Background
Since the discovery of gain of function mutations in the proto-oncogenes
NRAS [
1] and
BRAF [
2], thousands of human skin melanoma samples have been analyzed, and the estimated incidence of
NRAS and
BRAF mutations are 18 % and 41 %, respectively [
3]. These mutations are often mutually exclusive [
4,
5]. The
BRAF locus is localized on chromosome 7q, and most
BRAF mutations involve the kinase activation loop at the p.V600 position. The most common
BRAF mutation is a substitution of a valine to a glutamic acid (c.1799 T > A, p.V600E).
BRAF V600E accounts for 85 % of exon 15 mutations in the most recent studies [
6,
7]. Another mutation, V600K is present in 9 % of melanomas. These
BRAF mutations constitutively activate the MAPK signaling pathway [
8].
Two BRAF inhibitors, vemurafenib and dabrafenib, targeting the BRAF p.V600 mutated protein, have recently been shown to prolong the progression-free and/or the overall survival of
BRAF V600-mutated advanced melanoma, as compared to dacarbazine [
9‐
11]. However, both are limited by the development of acquired resistance in many patients, with a median progression-free survival (PFS) of 6.9 and 6.7 months for vemurafenib and dabrafenib, respectively [
10,
12].
Mechanisms underlying acquired resistance to BRAF inhibitors have been extensively studied, and most of them involve acquired mutations within the same RAS-RAF-ERK pathway [
13]. By contrast, only little data is available concerning biomarkers of good/prolonged response to BRAF inhibitors. Recently, a high ratio of mutant/wild-type alleles of
BRAF was reported to be associated with a good response to BRAF inhibitors [
14].
Like most oncogenes, somatic mutations of
BRAF are thought to be heterozygous in tumors. Some studies reported that
BRAF mutations are not heterozygous in some cases [
15]. Additionally, in contrast to wild-type BRAF, which is only active as a dimer, products of alleles with gain of function mutations are also active as monomers [
16].
We present herein a validated quantification of mutated BRAF in a large series of human skin melanoma samples, and demonstrate that several cases are not heterozygous. We also present the results of a genetic study on mechanisms of the BRAF mutant allele increase in melanoma.
Methods
Samples and nucleic acid extraction
All samples were obtained from the bank of biological resources of Ambroise Paré Hospital. All surgical or fine needle biopsies were performed for routine diagnosis or evaluation of disease progression. The research was performed in compliance with the ethical principles of the Helsinki Declaration (1964). In accordance with French ethics laws, all patients were informed that part of their samples could also be used for research purposes, and that they could refuse this. None of patients refused the use of samples for research. Tumor sample collection was declared to the French Ministry of Research (DC 2009-933) and CPP IDF 8 ethics committee approved the MelanCohort study (030209), which is registered with Clinicaltrials.gov (NCT00839410). Signed informed consent for translational research was obtained from patients still alive. All diagnoses were confirmed by pathology review.
For most of the nucleic acid quantification studies, the tumor DNA was extracted from formalin fixed paraffin embedded (FFPE) tissue. However, for mRNA extraction and high density SNP analysis, frozen samples were used. In all cases, a 4 micrometers-thick section was stained with hematoxylin & eosin and reviewed by a pathologist before extraction, to confirm the presence of melanoma and to select areas with the highest density of tumor cells for macrodissection. For all samples, tumors cell content was estimated in the percentage of tumor cells and the data was noted. To evaluate the accuracy of tumor cell content assessment, a series of 41 randomly selected samples was assessed by three independent pathologists.
For each sample, serial sections or punch sampling were then used for nucleic acid extraction. For DNA extraction, samples were digested by an overnight incubation in the presence of proteinase K, followed by the application of the QIAamp DNA mini kit (Qiagen, Courtaboeuf, France) as previously described [
17]. The RNeasy Mini Kit (Qiagen) was applied for RNA extraction. DNA and RNA were controlled with a spectrophotometer (ND-100, Nanodrop®).
Real-time PCR
Real-time PCR (rtPCR) was performed as previously described [
18]. 1 μL of DNA brought to 25 ng/μL was applied to each reaction mixture. The amplification reaction was performed in Applied Prism 7900 HT (Thermo Fisher Scientific, Illkirch, France). Each sample was analyzed in two different reaction mixes: in the first one, all
BRAF alleles present in the tumors were amplified; in the second one, only the mutated allele was detected by peptide nucleic acids (PNAs)-specific inhibition of wild-type (WT) allele amplification. Each PCR reaction was carried out in duplicate. The primers and probe sequences were published previously [
17]. The relative quantification method was used to compare expression levels of wild type allele and
BRAF V600E mutated allele using comparative Ct method as described by Livak
et al. [
19].
Picoliter-droplet digital PCR
Picoliter-droplet digital PCR (dPCR) testing was performed using previously described protocols [
20,
21] with the Raindrop Instrument (RainDance Technologies, Billerica, MA). Shortly, in a pre-PCR environment, 12.5 μL Taqman Genotyping Master Mix (Life Technologies, Saint Aubin, France) was mixed with the assay solution. The assay solution contained: 0.75 μL of 40 mM dNTP Mix (New England BioLabs, Evry, France), 0.5 μL of 25 mM MgCl2, 2.5 μL of 10x Droplet Stabilizer (RainDance Technologies), 1.25 μL of 20x Taqman® Assay Mix containing 8 μM of forward and reverse primers, 200 nM of 6-FAM and 200 nM of VIC Taqman® labelled-probes (Additional file
1) and target DNA template to a final reaction volume of 25 μL. A minimum of 280 ng of DNA was used in each assay. 5 millions highly monodispersed droplets were generated using the Raindrop source instrument following manufacturer’s instructions. The emulsions were submitted to thermocycling, starting with 2 min at 50 °C, 10 min at 95 °C, followed by 45 cycles of: 95 °C, 15 s and 60 °C, 1 min (using a 0.6 °C/min ramp rate). After completion, the end-point fluorescence signals from each droplet were measured using the Raindrop Sense instrument. Analyses of the data were performed using the Raindrop analyst software as previously described [
20,
21]. The reference sequence was B-RAF cDNA sequence (GenBank NM_00433.4).
Pyrosequencing
Pyrosequencing was performed as previously described [
22]. It is a method of DNA sequencing based on the "sequencing by synthesis" principle. The results are displayed in the form of peaks corresponding to the detection of pyrophosphate release after nucleotides incorporation (pyrogram). Peak area is proportional to the number of individual nucleotide incorporated to the sequence; thus allowing the relative quantifications of mutated and WT alleles. RNA was transcribed to cDNA and FFPE tumor DNA concentrations were brought to 10 and 20-25 ng/μL prior to PCR amplification. Primers used for FFPE tumor DNA amplification and pyrosequencing were published previously [
17]. Biotinylated amplicon was verified on agarose gel and analyzed with PyroMark 24 (Qiagen) according to manufacturer recommendations. Primers used for frozen tumor DNA/RNA amplification and pyrosequencing are presented in Additional file
1.
FISH and Immunohistochemistry
Tissue microarray (TMA) was performed for 141 samples of melanomas (140 patients) and 42 samples of melanocytic nevi (junctional, intradermal and compound) of more than 4 mm long axis. For each tumor, three cores of 0.6 mm diameter from distinct tumor regions were spotted onto the slides.
For immunohistochemistry, the VE1 antibody was used as previously described [
17,
23]. Detection of BRAF p.V600E mutated protein with VE1 has been shown to have a high sensitivity, specificity and reproducibility. Intensity of staining was evaluated by two independent observers on a semi-quantitative scale of 0–3. The VE1 antibody was scored as negative (0) when there was no staining, weak staining of single interspersed cells, or staining of monocytes/macrophages. Positive staining was scored as: weakly positive staining (1), moderately positive staining (2) and strongly positive staining (3) of melanoma cells. Cases were considered not interpretable when nuclear staining was present. Cases were scored as ambiguous if immunostaining could not be scored as positive or negative.
For FISH analysis, TMA section slides of 4 micrometers were stored at -20 °C and hybridization was performed within 2 weeks of cutting. All samples were analyzed with the RP11-121G9 BAC probe covering the BRAF gene. The chromosome 7 centromere probe was used as reference probe (Agilent Technologies, Les Ulis, France). A total of 53 tumor samples were also analyzed with the commercially available BRAF probe, SureFISH 7q34 BRAF (Agilent Technologies), together with the chromosome 7 centromere probe. For home-made probes, bacteria carrying a BAC vector were grown overnight onto solid agar medium, followed by an overnight proliferation in a LB medium. BAC DNA was extracted using NucleoBond PC 500 or NucleoBand Xtra BAC Kits (Macherey-Nagel, Hoerdt, France) as recommended by supplier. 1 μg of DNA was labeled by nick-translation reaction according to manufacturer’s instructions (Abbott Molecular Inc., Rungis, France). After overnight precipitation at -20 °C in the presence of human cot DNA, sodium acetate and ethanol, the probe was resuspended in the hybridization buffer (LSI/WCP Hybridization Buffer)(Abbott Molecular Inc). They were used at a final concentration of 40-50 ng/μL. FFPE slides were prepared for hybridization using Histology FISH accessory KIT (DAKO, Les Ulis, France). Commercial probes were applied according to the manufacturer’s recommendations (Agilent Technologies) and a co-denaturation of the probes and the tumor section were performed to create single-stranded DNA. The probes and the slides were denaturated separately when BRAF BAC probes were used together with commercial chromosome 7 centromere probe. Before an overnight incubation in a humidified chamber, a suppression of the repetitive sequences was performed for DNA BRAF BAC (45 min at 37 °C). After post-hybridization wash, and DAPI staining in the Vectashield® Mounting Media (Vector Laboratories, Les Ulis, France), fluorescence signals were analyzed using a Leica DM4000B microscope equipped with appropriate filters and a DFC300FX camera under the control of LAS V4.0 software (Leica). Two independent analyses were performed.
SNP analysis
DNA was extracted from frozen melanoma samples and hybridized on HumanCore BeadChip (Infinium Ilumina®, Evry, France) according to the manufacturer’s instructions by IntegraGen. This array contains more than 240,000 highly-informative genome-wide tag SNP and over 20,000 high-value markers. Chromosome 13p, 14p, 15p, 21p and 22p markers are not represented in this array. Chromosome Y and X were only used to control the gender of the patient. All genome positions were based upon NCBIGRCh37/hg19 from UCSC Genome Bioinformatics. The genotyping data were normalized by the IntegraGen commercial platform and analyzed (copy number aberrations (CNA) and allele disequilibrium (AD)) using GenomeStudio software (version 1) (Illumina Inc).
The Cancer Genome Atlas Dataset
For external validation of our results on an independent cohort, the skin melanoma dataset of The Cancer Genome Atlas (TCGA)(Provisional) was downloaded through the cBioPortal for Cancer Genomics website (
http://www.cbioportal.org/ date March 13th 2015). We selected “All Complete Tumors (278)” of the patient/case set and entered the
BRAF gene in the TCGA (Provisional) dataset. The cBioPortal website provided the data about mutation type, amino acid change and variant allele frequency. Analysis was performed as published [
24].
Statistic analyses
The correlations between the percentage of mutated BRAF in cDNA/gDNA and in rtPCR/pyrosequencing analysis were tested by estimating the coefficient of correlation. The chi2 test supplemented when necessary in Yates correction was used to analyze the differences between different BRAF-M% groups and chromosome 7 status. The t-test was used to compare tumors with less or more than 80 % of tumor cells. The results were considered significant when P < 0.05.
Discussion
The primary objective of this study was to assess the percentage of BRAF mutated allele (BRAF-M%) in a large series of human melanoma samples. For this purpose, we first validated quantification by pyrosequencing by comparing it with 2 other methods: real-time PCR and picoliter-droplet digital PCR. We also demonstrated a close correlation of BRAF-M% quantifications in genomic DNA and messenger RNA. The best inter-pathologist reproducibility of the estimation of the percentage of tumor cells in mutated melanoma was obtained with a cut-off at 80 %. Furthermore, the mean BRAF-M% was significantly lower for cases with <80 % of tumor cells, probably due to the presence of the wild type non-tumor cells. We therefore limited the analysis to samples containing at least 80 % of tumor cells.
Using this validated quantitative method, we analyzed BRAF-M% in a series of 368 melanoma samples and found that it was very heterogeneous. We then investigated the genetic cause of this high heterogeneity by FISH and high density SNP array, and demonstrated that chromosome 7 aneuploidy was the main mechanism of unbalanced BRAF allelic ratio. Finally we showed that, as opposed to melanomas, benign melanocytic tumors with BRAF mutations did not have any chromosome 7 instability.
Although oncogenic mutations are expected to occur in only one of the two parental alleles, we found that only two thirds of
BRAF mutated melanomas were heterozygous. Decreased (<30 %) and increased (>60 %) levels of V600E mutations were detected in 14.8 % and 19 % of mutated
BRAF p.V600E melanomas, respectively. Unbalanced
BRAF mutations have been previously reported to be frequent in smaller series of melanoma [
14,
16,
25]; however quantitative methods were not clearly validated, and correlation with mRNA levels was not established. Our results were further confirmed by analysis of the TCGA database. These results raise important questions concerning BRAF targeted therapies. Indeed, ATP-competitive RAF inhibitors have different effects in cells expressing wild-type or mutant BRAF forms, with up- or down-regulation of the ERK signaling pathway in certain
in vitro conditions [
26]. Furthermore, activation of the wild-type form depends on dimerization, while monomeric mutant forms have been shown to be active [
16]. As BRAF-M% appears to be highly heterogeneous in melanomas, one could expect differences in response to ATP-competitive RAF inhibitors between tumors. Data from a recent clinical study support this hypothesis, showing that the
BRAF V600 mutation level was significantly associated with a better response rate to vemurafenib during the first 10 months of treatment [
14]. In our series, two patients with increased
BRAF V600E levels (86.6 % and 86.7 %) had a prolonged disease-free survival during respectively 25 and 39 months of BRAF p.V600 inhibitors therapy. These results emphasize the appeal of mutant
BRAF quantification assessment prior to BRAF inhibitor treatment, to correlate with clinical response rate and survival. In clinical practice, the quantification of BRAF mutations deserves to be considered as a standard item in reporting the mutational status of BRAF. Furthermore,
in vitro screening of new drugs should include cell lines with different BRAF mutant levels, thus more closely representative of BRAF-M% in patients’ melanomas. Of note, an
in vitro study showed that 4/4 melanoma cell lines with homozygous
BRAF
V600E
mutations were sensible to vemurafenib, while 3/6 with heterozygous
BRAF
V600E
mutations were resistant [
27].
Our initial hypothesis was that amplification of the BRAF locus is responsible for high BRAF-M%. However, BRAF amplification was observed in only 5.6 % of samples (n = 7), including 5 BRAF mutated cases. By contrast, we detected a polysomy of chromosome 7 in the majority of cases with high BRAF-M%. A second FISH analysis with another probe specific for the BRAF locus and SNP analysis on chromosome 7 confirmed the quality of our FISH data.
The low BRAF-M% may be related to the presence of non-tumor cells. Indeed, we observed a relationship between percentage of tumor cells and BRAF-M%. However cases with low BRAF-M% were also detected when including only samples with >80 % of tumor cells. Tumor heterogeneity of
BRAF mutated melanomas, including some areas without
BRAF mutations, have been reported by a few groups [
28,
29], and could also be responsible for low BRAF-M%. Therefore, we performed an
in situ analysis of cases with <30 % of mutant allele by immunohistochemistry with the BRAF p.V600E-specific VE1 antibody on whole slide sections. In all cases available for analysis, no negative areas were detected. Thus the main cause of low BRAF-M% is probably similar to high BRAF-M%. Unfortunately, we cannot confirm this in the present study, because only 6 cases were analyzed by FISH. Interestingly,
BRAF mutated tumors with numerous copies of chromosome 7 displayed stronger VE1 staining, suggesting a higher expression of the
BRAF mutant allele.
We report herein that only 18.4 % of melanomas had no alterations of chromosome 7. Other groups have already shown that chromosome instability is not restricted to chromosome 7. Indeed DNA copy number alterations were widely studied in both primary melanomas [
30,
31] and melanoma cell lines [
31‐
34] and frequent gains of 6p, 7, 8, 17q and 20q and losses of 9p, 10, 21q were reported. However, these studies did not quantify the amounts of the mutated BRAF allele. Two groups correlated chromosome 7 copy numbers with BRAF mutational status [
15,
35]; however BRAF-M% was evaluated on sequence electropherogram peaks. Willmore-Payne et al. detected seven cases with chromosome 7 polysomy and two with
BRAF amplification [
15]. The percentage of
BRAF-mutated and WT alleles were compared with 100 K SNP chip data for chromosome 7 by Spittle and colleagues [
25]. However, this analysis, carried out in eight melanoma cell lines, showed the preferential amplification of mutant
BRAF as a mechanism of an increased ratio of mutant/WT
BRAF.
We detected the
BRAF p.V600E mutation in 78.6 % of dermal and/or junctional melanocytic naevi. These results were obtained through immunohistochemistry with the VE1 antibody, whose specificity, sensitivity and reproducibility were demonstrated in melanomas [
17,
22,
36]. A similar frequency of
BRAF mutations in the same types of nevi has already been reported [
37]. We then analyzed chromosome 7 aneuploidy in
BRAF mutated nevi with FISH of chromosome 7. As opposed to the results obtained in melanoma, no alterations were detected in the 27 cases of nevi available, thus excluding a causal link between
BRAF mutations and chromosomal instability in melanocytic tumors. These results are in keeping with previously published data [
38].
BRAF was previously proposed to be the driver of copy number increase in melanoma [
35]. The present data do not support this hypothesis. However, association of chromosome 7 aneuploidy with malignancy was significant in
BRAF mutated melanocytic tumors (35/66 versus 0/27,
P < 0.05). Our results suggest that, among the high number of genetic alterations present in melanomas, a frequent oncogenic pathway is characterized by an early gain of function mutation in
BRAF and a late transforming mutation in another gene responsible for chromosomal instability. However, this has to be confirmed in cell models.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
VT has received honoraria from RainDance Technologies. JFE received honoraria from Roche and Glaxo Smith Kline for counseling on patients with melanomas on the diagnosis and/or treatment with BRAF inhibitors. PS received honoraria for counseling on diagnosis and/or treatment of patients with melanomas from Roche and Glaxo Smith Kline.
Authors' contributions
Conception and design: PS and JFE. Development of methodology: ZHR, LB, CLG, VT and JFE. Study supervision: ZHR, PS and JFE. Writing of the manuscript: ZHR, EFB, JFE. All authors participated in the acquisition, analysis and interpretation of data (acquired and managed patients, provided facilities, etc.). All authors read and approved the final manuscript.