Oncogenic switch and single-agent MET inhibitor sensitivity in a subset of EGFR-mutant lung cancer

The clinical efficacy of epidermal growth factor receptor (EGFR)–targeted therapy in EGFR-mutant non–small cell lung cancer is limited by the development of drug resistance. One mechanism of EGFR inhibitor resistance occurs through amplification of the human growth factor receptor (MET) proto-oncogene, which bypasses EGFR to reactivate downstream signaling. Tumors exhibiting concurrent EGFR mutation and MET amplification are historically thought to be codependent on the activation of both oncogenes. Hence, patients whose tumors harbor both alterations are commonly treated with a combination of EGFR and MET tyrosine kinase inhibitors (TKIs). Here, we identify and characterize six patient-derived models of EGFR-mutant, MET-amplified lung cancer that have switched oncogene dependence to rely exclusively on MET activation for survival. We demonstrate in this MET-driven subset of EGFR TKI-refractory cancers that canonical EGFR downstream signaling was governed by MET, even in the presence of sustained mutant EGFR expression and activation. In these models, combined EGFR and MET inhibition did not result in greater efficacy in vitro or in vivo compared to single-agent MET inhibition. We further identified a reduced EGFR:MET mRNA expression stoichiometry as associated with MET oncogene dependence and single-agent MET TKI sensitivity. Tumors from 10 of 11 EGFR inhibitor–resistant EGFR-mutant, MET-amplified patients also exhibited a reduced EGFR:MET mRNA ratio. Our findings reveal that a subset of EGFR-mutant, MET-amplified lung cancers develop dependence on MET activation alone, suggesting that such patients could be treated with a single-agent MET TKI rather than the current standard-of-care EGFR and MET inhibitor combination regimens.

incubation in drug, cells were fixed in 10% formalin, and stained with 0.05% (w/v) crystal violet diluted in ultrapure water containing a final concentration 20% ethanol v/v. Following up to one hour of staining, wells were rinsed twice in ultrapure water, and placed upside-down to dry completely prior to imaging.
For active caspase 3/7 quantification, cells were seeded at 3,000-5,000 cells/well and treated the following day with TKIs and CellEvent Caspase-3/7 Green ReadyProbes reagent (Fisher Scientific). The confluence and caspase activation of cells were monitored using an IncuCyte FLR Live Cell Analysis system (Essen Bioscience). At the conclusion of the study, data were analyzed using IncuCyte image analysis software. Caspase-3/7-positive fluorescent signal count was normalized to confluence following 96 hours of TKI treatment.

Organoid Cell Viability Assay
The cells in 3D matrices were harvested and dissociated using TrypLE Express Enzyme (Thermo Fisher Scientific) at 37°C. The single cells were resuspended in cold RETM/10% Matrigel and plated into 384-well ultra-low attachment microplates (Corning) at 1000 cells per well. On the next day, cells were treated with DMSO, osimertinib and crizotinib dose gradients, or 500 nM osimertinib combined with crizotinib dose curve, as indicated. Each dose had more than three replicates. At the study endpoint, the representative images of each condition were captured under microscope (Olympus CKX41 with DP72 camera). The viability assay was performed using CellTiter-Glo 3D (Promega) according to the manufacturer's protocol. Statistical significance was accessed using ANOVA and Turkey's post-test for multiple comparisons.

Co-immunoprecipitation and Western blotting
Cells were seeded at 500,000 -750,000 cells/100 mm dish for Western blots, and 1.5-2 x 10 6 cells/150 mm dish for immunoprecipitation. Cells were treated the following day with drugs, concentrations, and durations specific to each study, and were lysed in RIPA Buffer or  Lysis Buffer (table S3). Lysates were quantified, diluted, and immunoprecipitated using previously described protocols (9). Briefly, for western blot, cells were lysed in RIPA buffer with Triton X (Boston BioProducts Inc.). Cells for immunoprecipitation and tumors for western blot were lysed and homogenized, as necessary, in 1X NP-40 buffer (Boston BioProducts Inc.).
Lysis buffers were supplemented with phosphatase inhibitor (Sigma Aldrich) and protease inhibitor tablets (Sigma Aldrich). Lysates were quantified by BCA assay (Thermo Fisher Scientific) and measured on a PolarSTAR Omega Plate Reader. Lysates were normalized according to a BCA standard curve run alongside samples. For Western blots, samples were diluted in SDS sample reducing buffer (Boston BioProducts Inc.), heated to 100°C, and stored at -80°C until analysis. For CO-IP, quantified lysates were uniformly diluted in NP-40 buffer to 300 μg -1 mg per IP in 250 -500 μL volume per IP in 1.7 mL microcentrifuge tubes (Thomas Scientific). Corresponding antibodies (table S3) were added at a 1:50 or 1:100 dilution, depending on product recommendations, and tubes were incubated overnight at 4°C on a rotator.
The following day, antibodies and attached proteins were incubated with A/G-Plus Agarose beads (Santa Cruz Biotechnology) for >1 hour at 4°C, and subsequently immunoprecipitated by centrifugation at 500xg for 5 minutes. Beads were washed three times in 1X NP-40 buffer prior to resuspension and boiling in reducing buffer. As with western blot lysates, samples were stored at -80°C until analysis following the addition of reducing buffer.
Samples for Western blot ands immunoprecipitation analysis were electrophoresed alongside a protein size standard ladder (Bio-Rad Laboratories) on precast 4-12% Bis-Tris gels (Life Technologies) and transferred to Immobulin-P membranes (Fisher Scientific), following established protocols (46) and using the reagents detailed in table S3. After protein transfer, Immobulin-P membranes were cut by size, blocked with 5% (m/v) non-fat milk (Bio-Rad Laboratories) in 1X Tris-Buffered Saline (Westnet Inc.)/0.1% Tween®20 (Sigma Aldrich), and incubated overnight on a rocker at 4°C in primary antibodies (table S3) diluted 1:1000 in 3% (m/v) BSA in TBS-T (1X Tris-Buffered Salinen/0.1% Tween20). The following day, membranes were washed in TBS-T, incubated for one hour at room temperature in HRP-conjugated secondary antibody diluted 1:5000 in 3% BSA (table S3), re-washed, and developed with HRP substrate (Life Technologies). Western blots were imaged and analyzed on an Amersham Imager 600 (GE Healthcare Life Sciences), as previously described (46). Immunoprecipitation studies were quantified using ImageQuant TL Image Analysis Software (GE Healthcare Life Sciences).
Uniform grids were used to ensure that an equal region was quantified for each cell line/protein.
For ERBB3 pulldowns, p85 bands were quantified and normalized to ERBB3 bands from same co-IP; ERBB3 quantification was normalized to p85 in reciprocal pulldown studies. Where applicable, Western blots were quantified using adobe photoshop following image inversion and selection of uniform areas. All quantified Western blot bands were normalized to coinciding loading control or coinciding loading control and total protein, as indicated in figure legends, prior to plotting.
qRT-PCR, and NGS RNA was extracted from cell lines and snap-frozen tumors using TRIzol or the Qiagen RNeasy RNA Isolation kit, abiding by manufacturers' protocols. cDNA was synthesized from RNA using the QuantiTect Reverse Transcription Kit from Qiagen, as previously described (48). Following cDNA synthesis, real-time qPCR was carried out according to manufacturers' instructions using TaqMan Gene Expression Master Mix from Applied Biosystems (Life Technologies), and primers for human EGFR (FAM reporter) and MET (Fam reporter). Gene expression was normalized to β-actin (VIC reporter) and GUSB (VIC reporter) measurements taken from each well (table S1). qRT-PCR was carried out in optical 96-well plates (Fisher Scientific) on a StepOne Real-Time PCR System (Applied Biosystems), and three technical replicates were measured for each cell line/gene. Each cell line and xenograft PCR quantification study was repeated for a total of at least three biological replicates, unless otherwise noted. Where relevant, EGFR:MET expression ratios were calculated as 2 -(ΔCt(EGFR)-ΔCt(MET)) , where ΔCt = CtEGFR or MET -CtACTB or GUSB. Data were visualized and statistical analyses were carried out using GraphPad Prism software. To determine allelic expression fraction, exons 19 and 21 of EGFR were amplified from cDNAs using Phusion High-Fidelity Polymerase (New England Biolabs) and cDNA specific primers (table S1) in accordance with manufacturer protocols and submitted to the Massachusetts General Hospital Center for Computational and Integrative Biology DNA core for targeted next-generation sequencing.
DNA FISH and BaseScope RNA ISH Analysis MET genomic FISH was carried out on 10% formalin fixed cell lines or formalin fixed paraffin embedded (FFPE) tumor tissues with centromere 7 (CEN7) as a control marker, using previously published protocols and probes (48). Mean MET copy number (CN) was based on analysis of at least 30 nuclei per specimen. MET amplification was defined as either MET-CN > 5 or MET/CEP7 > 2, in accordance with the criteria used for clinical trial enrollment for osimertinib and savolitinib (17). For BaseScope analysis, cell lines were pelleted prior to formalin fixation.
Cell line and tumor specimens were prepared by FFPE and arranged into tissue microarrays for submission to ACD Bio for processing and image scanning. BaseScope images were quantified as signal area/cell area using ImageJ image processing software.
RT-ddPCR cDNA-specific RT-ddPCR primers were designed to span exon-intron junctions to exclude genomic DNA. Probes were designed with HEX or FAM reporter dye and ZEN/Iowa Black FQ quencher (table S1). Primers and probes were designed specific to wild type MET or mutant EGFR, with mutation-specific probes used for distinct EGFR exon 19 deletions. Primers and probes were manufactured by Integrated DNA Technologies. RT and ddPCR reactions were prepared using One-Step RT-ddPCR Advanced Kit for Probes in a reaction (20 μL) including 1xSupermix, 20U/μL reverse transcriptase, 15mM DTT, 900 nM primers, 250 nM probes and 10 ng RNA. After droplets generation (Automated Droplet Generator, Bio-Rad), the reactions cycled through following conditions: 42°C/60 min; 95°C/10 min; 40 cycles of 95°C/30 sec followed by 55°C/1 min; 98°C/ 10 min; and 4°C hold. The thermocycler plate was read by a QX200 Reader (Bio-Rad) and data analyzed by QuantaSoft software.

Expression Vectors and Viral Transduction
EGFR L858R and EGFR Del19 ORFs were cloned into the pRetroX doxycycline inducible plasmid by gateway cloning, as described in Supplementary Methods. Plasmids were validated by sequencing and transfected into HEK293T/17 cells alongside pRetro-gag/pol and MDM2.G retroviral plasmids (table S1). The doxycycline-inducible Retro-X Tet-One system acquired from Clontech Laboratories, Inc. was cloned in accordance with manufacturer's user manual. EGFR Del19 and EGFR L858R open reading frames were PCR amplified from a donor vector using the primers described in table S1, and cloned into pRetroX-TetOne vector using the In-Fusion HD Cloning Plus Kit (Takara Bio), abiding by manufacturer protocols. Plasmids were transformed and grown in competent bacteria (Fisher Scientific), and isolated by plasmid miniprep (table S3). Following validation of cloned pRetroX-TetOne vectors by sequencing (table S1), EGFR expression plasmids were combined with Retro gag/pol packaging vector (Addgene) and envelope vector pMD2.G (Addgene), and transfected into HEK293/17 cells plated the previous day in complete DMEM media supplemented with tetracycline-free FBS.
Transfection to make virus was carried out using the Xfect transfection reagent (Takara Bio) in accordance kit protocols, and complete DMEM/tetracycline-free FBS media was replenished less than 12 hours after transfection, and every 24 hours subsequently. Media containing 36-72-hour retrovirus was harvested from the supernatant of HEK293T cells, passed through a 0.45 μm syringe filter (BD and Fisher Scientific), supplemented with a final concentration of 10 μg/mL Polybrene (Santa Cruz Biotechnology), and added to DFCI81 and DFCI161. Transduction was carried out overnight, and virus-containing media was removed from cells within 24 hours.
Following a two-day recovery period in complete RPMI-1640 supplemented with tetracyclinefree FBS, transduced DFCI81 and DFCI161 cells were selected for two weeks with 2 µg/mL puromycin (Fisher Scientific), according to standard protocols (46). DFCI81 cells ectopically overexpressing ERBB3 on a retroviral JP1540 vector or control empty vector plasmid ("FLAG") (33) were generated and selected through an identical workflow as the pRetroX cells. All assays were performed following one week of puromycin washout, and within one month of Tumor outgrowth was measured during short-term and long-term drug withdrawal after singleagent MET inhibitor versus combined EGFR and MET inhibitor treatment. Statistical analysis by one-way ANOVA followed by Dunnett's multiple comparisons test revealed that the EGFR/MET co-dependent HCC827GR6 cells exhibited a significantly higher Bliss synergy score compared to EGFR-dependent and MET-dependent models.           Fig. 5A were pooled, and a lack of statistical significance between treatment arms was determined by one-way ANOVA followed by Tukey's post-test for multiple comparisons.    Fig. 5F were pooled, and a lack of statistically significant difference between treatment arms was determined by one-way ANOVA followed by Tukey's post-test for multiple comparisons.  In contrast to EGFR-mutant lung cancers that exhibit EGFR dependence and EGFR-mutant, MET-amplified lung cancers that develop EGFR/MET co-dependence, a subset of EGFR-mutant NSCLC shows MET dependence accompanied by downregulated EGFR transcript and protein expression and preferentially MET-mediated activation of downstream kinase cascades.