Genomic and Molecular Screenings Identify Different Mechanisms For

Published OnlineFirst April 10, 2017; DOI: 10.1158/1535-7163.MCT-17-0104

Cancer Biology and Signal Transduction Molecular Cancer Therapeutics Genomic and Molecular Screenings Identify Different Mechanisms for Acquired Resistance to MET Inhibitors in Lung Cancer Cells Pol Gimenez-Xavier1, Eva Pros1, Ester Bonastre1, Sebastian Moran2, Ana Aza1, Osvaldo Grana~ 3, Gonzalo Gomez-Lopez3, Sophia Derdak4,5, Marc Dabad4,5, Anna Esteve-Codina4,5, Jose R. Hernandez Mora2, Diana Salinas-Chaparro1, Manel Esteller2,6, David Pisano3, and Montse Sanchez-Cespedes1

Abstract

The development of resistance to tyrosine kinase inhibitors groups of resistant cells. In one of these, the cells have acquired (TKI) limits the long-term efficacy of cancer treatments involving sensitivity to erlotinib, concomitantly with mutations of the them. We aimed to understand the mechanisms that underlie KIRREL, HDAC11, HIATL1, and MAPK1IP1L genes, among acquired resistance (AR) to MET inhibitors in lung cancer. EBC1 others. In the other group, some cells have acquired inactivation cells, which have MET amplification and are sensitive to TKIs of neurofibromatosis type 2 (NF2) concomitantly with strong against MET, were used to generate multiple clones with AR to a overexpression of NRG1 and a mutational profile that includes MET-TKI. Whole-exome sequencing, RNA sequencing, and global changes in LMLN and TOMM34. Multiple independent and DNA methylation analysis were used to scrutinize the genetic and simultaneous strategies lead to AR to the MET-TKIs in lung cancer molecular characteristics of the resistant cells. AR to the MET-TKI cells. The acquired sensitivity to erlotinib supports the known involved changes common to all resistant cells, that is, phenotypic crosstalk between MET and the HER family of receptors. For the modifications, specific changes in gene expression, and reactiva- first time, we show inactivation of NF2 during acquisition of tion of AKT, ERK, and mTOR. The gene expression, global DNA resistance to MET-TKI that may explain the refractoriness to methylation, and mutational profiles distinguished at least two erlotinib in these cells. Mol Cancer Ther; 16(7); 1–11. 2017 AACR.

Introduction growth following the administration of specific tyrosine kinase inhibitors (TKI) has been observed in lung cancer cells with Most novel therapeutic approaches in cancer are developed on alterations at various RTKs, including mutations, gene fusions, the basis of our increasing knowledge of underlying oncogenic and strong focal gene amplification (GA) at ALK, EGFR, ERBB2, mutations. In the case of lung cancer, our better understanding of FGFR1, MET, PDGFRA, RET, and ROS genes, among others (1–9). its genetic architecture has enabled novel treatments to be Inasmuch as most of these observations are based on screenings designed, most of which are based on drugs that target growth made in cancer cell lines, these have become essential preclinical factor receptors with tyrosine kinase activity (RTK) that are genet- tools for evaluating how the cancer genotype determines the ically activated in the tumors. A strong effect of decreasing cell response to these specific drugs (1–3). Despite the success of genotype-directed therapies, some

1 tumors with activated target RTK exhibit frequent failures in the Genes and Cancer Group, Cancer Epigenetics and Biology Program (PEBC), ﬁ Bellvitge Biomedical Research Institute-IDIBELL, Hospitalet de Llobregat, Bar- primary response to these speci c TKIs. This is the so-called celona, Spain. 2Cancer Epigenetics Group, Cancer Epigenetics and Biology intrinsic (i.e., primary) resistance, which has been associated with Program (PEBC), Bellvitge Biomedical Research Institute-IDIBELL, Barcelona, the simultaneous activation of parallel signal transduction path- Spain. 3Bioinformatics Unit, Structural Biology and BioComputing Programme, ways that circumvent the targeted pathway blockade (9). Further- 4 Spanish National Cancer Centre (CNIO), Madrid, Spain. CNAG-CRG, Centre for more, in most tumors that are initially sensitive to these therapies, Genomic Regulation (CRG), Barcelona Institute of Science and Technology acquired (i.e., secondary) resistance inevitably develops due to the (BIST), Barcelona, Spain. 5Universitat Pompeu Fabra (UPF), Barcelona, Spain. 6Institucio Catalana de Recerca i Estudis Avancats¸ (ICREA), Barcelona, Spain. acquisition of genetic alterations in the kinase target or in molecules acting in related signaling pathways (10–14). The origin of Note: Supplementary data for this article are available at Molecular Cancer the genetic alterations responsible for the acquired resistance (AR) Therapeutics Online (http://mct.aacrjournals.org/). could either be generated de novo or arise as clonal expansions P. Gimenez-Xavier and E. Pros contributed equally to this article. from preexisting minor subclones. In lung cancer, this has been Corresponding Author: Montse Sanchez-Cespedes, Bellvitge Biomedical extensively modeled, especially to study the responses to EGFR Research Institute (IDIBELL), Av. Gran Via de Hospitalet, 199-203, Hospitalet inhibitors, using isogenic pairs of drug-sensitive and drug-resis- de Llobregat, Barcelona 08908, Spain. Phone: 349-3260-7132; Fax: 349-3260- tant human cancer cell lines (10, 12–14). It is well established that 7219; E-mail: [email protected] resistance to TKIs against EGFR can be acquired through the doi: 10.1158/1535-7163.MCT-17-0104 p.T790M mutation at EGFR, which prevents the binding of the 2017 American Association for Cancer Research. inhibitor to the receptor, or the ampliﬁcation or overexpression of

www.aacrjournals.org OF1

Gimenez-Xavier et al.

RTKs such as MET, EGFR, AXL, and ERBB2 (10–15). Likewise, AR lysis buffer and absorbance was measured at 596 nm. Data were to crizotinib in EML4-ALK-positive lung cancer cell lines has been analyzed by nonlinear regression using GraphPad Prism Soft- associated with point mutations or amplification in ALK or KIT, ware version 5.2 (GraphPad Software, Inc.). Viabilities were among other genes (15, 16). expressed as a percentage of the untreated controls. The IC50 Regarding the MET oncogene, the genetic and molecular was determined from the dose–response curve. Results are mechanisms linking AR to its associated TKI are not fully under- presented as the median of at least three independent experi- stood. Amplification and overexpression of wild-type KRAS or ments performed in triplicate for each combination of cell line activation of BRAF and of the HER pathways are some of the and compound. mechanisms that have been proposed for the AR to MET-TKIs in For wound-healing assays, cells were grown in 6-well plates various types of cancer (17–21). Here, we have used the EBC1 lung with complete medium. After 90% confluence was reached, the cancer cells, which are intrinsically sensitive to the MET-TKIs due medium was replaced with FBS-free media for 24 hours. A wound to genetic activation of MET by GA, as a cancer cell model to was created with a germ-free 100-mL pipette tip in the monolayer. investigate AR. We obtained various clones from the EBC1 paren- The cells were washed with PBS and grown in FBS-free media for tal cells that had become refractory to the MET-TKIs and, taking 72 hours. The wounds were observed under a phase contrast advantage of next-generation sequencing technologies, scruti- microscope (IX81, Olympus). The images were analyzed by nized these various cell subpopulations for genetic and molecular drawing lines at the wound edges. The width of the scratch was alterations responsible for drug resistance. measured using TScratch software (25). For colony formation assays, cells were seeded at a density of 2,000 to 4,000 cells per well, in 6-well plates, in DMEM media Materials and Methods containing 10% (v/v) FBS, with or without the corresponding Generation of AR to PHA665752 and tyrosine kinase inhibitor treatment, and stained 14 to 15 days later with Crystal Violet treatments (Merck Millipore). Each assay was performed in triplicate. Col- The EBC1 lung cancer cell line was obtained from the JCRB onies were quantified using ImageJ software. Cell Bank (National Institutes of Biomedical Innovation, Health and Nutrition, Japan), grown under recommended FISH analysis conditions, and maintained at 37Cinahumidified atmo- To determine gene copy number of the MET gene, we per- sphere of 5% CO2. The cell line was obtained directly from the formed dual-color FISH analysis in interphase nuclei, following repository and passaged for fewer than 4 months after receipt, previously described protocols (26). Briefly, we used a BAC clone before generating the various derived resistant cells. The cells labeled in Spectrum Red for the MET gene (RP11-95I20) and were tested for mycoplasma infection prior to the generation of control BAC, labeled in Spectrum Green and located pericentro- the various resistant cells and periodically thereafter. All the merically (RP11-45N18). cells used in the experiments, tested negative for the infection. Genomic DNA and total RNA were extracted using standard Antibodies and Western blots protocols. The identity of the EBC1 cells, both the parental and The antibodies are indicated in the Supplementary Methods. each of the resistant pools and clones, was verified by deter- For Western blots, cells were lysed into ice-cold RIPA lysis cell mining the presence of the TP53 mutation (5) and amplifica- buffer supplemented with protease and phosphatase inhibitor tion of the MET gene in the cells. cocktails (Complete Mini and Phosphostop), followed by clear- To generate AR to MET inhibitors, the EBC1 cells were cultured ance of lysates by microcentrifugation. Supernatants were for about 16 weeks in a medium with an increasing concentration resolved on an SDS-PAGE gel, transferred onto nitrocellulose of PHA665752 (Ref. 2693, Tocris Bioscience), from a starting membranes (Amersham, GE Healthcare Life Science), and probed concentration of 62.5 nmol/L to a final concentration of 1 mmol/ with the indicated antibodies. Detection was done using a chemi- L. Individual cells capable of proliferation were allowed to grow luminescence system (Millipore). and were selected as individual clones. Individual clones or pools of them were selected for subsequent experiments, transferred to Whole-exome sequencing and RNA sequencing separate plates and treated with 100 nmol/L of PHA665752. The Genomic DNA was extracted from the EBC1-P, EBC1-PR2, EBC1 cells were also tested for sensitivity to the MET inhibitor, and EBC1-PR4 cells. The cells were collected and placed in 1% crizotinib (Ref. 4368, Tocris Bioscience), the FGFR1 inhibitor, SDS/proteinase K (10 mg/mL) at 58C for 1 hour and subjected PD173074 (Ref. 2499, Sigma-Aldrich), the EGFR inhibitor, erloti- to phenol–chloroform extraction and ethanol precipitation nib (sc-202154, Santa Cruz Biotechnology), and the AXL inhibitor, following standard protocols. DNA was assessed for quality R428 (BGB324, Selleckchem). Information about the chemical by visualization in an agarose gel and quantified with Nano- structure of the PHA665752, PD173074, and R428 compounds Drop and PicoGreen. For RNA sequencing (RNA-seq), total was previously published (22–24). RNA was extracted from EBC1, EBC1-C1, EBC1-V8, EBC1-PR1, EBC1-PR2, EBC1-PR3, EBC1-PR4 cells, using a TRIzol Plus RNA Cell viability, wound healing, and colony formation assays Purification Kit (Purelink RNA Mini Kit, Ambion). RNA quality For the cell-viability assays, cells were seeded at a density of and integrity (RIN) were measured with an Agilent 2100 2,000 cells/well on 96-well plates and incubated overnight Bioanalyzer (Agilent Technologies) and quantified with a before being exposed to different concentrations of the drug NanoDrop Spectrophotometer. for 72 hours. Ten microliters of a solution of 5 mg/mL MTT (3- Whole-exome sequencing (WES) and RNA-seq were carried out (4,5)-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; at the Spanish National Genome Analysis Center (CNAG-CRG; Sigma Chemical Co.) was added and incubated for 3 hours at ref. 27). For exome capture, the SureSelect Human All Exon Kit v4 37C. The precipitated formazan was dissolved with 100 mLof or v5 (Agilent Technologies) was used, and for RNA library

OF2 Mol Cancer Ther; 16(7) July 2017 Molecular Cancer Therapeutics

Acquired Resistance to MET Inhibitors in Lung Cancer

preparation the TruSeq RNA Library Preparation Kits (Illumina) Global methylation microarray analysis were used before deep sequencing in an HiSeq 2000 Illumina For DNA methylation microarrays, all DNA samples were Analyzer (Illumina) to generate 2 100 read pairs for WES and 2 assessed for integrity, quantity, and purity by electrophoresis in 75 read pairs for RNA-seq. a 1.3% agarose gel, PicoGreen quantification, and NanoDrop measurement. We performed bisulfite conversion of 500 ng of WES and RNA-seq and data analysis to identify variants specific genomic DNA using an EZ DNA Methylation Kit (Zymo to resistant cells Research). Bisulfite-converted DNA (200 ng) was used for hybrid- Data were analyzed at the Spanish National Cancer Centre ization on the MethylationEPIC BeadChip (Illumina). Raw fluo- (CNIO, Madrid, Spain) or at the CNAG-CRG. RNA-seq reads were rescence intensity values were normalized with Illumina Genome mapped against the human reference genome (hg19), with the Studio software (V2011.2) using "control normalization" with GEM RNA-seq pipeline (http://gemtools.github.io/; see supple- background correction. Normalized intensities were then used to mentary methods for details). calculate DNA methylation levels (b values). Likewise, values with The raw data produced by this study are available in the low statistical power (indicated by detection values of P > 0.01) following databases: RNA-seq and WES at the Sequence Read were excluded from the analysis. Genotyping probes present on Archive (SRA) under accession number SRP096814. the chip and DNA methylation probes overlapping with known SNPs were also removed. Probes were considered to be in a promoter CpG island if they were located within a CpG island qRT-PCR (UCSC database) and <2,000 bp away from a transcription start We used qRT-PCR to validate genes identified by RNA-seq site (outside chromosome X). Samples were clustered in an analysis. Total RNA was extracted with PureLink RNA Mini Kit unsupervised manner by using the 5,000 most variable values (Ambion, Life Technologies) and reverse-transcribed with for CpG methylation according to the SD for the CpG sites located SuperScript II Reverse Transcriptase (Invitrogen, Thermo Fisher in promoter regions by hierarchical clustering using the complete Scientific) and Random Primers (Promega), according to the method for agglomerating the Manhattan distances. Microarray manufacturers' instructions. qRT-PCR was performed in an methylation data are available from the Gene Expression Omni- Applied Biosystems 7900HT Fast Real-Time PCR System using bus (GEO) under accession code GSE92608. SYBR Green PCR Master Mix (Applied Biosystems, Life Tech- nologies). Three biological replicates were carried out. All data were normalized against the expression of gene control ACTB. Statistical analysis t – The sequences of the primers used are provided in Supplemen- Analysis consisted of two-tailed unpaired Student and Mann U tary Table S1. Whitney tests and ANOVA. Continuous variables were sum- marized as means and SDs. We considered group differences to be significant for values of P < 0.05. Genetic screening of NF2 To validate the inversion found at the neurofibromatosis type 2 (NF2) gene, exons 2 to 7 (NM_000268.3) were PCR-amplified at Results the DNA level and specific primers were designed at the mRNA Morphologic modifications and changes in the levels of cell level to amplify the fusion transcript (primer sequences are structure and signal transduction-related proteins underlie AR presented in Supplementary Table S1). PCR products were direct- to MET inhibitors in EBC1 cells ly sequenced using the BigDye Terminator v3.1 Cycle Sequencing The EBC1 cell line endures high levels of MET-GA and is very Kit (Applied Biosystems) in an Applied Biosytems 3070 DNA sensitive to MET inhibition (5, 18). To generate resistance to the Analyzer. Mutation Surveyor u3.97 software (SoftGenetics, LLC) MET inhibitor, PHA665752, we exposed the EBC1 cells to increas- was used to analyze sequences. The presence of an intragenic ing concentrations of the drug for sixteen weeks. By the end of this homozygous deletion was concluded when product amplifica- continuous exposure, various clones or pools of resistant cells had tions at NF2 were recurrently negative in multiplexed PCR reac- developed AR to the drug in vitro (Fig. 1A and B). We selected two tions with GAPDH. of these clones (EBC1-C1 and EBC1-V8) and four pools (EBC1- PR1, EBC1-PR2, EBC1-PR3, and EBC1-PR4) for various anal- Virus infection and shRNA experiments yses aiming to characterize the cells morphologically and To deplete NF2 expression in the EBC1-P cells we used molecularly. The levels of MET amplification remained similar shRNAs, which were purchased from SIGMA-MISSION (Len- in the parental and resistant cells, whereas the phosphorylation tiExpress Technology, Sigma-Aldrich) as a glycerol stock of 5 (activation) of MET (pMET) was abrogated in the resistant cells pLKO plasmids carrying NF2-specificshRNAsequencesand (Fig. 1B and C). Together, these observations indicate that, in scramble control (Supplementary Table S1). For the infections, all the resistant cells, the resistance to the TKI had been acquired lentiviruses carrying the short hairpin RNAs (shRNA) were through a mechanism that does not involve genetic changes in generated within the 293T cells. 293T cells were cotransfected MET.Weconfirmed that the various EBC1-resistant (henceforth with each specific shRNA and the packaging plasmids psPAX EBC1-R) cells were also refractory to crizotinib treatment, and pMD2.6 (Sigma Aldrich). After 48 hours, 293T filtered another TKI known to inhibit the tyrosine kinase activity of supernatants were collected and used in the culture media for MET (Supplementary Fig. S1). the different cell lines for an additional 24 to 48 hours. After The EBC1-R cells were very different, morphologically, from the virus infection, different cell populations were selected with EBC1-parental (henceforth EBC1-P) cells. Although the EBC1-P puromycin for 72 or 96 hours. To obtain stable expression of cells had an elongated appearance with large protrusions, all the the shRNAs, cell populations were selected with puromycin for EBC1-R cells were polygonal and exhibited patchy growth (Fig. 2to3weeks. 1D and Supplementary Fig. S1). The EBC1-R cells migrated slower

www.aacrjournals.org Mol Cancer Ther; 16(7) July 2017 OF3

Gimenez-Xavier et al.

Figure 1. Characterization of EBC1-R cells. A, Description of the EBC1-R cells generated. They arose from clonal expansion of individual EBC1-R cells or from pools of EBC1-R cells. B, Left, cell-growth inhibition upon administering PHA665752 treatment to EBC1 parental (EBC1-P) and resistant (EBC1-PR2) cells. Lines represent cell survival relative to untreated controls of the MTT assays in the cells treated with increasing concentrations of the inhibitor for 72 hours. Error bars indicate the standard deviation. Right, examples of metaphase nuclei from EBC1-P cells and the indicated resistant cells at the MET gene (probes in red). Control probe in green. C, Western blot anti-phospho-MET (pMETY1234/Y1235) and anti-MET (MET) in the indicated cell lines. Constitutive MET activation is present in the EBC1 parental cells (EBC1-P) cells before treatment with the PHA665752 (PHA, 100 nmol/L) inhibitor. The EBC1-R cells maintain the ability to inhibit MET phosphorylation in the presence of the drug. D, Phase contrast images showing the changes in cell morphology between EBC1 parental (EBC1-P) and the indicated resistant cells. Scale bar, 100 mm. E, Left, phase contrast microscopy images of the wound-healing migration assay. Wound closure was monitored at the indicated times. On the right, quantiﬁcation of the assay. The open wound area was quantiﬁed with TScratch software and the percentage calculated relative to the open area at 0 hour (100%). Error bars, SD of three replicates. F, Western blot analysis of the indicated proteins in the EBC1-P and EBC1-R cells. In the case of EBC1-P, extracts with (þ)andwithout() treatment with the PHA665752 inhibitor (PHA, 100 nmol/L) are shown. The panels above indicate the levels of different proteins related to cell adhesion and morphology. The panels below show the phosphorylation levels of proteins involved in signal transduction pathways. ACTIN, total protein loading controls.

than the EBC1-P cells in the wound-healing assay (Fig. 1E). EBC1 cells with AR to MET inhibitors bypass mTOR inhibition Because these modifications are indicative of strong molecular and undergo a shift in gene expression profiles and genetic shifts we determined the levels of various proteins Next, we examined the activation/inactivation of signal associated with cell shape. Two of the EBC1-R cells (-PR3 and transduction molecules acting downstream of the MET recep- -PR4) underwent a sharp drop in the levels of vimentin (Fig. tor. As expected, the administration of the PHA665752 drug 1F), suggesting the involvement of a mesenchymal-to-epithe- to the EBC1-P cells greatly reduced the phosphorylation of lial transition, associated with the resistance to PHA665752. AKT and ERK as well as the levels of phosphorylation of the We observed dramatic changes in the p120ctn protein (28) that ribosomal proteins S6, indicating the inactivation of mTOR. affected all the EBC1-R cells. Furthermore, there was a slight All the EBC1-R cells bypassed the inhibitory effect of the drug decrease in the levels of the GAPDH protein in all the EBC1-R and maintained activation of AKT, ERK, and mTOR, although cells (Fig. 1F). No changes in the levels of any of these proteins the levels of phosphorylated AKT were somehow decreased in were observed in the EBC1-P cells, following a short treatment most of the EBC1-R cells (Fig. 1F). Taken together, these with the TKI, indicating that these are stable molecular mod- results indicate that all the EBC1-R cells maintain activated ifications for the adaptation of cells to a constant exposure to mTOR, but in a manner that is independent of the level of MET inhibition. MET activity.

OF4 Mol Cancer Ther; 16(7) July 2017 Molecular Cancer Therapeutics

Acquired Resistance to MET Inhibitors in Lung Cancer

To examine the molecular characteristics of the resistant cells in the -PR1, -PR3, and -PR4 cells were closely related to each other greater detail we performed RNA sequencing (from herein RNA- (Supplementary Fig. S2). The EBC1-PR4 cells exhibited the most seq) of the EBC1-P and of the six populations of EBC1-R cells. We different proﬁle and the closest similarity to the parental cells. found marked changes in gene expression. Some transcripts showed greater than 100-fold changes in most of the EBC1-R Upregulation of tyrosine kinase-receptor ligands and cells, such as those of ABI3BP, CCL2, and NR3C2 (Supplementary de novo sensitivity to the EGFR inhibitor, erlotinib, Table S2A). Consistent with the changes in cell morphology and in multiple EBC1-R cells in cell growth, the EBC1-R cells revealed an upregulation of genes There was strong upregulation (by 50-fold) of the transcripts, associated with cell adhesion, growth, and organization function- GAS6 and PROS1 (Supplementary Table S2A), which encode for alities, whereas the downregulated genes were enriched in cell ligands of the RTKs named AXL, TYRO3, and MER, all of them epithelial differentiation and in glucose metabolism functions members of the TAM receptor subfamily (24, 29). The expression (Fig. 2A; Supplementary Table S2A). As shown in the MDS plot, levels of AXL were also slightly higher in the EBC1-R cells. These the gene expression proﬁles of the EBC1-PR2 and -C1 were very observations indicate activation of signaling through the TAM- similar to each other and closely related to the EBC1-V8, whereas family of receptors. We extended our search to genes encoding

Figure 2. Molecular characterization of EBC1-R cells. A, Enriched gene ontology (GO) classifications (P < 0.05 for all categories shown) among genes upregulated and downregulated in the gene expression profile of EBC1-R cells relative to EBC1-P cells (from Supplementary Table S2A). B, mRNA levels of indicated genes, relative to ACTB, in the EBC1-P and various EBC1-R cells. C, Colony formation assay for the EBC1-P and for the various EBC1-R cells. Bottom, quantification of the number of colonies (cells treated with erlotinib relative to untreated cells). D, Western blots of the indicated total and phosphorylated proteins in EBC1-P and EBC1-R cells. EBC1-P, extracts with (þ) and without () treatment with the PHA665752 inhibitor are shown. ACTIN, total protein loading controls. E and F, Western blots indicating phosphorylation levels of EGFR, AKT, and S6 are shown for the indicated treatments and cells. PHA, PHA665752 (100 nmol/L); ERLO, erlotinib (100 nmol/L); CRIZO, crizotinib (100 nmol/L). Error bars, SD of three replicates. Two-tailed unpaired Student t test for EBC1-P versus each EBC1-R cell type. , P < 0.05; , P < 0.005; , P < 0.001; , P < 0.0001.

www.aacrjournals.org Mol Cancer Ther; 16(7) July 2017 OF5

Gimenez-Xavier et al.

Table 1. List of validated SNVs from WES data, and frequency of reads in EBC1-PR2, EBC1-PR4, and EBC1-P cells Codon Amino acid Reads- Reads- Reads Gene name Ofﬁcial full name Effect change change PR2 PR4 EBC1-P HDAC11 Histone deacetylase 11 Modify Ggg/Agg p.G152R 14/30 0/24 0/47 HIATL1 Hippocampus abundant transcript-like 1 Moderate caG/caT p.Q168H 7/21 0/105 0/52 KIRREL Kin of IRRE-like Moderate Gca/Tca p.A398S 9/26 0/59 0/51 MAPK1IP1L Mitogen-activated kinase 1 interacting protein 1-l Modiﬁer cCt/cTt p.P39L 15/60 0/55 0/64 PLXND1 Plexin D1 Moderate atG/atA p.M845I 29/86 0/65 0/115 LMLN Leishmanolysin-like peptidase High Cga/Tga p.R322X 0/40 58/239 0/72 PGF Placental growth factor Moderate gTg/gGg p.V59G 3/14 11/34 0/35 TOMM34 Translocase of outer mitochondrial membrane 34 Moderate aaA/aaC p.K72N 0/113 13/66 1/34 TRPC4AP Transient receptor potential cation channel subfamily C member Moderate Caa/Gaa p.Q535E 0/81 13/71 1/116 4-associated protein

other RTKs or growth factors. Although we found a moderate searched for variants that occurred at >30% allele frequency in increase in the levels of FGFR1, ERBB4, and PDGFC, the signif- each of the EBC1-R cells but not in the EBC1-P cells. Using a set of icance of these increases in levels is probably limited given the optimized filtering criteria (see Materials and Methods) we iden- magnitude of these changes. It was of particular note that the tified more than 250 novel single nucleotide variants (SNV) that EBC1-PR4 cells experienced a 100-fold increase in the level of were present in EBC1-PR2 or EBC1-PR4 cells but not in EBC1-P expression of neuregulin (NRG1). We validated all these findings cells. From these, we further selected genes with inactivating by quantitative RT-PCR (Fig. 2B). mutations or those with nonsynonymous mutations that were Taking previous reports (18) and our current observations into also present at the mRNA level Table 1. Only one gene was consideration, we explored the possible changes in sensitivity to mutated in both EBC1-PR2 and EBC1-PR4 cells, but this was an the FGFR1 inhibitor, PD173074, to the EGFR inhibitor, erlotinib, intronic mutation. No changes in amino acid coding sequences and to the AXL inhibitor, R428. We measured the effects on cell were shared by the two resistant cells, at least at a frequency higher growth by assessing the IC50 of these compounds. None of the that 30% (Supplementary Table S2C). cells showed significant changes in sensitivity to the PD173074 There has been debate about the origin of the genetic alterations compound, but there was a decrease in the IC50 for erlotinib arising in cells that have AR to a given TKI. Previous papers have (about 10- to -20-fold) in many of the EBC1-R cells. The PR3 and reported the presence of these alterations in untreated specimens, PR4 cells were found to be the less sensitive (Supplementary Fig. albeit at low frequency, supporting the view that there is an S3A). The development of different degrees of secondary sensi- expansion of preexisting clones rather than de novo genetic muta- tivity to erlotinib, as a consequence to the AR to the MET-TKIs in tions (15, 21). Taking this into account, we manually searched for the EBC1 cells, was previously described (18). Furthermore, none these mutations in the EBC1-P and found that some of the of the cells tested showed significant changes in the sensitivity to changes were already present in the parental cells (Table 1). We R428 (Supplementary Fig. S3A). The de novo acquisition of sen- also searched for them in the other EBC1-R cells by examining sitivity to erlotinib, found in most of the EBC1-R cells, was RNA-seq data, and found that the variants were also present in confirmed in a cell colony formation assay (Fig. 2C). Interestingly, most of them, although at different frequencies (Fig. 3A). For the acquisition of sensitivity to erlotinib was not accompanied by example, the missense mutations of KIRREL and HDAC11 are a rise in phosphorylation at EGFR (pEGFR), at least at Tyr1068 present in about 50% of the -PR1, -PR2, -C1, and -V8 cells, and at Tyr1173 (Fig. 2D). Rather, the levels of pEGFR at these whereas their abundance is less than 10% in the -PR3 and -PR4 residues were higher in the EBC1-P and were partially reduced by cells. Together, the existence of identical mutations in different PHA665752 or crizotinib treatment, but not by erlotinib (Fig. 2E). EBC1-R cells indicates that these have emerged as an expansion of Considered together, these observations suggest that a hyperactive subclonal populations from the EBC1-P cells. MET directly or indirectly drives the phosphorylation of EGFR at these residues, and that this does not lead to erlotinib sensitivity. Genetic inactivation of the NF2 gene in one of the erlotinib- The treatment with R428 neither reduced the levels of pEGFR nor resistant EBC1-R cells inactivated the AKT/mTOR downstream pathway (Supplementa- In addition to SNVs, we used the WES and RNA-seq data to ry Fig. S3B). Furthermore, it was intriguing to observe that the search for structural variations, intragenic homozygous deletions suppression of cell growth triggered by erlonitib in the EBC1-R (HD), and GA using specific analytical pipelines (see Materials cells did not involve inhibition of either the AKT or mTOR and Methods). No regions were found with high levels of GA, pathways (Fig. 2F). This was also previously reported (18). whereas several fusions of unknown relevance were identified (Supplementary Table S2D). Interestingly, in the EBC1-PR4 cells Characterization of genetic events of EBC1-R cells reveals we detected an HD that affected between 5 and 16 exons of the selection of preexistent subclones in untreated parental cells tumor suppressor gene NF2. This alteration corresponded to an On the basis of the moderate changes observed in gene expres- inversion that predicted a truncated protein (Fig. 3B and C). As sion we concluded that GA of RTKs could not explain the resis- expected, the cells carrying the NF2 inversion showed no NF2 tance phenotypes observed in our cells. Thus, we decided to protein (also called Merlin; Fig. 3D). By contrast, the levels of the perform WES. We selected EBC1-PR2 and EBC1-PR4 cells, because NF2/Merlin protein were higher in the EBC1-PR1, PR2, and PR3 these were the two EBC1-R cells that differed the most in their gene cells than in the EBC1-P cells (Fig. 3D). expression profiles and their sensitivity to erlotinib. The exome Next, we attempted to determine whether the inactivation coverage report is presented in Supplementary Table S2B. We of NF2 was responsible for the AR of the EBC1-PR4 cells to the

OF6 Mol Cancer Ther; 16(7) July 2017 Molecular Cancer Therapeutics

Acquired Resistance to MET Inhibitors in Lung Cancer

Figure 3. WES analysis identifies inactivation of NF2 as one mechanism for AR to MET-TKI. A, Radial chart (left) and table (right) showing the percentage of mutated reads for each SNV relative to the total number of reads at the mRNA level in the indicated cells. SNVs were identified by WES (see Table 1). B, Top, schematic representation of the structure of the NF2 gene and the position of the different exons, highlighting the region affected by the inversion and the deleted exons; bottom, agarose gene with the PCR products of the indicated exons and EBC1 cells, depicting some of the deletions (red). Neg., negative control. C, Chromatogram depicting the inversion found in the EBC1-PR4 at the mRNA level. D, Western blot analysis showing the levels of NF2/Merlin protein in the indicated cells. The LOU-NH91 carry a mutation at NF2 (p.P257fs39), predicting no protein expression and was used as negative control. The H322, H1439, and MCF7- cancer cells were positive controls. The band observed in the LOU-NH91 and the lower bands in the EBC1-related cells are nonspecific. NUCLEOLIN, total protein loading control.

MET-TKI. First, we tested whether downregulation of NF2 in Notwithstanding these observations, the involvement of NF2 the EBC1-P cells mimicked the phenotypic and molecular inactivation in the AR to MET-TKI in EBC1-P cells cannot be characteristics found in EBC1-PR4 cells. To this end, we completely ruled out. Low levels of the Merlin protein were still depleted the expression of NF2 in the EBC1-P cells, using detectable in the EBC1-P cells after infection with shNF2 (Fig. various shRNAs. Compared with scramble RNA, cells infected 4B). These levels could be high enough for Merlin to exert its with different shNF2 underwent morphological changes, function and so prevent refractoriness to MET inhibition in the which closely resembled those of the EBC1-R cells, that is, EBC1-P cells. polygonal-shaped cells that grow in patches (Fig. 4A). How- ever, downregulation of NF2 in the EBC1-P cells did not trigger Changes in global DNA methylation discriminate various some of the molecular changes observed in the EBC1-PR4 cells, EBC1-R cells such as the shift in the expression of p120ctn isoforms or the To determine whether AR involves modiﬁcations in DNA reduced levels of vimentin (Fig. 4B). In contrast, depletion of methylation we performed genome-wide DNA methylation pro- NF2 did induce an increase in the levels of ABI3BP and NRG1 ﬁling (30). We selected the 5,000 CpGs with the most variable (Fig. 4C). Finally, the drop in NF2 levels did not affect the methylation levels plotted in an unsupervised manner (Fig. 5A growth inhibition capability of the PHA665752 compound in and Supplementary Table S2E). The analysis segregated the cells the EBC1-P cells (Fig. 4D). into two clusters that were consistent with the similarities in the

www.aacrjournals.org Mol Cancer Ther; 16(7) July 2017 OF7

Gimenez-Xavier et al.

Figure 4. Effects of downregulation of NF2 in EBC-P cells with various independent shRNAs targeting NF2 (sh#1, sh#2, sh#3, sh#4, sh#5). A, Phase contrast images showing the changes in cell morphology between EBC1-P cells infected with scramble RNA (scr) or with independent shRNAs targeting NF2. Scale bar, 100 mm. B, Western blot analysis showing the levels of NF2/Merlin and other indicated proteins related to cell adhesion and morphology in EBC1-P and EBC1-P cells after downregulating NF2 expression with various independent shRNAs targeting NF2 (sh#1, sh#2, sh#3, sh#4, and sh#5). Scr, scramble control. NUCLEOLIN OR ACTIN, total protein loading controls. C, mRNA levels of ABI3BP and NRG1, relative to ACTB, in EBC1-P and EBC1-P cells with scramble mRNA, and in EBC1-P cells after downregulating NF2 expression with four shRNAs (sh#1-4). Error bars, SD of three replicates. Two-tailed unpaired Student t tests for EBC1-P versus each EBC1-R cells. , P < 0.05; , P < 0.005; , P < 0.001. D, Representative examples of the colony formation assay for EBC1-P and EBC1-P cells after downregulating NF2 expression (sh#1 and sh#2) in untreated cells () and in cells treated with MET-TKI (þ).

genetic and gene expression profiles (Figs. 2B and 3A). One of activation of related RTKs or effectors that act as a bypass tracks the clusters included the resistant cells, -PR1, -PR2, -C1, and -V8, mediated secondary refractoriness. Furthermore, all the EBC1-R and the other included the parental cells (EBC1-P) and the -PR3 cells exhibited very similar shifts in cell morphology, from mes- and -PR4. enchymal to epithelioid-like features. All these morphological We next determined whether there was an association between modifications are consistent with the marked molecular changes changes in DNA methylation and in gene expression. We found observed in these cells, that is, enriched cell adhesion, decreased no major differences in promoter hypermethylation among the glucose metabolism, and cell proliferation-related transcripts, as upregulated or downregulated genes common to all the resistant well as a shift in the mobility of p120ctn protein, shown by cells (Supplementary Table S2A), indicating that the changes Western blot analysis, due to variations in the protein isoform in gene expression profiles did not arise as a consequence of abundances or in posttranslational modifications. The p120ctn large differences in global DNA methylation (Fig. 5B). We did protein plays a crucial role in the regulation of cell–cell adhesion observe a significant increase in global CpG methylation, at adherens junctions and in the epithelial-to-mesenchymal tran- specifically the bodies of genes that were upregulated in the sition (34). In previous works, morphological modifications EBC1-R cells (Fig. 5B). such as shifts from NSCLC to either SCLC or epithelial-to-mesenchymal transformations have been reported after AR to the EGFR-TKIs (32, 35). Discussion Another characteristic common to all the EBC1-R cells was a In lung cancers, de novo mutations acquired at EGFR or ALK are dramatic upregulation of the GAS6 transcript, which codes for the not always associated with the AR to their respective TKIs, and ligand of RTK, AXL, which itself was also slightly more abundant alterations at other RTKs or signal transduction molecules can also in all the resistant cells. Increased levels of GAS6 and AXL have account for the AR (15, 31, 32). Genetic alterations that have so far been reported in studies of AR to various TKIs, such as erlotinib been reported to confer AR on MET-TKIs include not only MET and PIK3 inhibitors, although its significance is still unclear point mutations and MET GA, but also a constitutively active because no genetic alterations at AXL or GAS6 have been found SND1-BRAF fusion protein and amplification and overexpression (13, 36–38). Interestingly, AXL belongs to the TAM receptor of wild-type KRAS (17, 19, 20, 33). None of these was observed in subfamily, together with TYRO3 and MER (24, 29), and GAS6 the EBC1-R cells that we generated in this study to model AR to has the ability to activate all these receptors. It is of note that all the MET-TKI inhibitors. Instead, as previously reported (18), we EBC1-R cells also showed strong upregulation of PROS1, which found that the PHA665752 compound readily inhibited MET codes for another ligand of the receptors TYRO3 and MER. This phosphorylation in all the EBC1-R cells, indicating that the indicates that signaling through the TAM receptor family is

OF8 Mol Cancer Ther; 16(7) July 2017 Molecular Cancer Therapeutics

Acquired Resistance to MET Inhibitors in Lung Cancer

Figure 5. Analysis of the changes in genome wide methylation, using an 850,000 CpG methylation microarray. A, Unsupervised hierarchical clustering in the indicated parental EBC1 cells (EBC1-P) and EBC1-R cells for the 5,000 most variable CpGs (from Supplementary Table S2E). The heatmap colors illustrate beta values representing the degree of methylation, from low (green) to high (red), as shown by the scale at the top- right of the Figure. B, Box-plot showing the delta beta values of the CpGs at the promoters (only CpG islands) or bodies of the upregulated (Up) and downregulated (Down) genes, from Supplementary Table S2A. Two-tailed unpaired Student t tests of the b values of EBC1-P versus each EBC1-R cell. , P < 0.05; , P < 0.005; , P < 0.001; , P < 0.0001; NS, not signiﬁcant.

strongly activated in these cells. In spite of this, the inactivation of Tyr1068 and Tyr 1173 of EGFR. This suggests that MET activity the AXL receptor did not have a significant effect on cell growth, induces EGFR phosphorylation either directly or through down- suggesting that AXL activation does not drive the AR to MET-TKI in stream targets. In spite of thoroughly examining the genetic and these cells. gene expression background of some EBC1-R cells, we have not Although changes in cell morphology and upregulation of AXL/ identified the alteration responsible for this feature. Because of GAS6/PROS1 were common to all the EBC1-R cells, others were that, we still cannot provide with candidate markers that could be more specific. We found at least two clusters of genetic variants useful for patient's selection. We found mutations that were more and gene methylation profiles among the EBC1-R cells, one that prevalent in the EBC1-R cells with sensitivity to erlotinib, such as contained -PR1, -PR2, -C1, and -V8 cells, and another featuring KIRREL (also called NEPH1), which is involved in cell–cell -PR3 and -PR4 cells. This highlights the expansion of minor interactions that modulate ERK signaling through phosphoryla- subclonal populations that are already present in the parental tion-dependent interaction with GRB2 (39). cells, following a population bottleneck as a result of the massive After exhaustive genetic screening of the EBC1-R cells in search reduction in population size due to the anti-MET treatment. of the event underlying AR to the MET-TKI, we identified, only in Interestingly, the former group was strongly associated with the EBC1-PR4 cells, a rearrangement of the NF2 gene, which leads marked de novo sensitivity to erlotinib. Although could be to an absence of the encoded protein, Merlin. Merlin is a scaffold further validated in in vivo models, our observations are con- protein that mediates RTK signaling events associated with cell sistent with previous reports (18), and evidence of the existence proliferation and survival. It also allows the connection between of multiple simultaneous mechanisms of AR within the same RTKs, actin, and intracellular effectors (40). Reduced NF2 levels in tumor specimen. Collectively, the findings highlight the impor- EBC1-P cells triggered some of the features of the resistant cells but tance of determining the type and degree of tumor heteroge- did not lead to resistance to the MET inhibitor. In spite of this, the neity within a primary tumor to predict and prevent recurrence involvement of NF2 inactivation in AR to MET-TKI cannot be with cancer treatments. ruled out. The Merlin protein could still be detected in the EBC1-P De novo sensitivity to erlotinib following AR to MET-TKIs had cells, although at levels possibly low enough to prevent refracto- previously been demonstrated but, although the mechanism riness to MET inhibition in the EBC1-P cells. The genetic alteration involved the upregulation of the EGFR ligand amphiregulin, the at NF2 detected in the EBC1-PR4 cells could also be the cause of underlying genetic event is unknown (18). Previous observations the refractoriness to erlotinib, since inactivation of NF2 is known have revealed de novo sensitivity to MET-TKI, due to amplification to be associated with AR to the EGFR inhibitors afatinib and of MET, following AR to EGFR TKI (11). Collectively, these cetuximab in lung adenocarcinoma patients (41). In this regard, observations demonstrate a functional connection between MET the EBC1-PR4 cells showed very high levels of NRG1. This codes and EGFR that is responsible for the reciprocal activation of the for a ligand of various HER-family of receptors, including HER4, receptors during AR. As further evidence of this relationship, we which are known to be highly active in human schwannomas and observed that treating EBC1-P cells with MET-TKIs, but not with in NF2-deficient cell lines. In addition, Merlin is rapidly phos- erlotinib, readily reduced the levels of phosphorylation of phorylated following NRG1 binding to HER-receptors, indicating

www.aacrjournals.org Mol Cancer Ther; 16(7) July 2017 OF9

Gimenez-Xavier et al.

that Merlin mediates signal transduction in response to NRG1– Administrative, technical, or material support (i.e., reporting or organizing HER4 binding (42). data, constructing databases): P. Gimenez-Xavier, E. Pros, S. Derdak, In conclusion, our results provide evidence that AR to MET M. Sanchez-Cespedes Study supervision: M. Sanchez-Cespedes inhibitors in EBC1 cells involves several simultaneously acting mechanisms. These include an acquired sensitivity to erlotinib in Acknowledgments fi some, but not all, resistant cells. Our ndings draw attention to The authors acknowledge the technical assistance of Patricia Cabral of the the importance of considering combinatorial therapeutics when Genes and Cancer Group at IDIBELL. designing strategies for second-line treatments in TKI-treated lung cancer patients. Grant Support This work was supported by Spanish grants SAF2014-54571-R, RTICC Disclosure of Potential Conflicts of Interest (RD12/0036/0045 to M. Sanchez-Cespedes); from the ISCIII under the Inte- grated Project of Excellence no. PIE13/00022 (ONCOPROFILE; to M. Esteller No potential conflicts of interest were disclosed. and M. Sanchez-Cespedes), cofunded by FEDER funds/European Regional Development Fund (ERDF)-A Way to Build Europe; from the Ministerio de Authors' Contributions Economía y Competitividad (PTA201409515 to M. Dabad); and from the Conception and design: P. Gimenez-Xavier, M. Sanchez-Cespedes AGAUR Agency to Grups de recerca consolidats (2014SGR641; to M. Sanchez- Development of methodology: P. Gimenez-Xavier, E. Pros, E. Bonastre, A. Aza, Cespedes). The CNAG-CRG laboratory is a member of the Spanish National D. Salinas-Chaparro Bioinformatics Institute (INB), PRB2-ISCIII, and is supported by grant PT13/ þ þ Acquisition of data (provided animals, acquired and managed patients, 0001, of the PE I D i 2013-2016, funded by ISCIII and FEDER (to provided facilities, etc.): E. Bonastre, S. Moran, M. Esteller A. Esteve-Codina). S. Derdak was supported by the Torres Quevedo subpro- Analysis and interpretation of data (e.g., statistical analysis, biostatistics, gramme (MICINN) under grant agreement PTQ-12-05391. computational analysis): P. Gimenez-Xavier, E. Pros, A. Aza, O. Grana,~ The costs of publication of this article were defrayed in part by the payment of advertisement G. Gomez-Lopez, S. Derdak, M. Dabad, A. Esteve-Codina, J.R. Hernandez Mora, page charges. This article must therefore be hereby marked in D. Pisano, M. Sanchez-Cespedes accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Writing, review, and/or revision of the manuscript: P. Gimenez-Xavier, E. Pros, E. Bonastre, S. Moran, A. Aza, S. Derdak, M. Esteller, D. Pisano, M. Sanchez- Received January 31, 2017; revised January 26, 2017; accepted April 3, 2017; Cespedes published OnlineFirst April 10, 2017.

References 1. McDermott U, Sharma SV, Dowell L, Greninger P, Montagut C, Lamb J, 12. Takezawa K, Pirazzoli V, Arcila ME, Nebhan CA, Song X, de Stanchina E, et al. Identification of genotype-correlated sensitivity to selective kinase et al. HER2 amplification: a potential mechanism of acquired resistance to inhibitors by using high-throughput tumor cell line profiling. Proc Natl EGFR inhibition in EGFR-mutant lung cancers that lack the second-site Acad Sci U S A 2007;104:19936–41. EGFRT790M mutation. Cancer Discov 2012;2:922–33. 2. Lynch TJ, Bell DW, Sordella R, Gurubhagavatula S, Okimoto RA, Brannigan 13. Zhang Z, Lee JC, Lin L, Olivas V, Au V, LaFramboise T, et al. Activation of the BW, et al. Activating mutations in the epidermal growth factor receptor AXL kinase causes resistance to EGFR-targeted therapy in lung cancer. Nat underlying responsiveness of non-small-cell lung cancer to gefitinib. Genet 2012;44:852–60. N Engl J Med 2004;20:2129–39. 14. Jia P, Jin H, Meador CB, Xia J, Ohashi K, Liu L, et al. Next-generation 3. Paez JG, J€anne PA, Lee JC, Tracy S, Greulich H, Gabriel S, et al. EGFR sequencing of paired tyrosine kinase inhibitor-sensitive and -resistant mutations in lung cancer: correlation with clinical response to gefitinib EGFR mutant lung cancer cell lines identifies spectrum of DNA therapy. Science 2004;304:1497–500. changes associated with drug resistance. Genome Res 2013;23: 4. McDermott U, Iafrate AJ, Gray NS, Shioda T, Classon M, Maheswaran S, et al. 1434–45. Genomic alterations of anaplastic lymphoma kinase may sensitize tumors 15. Camidge DR, Pao W, Sequist LV. Acquired resistance to TKIs in solid to anaplastic lymphoma kinase inhibitors. Cancer Res 2008;68:3389–95. tumours: learning from lung cancer. Nat Rev Clin Oncol 2014;8:473–81. 5. Blanco R, Iwakawa R, Tang M, Kohno T, Angulo B, Pio R, et al. A gene- 16. Katayama R, Shaw AT, Khan TM, Mino-Kenudson M, Solomon BJ, Halmos alteration profile of human lung cancer cell lines. Hum Mut 2009; B, et al. Mechanisms of acquired crizotinib resistance in ALK rearranged 30:1199–206. lung cancers. Sci Transl Med 2012;4:120ra17. 6. Onozato R, Kosaka T, Kuwano H, Sekido Y, Yatabe Y, Mitsudomi T. 17. Cepero V, Sierra JR, Corso S, Ghiso E, Casorzo L, Perera T, et al. MET and Activation of MET by gene amplification or by splice mutations deleting KRAS gene amplification mediates acquired resistance to MET tyrosine the juxtamembrane domain in primary resected lung cancers. J Thorac kinase inhibitors. Cancer Res 2010;70:7580–90. Oncol 2009;4:5–11. 18. McDermott U, Pusapati RV, Christensen JG, Gray NS, Settleman J. Acquired 7. Weiss J, Sos ML, Seidel D, Peifer M, Zander T, Heuckmann JM, et al. resistance of non-small cell lung cancer cells to MET kinase inhibition is Frequent and focal FGFR1 amplification associates with therapeutically mediated by a switch to epidermal growth factor receptor dependency. tractable FGFR1 dependency in squamous cell lung cancer. Sci Transl Med Cancer Res 2010;70:1625–34. 2010;2:62ra9. 19. Rosen N, Solit DB, Weiser MR. HER kinase activation confers resistance to 8. McDermott U, Ames RY, Iafrate AJ, Maheswaran S, Stubbs H, Greninger P, MET tyrosine kinase inhibition in MET oncogene-addicted gastric cancer et al. Ligand-dependent platelet-derived growth factor receptor (PDGFR)- cells. Mol Cancer Ther 2008;7:3499–508. alpha activation sensitizes rare lung cancer and sarcoma cells to PDGFR 20. Lee NV, Lira ME, Pavlicek A, Ye J, Buckman D, Bagrodia S, et al. kinase inhibitors. Cancer Res 2009;69:3937–46. A novel SND1-BRAF fusion confers resistance to c-Met inhibitor 9. Rosell R, Bivona TG, Karachaliou N. Genetics and biomarkers in perso- PF-04217903 in GTL16 cells through APK activation. PLoS One nalisation of lung cancer treatment. Lancet 2013;382:720–31. 2012;7:e39653. 10. Pao W, Miller VA, Politi KA, Riely GJ, Somwar R, Zakowski MF, et al. 21. Turke AB, Zejnullahu K, Wu YL, Song Y, Dias-Santagata D, Lifshits E, et al. Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is Preexistence and clonal selection of MET amplification in EGFR mutant associated with a second mutation in the EGFR kinase domain. PLoS Med NSCLC. Cancer Cell 2010;17:77–88. 2005;2:e73. 22. Christensen JG, Schreck R, Burrows J, Kuruganti P, Chan E, Le P, et al. 11. Engelman JA, Zejnullahu K, Mitsudomi T, Song Y, Hyland C, Park JO, et al. A selective small molecule inhibitor of c-Met kinase inhibits c-Met-depen- MET amplification leads to gefitinib resistance in lung cancer by activating dent phenotypes in vitro and exhibits cytoreductive antitumor activity in ERBB3 signaling. Science 2007;316:1039–43. vivo. Cancer Res 2003;63:7345–55.

OF10 Mol Cancer Ther; 16(7) July 2017 Molecular Cancer Therapeutics

Acquired Resistance to MET Inhibitors in Lung Cancer

23. Mohammadi M, Froum S, Hamby JM, Schroeder MC, Panek RL, Lu GH, 33. Tiedt R, Degenkolbe E, Furet P, Appleton BA, Wagner S, Schoepfer J, et al. et al. Crystal structure of an angiogenesis inhibitor bound to the FGF A drug resistance screen using a selective MET inhibitor reveals a spectrum receptor tyrosine kinase domain. EMBO J 1998;17:5896–904. of mutations that partially overlap with activating mutations found in 24. Holland SJ, Pan A, Franci C, Hu Y, Chang B, Li W, et al. R428, a selective cancer patients. Cancer Res 2011;71:5255–64. small molecule inhibitor of Axl kinase, blocks tumor spread and prolongs 34. Martin-Belmonte F, Perez-Moreno M. Epithelial cell polarity, stem cells survival in models of metastatic breast cancer. Cancer Res 2010;70: and cancer. Nat Rev Cancer 2011;12:23–38. 1544–54. 35. Suda K, Mizuuchi H, Maehara Y, Mitsudomi T. Acquired resistance 25. Geb€ack T, Schulz MM, Koumoutsakos P, Detmar M. TScratch: a novel and mechanisms to tyrosine kinase inhibitors in lung cancer with activating simple software tool for automated analysis of monolayer wound healing epidermal growth factor receptor mutation-diversity, ductility, and desti- assays. Biotechniques 2009;46:265–74. ny. Cancer Metastasis Rev 2012;31:807–14. 26. Pros E, Lantuejoul S, Sanchez-Verde L, Castillo SD, Bonastre E, Suarez- 36. Elkabets M, Pazarentzos E, Juric D, Sheng Q, Pelossof RA, Brook S, et al. AXL Gauthier A, et al. Determining the profiles and parameters for gene mediates resistance to PI3Ka inhibition by activating the EGFR/PKC/ amplification testing of growth factor receptors in lung cancer. Int J Cancer mTOR axis in head and neck and esophageal squamous cell carcinomas. 2013;133:898–907. Cancer Cell 2015;27:533–46. 27. Pereira C, Gimenez-Xavier P, Pros E, Pajares MJ, Moro M, Gomez A, et al. 37. Brand TM, Iida M, Stein AP, Corrigan KL, Braverman CM, Luthar N, et al. Genomic profiling of patient-derived xenografts for lung cancer identifies AXL mediates resistance to cetuximab therapy. Cancer Res 2014;74: B2M inactivation impairing immunorecognition. Clin Cancer Res 2017 5152–64. March 16. [Epub ahead of print]. 38. Scaltriti M, Elkabets M, Baselga J. Molecular pathways: AXL, a membrane 28. Castillo SD, Angulo B, Suarez-Gauthier A, Melchor L, Medina PP, Sanchez- receptor mediator of resistance to therapy. Clin Cancer Res 2016;22: Verde L, et al. Gene amplification of the transcription factor DP1 and 1313–7. CTNND1 in human lung cancer. J Pathol 2010;222:89–98. 39. Harita Y, Kurihara H, Kosako H, Tezuka T, Sekine T, Igarashi T, et al. Neph1, 29. Stitt T N, Conn G, Gore M, Lai C, Bruno J, Radziejewski C, et al. The a component of the kidney slit diaphragm, is tyrosine-phosphorylated by anticoagulation factor protein S and its relative, Gas6, are ligands for the the Src family tyrosine kinase and modulates intracellular signaling by Tyro 3/Axl family of receptor tyrosine kinases. Cell 1995;80:661–70. binding to Grb2. J Biol Chem 2008;283:9177–86. 30. Sandoval J, Heyn H, Moran S, Serra-Musach J, Pujana MA, Bibikova M. 40. Petrilli AM, Fernandez-Valle C. Role of Merlin/NF2 inactivation in tumor Validation of a DNA methylation microarray for 450,000 CpG sites in the biology. Oncogene 2016;35:537–48. human genome. Epigenetics 2011;6:692–702. 41. Pirazzoli V, Nebhan C, Song X, Wurtz A, Walther Z, Cai G, et al. Acquired 31. Azam M, Latek RR, Daley GQ. Mechanisms of autoinhibition and STI-571/ resistance of EGFR-mutant lung adenocarcinomas to afatinib plus cetux- imatinib resistance revealed by mutagenesis of BCR-ABL. Cell 2003; imab is associated with activation of mTORC1. Cell Rep 2014;7:999–1008. 112:831–43. 42. Thaxton C, Lopera J, Bott M, Fernandez-Valle C. Neuregulin and laminin 32. Sequist LV, Waltman BA, Dias-Santagata D, Digumarthy S, Turke AB, Fidias stimulate phosphorylation of the NF2 tumor suppressor in Schwann cells P, et al. Genotypic and histological evolution of lung cancers acquiring by distinct protein kinase A and p21-activated kinase-dependent pathways. resistance to EGFR inhibitors. Sci Transl Med 2011;3:75ra26. Oncogene 2008;27:2705–15.

www.aacrjournals.org Mol Cancer Ther; 16(7) July 2017 OF11

Genomic and Molecular Screenings Identify Different Mechanisms for Acquired Resistance to MET Inhibitors in Lung Cancer Cells

Pol Gimenez-Xavier, Eva Pros, Ester Bonastre, et al.

Mol Cancer Ther Published OnlineFirst April 10, 2017.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-17-0104

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2017/04/08/1535-7163.MCT-17-0104.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/early/2017/06/14/1535-7163.MCT-17-0104. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.