Acquired Resistance to the Hsp90 Inhibitor, Ganetespib, in KRAS-Mutant NSCLC Is Mediated via Reactivation of the ERK–p90RSK–mTOR Signaling Network

This article is featured in Highlights of This Issue, p. 773

In 2015, approximately 160,340 lung cancer–related deaths occurred in the United States alone (1). Despite recent advances in lung cancer treatment, it remains the leading cause of cancer-related deaths in the United States and worldwide, and the 5-year survival rate is still below 20%. Although the discovery of driver oncogenes has revolutionized lung cancer therapeutics, no effective therapies exist for the oncogenic driver KRAS mutation (2–4), which is present in approximately 25% of non–small cell lung cancer (NSCLC) patients’ tumors. Patients with KRAS-mutant lung cancer have worst outcomes in early-stage disease and a poor prognosis in the metastatic setting (3). There is a critical need for novel agents targeting KRAS-mutant NSCLC.

Hsp90 is an ATP-dependent molecular chaperone required for the stability of its “client” oncoproteins, many of which are effectors of KRAS, such as members of RAF/ERK and PI3K/AKT/mTOR pathways (5). The critical role of Hsp90 in tumorigenesis led to the development of several first- and second-generation Hsp90 inhibitors (Hsp90i), which demonstrated promising responses in oncogene-driven cancers, such as HER2⁺ breast cancer (6). Ganetespib, a second-generation small-molecule Hsp90i, has shown significant single-agent activity in ALK-driven disease, but only transient, unconfirmed responses were observed in patients with KRAS-mutant tumors (7, 8). Therefore, identifying acquired resistance mechanism(s) to ganetespib in KRAS-mutant NSCLC may allow for the development of an effective therapeutic strategy for KRAS-mutant NSCLC.

Here, we have characterized the mechanism(s) of acquired resistance to ganetespib by generating and analyzing multiple ganetespib-resistant KRAS-mutant NSCLC cell lines. We found that hyperactivation of RAF/MEK/ERK and PI3K/AKT/mTOR signaling is a key component of acquired ganetespib resistance. Moreover, our results strongly implicate the ERK pathway and the ERK target, p90RSK, an important serine/threonine kinase, as central mediators of this resistance. We have also examined potential combinatorial activity of ganetespib with inhibitors targeting critical signaling molecules of these pathways. Overall, our preclinical data provide the justification for future clinical trials involving rationally designed Hsp90i combinations that may be effective against KRAS-mutant NSCLC.

Seven human KRAS-mutant NSCLC cell lines (A549, H460, H358, H727, H23, Calu-6, and SW1573) and embryonic kidney cell line HEK 293T were acquired in 2013 from the ATCC and maintained in ATCC-specified growth medium. All cell lines were authenticated by autosomal STR (short tandem repeat) profiling done at University of Arizona Genetics Core. Z-VAD-FMK was purchased from R&D Systems. Ganetespib and PX-866 were generously gifted by Synta Pharmaceutical Corp. and Dr. Peter Wipf's laboratory (University of Pittsburgh, Pittsburgh, PA), respectively. 17-AAG and NVP-BEZ235 were purchased from LC Laboratories, onalespib (AT13387), gedatolisib (PF-05212384, PKI-587), SCH772984, and BI-D1870 from Selleck Chemicals, SL0101 from Calbiochem, and docetaxel from Sigma-Aldrich.

To derive our ganetespib-resistant cell lines, we used low-dose (IC₂₅ and below) continuous treatments with gradually increasing doses, which allowed the development of ganetespib-resistant cell lines in approximately 3 months. The maximum dose of ganetespib at which A549-GR100 cells survive with negligible cell death was 100 nmol/L. H460-GR10 and H358-GR10 could not survive beyond 10 nmol/L ganetespib dose and are hence named GR10s. Ganetespib resistance of these ganetespib-resistant cells was tested in comparison with their parental stocks every second week. Once resistance was confirmed, these cells were always maintained in their respective ganetespib doses.

Cell viability after treatment was determined using the CellTiter96 Aqueous One Solution Cell Proliferation Assay Kit (Promega) according to the manufacturer's protocol. Quadruplets were used for each treatment group, and data were normalized to percentage of controls. IC₅₀ values were calculated using Prism V5.0 (GraphPad software). Colony formation assays were performed as described previously (9). For cell proliferation studies, at least three independent experiments were performed.

After being treated with specific drugs for defined periods of times, cell collection, protein preparation, concentration measurements, and Western blotting were performed as described previously (9). Information on all antibodies used in this report is provided in Supplementary Table S1.

293T cells (4 × 10⁶) were seeded in 25-cm² flasks and were transfected to generate lentiviral particles using a four-plasmid system as per the TRC Library Production and Performance protocols, RNAi Consortium, Broad Institute (10), and as described previously (9). p90RSK shRNA constructs were purchased from Johns Hopkins University HiT Center (Baltimore, MD), and their respective sequences are listed in Supplementary Table S2. pLKO.1-shRNA scramble vector was obtained from Dr. David M. Sabatini (Whitehead Institute for Biomedical Research, Cambridge, MA) through Addgene (Addgene plasmid 1864) as described previously (11). The pLenti CMV Puro DEST (w118-1) vector was obtained from Eric Campeau (University of Massachusetts Medical School, Worchester, MA) through Addgene (Addgene plasmid 17452). The Ultimate ORFs (Invitrogen) for RSK1-4 were obtained from the Johns Hopkins University HiT Center, and an LR reaction (Invitrogen) was performed to construct pLenti Puro DEST (w118-1)-RSK 1-4. All constructs were sequence verified. The ORF clone IDs of the constructs are IOH46696 (RSK1.a, variant 1), IOH12130 (RSK1.b, variant 2), IOH63248 (RSK2), IOH3648 (RSK3), and IOH36120 (RSK4).

For A549 xenografts, a total of 5 × 10⁶ viable cells were suspended in equal volumes of PBS and Matrigel and subcutaneously injected in adult 6- to 8-week-old athymic nude mice [Crl:NU (NCr)-F; Charles River Laboratories). Once the tumors reached an average size of ≥150 mm³ (range, 100–250 mm³), mice were distributed among control and treatment arms and were intravenously dosed with either vehicle control or ganetespib 100 mg/kg once a week. Tumor sizes [1/2(length × width²)] were measured by digital caliper twice a week. In addition, we used a KRAS-mutant human patient-derived xenograft (PDX) model established from a specimen of brain metastasis (BM012-15) from a patient with KRAS G12C mutation. Tumor tissues (2 mm²) cut with sterile blade were implanted subcutaneously. Once reaching ≥150 mm³, animals were randomized to 2 arms and intravenously dosed with vehicle control or ganetespib (50 mg/kg). Animals were sacrificed once tumors reached approximately 2,000 mm³.

Group comparisons in cell line studies were conducted using Student t test. Growth patterns in animal studies were summarized graphically by plotting the mean and SE for each treatment group at each tumor assessment time. Random intercept linear mixed models were used to predict log (base 2) of tumor volume. Residual plots and influence statistics were examined to ensure that model assumptions, such as log-linear growth, were not violated. Statistical analyses were conducted using SAS/STAT software, version 9.4 (SAS Institute, Inc.), and reported P values were two-sided.

We first established the relative efficacy of the drug in vitro in a panel of KRAS-mutant NSCLC cell lines. Representative viability curves after ganetespib treatment are shown for all 7 lines at 72 hours in Fig. 1A (left). In accordance with previous studies (12), ganetespib exhibited potent cytotoxicity with low nanomolar range of IC₅₀ values (31–67 nmol/L) tabulated in Fig. 1A (right). Ganetespib was also efficacious in two distinct in vivo models harboring KRAS mutation. Ganetespib induced notable tumor growth inhibition both in the KRAS-mutant (G12S) A549 xenograft model (Fig. 1B, left), as well as in our KRAS-mutant (G12C) PDX model (Fig. 1B, right). Using a random intercept linear mixed model predicting log (base 2) of tumor volume, the predicted average doubling time for the A549 model was 24 days [95% confidence interval (CI), 17–44] for ganetespib and 9 days (95% CI, 8–11) for the vector (Wald test; ***, P < 0.001). For the PDX, the fitted doubling times were 6 days (95% CI, 5–6) for ganetespib and 4 days (95% CI, 4–5) for the vector (Wald test; ***, P < 0.001).

Next, we examined the effect of ganetespib on the expression of several Hsp90 clients in the RAF/MEK/ERK and PI3K/AKT/mTOR pathways. Three representative KRAS-mutant cell lines A549, H460, and H358 were treated with increasing doses of ganetespib and expression of Hsp90 clients, and other related proteins were profiled by immunoblotting. Our results suggest that the expression and activity of RAF/ERK and PI3K/AKT/mTOR family members are notably diminished after ganetespib treatment (Fig. 1C; Supplementary Fig. S1). In addition, crucial upstream mTOR regulators, such as PI3K p110 subunits α and β, P-AKT, P-TSC2, and P-AMPKα, were also markedly inhibited by ganetespib.

To derive ganetespib-resistant cell lines, we initiated low-dose (IC₂₅ and below) continuous treatment with gradually increasing doses of ganetespib treatment. As shown in Fig. 2, we have successfully developed A549-GR100 (resistant up to 100 nmol/L ganetespib), H460-GR10, and H358-GR10 (resistant up to 10 nmol/L ganetespib) cell lines. Our Hill slope curve analysis indicated IC₅₀ values of 61.7 nmol/L for the parental A549 and 337.1 nmol/L for the A549-GR100 cells, 31 nmol/L for the parental H460 and 178.8 nmol/L for the H460-GR10 cells, 67.1 nmol/L for the parental H358 and 193.7 nmol/L for the H358-GR10 cell lines, respectively (Fig. 2A; Supplementary Table S3). To examine the durability of our ganetespib-resistant cells, we performed long-term colony formation assay and observed increased viability of ganetespib-resistant cells compared with parental cells with increasing doses of ganetespib (Fig. 2B). Differential sensitivity of specific cancer types to Hsp90is owing to their differential drug metabolism properties has been reported (13). Therefore, to rule out that the observed resistance was not simply due to altered ganetespib metabolism, we examined the responses of our ganetespib-resistant cell lines to Hsp90is with different metabolic pathways. We found that in addition to being resistant to ganetespib (Fig. 2A, left), ganetespib-resistant cell lines showed cross-resistance to the first-generation ansamycin compound, 17-AAG (Fig. 2A, middle) as well as to another second-generation Hsp90i, AT13387 (Fig. 2A, right). These results suggest that resistance is not just due to differential drug metabolism in our resistance cells. We next asked whether the Hsp90 chaperone machinery was still functional in our ganetespib-resistant cells and whether ganetespib was still able to engage the Hsp90 machinery. As the induction of Hsp90 and Hsp70 expression in response to Hsp90 inhibition is a well-established compensatory mechanism within the cell (6), we examined whether this response was intact in our ganetespib-resistant cells. After ganetespib treatment, we observed increased expression of both Hsp90 and Hsp70 in the parental and ganetespib-resistant cells (Supplementary Fig.S2). Often times, this is accompanied by an increase in the heat shock factor 1 (HSF1) transcription factor (6). In our cells, we observed a modest increase in both the parental and ganetespib-resistant A549 cells but failed to see an increase in our H460 ganetespib-resistant cells after ganetespib likely secondary to marked increased baseline levels that were present. We also examined whether selected Hsp90 client proteins were still degraded in the presence of ganetespib in our resistant cells. Although the baseline expression of active GSK-3-β, a well-known Hsp90 client protein (14, 15), was significantly higher in ganetespib-resistant cells, in both control and ganetespib-resistant cells, the GSK-3-β activity reduced in a dose-dependent manner, further corroborating the fact that the Hsp90 chaperonage is active in our ganetespib-resistant system.

As we observed marked reduction in expression of the ERK and mTOR pathway members in our parental KRAS-mutant cells in response to ganetespib (Fig. 1C), we analyzed the expression and activity of the members of these pathways in our control and ganetespib-resistant KRAS-mutant cells. Surprisingly, we observed notable reactivation/stabilization of RAF/MEK/ERK (Fig. 3) and PI3K/AKT/mTOR (Fig. 4; Supplementary Fig. S3) pathway members in our ganetespib-resistant lines. Although KRAS is not a known Hsp90 client and has been previously reported to be unaffected by ganetespib treatment (12), for the first time, here we report that mutant KRAS expression is also altered by ganetespib treatment. Utilizing KRAS-mutant–specific antibodies, we observed elevated expressions of mutant KRAS (G12S in A549 and Q61H in H460) compared with controls (Fig. 3A). Our results suggest that upregulation of the RAF/MEK/ERK as well as the PI3K/AKT/mTOR pathways may mediate resistance to ganetespib.

We next examined the requirement for RAF/MEK/ERK activity in ganetespib resistance using the potent, ATP-competitive and noncompetitive inhibitor of ERK1/2, SCH772984 (16, 17). Both A549 and H460 ganetespib-resistant cells, when treated with increasing concentration of the drug, compared with parental cells, showed greater sensitivity to SCH772984 (Fig. 3B). SCH772984 IC₅₀ was determined to be 5.7 μmol/L in A549C cells compared with 1.7 μmol/L in A549-GR100 cells and 6.1 μmol/L in H460C cells compared with 2.0 μmol/L in H460-GR10 cells (Supplementary Table S3). These results suggest that ganetespib resistance leads to increased dependence on the ERK signaling pathway. We next investigated whether the combination of Hsp90 and ERK inhibition in parental cells would lead to synergistic growth inhibition. We failed to detect a synergistic interaction, but rather, a strong additive combinatorial activity was observed in Loewe excess matrices for ganetespib with SCH772984 (data not shown). To validate the combinatorial activity of ganetespib and SCH772984, we selected specific dose combinations on the basis of the Loewe excess matrices results. We observed a strong additive effect in A549 parental cells for the combination of 500 nmol/L SCH772984 and 50 to 75 nmol/L ganetespib as compared with control (P < 0.05; Fig. 3C). In addition to other important kinases like AKT (18, 19), ERK1/2 is also known to lead to activation of the mTOR pathway (Supplementary Fig. S4A; ref. 20). Therefore, we assessed the ability of SCH772984 to inhibit mTOR signaling in control versus ganetespib-resistant cells. Surprisingly, even in the presence of uninhibited AKT signaling, SCH772984 was able to affect the expression of several mTOR pathway proteins (Supplementary Fig. S4B), supporting the potential efficacy of the combination of SCH772984 and ganetespib.

We next examined the dependence of ganetespib-resistant cells on PI3K/AKT/mTOR pathway by treating the parental and ganetespib-resistant cells with the dual PI3K/mTOR inhibitors BEZ235 (Fig. 4B) and PF-05212384 (Supplementary Fig. S3B), or the PI3K inhibitor, PX-866 (Fig. 4C). Remarkably, we observed increased dependence on this pathway as well as synthetic lethality to these drugs in the ganetespib-resistant cells (respective IC₅₀ values are tabulated in Supplementary Table S3). As our panel of ganetespib-resistant cells showed synthetic lethality to these drugs, we decided to further analyze the efficacy of PF-05212384, which is currently in clinical development, in combination with ganetespib in our control parental cells. On the basis of our preliminary results of Loewe excess matrices (data not shown) for ganetespib with PF-05212384, we tested the efficacy of the combination in A549-naïve cells. We selected specific dose combinations on the basis of the Loewe excess matrices. Our results indicated a strong additive effect of the 10 nmol/L PF-05212384 in combination with either 50 or 75 nmol/L ganetespib, compared with ganetespib treatment alone (Supplementary Fig. S3C).

Simultaneous increased activity of the RAF/MEK/ERK pathway as well as mTOR pathway members in ganetespib-resistant KRAS-mutant cells in the presence of ganetespib led to the question of what signaling molecule mediates the cross-talk between the two pathways. Upon activation, ERK1/2 targets multiple downstream mediators, including the p90RSK (90 kDa ribosomal S6 kinase) family proteins (21, 22). This protein has four human isoforms (p90RSK1–4), among which both RSK1 and RSK2 have been reported to activate/phosphorylate the regulatory associated protein of mTOR (RAPTOR), which is required for the mTORC1 complex activity (22, 23). Hence, we investigated the involvement, if any, of this protein in ganetespib resistance. Remarkably, we observed significantly higher p90RSK activity as indicated by the levels of S380 phosphorylation in ganetespib-resistant cells compared with controls (Fig. 5A). This evolutionary conserved protein is not only activated by sequential phosphorylation by ERK1/2, but also requires PDK1 phosphorylation, which is essential for its activation (Fig. 5B; ref. 24). We observed increased levels of total and P-PDK in the ganetespib-resistant cells as well (Supplementary Fig. S3A). Next, we investigated whether the expression and activity of individual isoforms differed in our control versus ganetespib-resistant cell lines (Fig. 5C). Although, the basal expression levels of the p90RSK isoforms in ganetespib-resistant cells were not markedly higher than that in their respective control cells, the levels of expression for both total and phosphorylated forms were increased for a subset of p90RSK isoforms in ganetespib-resistant cells compared with their respective controls. In addition, a subset of p90RSK isoforms is induced in a ganetespib dose–dependent manner, suggesting that the p90RSK isoforms are primed to be induced in response to increasing Hsp90 inhibition. These findings are concordant with the observed increased activity of P-ERK1/2 (Fig. 3) and P-PDK1 (Supplementary Fig. S3A) in ganetespib-resistant cells in the presence of ganetespib and place p90RSK as the connecting hub between these two signaling pathways. Further supporting these findings was the observation that ERK1/2 inhibition diminished all forms of p90RSK expression, followed by decreased mTOR signaling (Supplementary Fig. S4B).

Our results strongly suggest that p90RSK activity and expression were critical for mediating the observed resistance. To validate our findings, we first expressed shRNAs to silence individual p90RSK isoforms (Supplementary Fig. S5A) to determine the relative contribution of each isoform, as we observed increased activity of all isoforms in our ganetespib-resistant cells. We observed that although the ganetespib-resistant cells can tolerate the loss of a single isoform in the absence of ganetespib, targeting even one RSK isoform is sufficient to sensitize them to ganetespib (Fig. 6A) compared with control cells. Next, to further corroborate our finding, ganetespib-resistant cells were treated with two well-characterized and specific inhibitors of p90RSK, BI-D1870 (25) and SL0101 (26). Interestingly, although these inhibitors had little activity as monotherapy in the parental cells, they demonstrated significant cytotoxicity in A549-GR100 and H460-GR10 (Fig. 6B; Supplementary Table S3). To further support that the observed cytotoxicity of BI-D1870 was not due off-target effects of this agent, we examined the effect of BI-D1870 on RSK activity at the doses at which we observed growth inhibition. We observed dose-dependent decrease in the activities of all four isoforms in control and ganetespib-resistant cells, which directly correlated with the observed cytotoxicity (Supplementary Fig. S6). To validate p90RSK's involvement in resistance mechanism, we overexpressed individual p90RSK isoforms in parental A549 cells (Supplementary Fig. S5B) and examined whether they could induce ganetespib resistance. Interestingly, overexpression of a subset of isoforms (1, 2, 4) led to significant ganetespib resistance (Fig. 6C), suggesting that p90RSK is one of the key factors mediating the acquired resistance to ganetespib.

To establish the role of p90RSK in ganetespib resistance in vivo and to examine whether overexpression of one of the isoform would be sufficient enough to induce the resistance, we implanted A549-W118Δ (control) and A549-RSK1A–overexpressing cells in mice to form tumors, which we subsequently treated with ganetespib or vehicle (DMSO; Fig. 6D). A linear mixed effects model with random intercept predicted tumor volume (log, base 2) by cell line (A549-W118Δ vs. A549-RSK1A), treatment (vehicle vs. ganetespib), and time (continuous, days 1, 4, 7, 10, 14, 18, and 21) was used. With only vehicle treatment, there was not a significant difference in growth rates between cell lines as the predicted average doubling time for both cell lines was 9 days (95% CI, 8–10). However, with ganetespib treatment, there was a significant difference in growth between the A549-W118Δ and A549-RSK1A cell lines as the average doubling time for A549-RSK1A tumors was 10 days (95% CI, 9–11; Wald P = 0.11 for treatment effect within cell line) and for A549-W118Δ was 17 days (95% CI, 14–20; P < 0.001). These results strongly suggest that overexpression of even one isoform of the protein can induce significant ganetespib resistance both in vitro and in vivo.

Clinical advances in the treatment of lung cancer have come from the recognition that NSCLC is a collection of distinct molecularly driven neoplasms, not a single disease entity. Such approaches based on the molecular background of the tumor in NSCLC patients have led to the approval by the FDA of EGFR TKIs in EGFR-mutant lung adenocarcinomas and ALK inhibitors in ALK-rearranged lung adenocarcinomas (27). Unfortunately, there is no effective therapy for the most common oncogenic driver in NSCLC mutant KRAS. Thus, there is a critical need for novel agents targeting KRAS-mutant NSCLC. In this report, we focused on determining the molecular mechanisms responsible for acquired ganetespib resistance in KRAS-mutant NSCLC and propose a rational approach to overcome such resistance. These studies are essential for Hsp90 inhibitors to go forward in clinical development, as Hsp90 inhibitor combinations tested in the clinic to date have been empirically paired with standard-of-care chemotherapy. Previous combinations not based upon an understanding of mechanisms of Hsp90 resistance have resulted in disappointing and discouraging negative phase III trials, including a recent large phase III trial looking at the combination of ganetespib with docetaxel in advanced adenocarcinoma of the lung (5).

Therefore, we explored the molecular mechanisms underlying ganetespib resistance studying three ganetespib resistant KRAS-mutant NSCLC cell lines. We have established the hyperactivation of RAF/MEK/ERK and PI3K/AKT/mTOR signaling as the foundation of the resistance mechanism. These findings provided us with a number of molecular targets we could select to combine with ganetespib monotherapy to overcome the resistance (Fig. 7). mTOR serves as a master switch of cell growth and metabolism and many other signaling pathways including RAS/RAF/ERK lead to mTOR activation. Ganetespib-resistant cells also exhibited increased dependence on these pathways, as these cells showed synthetic lethality to the targeted pharmacologic inhibitions by PF-05212384, BEZ235, PX-866, and SCH772984. Interestingly, several previous reports have suggested that the combination of an Hsp90i and mTOR or dual PI3K/mTOR inhibitor may be efficacious in vitro and in vivo (12, 28).

In addition, we observed significant activity with the ERK inhibitor SCH772984. Currently, at least two ERK1/2 inhibitors are in phase I trials, MK8353 (NCT01358331) and BVD523 (NCT01781429). Our preclinical studies on SCH772984 suggest that the combination of an Hsp90 inhibitor with an ERK inhibitor would be a rational combination to explore in the clinic. Alternatively, several MEK inhibitors are FDA approved for melanoma and are being tested currently in NSCLC. Interestingly, the MEK inhibitor selumetinib has demonstrated significant activity in combination with docetaxel in KRAS-mutant NSCLC (29). In addition, the dual inhibitor of PI3K/mTOR, PF-05212384, is currently being investigated in several phase I and II studies (NCT00940498, NCT02920450, NCT01920061), where it is being tested in combination with a variety of cytotoxic and targeted agents.

The most striking part of our finding is the involvement of the p90RSK family of proteins in mediating the acquired resistance to ganetespib in KRAS-mutant NSCLC. Upon activation by ERK1/2 and PDK1, p90RSK proteins proceed to activate a vast array of substrates. Although there is evidence suggesting that RAF/MEK/ERK pathway activation may inhibit AKT signaling (30), cross-activation of PI3K/AKT/mTOR either directly by ERK activation or via p90RSK activation leading to activation of mTORC1 has also been well documented (31, 32). Our results demonstrating increased activity and expression in some cases for all four p90RSK isoforms in our ganetespib-resistant cell lines (Fig. 5) suggest that p90RSK is an important mediator of resistance. This is further supported by the ability of isoform-specific p90RSK knockdown in ganetespib-resistant cells to resensitize to ganetespib, and conversely, overexpression of individual RSK isoforms led to de novo ganetespib resistance in vitro and in vivo. Furthermore, targeted pharmacologic inhibition of p90RSK with BI-D1870 or SL0101 resulted in synthetic lethality in ganetespib-resistant cells (Fig. 6). In this article, we have not identified the key p90RSK substrate(s) that are responsible for inducing resistance to ganetespib; however, we are actively pursuing these targets. p90RSK family has been implicated in the regulation of cell growth and protein synthesis, cell migration and survival, cell proliferation, and cell-cycle progression and in some cases drug efflux (24, 33, 34). Although it is less likely that p90RSK is mediating resistance through increased drug efflux given our results in Supplementary Fig. S2, it could still be possible that p90RSK1-induced upregulation of P-glycoprotein may contribute to the observed resistance via drug efflux as recently described by Katayama and colleagues (34). These findings not only secure its role in mediating ganetespib resistance in KRAS-mutant NSCLC, but also identify p90RSK as a central targetable node to prevent or overcome resistance.

In conclusion, our preclinical work suggests that the combination of ganetespib with an ERK1/2 inhibitor, PI3K inhibitor, dual PI3K/mTOR inhibitor, or an RSK inhibitor would be an effective strategy to test in the clinic (Fig. 7). Information on the in vivo use of the p90RSK inhibitors is lacking; therefore, further preclinical development of these agents is necessary before they can be tested in the clinic. The combination of ganetespib or even other Hsp90i currently in clinical trials (e.g., AT13387) with any one of the inhibitors tested in this study may prevent ganetespib resistance and/or help overcome the resistance after single-agent treatment. These studies provide preclinical rationale for a future phase I/II trial in KRAS-mutant NSCLC testing these therapeutic combinations.

No potential conflicts of interest were disclosed.

Conception and design: S. Chatterjee, T.F. Burns

Development of methodology: S. Chatterjee, T.F. Burns

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Chatterjee, E.H.-B. Huang, I. Christie, T.F. Burns

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Chatterjee, E.H.-B. Huang, I. Christie, B.F. Kurland, T.F. Burns

Writing, review, and/or revision of the manuscript: S. Chatterjee, B.F. Kurland, T.F. Burns

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Chatterjee, E.H.-B. Huang, I. Christie

Study supervision: T.F. Burns

Other (animal tumor PDX and xenograft experiment and drug evaluation): E.H.-B. Huang

The authors thank Mark Socinski, MD; James G. Herman, MD; Laura Stabile, PhD; Deborah L. Galson, PhD; Peter Wipf, PhD; Frank P. Vendetti, PhD; Juraj Adamik, PhD; Zachary A. Yochum, B.S., at the University of Pittsburgh and David Proia formerly of Synta Pharmaceuticals for the discussion, advice regarding this work, and supply of resources when applicable. They also thank Synta Pharmaceuticals Corp. for the constant supply of ganetespib.

T.F. Burns and S. Chatterjee have received research funding for this project from a LUNGevity Foundation Career Development Award and a UPCI LUNG SPORE CDA P50CA090440. This project used the UPCI Animal and Biostatistics Facilities that are supported in part by award P30CA047904.

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.