma since 0 out of 26 Spitz naevi were mutant. Additionally, in this particular diagnostic setting, other BRAF mutant benign tumors, such as CAN, would be easily excluded using standard histologic criteria. Unfortunately, this test would have poor sensitivity because a significant proportion of melanomas have wild-type BRAF. However, our study has only assessed histologically typical melanomas and Spitz naevi, and these would not normally be expected to cause diagnostic problems in everyday practice. It remains unknown whether a similar mutation frequency would be seen if samples had been chosen on the basis of those that cause diagnostic problems, namely, atypical Spitz naevi and Spitzoid melanomas.
We found that NRAS mutant cases tended to have increased phospho-ERK1/2, which is entirely plausible given that NRAS-BRAF-MEK1/2-ERK1/2 represents a signaling cascade. However, the finding that BRAF mutation was associated with absent or weak phospho-ERK1/2 expression appears to be at odds with this. While this may indicate that the antibody against dually phosphorylated ERK1/2 is not specific, an alternative possibility is that this relates to differences in kinase activity of BRAF and NRAS mutants. Davies et al.2 looked at the effect of BRAF V599E/HRAS G12V interaction on kinase activity but did not compare them individually. However, these mutant proteins showed striking differences in biologic effect, with BRAF V599E being far less effective at transforming mouse fibroblasts than HRAS G12V.2 Nevertheless, we found 2 melanomas with BRAF mutation in association with wild-type NRAS that bucked the trend by having strong cytoplasmic phospho-ERK1/2 staining. A possible explanation for this is that BRAF mutation is insufficient to cause constitutive MAPK pathway activation and resultant strong phospho-ERK1/2 expression on its own, but requires other factors. The development of autocrine growth factor stimulation, which is not seen in CAN, may be one of the other factors, as demonstrated by Satyamoorthy et al.10 In addition, the latter looked at phospho-ERK1/2 immunostaining in CANs, vertical growth phase and metastatic melanomas. They stated that CANs were negative while the other samples stained intensely, but it is difficult to draw comparisons with our data because the numbers of cases they assessed and how the staining intensity was graded is not documented. In addition, it is not stated whether the samples were paraffin-embedded.
In summary, we conclude that a high frequency of BRAF and NRAS mutations is not a characteristic feature of all melanocytic tumors and that the majority of Spitz and blue naevi must arise through as yet unknown molecular mechanisms. Further studies are needed to ascertain the precise alterations involved.