Published in Volume
124, Issue 4
(April 1, 2014)J Clin Invest.
Copyright © 2014, American Society for Clinical Investigation
Oncogenic and sorafenib-sensitive ARAF mutations in lung adenocarcinoma
1Molecular Pathology Unit, Massachusetts General Hospital, Harvard Medical School, Charlestown, Massachusetts, USA.
2Cancer Program, Broad Institute of Harvard and MIT, Cambridge, Massachusetts, USA.
3Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts, USA.
4Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts, USA.
5James Thoracic Center, The Ohio State University Wexner Medical Center, Columbus, Ohio, USA.
6Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, Tennessee, USA.
7Department of Medicine, University of Texas Southwestern Medical Center, Dallas, Texas, USA.
8Department of Pathology, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts, USA.
Address correspondence to: David P. Carbone, James Thoracic Center, Department of Medicine, The Ohio State University Medical Center, 410 W. 10th Ave., Columbus, Ohio 43210, USA.Phone: 614.685.4478; Fax: 614.293.4372; E-mail:
firstname.lastname@example.org. Or to: Matthew L. Meyerson, Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, 450 Brookline Ave., Boston, Massachusetts 02115, USA.Phone: 617.632.4768; Fax: 617.582.7880; E-mail:
Authorship note: Marcin Imielinski and Heidi Greulich contributed equally to this work. Matthew Meyerson and David P. Carbone are co–senior authors.
First published February 24, 2014
Submitted: August 19,
2013; Accepted: December 17,
Targeted cancer therapies often induce “outlier” responses in molecularly defined patient subsets. One patient with advanced-stage lung adenocarcinoma, who was treated with oral sorafenib, demonstrated a near-complete clinical and radiographic remission for 5 years. Whole-genome sequencing and RNA sequencing of primary tumor and normal samples from this patient identified a somatic mutation, ARAF S214C, present in the cancer genome and expressed at high levels. Additional mutations affecting this residue of ARAF and a nearby residue in the related kinase RAF1 were demonstrated across 1% of an independent cohort of lung adenocarcinoma cases. The ARAF mutations were shown to transform immortalized human airway epithelial cells in a sorafenib-sensitive manner. These results suggest that mutant ARAF is an oncogenic driver in lung adenocarcinoma and an indicator of sorafenib response.
Lung adenocarcinomas harbor recurrent activating oncogenic mutations and fusions in receptor tyrosine kinase pathway genes, some of which (EGFR, EML4-ALK, CD74-ROS1) have been associated with clinical response to small-molecule inhibition (1–3). Despite these advances, most lung adenocarcinoma cases lack a clinically actionable genetic alteration and over 50% lack a plausible oncogenic “driver,” as demonstrated by recent large-scale genome surveys (3). One approach for the discovery of clinically actionable drivers is genomic analysis of exceptional drug responses (4). We used next-generation sequencing to investigate the genetic basis of a sustained “outlier” response to sorafenib in lung adenocarcinoma.
Results and Discussion
A 66-year-old light former smoker (<5 packs per year smoking history) was diagnosed in April 2002 with stage IV lung adenocarcinoma. She failed multiple therapy regimens (gemcitabine and vinorelbine, gefitinib, bortezomib) between 2002 and 2005 and received a palliative lobectomy in early 2006 for worsening hypoxia. She began treatment with oral sorafenib, a broad-spectrum kinase inhibitor with activity against BRAF, RAF1, RET, PDGFRA, and KIT, among others (5, 6), in June 2006 as part of the ECOG 2501 trial (7). Within 2 months, her CT scans demonstrated a near-complete response (Figure 1). She remained progression free and asymptomatic for the next 5 years while continuing sorafenib treatment. In July 2011, a CT scan demonstrated enlargement of a right lower lobe mass meeting Response Evaluation Criteria In Solid Tumors criteria for progression. Sorafenib was discontinued, and she was started on carboplatin, paclitaxel, and bevacizumab. Therapy was discontinued after 2 cycles due to side effects, worsening fatigue, and oxygen requirements. She was admitted to hospice and died in November 2011. At the time of relapse, she was the last remaining study participant receiving sorafenib and 1 of only 9 responders among 306 evaluable patients. A more detailed time line of her case is shown in Figure 1 and described in the Supplemental Methods (supplemental material available online with this article; doi:
Time line of patient’s lung adenocarcinoma diagnosis, treatment, and response. Colored rectangles near the time line represent durations of targeted therapy (blue, green) and chemotherapy (pink). Original magnification, ×200 (Apr. 2002). Dx, diagnosis; RLL, right left lobe; RML, right middle lobe.
We used massively parallel DNA sequencing of this patient’s tumor resection and peripheral blood samples before sorafenib treatment (Figure 1) to determine possible genetic alterations underlying her sustained sorafenib response. Whole-genome sequencing (WGS) of primary tumor (37.9X) and normal (37.7X) tissue revealed 25,150 somatic mutations (8.7 mutations per Mb), including 101 nonsynonymous mutations affecting the coding regions of 99 genes. The spectrum of somatic DNA variants (with respect to mutation, rearrangement, and copy number alteration) was consistent with that of other lung adenocarcinomas profiled in large-scale genome surveys (refs. 3, 8–11; Supplemental Figures 1 and 2; Supplemental Tables 1–3; and Supplemental Results).
The genome sequence data were notable for the absence of hot spot mutations in KRAS, EGFR, BRAF, ERBB2, or PIK3CA; gene fusions involving ALK, ROS1, or RET (a sorafenib target); and any other known oncogenic alterations (Supplemental Results). WGS analysis revealed numerous low-level (1–4 copy) broad DNA gains involving sorafenib targets (Supplemental Figure 1), without high-level focal amplifications. Among these was an approximately 60-Mb alteration, involving 300 genes on chromosome 4 and including 1–2 copy gains of canonical sorafenib targets PDGFRA, KIT, and KDR on 4q12 (6). 4q12 gain has been observed previously in non–small-cell lung cancer lines, in which it was not associated with sensitivity to tyrosine kinase inhibition (imatinib, sunitinib) (12). Germ line variant analysis of peripheral blood WGS data did not reveal any rare deleterious germ line coding mutations in 29 known sorafenib target genes (ref. 5 and Supplemental Table 4).
Among 101 somatic coding mutations and 2 in-frame fusions predicted by WGS analysis, only 15 variants were detected in RNA sequencing (RNA-seq) data with more than 2 supporting reads (Table 1). Among expressed coding variants, the most likely candidate oncogenic driver was ARAF S214C. ARAF encodes a serine-threonine kinase in the Raf protein family, which also includes RAF1 (also known as CRAF) and BRAF, a lung adenocarcinoma and melanoma oncogene. Like other Raf family proteins, ARAF transduces MAP kinase pathway signals from Ras to MEK and ERK and is a sorafenib target; however, unlike its paralogs, ARAF has never been implicated in tumorigenesis (6, 13).
Summary of genes harboring expressed nonsilent somatic coding variants
ARAF S214C was detected in 11 of 23 tumor and 0 of 18 normal DNA reads and 48 of 56 tumor RNA reads (Table 1 and Figure 2A). ARAF was the second most highly expressed sorafenib target gene in our analysis (after KIT) and the only one harboring somatic sequence alterations (Supplemental Table 4). We identified 3 ARAF codon 214 somatic mutant cases across 564 lung adenocarcinomas profiled by The Cancer Genome Atlas (TCGA) and Imielinski et al. (3) and 3 additional cases with somatic mutations in a paralogous (p.S259) or nearly paralogous residue (p.S257) in the related gene RAF1 (Figure 2B). Of these 6 lung adenocarcinoma cases, 4 lacked somatic mutations in any known lung adenocarcinoma oncogenes; 1 RAF1 p.S257W mutant tumor harbored a KRAS G12V mutation, and a 1 RAF1 p.S259F mutant tumor harbored a BRAF p.G469V somatic mutation (Supplemental Figure 3). Additional mutations in this putative hot spot were discovered among 21 tumor types (4,608 cases) profiled by TCGA, including colorectal adenocarcinoma (4 of 217 cases), gastric adenocarcinoma (2 of 264 cases), and cutaneous melanoma (3 of 269 cases) (Figure 2B and Supplemental Table 5). RAF1 p.S257 and p.S259 mutations have also been reported in ovarian cancer via the COSMIC database (
http://cancer.sanger.ac.uk/cancergenome/projects/cosmic/) and Noonan syndrome, a congenital disorder associated with constitutive Ras pathway signaling (14).
ARAF mutation found in a sorafenib-responsive lung adenocarcinoma defines a novel Raf family somatic mutation hot spot. (A) Read alignments in tumor DNA (TD, pink) and normal DNA (ND, green) and tumor RNA (TR, aqua), supporting ARAF S214C somatic mutation call (variant bases: brown, G; green, A; blue, C; red, T) in sorafenib-responder case. (B) Aligned ARAF, RAF1, and BRAF protein domain models overlaid with publicly available (
http://cancergenome.nih.gov/), published (3), and sorafenib-responder somatic mutation data (SR-12), colored by tumor of origin. Red labels denote the putative RAF1/ARAF hot spot. BRAF p.V600E mutations are excluded for simplicity. luad, lung adenocarcinoma; skcm, cutaneous melanoma; coad, colorectal adenocarcinoma; stad, gastric adenocarcinoma; hnsc, head and neck squamous cell carcinoma; kirp, kidney renal papillary cell carcinoma; blca, bladder carcinoma; read, rectal adenocarcinoma; lgg, lower-grade glioma; thca, thyroid carcinoma; ucec, uterine corpus endometrial carcinoma; lusc, lung squamous cell carcinoma; gbm, glioblastoma multiforme; brca, breast adenocarcinoma; prad, prostate adenocarcinoma; cesc, cervical squamous cell carcinoma; kirc:kirp, kidney renal clear cell carcinoma; ov, ovarian carcinoma.
To examine the functional impact of the ARAF and RAF1 mutations observed in lung adenocarcinoma, we ectopically expressed mutant alleles in immortalized tracheobronchial epithelial (AALE) cell lines. Ectopic expression of all 3 ARAF p.S214 lung adenocarcinoma variants substantially enhanced soft agar colony formation (Figure 3, A and C) and phospho-MEK levels (Figure 3B) relative to vector control and a kinase-dead (D429A) variant. Sorafenib treatment inhibited ARAF-induced AALE soft agar colony formation (Figure 3, C and D) and MEK phosphorylation (Supplemental Figure 3A) at concentrations (IC50 1.0–1.3 μM, 95% CI 0.7–1.6) lower than those achievable in serum (6 μM) using standard oral sorafenib doses (ref. 15 and Supplemental Table 6). Because MEK1 and MEK2 are downstream substrates of ARAF in the MAP kinase signaling pathway (13), we additionally tested trametinib, a MEK1 and MEK2 inhibitor (16). Trametinib also inhibited colony formation and decreased ERK phosphorylation in cells expressing the oncogenic ARAF mutants (IC50 1–2 nM) (Supplemental Figure 4, B–D, and Supplemental Table 6). Similar to the ARAF variants, RAF1 p.S257L and p.S259A induced anchorage-independent growth and increased MEK/ERK phosphorylation in AALE and NIH-3T3 cells (Supplemental Figure 5) in a sorafenib- and trametinib-sensitive manner (Supplemental Figure 6).
ARAF p.S214 mutations found in lung adenocarcinoma are oncogenic and sensitive to sorafenib.
(A) Soft agar colony formation by AALE cells expressing variants of ARAF. pBp, empty vector; S214C and S214F, ARAF p.S214 variants; D429A, kinase-dead ARAF. (B) Immunoblot indicating protein levels and phosphorylation status of MEK. (C) Response of ARAF mutant AALE soft agar colony formation to the indicated concentrations of sorafenib. (D) Dose-response curves for data shown in C. Data and error bars in A, C, and D represent mean ± SEM, respectively, obtained from triplicate experiments.
In summary, our genomic and functional results suggest the transforming ARAF codon 214 substitution as the most likely driver of this patient’s tumor and determinant of its sorafenib response. The absence of known recurrent oncogenic alterations or alterations in other sorafenib targets (Supplemental Table 4) in our comprehensive profiling data supports this conclusion. Moreover, the discovery of additional patients with transforming ARAF codon 214 and RAF1 codon 257 and 259 mutations in independent lung adenocarcinoma and pan-cancer data sets suggests somatic selection for a novel oncogenic hot spot in ARAF and RAF1, both which encode sorafenib targets (6). Inhibition of colony formation by sorafenib or trametinib in cells overexpressing wild-type ARAF or RAF1 suggests that sorafenib and trametinib do not exhibit increased activity toward the mutant gene products, but rather that ARAF and RAF1 mutations may confer an inhibitor-sensitive oncogene dependency. This is consistent with the current understanding of the mechanisms underlying the specificity of established targeted therapies (e.g., imatinib) (17). To the best of our knowledge, the ARAF and RAF1 mutations profiled in this study have not been characterized previously as oncogenic somatic mutation hot spots in clinical cancer samples. Interestingly, one group has recently associated derived RAF1 p.S257P mutations with in vitro PLX-4720 resistance in BRAF p.V600E mutant melanoma cells (18); however, all of the TCGA samples from patients with lung cancer analyzed in the current study were treatment naive.
The ARAF/RAF1 mutations characterized in this study lie in a Raf CR2-domain phosphorylation site that negatively regulates Ras binding and RAF1 activation via binding of 14-3-3 (13, 19). This region is distinct from the kinase domain hot spots of BRAF (near p.G469 and p.V600), which are mutated in approximately 5% of patients with lung adenocarcinoma (3), suggesting that ARAF p.S214/RAF1 p.S259 may function by a distinct biochemical mechanism. Our preliminary biochemical data suggests that ARAF also bound 14-3-3, and this binding was markedly attenuated in the setting of p.S214C mutation (Supplemental Figure 7). These results suggest that ARAF p.S214 mutations may potentiate Ras/Raf signaling by abrogating 14-3-3 and ARAF binding.
Clinical trials of sorafenib in patients with advanced non–small-cell lung cancer have demonstrated modest activity, with no survival advantage (7, 20). Though it is attractive to consider ARAF/RAF1 mutations as possible biomarkers of sorafenib response in lung adenocarcinoma, additional profiling and functional characterization studies will be required to establish this link. This includes mutation data from additional sorafenib responders and/or sorafenib-response data from ARAF p.S214/RAF1 p.S259 mutated cases. Our initial sequencing of 3 transient responders (<6-month progression-free survival) from the ECOG 2501 trial did not reveal additional Raf family mutations. Identification of additional ARAF p.S214/RAF1 p.S257 mutant cases in lung adenocarcinoma will likely require screening of large sets of patients, given the low frequency of these variants (1%). Though we believe that the cell line model used in this study should adequately recapitulate the observed overexpression of mutant ARAF in this patient’s tumor (Table 1 and Supplemental Table 4), knockdown and inhibitor studies of cell lines with naturally occurring ARAF/RAF1 mutations will be useful in definitively establishing essentiality and oncogene addiction in an endogenous context. Our preliminary analysis of the Cancer Cell Line Encyclopedia sequencing data (
http://www.broadinstitute.org/ccle/home) did not reveal any ARAF p.S214/RAF1 p.S257 mutant cell lines.
If recurrent but rare mutations underlie the oncogenicity and responsiveness of “driver-negative” lung adenocarcinomas, they are not likely to be nominated by statistical analysis of several hundred (or even thousands) of genome-sequenced cases. Our study suggests that a powerful, alternate approach to driver mutation discovery may be through the analysis of outlier patient responses and the identification of driver mutations through the preponderance of genomic, biochemical, and functional evidence.
Sequencing and bioinformatics. WGS of fresh tumor and peripheral blood DNA was performed using standard Broad Institute Illumina-based sequencing and analysis protocols (see Supplemental Methods). RNA-seq of the primary tumor specimen was performed using poly-A enrichment and standard Illumina protocols (see Supplemental Methods). Bioinformatics analyses and visualizations were generated using R Bioconductor packages (
http://www.bioconductor.org) and CIRCOS (
Retroviral transduction and soft agar assays. Ectopic expression of mutant constructs and assessment of anchorage-independent proliferation were performed as described previously (21). See the Supplemental Methods for experimental details.
Immunoblotting. Cells were lysed in a buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 2.5 mM EDTA, 1% Triton X-100, and 0.25% IGEPAL CA630. Protease inhibitors (Roche) and phosphatase inhibitors (Calbiochem) were added prior to use. Immunoblotting was performed as described previously (21) using antibodies specified in the Supplemental Methods.
Statistics. IC50 estimates and 95% CIs were obtained from sorafenib and trametinib inhibition dose-response data using least-squares nonlinear regression on a standard 4-parameter logistic model (see Supplemental Methods for additional details).
Study approval. The collection and genomic analysis of this patient sample was carried out in accordance with protocols approved by Institutional Review Boards at the Broad Institute of Harvard and MIT (Broad COUHES 1103004402) and Vanderbilt University (VICC THO 0547). Informed consent for genomic analysis was obtained from the patient at the time of sample collection.
View Supplemental data
View Supplemental Table 1
View Supplemental Table 2
View Supplemental Table 3
View Supplemental Table 5
M. Imielinski is supported by NCI training grant T32 CA9216-31. L. Araujo is supported by a Conquer Cancer Foundation of ASCO Long-Term International Fellowship. This work was supported by NCI U01CA114771 (to D.P. Carbone), Uniting Against Lung Cancer (to M. Meyerson and H. Greulich), the Lung Cancer Research Foundation (to M. Meyerson), Novartis (to M. Meyerson), Department of Defense Congressionally Directed Medical Research Programs Lung Cancer Research Program W81XWH-12-1-0269 (to M. Meyerson), and the American Lung Association (to M. Meyerson). We thank Mark Bray and members of the Broad Institute Imaging Platform for development of the CellProfiler pipeline, used to quantify the soft agar photographs.
Conflict of interest: Matthew Meyerson is a founder and equity holder of Foundation Medicine, a for-profit company that provides next-generation sequencing diagnostic services.
Citation for this article:J Clin Invest. 2014;124(4):1582–1586. doi:10.1172/JCI72763.
Buettner R, Wolf J, Thomas RK. Lessons learned from lung cancer genomics: the emerging concept of individualized diagnostics and treatment. J Clin Oncol. 2013;31(15):1858–1865.
Oxnard GR, Binder A, Janne PA. New targetable oncogenes in non-small-cell lung cancer. J Clin Oncol. 2013;31(8):1097–1104.
Imielinski M, et al. Mapping the hallmarks of lung adenocarcinoma with massively parallel sequencing. Cell. 2012;150(6):1107–1120.
Iyer G, et al. Genome sequencing identifies a basis for everolimus sensitivity. Science. 2012;338(6104):221.
Anastassiadis T, Deacon SW, Devarajan K, Ma H, Peterson JR. Comprehensive assay of kinase catalytic activity reveals features of kinase inhibitor selectivity. Nat Biotechnol. 2011;29(11):1039–1045.
Wilhelm S, et al. Discovery and development of sorafenib: a multikinase inhibitor for treating cancer. Nat Rev Drug Discov. 2006;5(10):835–844.
Wakelee HA, Lee JW, Hanna NH, Traynor AM, Carbone DP, Schiller JH. A double-blind randomized discontinuation phase-II study of sorafenib (BAY 43-9006) in previously treated non-small-cell lung cancer patients: eastern cooperative oncology group study E2501. J Thorac Oncol. 2012;7(10):1574–1582.
Weir BA, et al. Characterizing the cancer genome in lung adenocarcinoma. Nature. 2007;450(7171):893–898.
Ding L, et al. Somatic mutations affect key pathways in lung adenocarcinoma. Nature. 2008;455(7216):1069–1075.
Kan Z, et al. Diverse somatic mutation patterns and pathway alterations in human cancers. Nature. 2010;466(7308):869–873.
Govindan R, et al. Genomic landscape of non-small cell lung cancer in smokers and never-smokers. Cell. 2012;150(6):1121–1134.
Ramos AH, et al. Amplification of chromosomal segment 4q12 in non-small cell lung cancer. Cancer Biol Ther. 2009;8(21):2042–2050.
Wellbrock C, Karasarides M, Marais R. The RAF proteins take centre stage. Nat Rev Mol Cell Biol. 2004;5(11):875–885.
Pandit B, et al. Gain-of-function RAF1 mutations cause Noonan and LEOPARD syndromes with hypertrophic cardiomyopathy. Nat Genet. 2007;39(8):1007–1012.
Awada A, et al. Phase I safety and pharmacokinetics of BAY 43-9006 administered for 21 days on/7 days off in patients with advanced, refractory solid tumours. Br J Cancer. 2005;92(10):1855–1861.
Flaherty KT, et al. Improved survival with MEK inhibition in BRAF-mutated melanoma. N Engl J Med. 2012;367(2):107–114.
Druker BJ, et al. Effects of a selective inhibitor of the Abl tyrosine kinase on the growth of Bcr-Abl positive cells. Nat Med. 1996;2(5):561–566.
Antony R, Emery CM, Sawyer AM, Garraway LA. C-RAF mutations confer resistance to RAF inhibitors. Cancer Res. 2013;73(15):4840–4851.
Abraham D, et al. Raf-1-associated protein phosphatase 2A as a positive regulator of kinase activation. J Biol Chem. 2000;275(29):22300–22304.
Dingemans AM, et al. A phase II study of sorafenib in patients with platinum-pretreated, advanced (Stage IIIb or IV) non-small cell lung cancer with a KRAS mutation. Clin Cancer Res. 2013;19(3):743–751.
Greulich H, et al. Functional analysis of receptor tyrosine kinase mutations in lung cancer identifies oncogenic extracellular domain mutations of ERBB2. Proc Natl Acad Sci U S A. 2012;109(36):14476–14481.