Kinome RNAi Screens Reveal Synergistic Targeting of MTOR and FGFR1 Pathways for Treatment of Lung Cancer and HNSCC

Our studies and those of others demonstrate that overexpressed, nonmutated FGFR1 participates as an oncogenic driver via autocrine FGFs in cell lines derived from lung cancers of all histologies (1–5), head and neck squamous cell carcinomas (HNSCC; refs. 6, 7) and malignant pleural mesothelioma (8). As a result, multiple early-phase clinical trials of FGFR-targeting tyrosine kinase inhibitors (TKI) are now under way, including a study of the multikinase TKI ponatinib (9) in lung cancer at our institution (NCT01935336).

The clinical efficacy of FGFR TKIs as single anticancer agents is not fully realized. Yet, the problem of intrinsic and acquired resistance to TKI monotherapy has emerged as a major limitation to long-term control or cure of solid tumors (10–13) and portends similar difficulties with single FGFR TKIs as therapeutics. Defining mechanisms of acquired resistance to targeted therapeutics is an ongoing subject of intense investigation and sets the stage for strategies to deploy inhibitors of the resistance mechanisms following treatment failure of the initial drug. Thus, “serial monotherapy” has emerged as a logical approach in clinical oncology for solid tumors, including lung cancer. In this regard, however, it is important to review the lessons learned from acquired resistance to antimicrobial and antiviral monotherapy over the past 60 years (reviewed in ref. 12). The present strategy to combat acquired resistance to monotherapy in cancer by deploying sequential therapies to block emergent resistance pathways (i.e., MET inhibitors after resistance to EGFR-specific TKIs) failed as a strategy to cure TB and HIV infections.

Importantly, therapeutic success in HIV and TB infections was only achieved when combinations of inhibitors were deployed that induced rapid and synergistic suppression of the infectious agent at the onset of therapy, thereby preventing the emergence of drug resistance (12). We hypothesize that the development of rational, mechanism-based combinations of inhibitors that simultaneously inhibit multiple elements within transforming receptor tyrosine kinase (RTK) coactivation networks (14) active in cancer cells may achieve a similar impact on cancer cure or control. In this study, we deployed functional genomics screens with a kinome-targeting shRNA library to identify auxiliary pathways that co-signal with FGFR1 in lung cancer and HNSCC cell lines. Our studies establish MTOR as a protein kinase with essential properties in some FGFR1-dependent cancer cell lines as well as auxiliary, synthetic lethal (SL) properties in the context of FGFR inhibitors in other cell lines. In sum, our findings identify MTOR as a protein kinase that contributes to the intrinsic sensitivity of cancer cells to FGFR TKIs such that combined treatment with MTOR inhibitors and FGFR TKIs elicits synergistic growth inhibition. Thus, direct MTOR kinase inhibitors are attractive agents to consider combining with FGFR-specific TKIs for treatment of FGFR1-dependent lung cancers and HNSCCs.

All cancer cell lines used in this study were submitted to fingerprint analysis by the University of Colorado Cancer Center DNA Sequencing and Analysis Core to confirm their authenticity. Cell lines were routinely cultured in RPMI-1640 growth medium (Invitrogen) supplemented with 10% FBS with 1% penicillin–streptomycin (Sigma-Aldrich) at 37°C in a humidified 5% CO₂ incubator.

The Human Kinase TRC shRNA library (obtained from the Functional Genomics Shared Resource within the University of Colorado Cancer Center) was packaged in 293T cells as follows. 293T cells were incubated overnight at 37°C in a 5% CO₂ humidified incubator with Turbofect transfection reagent, 5.0 μg pΔ8.9, 5.0 μg pCMV-VSV-G and 3.0 μg kinome library. The virus-containing media from the 293T cells was then filtered through a 0.45-μm filter after adding 1 μg/mL polybrene and either used immediately as described or stored at 4°C until ready for use. In experiments where MTOR was validated as an essential or SL kinase, two MTOR shRNAs in the pLKO.2 lentiviral vector (TRCN0000332888 and TRCN0000363722) distinct from those shRNAs included in the library or an shRNA-targeting GFP as a negative control were packaged with the pCMV-VSV-G and pΔ8.9 component vectors. The virus was titered on NIH3T3 fibroblasts and the effect on cell growth was measured by clonogenic growth assay as described previously.

For details on functional genomics screens, see Supplementary Information.

To measure the effect of inhibitors or shRNA-mediated knockdown on cell growth, cells were seeded at 100 cells per well in 6-well tissue culture plates in full media. After 24 hours, cells were treated as indicated and cultured for 14 days with the addition of fresh media containing inhibitors every 7 days. Plates were rinsed twice with PBS, fixed, and stained with a solution of 0.5% (w/v) crystal violet in 6.0% (v/v) gluteraldehyde for 30 minutes at room temperature. Plates were rinsed in distilled H₂O and photographed. For measurement of anchorage-independent growth in soft agar, 20,000 cells were suspended in 1.5 mL media and 0.35% noble agar and overlaid on base layers containing 1.5 mL media and 0.5% noble agar in 6-well plates. Wells overlaid with 2 mL media containing drugs were fed once a week and allowed to grow for 14 to 21 days. Viable colonies were stained for 24 hours with 250 μL 1 mg/mL nitroblue tetrazolium. Digital photographs of both clonogenic and soft agar wells were used to quantify total colony area by Metamorph imaging software (Molecular Devices).

Cells were plated at 100 cells per well in 96-well tissue culture plates and treated with inhibitors at various doses. When the DMSO-treated control wells became confluent (1,2, weeks), cell numbers were assessed using a CyQUANT Direct Cell Proliferation Assay (Invitrogen) according to the manufacturer's instructions.

H1581 and Colo699 cells were suspended in 50% Matrigel/PBS at 10 million cells per mL and 1 million cells were injected s.c. in both flanks of female nu/nu mice according to protocols approved by the University of Colorado Institutional Animal Care and Use Committee (IACUC). When at least one of the tumors reached a volume of 100 mm³, the mice were randomized into treatment groups (n = 9,10/group) of diluent control (1% Polysorbate 80), AZD4547 (12.5 mg/kg), AZD2014 (10 mg/kg) or both AZD4547 and AZD2014. Drugs were delivered daily by oral gavage (∼0.25 mL/mouse). Peak plasma levels of AZD4547 following similar dosing are approximately 1 to 2 μmol/L (15). Also, a total C_max of 11 μmol/L (free C_max of 0.51 μmol/L) is achieved 30 minutes after a single 10 mg/kg dose of AZD2014 (S. Cosulich, AstraZeneca; unpublished data). Tumor volumes were determined by caliper measurements of the long and short diameter performed twice per week using the modified ellipsoid formula for volume, V = 1/2(length × width²). Mice were euthanized when tumors reached a volume greater than 2 cm³ or they exhibited signs of morbidity specified in the IACUC protocol.

Phospho-ERK, total ERK, phospho-AKT S473, total AKT, p-p70S6K T389, total p70S6K, p-S6 S235/236, total S6, p-Rictor T1135, total Rictor, MTOR, and PARP1 were measured by immunoblotting using antibodies obtained from Cell Signaling Technology, Inc. Aliquots of cell extracts prepared in lysis buffer (0.5% Triton X-100, 50 mmol/L β-glycerophosphate (pH 7.2), 0.1 mmol/L Na₃VO₄, 2 mmol/L MgCl₂, 1 mmol/L EGTA, 1 mmol/L DTT, 0.3 mol/L NaCl, 2 μg/mL leupeptin, and 4 μg/mL aprotinin) were submitted SDS-PAGE. After electrophoretic transfer to nitrocellulose, filters were blocked in 3% BSA (Cohn Fraction V; ICN Biomedicals, Inc.) in TBS with 0.1% Tween 20 (TTBS). The filters were then incubated overnight at 4°C with antibodies, washed three times in TTBS, and incubated for 1 hour at room temperature with alkaline phosphatase–coupled goat anti-rabbit antibodies. The filters were developed using Luminata Classico substrate (Millipore Corporation) according to the manufacturer's instructions. When blotting phosphorylated proteins, the filters were stripped and probed for the corresponding total signaling enzyme level or Na/K-ATPase α-subunit (Santa Cruz Biotechnology, Inc.) as a loading control.

H1581 cells were plated at 200,000 cells per well in 6-well tissue culture plates in full media. After 24 hours, cells were treated with DMSO, 300 nmol/L AZD4547, 100 nmol/L AZD8055, or combination in triplicate and cultured for 3 days. Cells were harvested and caspase-3 activity was assessed using the Caspase-3 Cellular Activity Assay Kit PLUS (Enzo Life Sciences, Inc.) according to the manufacturer's instructions.

Ponatinib was obtained by material transfer agreement from Ariad Pharmaceuticals, Inc. AZD4547 and AZD2014 were obtained by material transfer agreement from AstraZeneca. AZD8055, GSK690693 and MK2206 were purchased from Selleck Chemicals. Stocks of the drugs were prepared in DMSO at concentrations such that the final concentration of DMSO was 0.1% v/v.

Our recent studies demonstrate through molecular and pharmacologic approaches an autocrine role for nonmutated FGFR1 in multiple cancers, including lung cancer, malignant mesothelioma, and HNSCC (1,2,6,8). On the basis of precedent from monotherapy with TKIs in other oncogene-driven lung cancers, including those bearing mutant EGFR or rearranged ALK (10,11,13), intrinsic and/or acquired resistance mechanisms are predicted to limit the clinical response of FGFR inhibitors as well. To screen for signal pathways whose activity reduces intrinsic sensitivity of lung cancer cell lines to FGFR inhibitors, we performed functional genomics-based SL screens with a kinome-targeting shRNA library (see Materials and Methods) and lung cancer cell lines exhibiting high sensitivity (Colo699, H520, and H1703) and moderate/low sensitivity (H1299 and H157) to ponatinib (Table 1; ref. 2). Although ponatinib is a multikinase inhibitor with IC₅₀ values ranging from 0.2 to 8 nmol/L on ABL and SRC family kinases, FGFR1, 2 and 4, PDGFRs, VEGFRs and RET (16), our recent studies in lung cancer and mesothelioma cell lines indicate that sensitivity to the TKI is closely associated with FGFR1 expression and function (2,8). The cell lines were transduced with lentiviruses encoding the kinome-targeting shRNAs in the pLKO.1 vector at a multiplicity of infection <1. Following selection for puromycin resistance, which eliminates nontransduced cells and cells expressing an shRNA-targeting an essential gene, the cells were treated with or without ponatinib for 3 days followed by an additional 3 days of culture in medium lacking the drug. The shRNAs were amplified from genomic DNA by PCR and submitted to massively parallel deep sequencing, and the reads were analyzed by BiNGS!SL-seq (17) to determine the count frequency of the individual shRNAs in the control and treated samples. Kinase genes that are SL with respect to ponatinib treatment are defined herein by significantly decreased shRNA counts in the treated samples with at least two independent shRNAs.

The functional genomics analysis identified MTOR as the top-ranking SL hit in both H157 and H1299 cells, but not in the more ponatinib-sensitive H520 and H1703 cells (Supplementary Table S2) or Colo699 cells (data not shown). To validate MTOR as an SL gene with respect to ponatinib in H157 and H1299 cells, lentiviral-encoded shRNAs distinct from those used in the kinome shRNA library were transduced into H157 and H1299 cells. As shown in Fig. 1A, both MTOR-targeting shRNAs reduced MTOR protein levels relative to a nonsilencing control shRNA-targeting GFP. Although MTOR silencing, alone, exerted little or no effect on clonogenic growth of H157 or H1299 cells, enhanced growth inhibition by ponatinib upon MTOR knockdown was observed in both cell lines (Fig. 1B-D). Finally, treatment of H157 cells with ponatinib and the MTOR inhibitor, AZD8055 (18), yielded significantly greater clonogenic growth inhibition than ponatinib alone (Supplementary Fig. S3). These results validate MTOR as an SL gene with ponatinib.

MTOR was not identified as an SL protein kinase in H520, H1703, or Colo699 cells (Supplementary Table S2 and data not shown). Because the SL screen format (see Supplementary Fig. S1) eliminates shRNAs targeting essential kinases due to the selection for stable puromycin resistant cells, it is possible that MTOR shRNAs were eliminated from the transduced cell population in Colo699, H520, and H1703 cells. We performed an essential kinase screen in FGFR1-dependent Colo699 lung adenocarcinoma cells and FGFR1-dependent 584-A2 HNSCC cells (see Supplementary Materials and Methods). The count frequency of multiple MTOR shRNAs was determined 2 and 7 days after transduction by sequencing of amplified shRNAs and the results are presented graphically in Fig. 2A. Five of six independent MTOR-targeting shRNAs were eliminated from the pool of recovered shRNAs in Colo699 cells 7 days after transduction. By contrast, none of the six shRNAs were eliminated from 584-A2 cells after 7 days of incubation. Thus, the results are consistent with an essential role of MTOR in Colo699 cells, but not 584-A2 cells. The essential function of MTOR in Colo699 cells was further validated with two shRNA distinct from those used in the kinome library where strong reduction of clonogenic growth resulted from transduction of MTOR shRNAs relative to the GFP-targeting shRNA control (Fig. 2B).

As a pharmacologic approach to define the relative requirement of MTOR for growth and survival in FGFR1-dependent cancer cell lines, clonogenic cell growth was measured in a panel of cell lines in the presence of increasing concentrations of AZD8055. AZD8055, and the congener, AZD2014 are highly selective for MTOR in either the TORC1 or TORC2 complexes, with little activity on other members of the phosphatidylinositol kinase superfamily (18). In clonogenic/anchorage-independent growth assays, H157 and 584-A2 cells exhibited low sensitivity (IC₅₀ > 600 nmol/L) and H1299 cells intermediate sensitivity (IC₅₀ ∼180 nmol/L) to AZD8055 (Fig. 2C; Table 1). By contrast, Colo699, H520, and H1703 as well as the FGFR1-dependent lung cancer cell line, H1581, and the HNSCC cell line, CCL30, were uniformly sensitive to AZD8055 with IC₅₀ values <80 nmol/L. This rank order of sensitivity among the cancer cell lines to AZD8055 was confirmed with a cell proliferation assay (Supplementary Fig. S4) and was similar to that observed with AZD2014 (Table 1). The panel of cell lines was very sensitive to the TORC1-specific inhibitor, rapamycin, although little or no difference in relative sensitivity between the cell lines was observed with IC₅₀ values ranging from 0.3 to 2 nmol/L (Supplementary Fig. S5). Together, the studies indicate that FGFR1-dependent lung cancer and HNSCC cell lines exhibit variable degrees of MTOR dependency for growth and survival and explain the observation that MTOR was not identified as SL with ponatinib in Colo699, H520, and H1703 cells because it exerts an essential function in these cell lines relative to H157 and H1299 cells.

Despite the range of MTOR dependencies measured in the panel of cancer cell lines, we tested whether FGFR and MTOR inhibitors might still yield additive or synergistic growth inhibition in cell lines even when MTOR exhibits an essential phenotype. As shown in Supplementary Fig. S6, combination treatment with either ponatinib or AZD4547, a specific FGFR1,2,3 inhibitor (15), and AZD8055 resulted in significantly greater clonogenic growth inhibition relative to FGFR inhibitor, alone, in Colo699, H1703, and H520 cells. To rigorously test for synergistic growth inhibition resulting from simultaneous blockade of FGFR1 and MTOR, FGFR1-dependent lung cancer cell lines (H1581 and Colo699) or an HNSCC cell line (CCL30) exhibiting high sensitivity to AZD8055 (Table 1) were treated in a 96-well plate format (see Materials and Methods) with multiple concentrations of AZD4547, alone and in combination with the MTOR inhibitor, AZD8055. The effects on cell growth measured by the CyQUANT assay are shown in Fig. 3 and Supplementary Fig. S7, and analysis of the resulting data by the method of Chou and Talalay (19) revealed synergistic growth inhibition over multiple concentrations of the two drugs. The greater than additive growth inhibition by combined AZD4547 and AZD8055 was confirmed with the distinct MTOR inhibitor, AZD2014 (18), in H1581, Colo699, and 584-A2 cells (Fig. 4) and with rapamycin and AZD4547 in Colo699 and H1581 cells (Supplementary Fig. S8). Thus, the identification of MTOR as an SL pathway with FGFR1 is observed pharmacologically in multiple FGFR1-dependent cancer cell lines.

To explore the mechanism by which simultaneous inhibition of FGFR1 and MTOR yield synergistic growth inhibition, the activity of signaling pathways known to be regulated by these protein kinases was monitored by immunoblot analyses. Figure 5A reveals that AZD4547 inhibited ERK phosphorylation, but had little or no effect on phosphorylation of the TORC2 site (S473) of AKT in H1581 and Colo699 cells. In addition, none of the TORC1 targets (p70S6K, S6, and Rictor) showed altered phosphorylation by FGFR1 inhibition with AZD4547. By contrast, AZD8055 inhibited phosphorylation of AKT-Ser473, p70S6K, S6, and Rictor, but not phosphorylation of ERK (Fig. 5A). Finally, combined AZD4547 and AZD8055 treatment inhibited phosphorylation of ERKs and multiple targets of TORC1 and TORC2, consistent with the functioning of an RTK coactivation network (14) in which MTOR is regulated in parallel, not distal to FGFR1. These results are also observed with rapamycin and AZD4547 except that no inhibition of AKT Ser473 phosphorylation is observed (Supplementary Fig. S9). Dual blockade of FGFR1 and MTOR, but not inhibition of either target alone, induced PARP1 cleavage in H1581 cells as measured by immunoblot analysis, suggesting the induction of apoptosis with the combination inhibitor treatment (Fig. 5B). As an independent biochemical measure of apoptosis, the activity of caspase-3 was assayed in extracts from H1581 cells similarly treated. Figure 5C shows an approximately 5-fold increase in caspase-3 activity in cells treated with combined AZD4547 and AZD8055 relative to control cells or cells treated with the single agents.

It is noteworthy that neither AKT1, 2, nor 3 were identified as high-ranking hits in the SL screens (Supplementary Table S2), despite the findings in Fig. 5 showing marked inhibition of AKT phosphorylation by MTOR kinase inhibitors. It is possible that functional redundancy occurs among the distinct AKT gene products such that silencing of any one AKT gene fails to exert a phenotype. We tested the ability of two AKT inhibitors, MK2206 (20) and GSK690693 (21) to exert synergistic growth inhibition with FGFR inhibitors. The sensitivity of Colo699 cells to these AKT inhibitors is shown in Supplementary Fig. S10A and indicates only modest sensitivity to the single agents. As shown in Supplementary Figs. S10B and S11, strong synergy of MK2206 and ponatinib in Colo699 cells and GSK690693 and ponatinib in H1581 cells was observed. Thus, the findings in Figs. 3–5 and Supplementary Figs. S10 to S11 support the parallel activation of the MTOR–AKT signaling pathway as a modulator of the intrinsic sensitivity of multiple FGFR1-dependent cancer cell lines to FGFR TKIs.

H1581 and Colo699 cells were implanted in the flanks of female nu/nu mice as described in Materials and Methods. When the tumors reached approximately 100 mm³, the mice were randomized into treatment groups; diluent control, AZD4547 (12.5 mg/kg), AZD2014 (10 mg/kg), or the combination of both drugs. As shown in Fig. 6B and C, treatment of flank H1581 tumors with either AZD4547 or AZD2014 alone yielded little growth inhibition relative to diluent control (Fig. 6A). In fact, the modest effect of AZD4547 monotherapy is surprising, considering the potency with which this cell line is inhibited by AZD4547 in vitro (Fig. 3 and Table 1). However, in combination, AZD4547 and AZD2014 yielded significant tumor growth inhibition (Fig. 6D and E) and significantly prolonged survival (Fig. 6F), consistent with the synergistic growth inhibition observed in vitro. Similar relative activities of AZD4547 and AZD2014 as monotherapies, and in combination were observed following treatment of flank Colo699 xenografts (Supplementary Fig. S12A–S12F), although the survival benefit afforded by the combination therapy was not statistically significant (P = 0.06).

Using an unbiased RNAi screen, our study highlights MTOR as an actionable protein kinase that can be targeted in combination with FGFR1 to achieve synergistic growth inhibition in FGFR1-dependent cancer cell lines. The molecular basis for the synergism appears to involve the collapse of a greater signaling network by the combination therapy than that achieved by either FGFR1 or MTOR inhibition alone. The findings in Fig. 5 support the dominant regulation of the ERK pathway downstream of FGFR1 and activation of TORC1 targets (p70S6K and S6) and TORC2 targets (pAKT S473) by MTOR. Moreover, this rationally derived combination of MTOR inhibitors with FGFR inhibitors is consistent with an extensive literature demonstrating benefit of adding MTOR inhibitors to various targeted therapeutics. For example, a rapamycin analogue increased growth inhibition by ponatinib in FGFR2-driven endometrial cancer cell lines (22), and combining a dual PI3K–MTOR inhibitor with TKIs active on BCR-ABL yielded increased growth inhibition of CML cell lines (23). Combination of a BTK inhibitor with the MTOR inhibitor, AZD2014, induced synergistic killing of diffuse large B-cell lymphoma cells (24), and the benefit of combining IGF inhibitors with MTOR inhibitors for reducing growth of Ewing sarcoma cell lines has been documented (25). The generality of the synergistic growth suppression achieved with addition of MTOR inhibitors indicates the degree to which MTOR participates in cancer signaling networks.

The identification of MTOR as an SL protein kinase in the setting of ponatinib treatment of H157 and H1299 cells as well as the ability of AZD8055 and AZD2014 to synergize with AZD4547 in Colo699 and H1581 cells supports a model where MTOR signals in parallel with, not downstream of FGFR1. In this regard, our results are consistent with the existence of RTK coactivation networks as reviewed by Xu and Huang (14) where the ERK pathway distal to FGFR1 and MTOR represents distinct fragile points. The ability of the MEK inhibitor, selumetinib, to synergize with the MTOR inhibitor, AZD8055, supports this hypothesis (26). The identification of receptors or RTKs that reside upstream of MTOR in FGFR1-dependent cancer cells remains to be defined. Also, MTOR exists within two distinct signaling complexes, TORC1 and TORC2 (27) and the identity of the precise complex mediating the SL, and essential activities measured in our study were not resolved. The kinome shRNA library lacked shRNAs to Raptor and Rictor that would distinguish TORC1 and TORC2 and specific silencing of Rictor or Raptor yielded equivocal results (data not shown). A likely scenario is that both complexes participate in growth regulation in lung and HNSCC cells and clearly, both TORC1 and TORC2 are inhibited by AZD8055 and AZD2014 (18). Also, rapamycin strongly inhibited growth (Supplementary Figs. S5 and S8), supporting the critical involvement of TORC1. It is noteworthy that AKT1, 2, or 3 were not high-ranking hits in the screens, although redundancy among the three gene products could preclude identification of any with our RNAi-based approach. Like MTOR inhibitors, two independent AKT inhibitors synergized with ponatinib for growth inhibition of multiple cell lines (Supplementary Fig. S10 and S11), providing support for AKT as at least one important target of MTOR in these cancer cells.

In light of the potent in vitro sensitivity of Colo699 and H1581 cells to FGFR TKIs (Figs. 3 and 4), we were surprised by the modest degree of growth inhibition achieved with AZD4547 alone when these cell lines were propagated as flank xenografts in nu/nu mice (Fig. 6 and Supplementary Fig. S12). Still, this degree of growth inhibition is not inconsistent with the early results of clinical trials of AZD4547 and BGJ398 in squamous cell lung cancer where only partial responses have been observed thus far in less than 20% of patients (28,29). On the basis of our present studies showing strong in vivo responses with flank xenografts that are limited to combination treatment with AZD4547 and AZD2014, we wonder whether targeting auxiliary pathways such as MTOR will be required to observe significant clinical responses in FGFR1-dependent lung cancers and HNSCCs. If so, it is important that FGFR1 not be immediately abandoned as a therapeutic target in these settings based solely on a marginal clinical response to FGFR TKI monotherapy. Although the frequent and profound tumor shrinkage responses observed in lung cancers bearing mutant EGFR or rearranged ALK treated with TKI monotherapies have established a new expectation for clinical responsiveness, we hypothesize that cancers driven by nonmutated drivers like FGFR1 may inherently depend more on RTK coactivation networks for full-transforming potential (14). Admittedly, clinical investigation of combination therapies is unwieldy relative to monotherapy and bears increased concern of drug toxicity. Still, the fact that many lung cancers and the majority of HNSCC will present with nonmutated oncogene drivers necessitates careful consideration of combination therapies as a starting point in treatment design.

L.E. Heasley reports receiving a commercial research grant from ARIAD. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K.R. Singleton, A.C. Tan, L.E. Heasley

Development of methodology: K.R. Singleton, A.C. Tan, L.E. Heasley

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.R. Singleton, T.K. Hinz, E.K. Kleczko, L.A. Marek, J. Kwak, A.C. Tan, L.E. Heasley

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.R. Singleton, T.K. Hinz, E.K. Kleczko, L.A. Marek, J. Kwak, J. Kim, A.C. Tan, L.E. Heasley

Writing, review, and/or revision of the manuscript: K.R. Singleton, J. Kwak, J. Kim, A.C. Tan, L.E. Heasley

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L.E. Heasley

Other [assisted with execution of experiments and maintenance of laboratory materials (making buffers, media, cleaning supplies, etc.) and presentation of the data collected (generation of graphs and images] for experiments): T. Harp

The authors thank the Next-Generation Sequencing Production Core and staff within the University of Colorado School of Medicine, Department of Biochemistry and Molecular Genetics for expert assistance with the Illumina Genome Analyzer II sequencing of the kinome shRNA library. In addition, the authors acknowledge AstraZeneca for the supply of AZD4547 and AZD2014 as well as ARIAD Pharmaceuticals for the supply of ponatinib.

The studies were supported by the NIH (Lung SPORE P50 CA58187, UC Cancer Center Support grant P30 CA046934) and the VA (Merit Award 1BX001994-01).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.