|Home | About | Journals | Submit | Contact Us | Français|
G protein-coupled receptors (GPCRs), the largest human gene family, are important regulators of signaling pathways. However, knowledge of their genetic alterations is limited. In this study, we used exon capture and massively parallel sequencing methods to analyze the mutational status of 734 GPCRs in melanoma. This investigation revealed that one family member, GRM3, was frequently mutated and that one of its mutations clustered within one position. Biochemical analysis of GRM3 alterations revealed that mutant GRM3 selectively regulated the phosphorylation of MEK, leading to increased anchorage-independent growth and migration. Melanoma cells expressing mutant GRM3 had reduced cell growth and cellular migration after short hairpin RNA–mediated knockdown of GRM3 or treatment with a selective MEK inhibitor, AZD-6244, which is currently being used in phase 2 clinical trials. Our study yields the most comprehensive map of genetic alterations in the GPCR gene family.
Melanoma is a common form of skin cancer and, despite years of research, metastatic disease is often fatal1. Although some targeted treatment has been effective2, the results of these treatments are not long lasting, showing a need for new clinically relevant targets. GPCRs, which are activated by ligands, are integral in regulating various signaling pathways3,4. The importance these molecules play in human diseases is evident by the fact that 50–60% of the US Food and Drug Administration–approved therapeutics target GPCRs5. As GPCRs regulate pathways that affect cell growth, their genetic analysis in cancer is warranted.
To identify somatic mutations in GPCRs, we performed a GPCR-targeted mutational analysis in tumor DNA derived from 11 melanoma samples. For each sample, we performed DNA capture using molecular inversion probes of 7,059 regions corresponding to 2,400 GPCR exons encoding 734 GPCR genes6. We then analyzed DNA samples enriched for GPCR exons by massively parallel sequencing using the Illumina GAII platform. We identified 755 potential non-synonymous mutations when comparing the molecular inversion probe results to the known HapMap sequence. To determine which alterations were somatic, we examined the genomic DNA derived from matched normal tissue. From these alterations, we confirmed 106 to be somatic in 94 different genes (Supplementary Table 1). Eleven of these genes harbored at least two somatic mutations (Supplementary Table 2). The coding sequence of these 11 genes was analyzed for somatic mutations in a total of 80 melanoma samples, as described previously7. Supplementary Figure 1 depicts the stages described above.
From the ~3.4 Mb of sequencing information obtained, we identified 115 non-synonymous, somatic mutations in 42 of the 80 tumors (Table 1). The number of C>T mutations was significantly greater than the number of other substitutions, resulting in a high prevalence of C:G>T:A transitions (P < 1 × 10−6) (Supplementary Fig. 2), consistent with melanoma mutation signatures8. We found a total of seven nonsense mutations and two splice site alterations, potentially resulting in aberrant or truncated proteins for five of the genes. We identified recurring alterations in CHRM3, RXFP1, OR8B8 and OR1J2, which harbored p.Pro421Leu/p.Pro421Ser, p.Ser269Phe, p.Ser74Phe and p.Asp109Asn, respectively, in two individuals. Using SIFT (sorting intolerant from tolerant)9, we determined a computational estimation of the effects of the mutations (Supplementary Table 3). A description of these genes and the likely nature of their mutations is given in the Supplementary Note.
GRM3 and GPR98 were the most frequently mutated genes in our screen. GRM3 had a 16.3% mutation rate, with 18 non-synonymous mutations in 13 of 80 tumors. GPR98 had a total of 42 non-synonymous mutations in 22 of 80 tumors (27.5%). As GRM3 belongs to the metabotropic family and was previously shown to be linked to tumorigenesis10, it was evaluated genetically in an additional panel consisting of 57 melanoma specimens11. In this screen, we discovered 11 non-synonymous alterations in nine tumors affecting 15.7% of the individuals analyzed and yielding a non-synonymous to synonymous ratio of 29:7, which is significantly higher than the non-synonymous to synonymous ratio of 2:1 predicted for nonselected passenger mutations (P < 0.05). This investigation allowed the identification of a mutational hotspot in GRM3; we found p.Glu870Lys in four different individuals with melanoma (one individual from the original panel and three from the second panel) (Supplementary Table 4 and Supplementary Fig. 3). The likelihood for the occurrence of four identical mutations is approximately 1.8 × 10−12, suggesting that the GRM3 hotspot mutation is functionally important. Clinical information for all tumors harboring GPCR mutations is given in Supplementary Table 5.
We focused on GRM3 (the group II metabotropic glutamate receptor-3 gene, or mGluR3), as two genetic observations suggested that the mutations in this gene may be functionally important for melanoma tumori-genesis: (i) GRM3 was one of the most highly mutated genes in the screen, and (ii) it contained a mini-hotspot (p.Glu870Lys).
We functionally characterized four somatic mutations (resulting in p.Gly561Glu, p.Ser610Leu, p.Glu767Lys and p.Glu870Lys) discovered in GRM3 based on sequence conservation and their location within particular functional domains. To examine the biological effects of GRM3 mutations, we established stable pooled clones expressing either a vector control or wild-type or mutant (p.Gly561Glu, p.Ser610Leu, p.Glu767Lys or p.Glu870Lys) GRM3. We selected the melanoma cell lines A375 or Mel-STR, as they express wild-type GRM3. We saw similar levels of expression of GRM3 protein in both the A375 and Mel-STR stable clone cell lines, except for in the p.Glu767Lys and p.Glu870Lys mutants, which had reduced protein expression (Supplementary Fig. 4a,b).
To examine the effects of GRM3 mutations on cell growth, we investigated growth rate on plastic (Fig. 1a). In the presence of media containing 10% serum, all clones grew similarly (Supplementary Fig. 5a,b). However, if we reduced the serum concentration, wild-type clones grew at a lower rate than mutant clones, except for the clone expressing the p.Glu870Lys alteration (Fig. 1a and Supplementary Fig. 5c). We also observed this difference in cell growth when we assessed the cells for anchorage independence, where cells expressing mutant GRM3 formed a significantly higher number of colonies compared to wild type or empty vector (Fig. 1b; P < 0.05 t-test).
The C-terminal region of GPCR proteins is important for binding signaling molecules involved in pathways such as the RAS-RAF-MEK pathway3,4. Mutations in or near this region of GRM3 may therefore affect signal transduction leading to increased cell proliferation, thus providing the biochemical basis for the growth differences described above. As group 2 metabotropic glutamate receptors can be activated by agonists such as DCG-IV12–14, we tested the biochemical effects of GRM3 alterations in the presence and absence of DCG–IV. We saw striking differences in MEK1/2 phosphorylation. When we stimulated Mel-STR clones with DCG-IV, there was a sevenfold to tenfold increased phosphorylation of MEK1/2 compared to wild-type-GRM3–expressing cells (Fig. 1c). We observed similar results in the A375 clones (Fig. 1d). Notably, Mel-STR cells harbor mutant RAS and A375 cells harbor mutant BRAF, both of which are known to activate MEK15, and so mutant GRM3 thus allowed further activation of the MEK pathway.
Previous studies reported that activation of the MEK pathway increases cell migration16,17. As GRM3 variants activate the MEK pathway, we determined whether these variants also affect migration. To test this, we seeded A375 or Mel-STR pooled clones in serum-free medium in the presence or absence of DCG-IV and looked for migration (Fig. 2a,b). Mutant GRM3 expression increased migration compared to wild-type GRM3 or an empty vector containing cells in the absence of agonist (Fig. 2a,b; P < 0.05 t-test). Upon stimulation with DCG-IV, vector and wild-type–expressing cells migrated similarly to mutant-GRM3–expressing cells in the absence of stimuli (Fig. 2a,b). Our results suggest that expression of mutant GRM3 increases migration in the absence of growth factors or receptor agonists.
To determine whether these phenotypes occur in vivo, A375 pooled clones expressing vector, wild-type or mutant GRM3 were administered to NOD/SCID mice by tail vein injection. Nine weeks after injection, macroscopic assessment of lung colonization showed that the groups injected with cells expressing vector, wild-type or mutant p.Gly561Glu had two to three mice with gross lung tumors. In contrast, most of the mice injected with cells expressing the p.Ser610Leu, p.Glu767Lys or p.Glu870Lys GRM3 alterations had pulmonary macrometastases (Fig. 2c). Microscopic examination allowed for the detection of micrometastases, with no significant difference being seen in their number or size (Supplementary Fig. 6a,b). Thus, expression of mutant forms of GRM3 in melanoma cells affects growth in vivo once the lung is colonized.
To assess if melanoma cells with endogenous GRM3 mutations are dependent on GRM3 signaling for proliferation and migration, we used short hairpin RNA (shRNA) to stably knock down GRM3 protein levels in melanoma cells that harbor either wild-type GRM3 or endogenous mutant GRM3. We confirmed specific targeting of GRM3 by transient transfection in HEK293T cells and immunoblotting, as well as by quantitative RT-PCR analysis (Fig. 3a–c). The shRNA had little effect on cells harboring wild-type GRM3 but significantly reduced the growth of cells harboring mutant forms of GRM3 (Fig. 3d–h and Supplementary Fig. 7a,b; P < 0.005, P < 0.008 and P < 0.004 for Fig. 3f,g,h, respectively). Depletion of GRM3 by shRNA in mutant-GRM3–expressing cells reduced their ability to migrate significantly compared to wild-type GRM3 cells targeted with GRM3 shRNA (Fig. 3i–m and Supplementary Fig. 7c,d). To determine whether similar results occur in vivo, melanoma cells harboring either wild-type or mutant GRM3 targeted with GRM3 or control shRNA were administered into Nu/Nu mice by subcutaneous injection. Nineteen days after injection, depletion of GRM3 had little effect on in vivo growth of cells harboring wild-type GRM3. In contrast, GRM3 knockdown significantly reduced the tumor growth induced by cells harboring mutant GRM3 (Fig. 3n–q; P < 0.0005 and P < 0.02 for Fig. 3o,q, respectively). As the shRNA-mediated phenotypes could be caused by specific or nonspecific effects, we engineered an exogenous, non-targetable wild-type GRM3 construct that harbors silent mutations in the region of GRM3 targeted by shRNA #3 to rescue the effects of knockdown of endogenous GRM3. Melanoma cells harboring the p.Glu573Lys alterations stably expressing either control or GRM3 shRNA #3 construct were transduced either with the lentiviral non-targetable GRM3 construct or with the empty vector as a control. To show that the non-targetable GRM3 is not knocked down in the presence of GRM3 shRNA #3, we transiently transfected HEK293T cells and immunoblotted for FLAG-GRM3 and GAPDH as a loading control (Supplementary Fig. 8a). Importantly, non-targetable GRM3-reconstituted cells showed significantly more migration than cells infected with the control vector (Supplementary Fig. 8b). These results suggest that certain GRM3 mutations are essential for cellular proliferation as well as for cell migration in melanoma cells.
To evaluate whether inhibition of GRM3 signaling would result in a similar phenotype to depletion of endogenous GRM3, we exposed melanoma cells harboring either wild-type GRM3 or mutant GRM3 to AZD-6244 (Selumetinib, ARRY-142886), which is a selective, non-ATP–competitive small molecule inhibitor of MEK1/2 that is being tested in phase 2 clinical trials (see URLs). Exposure of melanoma cells to AZD-6244 inhibited MEK in mutant as well as in wild-type cells (Fig. 4a). However, the relative degree of inhibition was concentration dependent, resulting in greater inhibition of the mutant cells compared to wild-type cells. This suggested that mutant cells would be more sensitive to growth inhibition by AZD-6244 than wild-type cells. Indeed, exposure of most melanoma cells expressing mutant forms of GRM3 were 2–200-fold more sensitive to AZD-6244 compared to wild-type GRM3 cells (Fig. 4b and Table 2). The GRM3 mutations may be dependent on MEK signaling via BRAF, as evidenced by the genotypes of the investigated cells (Supplementary Table 6). To further investigate the relevance of mutant GRM3 to AZD-6244 sensitivity, we tested whether the sensitivity to MEK inhibition can be altered by modulating GRM3. To do this, we established a stable cell line overexpressing wild-type GRM3 in a mutant GRM3 background (Supplementary Fig. 9a). Exposure of these stable pools to AZD-6244 resulted in reduced cell proliferation, with a fourfold increased resistance in cells overexpressing wild-type GRM3 compared to control infected cells (Supplementary Fig. 9b,c). These results further suggest that AZD-6244 preferentially inhibits the signaling of cells expressing mutant GRM3.
The decreased growth of the mutant cells in the presence of AZD-6244 could have arisen either through alteration of the cell cycle or through increased cell death. To distinguish between these two possibilities, we performed FACS analysis and found that cells harboring mutant GRM3 showed a substantial increase in the levels of subG1-population cells (apoptotic cells) compared to wild-type cells (Fig. 4c). We observed similar results when we performed the experiment on a larger panel of mutant GRM3 cell lines (Fig. 4d; P < 0.05 t-test). We confirmed the apoptotic events by protein blot analysis of the cell lysates analyzed by FACS, which showed increased levels of cleaved PARP in mutant cells compared to wild-type cells (Fig. 4e). Thus, melanoma cell lines harboring mutant GRM3 are markedly more sensitive to MEK1/2 inhibition by AZD-6244, leading to increased cell death.
In this study, we used a systematic approach combining exon capture and massively parallel sequencing to genetically characterize the GPCR gene family, allowing for the identification of GRM3 somatic mutations in melanoma. The high frequency of mutations found in GRM3, the finding of a mutational hotspot, as well as the functional assays, suggest GRM3 to be a driver in melanoma.
Taken together, our data highlight a model for melanoma pathogenesis in which activation of MEK by GRM3 alterations promotes the proliferation and migration of melanoma cells. Although further investigation into the mechanism of GRM3 activation of the MEK pathway is required, our study suggests that the presence of GRM3 mutations is expected to indicate subpopulations of individuals whose tumors are dependent on MEK signaling. The prior failure of MEK inhibitors to obtain significant tumor responses in many BRAF p.Val600Glu melanomas18,19 may have resulted at least in part from the absence of additional mutations that activate the MEK pathway, such as those in GRM3. Therefore, targeting MEK signaling in the presence of GRM3 mutations may have a role in the treatment of melanoma.
Tissue and melanoma cell lines used for the discovery and first validation in this study were described previously20.
For the validation set of melanomas, optimum cutting temperature– embedded frozen clinical specimens were obtained from the University of Texas MD Anderson Cancer Center under institutional review board–approved protocols. Hematoxylin and eosin–guided dissection and isolation of DNA from the tumor-enriched isolates has been described previously11.
PCR and sequencing was done as previously described7,20,21. The primers used for PCR amplification are listed in Supplementary Table 8. The primary phase mutation screen was analyzed using Consed (see URLs). Variants were called using Polyphred 6.11 and DIPDetector (N. Hansen, unpublished data), an indel detector for improved sensitivity in finding insertions and deletions. Sequence traces of the secondary screen were analyzed using the Mutation Surveyor software package (SoftGenetics).
Human GRM3 (NM_000840) was cloned by PCR into pCDF1 or pcDNA3.1 as previously described7,20 using a clone (cat # MHS1010-9203778) purchased from Open Biosystems and with the primers listed in Supplementary Table 8. The p.Gly561Glu, p.Ser610Leu, p.Glu767Lys and p.Glu870Lys point mutants were made using Phusion PCR for site-directed mutagenesis. Amplification of a non-targetable insert using the primers listed in Supplementary Table 9 was done as previously described7,20 and cloned into pCDF1 vector.
Metastatic melanoma tumor, HEK 293T, A375, Mel-STR cell lines were maintained as previously described7,20. HEK 293T cells were transfected with Lipfectamine2000 reagent (Invitrogen) at a 6:1 ratio with DNA (μl:μg) using 3–5 μg of plasmid DNA.
Production of GRM3 lentivirus was done as previously described7. A375 and Mel-STR cells were seeded at 1.5 × 106 per T75 flask 24 h before infection. Lentivirus for GRM3 (wild type, p.Gly561Glu, p.Ser610Leu, p.Glu767Lys and p.Glu870Lys) or empty vector control were used to infect cells as previously described7. A375 or Mel-STR pooled clones were selected in the presence of 3 μg/ml puromycin or 300 μg/ml G418 for several days, respectively. Stable expression of GRM3 proteins (wild type and mutants) was determined by SDS-PAGE analysis followed by immunoblotting with anti-FLAG and anti–α-tubulin to show equivalent expression among the pools.
A375 or Mel-STR pooled GRM3 clones were seeded in 6-well plates at 1–2 × 105 cells per well the day before transfection. Cells were serum starved for 16 h before stimulation with 2.5 μM DCG-IV (Tocris Biosciences-Ellisville) for 10 min. Cell lysates were generated by direct lysis into 2× SDS sample buffer and then subsequently analyzed by protein blotting using anti–phopho-MEK1/2 (cat # 9154), anti-MEK1/2 (cat # 9126), anti-GAPDH (cat # CB 1001) (Millipore-Billerica) or anti–α-tubulin (cat # CP06) (EMD-Calbiochem-Gibbstown) as a loading control.
To examine growth potential, pooled A375 and Mel-STR GRM3 clones were seeded into 96-well plates at 250 cells per well in either 1%, 2.5% or 10% serum-containing medium and incubated for 13–17 days. Samples were analyzed as previously described7.
Mel-STR pooled GRM3 clones were plated in triplicate at 1,000 cells per well and in top plugs consisting of sterile 0.33% Bacto-Agar (Becton, Dickinson and Company) and 10% FBS (HyClone) in a 24-well plate. The lower plug contained sterile 0.5% Bacto-Agar and 10% FBS. After two weeks, the colonies were photographed and quantitated using ImageJ (NIH software).
Mel-STR or A375 pooled clones were seeded at 30,000 cells per well in serum-free medium or serum-free medium plus 2.5 μM DCG-IV and incubated for 16–18 h before harvesting. Inserts were analyzed as previously described20.
Constructs for stable depletion of GRM3 (cat # RHS4533-NM_000840) were obtained from Open Biosystems and were confirmed to efficiently knockdown GRM3 at the protein level. Lentiviral stocks were prepared as previously described20. Melanoma cell lines (31T, 34T, 36T, 63T and 76T) were infected with shRNA lentiviruses for each condition (vector and two different GRM3-specific shRNAs). Selection of stable pooled clones was done in the presence of 3 μg/ml puromycin containing normal medium for 3–5 days before determining knockdown efficiency. Stably infected pooled clones were tested in functional assays. For the shRNA, inducible system constructs for stable depletion of GRM3 (cat # RHS4740-NM_000840) were obtained from Open Biosystems and were confirmed to efficiently knockdown GRM3 at the message level. Lentiviral stocks were prepared as previously described20. Melanoma cell lines (31T or 36T) were infected with shRNA lentiviruses for each condition (vector and two different GRM3-specific shRNAs). Selection and growth were done as described above. Stably infected pooled clones were tested in functional assays. Sequences of stable shRNA knockdown plasmids for GRM3 are listed in Supplementary Table 10. Doxycycline-inducible shRNA knockdown clones were grown in the presence of 0.2 μg/ml of doxy-cycline for 7 days before any functional assays. The clones were kept in the presence of 0.2 μg/ml doxycycline throughout the experiment.
To rescue shRNA-mediated knockdown of GRM3 in melanoma cell lines, the non-targetable GRM3-expressing lentivirus was made as described above and used to infect the melanoma cell line 63T. After infection, cells were given 96–120 h to recover from infection before being tested in functional assays.
Total RNA was extracted from pooled clones of melanoma cells 36T and 34T stably knocked down for endogenous GRM3 or from the doxycycline-induced clones 31T and 63T targeting GRM3 (in the presence of 0.2 μg/ml doxycyline for 7 days) following the manufacturer’s protocol for the RNeasy Mini Kit (QIAGEN #74101). Total RNA was eluted in 30 μl diethylpyrocarbonate (DEPC)-treated distilled H2O. A total of 1 μg of total RNA was used for single-strand complementary DNA (cDNA) synthesis using a SuperScript III First Strand kit (Invitrogen #18080-051). cDNA was amplified using the olido dT20 primer supplied in the kit. To test for loss of GRM3 message, we used 2 μl of cDNA in the PCR with either GRM3 primers or GAPDH primers (Supplementary Table 11) mixed with 2× Fast SYBR Green PCR mix at a final volume of 20 μl in triplicate (Applied Biosystems cat # 4355612). qRT-PCR analysis was done using the ABI 7900HT Fast Real-Time PCR system (with a standard program of stage 1: 50 °C for 2 min; stage 2: 95 °C for 10 min; stage 3: 40 cycles of 95 °C for 15 s and 60 °C for 1 min). Results were analyzed using Microsoft Excel.
NOD/SCID mice were purchased from Jackson Labs. All mice were housed in a pathogen-free facility and were given autoclaved food and water. A375 pooled clones with empty vector or wild-type GRM3 or mutant GRM3 were grown in T-175 flasks to 70–80% confluency. We resuspended 1 × 106 cells in 100 μl of sterile PBS and injected them intravenously (in the tail) into 7–10-week-old NOD/SCID mice. Mice were monitored biweekly, and lungs were examined and quantified for tumor formation by excision from mice killed at day 60 (9 weeks) after injection.
Nu/Nu mice were purchased from Charles River Labs. All mice were housed in a pathogen-free facility and were given autoclaved food and water. 31T (pLKO.1, shRNA #1 or shRNA #3) or 36T (pLKO.1, shRNA #1 or shRNA #3) cell lines were grown in 4–6 T-175 flasks, each to 90% confluency. We resuspended 2.5 × 106 cells in 200 μl of sterile 1× PBS and matrigel solution (1:1 v/v) (BD Biosciences #354234) and injected them subcutaneously into 6–8-week-old female Nu/Nu mice. Mice were monitored biweekly, and tumor diameters were measured using precision calipers for 17–20 days. For the shRNA inducible system, all mice were given sterile doxycycline food (5 g/mouse/day) and water. 31T (TRIPz-NC, sh639 or sh742) or 63T (TRIPz-NC, sh639 or sh742) cell lines were grown in 4–6 T-175 flasks, each to 90% confluency. We resuspended 2.5 × 106 cells in 200 μl of sterile 1× PBS and matrigel solution (1:1 v/v) (BD Biosciences #354234) and injected subcutaneously into 6–8-week-old female Nu/Nu mice. Mice were monitored biweekly, and tumor diameters were measured using precision calipers for 17–20 days.
To test AZD-6244 (cat # S1008) (Selleck Chemicals LLC) inhibition on melanoma cell lines, we seeded 96-well plates at 2,000–5,000 cells per well and incubated them for 24 h before addition of inhibitor at concentrations from 2 nM to 30 μM. Once the inhibitor was added, cells were incubated for 72 h at 37 °C. Cells were then analyzed using CellTiter-Glo according to the manufacturer’s protocol (cat# G7571) (Promega). Plates were read on a Luminoskan Ascent (Thermo Scientific) plate reader and analyzed using Microsoft Excel and GraphPad Prism v5.
Melanoma cells were seeded into T-25 flasks at densities of 3 × 105 cells per flask in normal complete medium and incubated at 37 °C for 24 h before addition of AZD-6244. AZD-6244 or DMSO was added for 72 h at a concentration of 2 μM. Cells were then analyzed by FACS as previously described7.
Statistical analyses were performed using the R statistical package (see URLs) and Microsoft Excel.
We thank S. Gutkind, J. Cronin, H. Abaan, P. Cruz, J. Mullikin, N. Hansen and members of the US National Institutes of Health Intramural Sequencing Center Comparative Sequencing Program for generating the sequence data analyzed here. We thank S. Hoogstraten-Miller and I. Ginty for assistance with the mouse experiments and S. Anderson for assistance with FACS analysis. This work was supported by the Intramural Research Programs of the National Human Genome Research Institute and National Cancer Institute, US National Institutes of Health, USA. Grant support was also provided by the University of Texas MD Anderson Cancer Center Melanoma Informatics, Tissue Resource and Pathology Core and the Melanoma Specialized Programs of Research Excellence (P50 CA93459). M.A.D. is supported by funding from the Melanoma Research Alliance, the American Society of Clinical Oncology and the MD Anderson Physician-Scientist Program. A.M. is supported by the Intramural Research program of the National Institute of Dental and Craniofacial Research, NIH.
URLs. Short Read Archive (SRA), http://trace.ncbi.nlm.nih.gov/Traces/sra/; phase 2 clinical trials, http://clinicaltrials.gov/; Consed, http://www.phrap.org/consed/consed.html; R, http://www.r-project.org/; ImageJ, http://rsbweb.nih.gov/ij/.
Accession codes. Accession codes: GRM3, CCDS5600.1; CHRM3, CCDS1616.1; LPHN2, CCDS1616.1; RXFP1, CCDS43276.1; GRM8, CCDS5794.1; CNR1, CCDS5015.1; OR1J2, CCDS35121.1; OR8B8, CCDS8446.1; OPN5, CCDS4923.1; OR8K1, CCDS31528.1; GPR98, CCDS47246.1; MEK, CCDS10216.1; NRAS, CCDS877.1; BRAF, CCDS5863.1; ERK, CCDS10672.1. Somatic variants are listed in Supplementary Table 2 and are deposited to the SRA (see URLs) with accession number SRA024490 and study number SRP003752.
Note: Supplementary information is available on the Nature Genetics website.
AUTHOR CONTRIBUTIONST.D.P., I.C.-N., X.W., E.H.M. and Y.S. designed the study. K.S.-H., M.A.D., J.E.G. and S.A.R. collected and analyzed the melanoma samples. X.W., I.C.-N., J.K.T., J.G., P.F.C. and J.C.L. analyzed the genetic data. T.D.P., A.M., J.J. and V.W. produced and analyzed the functional data. All authors contributed to the final version of the paper.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Reprints and permissions information is available online at http://www.nature.com/reprints/index.html.