Identifying genotype-dependent efficacy of single and combined PI3K- and MAPK-pathway inhibition in cancer

  1. Martin L. Sosa,1,
  2. Stefanie Fischera,1,
  3. Roland Ullricha,1,
  4. Martin Peifera,1,
  5. Johannes M. Heuckmanna,
  6. Mirjam Kokera,
  7. Stefanie Heyncka,
  8. Isabel Stückratha,
  9. Jonathan Weissa,
  10. Florian Fischera,
  11. Kathrin Michela,
  12. Aviva Goelb,
  13. Lucia Regalesb,
  14. Katerina A. Politib,
  15. Samanthi Pererac,
  16. Matthäus Getlikd,
  17. Lukas C. Heukampe,
  18. Sascha Ansénf,
  19. Thomas Zanderf,
  20. Rameen Beroukhimc,g,
  21. Hamid Kashkarh,
  22. Kevan M. Shokati,j,
  23. William R. Sellersk,
  24. Daniel Rauhd,
  25. Christine Orrl,
  26. Klaus P. Hoeflichl,
  27. Lori Friedmanl,
  28. Kwok-Kin Wongc,m,
  29. William Paon and
  30. Roman K. Thomasa,d ,f,2
  1. aMax Planck Institute for Neurological Research and Klaus Joachim Zülch Laboratories, Max Planck Society and Medical Faculty, University of Cologne, Gleueler Strasse 50, 50931 Cologne, Germany;
  2. bHuman Oncology and Pathogenesis Program, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021;
  3. cDepartment of Medical Oncology, Dana-Farber Cancer Institute, 44 Binney Street, Boston, MA 02115;
  4. dChemical Genomics Center, Max Planck Society, Otto Hahn Strasse 15, 44227 Dortmund, Germany;
  5. eDepartment of Pathology, University Hospital Bonn Medical School, Sigmund Freud Strasse 25, 53127 Bonn, Germany;
  6. fDepartment I of Internal Medicine and Center of Integrated Oncology Cologne–Bonn and
  7. hInstitute for Medical Microbiology, Immunology, and Hygiene, Medical Faculty, University of Cologne, 50924 Cologne, Germany;
  8. gBroad Institute of Harvard and Massachusetts Institute of Technology, 320 Charles Street, Cambridge, MA 02141;
  9. iHoward Hughes Medical Institute and Department of Cellular and Molecular Pharmacology, University of California, 600 16th Street, San Francisco, CA 94158;
  10. jDepartment of Chemistry, University of California, Room 419 Latimer Hall, Berkeley, CA 94720;
  11. kNovartis Institutes for BioMedical Research, 250 Massachusetts Avenue, Cambridge, MA 02139;
  12. lDepartment of Cancer Signaling and Translational Oncology, Genentech Inc., 1 DNA Way, South San Francisco, CA 94070;
  13. mDepartment of Medicine, Brigham and Women's Hospital and Harvard Medical School, 75 Francis Street, Boston, MA 02115; and
  14. nVanderbilt-Ingram Cancer Center, 777 Preston Research Building, 2220 Pierce Avenue, Nashville, TN 37232

  1. Edited by Charles R. Cantor, Sequenom, Inc., San Diego, CA, and approved August 27, 2009

  2. ↵1M.L.S., S.F., R.U., and M.P. contributed equally to this work. (received for review July 6, 2009)

 
Next Section

Abstract

In cancer, genetically activated proto-oncogenes often induce “upstream” dependency on the activity of the mutant oncoprotein. Therapeutic inhibition of these activated oncoproteins can induce massive apoptosis of tumor cells, leading to sometimes dramatic tumor regressions in patients. The PI3K and MAPK signaling pathways are central regulators of oncogenic transformation and tumor maintenance. We hypothesized that upstream dependency engages either one of these pathways preferentially to induce “downstream” dependency. Therefore, we analyzed whether downstream pathway dependency segregates by genetic aberrations upstream in lung cancer cell lines. Here, we show by systematically linking drug response to genomic aberrations in non-small-cell lung cancer, as well as in cell lines of other tumor types and in a series of in vivo cancer models, that tumors with genetically activated receptor tyrosine kinases depend on PI3K signaling, whereas tumors with mutations in the RAS/RAF axis depend on MAPK signaling. However, efficacy of downstream pathway inhibition was limited by release of negative feedback loops on the reciprocal pathway. By contrast, combined blockade of both pathways was able to overcome the reciprocal pathway activation induced by inhibitor-mediated release of negative feedback loops and resulted in a significant increase in apoptosis and tumor shrinkage. Thus, by using a systematic chemo-genomics approach, we identify genetic lesions connected to PI3K and MAPK pathway activation and provide a rationale for combined inhibition of both pathways. Our findings may have implications for patient stratification in clinical trials.

The past decade has witnessed the advent of targeted cancer therapeutics targeting mutationally activated proto-oncogenes. When targeted to the right patient population, such approaches have proven efficacious with, sometimes dramatic, responses and improvement in survival (1–3). Given the pace of the currently ongoing efforts to fully characterize cancer genomic aberrations, a comprehensive genetic compendium of all human cancers is within reach. Although initial studies suggested that most human tumors are dominated by an array of individual, or “private” mutations (4), more recent studies imply that most human cancer genome aberrations converge on activation of a limited repertoire of “downstream” oncogenic signaling pathways (5–8). Importantly, among the most heavily affected oncogenic pathways were the PI3K and the MAPK signaling pathways. Thus, rather than providing therapeutic strategies for each individual mutation, targeting key modulators of downstream pathways appears increasingly attractive. Importantly, small synthetic molecules targeting these pathways have been developed and are currently undergoing clinical testing.

Previously, mutations in BRAF have been linked to downstream dependency on MEK (9), the kinase phosphorylating MAPK (or ERK), KRAS-mutant lung cancers depend on both PI3K and MAPK signaling (10), and resistance to EGFR inhibition appears to involve mechanisms that maintain PI3K signaling (11, 12). However, in the vast majority of the cases, the inhibition of either the PI3K or MAPK pathway alone is not sufficient to robustly induce tumor shrinkage (13), in part explained through release of negative feedback loops resulting in the activation of the alternate pathway (14–18). By contrast, studies analyzing combinations of pathway inhibitors showed favorable results (13, 14). Thus, a genetically defined framework that would allow predicting which of these pathways is primarily affected and whether the combinatorial inhibition of both pathways is superior to single-agent treatment would greatly impact future clinical strategies in trials involving such therapeutics.

A major hurdle in the transition from preclinical drug discovery to clinical trials lies in the genomic diversity of human tumors and the lack of preclinical models that capture this diversity. Given the impact of genomic aberrations on therapeutic response, such models are necessary to identify lesions connected to individual drug response. To overcome these limitations, we have recently collected a panel of 84 non-small-cell lung cancer (NSCLC) cell lines that we have characterized in depth in gene copy number, gene expression, and mutation space (19). This panel yielded robust genomic predictors for several preclinical and clinical compounds. Here, we asked whether chemical perturbation of this panel with both PI3K and MEK inhibitors might help to systematically reveal downstream dependencies on these pathways as a function of genetic lesions.

Previous SectionNext Section

Results

Dissecting PI3K Signaling Pathway Dependency in Cancer.

As a first step in determining the role of PI3K signaling in NSCLC, we screened our cell line panel against the dual specific PI3K/mTOR inhibitor PI-103 (20–23). We found PI-103 to be active in nanomolar range against the majority of our cell lines (Fig. S1A) (17). Growth-inhibitory activity of PI-103 was largely independent of the mutational status of EGFR, PTEN, and PIK3CA; all of these lesions are known activators of PI3K signaling (24). When applying different prediction models using all significant lesions found in our cell line panel, no robust predictor could be determined (Fig. S1A).

We reasoned that cell lines with a dependency on PI3K signaling predominantly undergo apoptosis on inhibition of PI3K, and therefore, systematically screened our cell line panel for induction of apoptosis after treatment with 0.5 μM of PI-103 (Fig. 1A) and with 1 μM of PI-103 (Fig. S2 and Table S1). We found that in a fraction of cell lines (n = 10; apoptosis rate >18%), dual inhibition of PI3K and mTOR robustly induced apoptotic cell death (Fig. 1A). The cell lines most susceptible to induction of apoptosis were enriched for cells known to be dependent on activated receptor tyrosine kinases (RTK; EGFRmut, EGFRamp, METamp, HER2mut, and HER2amp) (Fig. 1A), suggesting a key role of PI3K signaling in transmitting survival signals downstream of mutant RTKs in these cells. When grouping these different genotypes to a subclass of RTK-dependent tumors, we could now robustly predict PI3K dependency in our cell line panel (P = 0.0229) (Fig. 1A; Fig. S1B). By contrast, activating mutations in the RAS/RAF pathway (RAS, KRASmut, NRASmut, and BRAFmut) were predominantly found in cell lines resistant to PI3K inhibition; thus, suggesting that RAS signaling-dependent tumors are not susceptible to inhibition of PI3K (10). We confirmed primary dependency on the genetically activated “upstream” RTK by demonstrating exquisite sensitivity of the respective cell lines to selective inhibitors targeting the respective activated kinase (19).

Fig. 1.
View larger version:
Fig. 1.

Inhibition of PI3K signaling in cancer. (A) All cell lines were screened for induction of apoptosis using Annexin-V/PI staining after 72-h treatment with PI-103. Bars represent the fraction of apoptotic cells and are sorted from the most sensitive cell line (Left) to the most resistant cell line (Right), and grouped according to the presence of RTK- (EGFR, ERBB2, MET) or RAS-lesions (KRAS, NRAS, BRAF). A two-by-two table highlights the distribution of apoptotic cell lines in the two different genetically defined groups (RTK, RAS). (B) Two PI-103 sensitive (H1975, HCC827) and two resistant cell lines (H441, H460) were treated with PI-103 either in a dilution series (Left) or over time at 1 μM (Right). Pharmacodynamic markers (pAKT, AKT, pS6K, S6K, pERK, ERK, p4EBP1) were assessed by immunoblotting for all cell lines and all conditions. Black bars indicate splicing of noncontiguous bands run on the same gel. (C) Nude mice were s.c. engrafted with H1975, HCC827, AN3CA, and MKN45 cells, and tumors were either treated with vehicle control or the PI3K inhibitor GDC-0941 at 75 to 150 mg/kg. The tumor volume change relative to the tumor volume at day 0 (y axis) are plotted over time (x axis). (D) Growth of lung tumors was induced in the ERBB2YVMA transgenic mice. Tumors detectable by MRI were treated with either vehicle (n = 4) or GDC-0941 (n = 3) at 75–150 mg/kg for 2–4 weeks. Tumor growth was measured by serial MRI (Lower) and tumor volumes were calculated using Image-J (SI Methods).

Activity of PI-103 could be linked to inhibition of the PI3K/Akt pathway, because ectopic expression of a constitutively active allele of Akt [myristoylated (Myr)-Akt] rescued cells from PI-103-induced apoptosis (Fig. S3). Also, hierarchical clustering of the activity of compounds with enhanced selectivity against the different isoforms of PI3K, as well as of the mTOR inhibitor rapamycin across all of the cell lines validated p110α as the critical target of PI-103 (Fig. S4 A and B) (25, 26). Also, these experiments revealed a high degree of synergy between PI3K and mTOR inhibition, as previously reported (Fig. S4C) (26).

In sensitive cells, PI-103 induced sustained suppression of phosphorylated (p-)Akt at submicromolar concentrations (Fig. 1B; Fig. S3). In resistant cells, p-Akt levels were also extinguished, but returned to almost baseline levels after 24–48 h of treatment. By contrast, in all cell lines tested, levels of p-ERK were either induced or failed to be reduced by PI-103 treatment (Fig. 1B; Fig. S3), presumably due to release of negative feedback loops (15–17). Thus, although RTK-driven cancers exhibit a therapeutically exploitable dependency on PI3K signaling, treatment-induced activation of the MAPK signaling pathway may limit the overall activity of single-agent PI3K inhibition.

We next transplanted a series of cell lines of different tumor types onto nude mice (SI Methods) and treated the mice with GDC-0941, a pharmalog of PI-103 with superior pharmacokinetic properties (27). The panel studied in vivo comprised cell lines derived from EGFR-mutant lung cancer, MET-amplified gastric cancer, and FGFR2-mutant endometrial cancer. Tumor growth was halted when treated with 150 mg/kg of GDC-0941, and resulted in tumor shrinkage in the case of the MET-amplified gastric cancer cell line MKN45 and the EGFR-mutant lung cancer cell line HCC827, in the latter case even when only 75 mg/kg of GDC-0941 were administered (Fig. 1C). Tumor growth inhibition was paralleled by decreased phosphorylation of AKT, as evidenced by immunohistochemical analysis of explanted tumors (Fig. S5). Remarkably, even the growth of tumors expressing the T790M resistance mutation of EGFR was inhibited by single-agent treatment with GDC-0941 (H1975; Fig. 1C Upper Right). By contrast, mice receiving placebo exhibited massive growth of all tumors (Fig. 1C). We next assessed the efficacy of GDC-0941 in two transgenic mouse models of RTK-driven NSCLC. In one model (28), lung cancer is driven by inducible expression of the insertion mutation YVMA of ERBB2 (Her2/neu). In the other model (29), lung-specific induction of the double-mutant EGFRL858R/T790M (LTM) leads to erlotinib-resistant lung cancer growth in mice. In the ERBB2YVMA mice, treatment with 150 mg/kg of GDC-0941 led to pronounced tumor shrinkage, whereas the lower dose (75 mg/kg) induced inhibition of tumor growth compatible with stable disease (Fig. 1D; Fig. S6A). In the LTM mice, 150 mg/kg of GDC-0941 halted tumor growth in four out of five mice, compatible with stable disease (Fig. S6B). These findings validate PI3K signaling as the predominant downstream signaling pathway regulating survival in RTK-driven cancers. However, in some cases, tumor growth was only stopped, compatible with release of negative feedback loops limiting the single agent activity of PI3K inhibition.

Dissecting MAPK Dependency in Cancer.

To identify MAPK signaling dependency in NSCLC, we systematically screened our cell line panel for apoptosis induction after treatment with the potent and selective MEK1/2 inhibitor PD0325901 at clinically achievable doses of 0.25 μM (Fig. S7) and 0.1 μM (Fig. S2 and Table S1) (30). Due to its high potency and selectivity, the MEK inhibitor PD0325901 was used to interrogate the MAPK pathway. This analysis indicated enrichment of cell lines with RAS pathway mutations among the top scoring cell lines displaying robust induction of apoptosis (P = 0.0165) (Fig. S7). We next grouped the cells according to their genotype and the fraction of apoptotic cells after treatment with PD0325901 and observed an enrichment of cells with MAPK lesions among the top scoring cell lines (P = 0.0437) (Fig. 2A). Interestingly, BRAF- and NRAS-mutant cells were predominantly found among the top 10 sensitive cell lines, but did not reach statistical significance due to the low prevalence of BRAF- and NRAS-mutations in our cell line panel (BRAFmut, 6%; NRASmut, 5%).

Fig. 2.
View larger version:
Fig. 2.

Inhibition of MAPK signaling in cancer. (A) All cell lines were screened for induction of apoptosis using Annexin-V/PI staining after 72 h of treatment with PD0325901. Bars represent the fraction of apoptotic cells and are sorted from the most sensitive cell line (Left) to the most resistant cell line (Right), and grouped according to the presence of RTK- (EGFR, ERBB2, MET) or RAS-lesions (KRAS, NRAS, BRAF). Two-by-two table highlights the distribution of apoptotic cell lines in the two different genetically defined groups (RTK, RAS). (B) Two PD0325901 sensitive (HCC364, Calu6) and two less sensitive cell lines (A549, H441) were treated with PD0325901 either in a dilution series (Left) or over time at a fixed concentration (0.5 μM; Right). Pharmacodynamic markers (pAKT, AKT, pS6K, S6K, pERK, ERK) were assessed by immunoblotting for all cell lines and all conditions. (C) Nude mice were s.c. engrafted with H2122 and A549 cells, and tumors were treated with either the vehicle control, GDC-0941 or PD0325901 at the indicated concentration. Both compounds were administered every other day in the case of H2122 or every day in the case of A549. Similar results were obtained with other doses and other schedules (Fig. 4 D and E). The mean tumor volumes (y axis) are plotted over time (x axis).

Again, treatment with an inhibitor of a single downstream pathway, PD0325901, led to induction of the other signaling pathway: p-Akt was induced both in highly sensitive and cells of limited sensitivity (Fig. 2B). Thus, our findings confirm and extend previous observations of predominant engagement of MAPK signaling by RAS-/RAF-mutant cancers (9, 31). However, release of negative feedback loops is likely to limit the overall efficacy of single-agent therapeutic MEK inhibition.

We next transplanted two KRAS-mutated cell lines onto nude mice and treated the mice with either GDC-0941 or PD0325901. Consistent with our in vitro results, H2122 tumor growth was not suppressed when mice where treated with the PI3K inhibitor at the maximal tolerated dose (MTD) of 150 mg/kg (Fig. 2C). However, tumor growth was inhibited when mice were treated with the MTD of 25 mg/kg of the MEK inhibitor PD0325901 (Fig. 2C). Similar results were obtained with A549 xenografts treated with lower doses of both inhibitors (Fig. 2D) and with other dosing schedules as well (Fig. 4 D and E). These findings support the notion that MAPK signaling is the predominant downstream signaling pathway regulating survival in cancers with activating mutations in the RAS/RAF-pathway. The effect of MAPK signaling inhibition is compatible with stable disease, and thus again, reactivation of the PI3K signaling might limit the single agent activity of MEK inhibition.

Enhanced Cell Killing by Dual PI3K and MAPK Blockade as a Function of Genetic Lesions.

Biochemical analyses of response to both PI3K and MEK inhibition had shown that both inhibitors lead to induction of the alternate signaling pathway, presumably by release of negative feedback loops (Figs. 1 and 2) (15–17). We hypothesized that dual blockade of both pathways might be able to suppress release of these negative feedback loops and, thus, induce apoptosis more potently. Given the experimental and analytical challenges in performing high-throughput combination compound screens, we performed a previously underscribed algorithm to define synergy of two compounds. Thus, synergy strength can be deduced as the difference of experimental data from a null model separating synergy from antagonism. For this analysis, we used growth inhibition measurements obtained in high-throughput cellular screens with both compounds and the combination profiled across the cell line panel at different concentrations (SI Methods). Clustering of cell lines according to the strength of synergy revealed that cell lines with RAS-pathway mutations exhibited the highest synergy score (Fig. 3A). By contrast, the combination index (32, 33) method yielded incomplete results (Fig. S8)

Fig. 3.
View larger version:
Fig. 3.

Combined inhibition of PI3K- and MAPK-signaling in cancer. (A) All cell lines were screened for synergistic response using seven different combinations of PI103 and PD0325901 (C1 = 0.025 μM PI103 + 0.025 μM PD0325901; C2 = 0.25 μM PI103 + 0.25 μM PD0325901; C3 = 0.1 μM PI103 + 0.1 μM PD0325901; C4 = 0.1 μM PI103 + 0.5 μM PD0325901; C5 = 0.5 μM PI103 + 0.1 μM PD0325901; C6 = 1.0 μM PI103 + 1.0 μM PD0325901; C7 = 0.5 μM PI103 + 0.5 μM PD0325901) in a viability assay (SI Methods). Positive values (red colors) of the synergy strength metric indicate synergy whereas negative values (blue colors) indicate antagonistic drug response. (B) The induction of apoptosis in all NSCLC cell lines (x axis) after 72 h of single PD0325901 (0.25 μM) treatment, single PI-103 (0.5 μM) treatment or combinatorial treatment with both inhibitors is displayed. Apoptosis was assessed by flow cytometry using Annexin-V/PI staining. Bars represent the fraction of apoptotic cells and are sorted from the most sensitive cell line (Left) to the most resistant cell line (Right) for the combined PI3K and MEK treatment. (C) All cell lines were screened for induction of apoptosis using Annexin-V/PI staining after 72-h treatment with either PD0325901 (0.25 μM), PI-103 (0.5 μM), or a combination of both compounds (0.25 μM PD0325901 + 0.5 μM PI-103; x axis) and the resulting fractions of apoptotic cells (y axis) were plotted as box-plots according to the presence of RAS- (Left) or RTK-lesions (Right).

To test whether this observation made in growth-inhibition assays could be extended to apoptosis assays, we determined the fraction of apoptotic cells after combined treatment with PI103 (500 nM) and PD0325901 (250 nM). As expected, inhibition of both pathways resulted in increased apoptosis in virtually all cell lines analyzed (Fig. 3B). To analyze the combined effect of PI3K and MEK inhibition as a function of genotypes, we separated the defined subgroups of cell lines in our panel that are defined by either RTK lesions or lesions affecting the RAS-/RAF-signaling pathway and compared the average rate of induction of apoptosis of the different treatment regimens (Fig. 3C). We observed a significant increase of apoptosis after combined treatment when compared with either single inhibition of MEK (n = 37; P = 0.003) or single inhibition of PI3K (n = 37; P = 2.4*10−5) in the RAS-mutation subgroup (Fig. 3C). In the subgroup of cells with RTK lesions, combined inhibition of both pathways was also significantly superior to single inhibition of MEK (P = 0.0044) and PI3K, although the latter comparison did not reach significance (P = 0.188). This result confirmed our observation obtained using our synergy score approach (Fig. 3A). We sought to recapitulate this finding in an independent cell line panel (Table S2) of tumors of various histologies harboring lesions in RTK and RAS pathways (n = 15). We again found that cell lines with RTK lesions primarily depended on PI3K signaling and cell lines with RAS-mutations were predominantly driven by MAPK signaling (Fig. S9). Again, the combined inhibition of both pathways was superior to single agent inhibition in both genotypes (Fig. S9). Thus, combined inhibition of PI3K and MAPK signaling induces enhanced cell killing in cancers driven by genetic lesions in RTKs and RAS/RAF oncoproteins. However, synergy was more pronounced in RAS/RAF-mutant tumors.

Suppression of Feedback Loop-Mediated Pathway Reactivation by Combined Blockade of MAPK and PI3K Pathways.

We hypothesized that enhanced cell killing by dual pathway inhibition (Fig. 3) might be due to suppression of release of negative feedback loops and analyzed the impact of these combinations on pathway activation. Biochemical analyses of response indicated that although the dual PI3K/mTOR inhibitor, PI-103, the p110α inhibitor, PIK-90, the mTOR inhibitor, rapamycin, and the MEK inhibitor, PD0325901, all led to inhibition of signaling downstream of the respective targets, all of these inhibitors led to induction of at least one signaling mediator in the alternate pathway (Fig. 4 A and B). By contrast, combined blockade of both PI3K and MAPK signaling potently suppressed activation of the other pathway in all cell lines tested (Fig. 4 A and B). Thus, combined inhibition of both PI3K and MAPK signaling pathways can suppress feedback loop-induced activation of other oncogenic signaling pathways, resulting in more potent induction of apoptosis.

Fig. 4.
View larger version:
Fig. 4.

Suppression of feedback loops by dual PI3K/MAPK-inhibition enhances tumor shrinkage in vivo. (A) Four NSCLC cell lines (H1975, HCC2429, HCC364, A549) with different genetic lesions were treated for 24 h with PI-103, PIK90, PD0325901 and rapamycin, in various combinations at fixed concentrations (PI-103, 1 μM; PIK90, 5 μM; PD0325901, 0.5 μM; rapamycin, 0.01 μM). Pharmacodynamic markers (pAKT, AKT, pS6K, S6K, pERK, ERK, p4EBP1) were assessed by immunoblotting. (B) Three different cell lines of non-NSCLC cancer type (AN3CA, BT474, MKN45) with different genetic lesions were treated for 6 h with PI-103, PIK90, PD0325901 and rapamycin, in different combinations at fixed concentrations (PI-103 1 μM; PIK90 5 μM, PD0325901 0.5 μM; rapamycin 0.01 μM). Pharmacodynamic markers (pAKT, AKT, pS6K, S6K, pERK, ERK, p4EBP1) were assessed by immunoblotting. The biochemical response to specific inhibitors targeting the primary genetic lesion in the respective cell line is shown as a reference (PD173074 targets FGFR, PD168393 targets ERBB2, PHA665752 targets MET). Black bars indicate splicing of noncontiguous bands run on the same gel. (C and D) Nude mice were s.c. engrafted with AN3CA or H2122 cells, and tumors were treated daily with vehicle control, GDC-0941, PD0325901, or a combination of both at the indicated dose. The tumor volumes (y axis) are plotted over time (x axis). (E) H2122 tumors were grown on nude mice as in D, and mice were treated with an intermittent schedule of the combination of GDC-0941 and PD0325901, both dosed at their MTD (the combination was administered every fourth day; GDC-0941 dose, 150 mg/kg; PD0325901 dose, 25 mg/kg).

Last, we transplanted the FGFR2-mutant cell line AN3CA into nude mice. Although single-agent treatment with the MEK inhibitor PD0325901 had no impact on tumor growth and the treatment with the PI3K inhibitor GDC-0941 halted tumor growth (Fig. 1C), only the combination of both compounds led to robust tumor shrinkage (Fig. 4C). We next analyzed the KRAS-mutant cell line H2122. As expected, only the combination treatment and not single-agent treatment led to significant tumor size reduction in vivo (Fig. 4D). Also, an alternating schedule where both the MEK inhibitor PD0325901 and the PI3K inhibitor GDC-0941 were administered at their MTD every fourth day was similarly effective (Fig. 4E), potentially being more tolerable.

Previous SectionNext Section

Discussion

A critical determinant for the success of molecularly targeted drugs will be to identify those tumors that are connected with a therapeutically amenable dependency and to define the optimal therapeutic strategy for treating these tumors. Here, we applied a chemo-genomics approach to link dependency on the PI3K and MAPK pathways to subsets of genomic aberrations in cancer using an NSCLC, as well as non-NSCLC cell line model. Notably, we found RTK-driven tumors to largely depend on PI3K and RAS-/RAF-driven tumors to be addicted to the MAPK signaling pathways, respectively. However, in all settings tested, release of negative feedback loops led to activation of the alternate pathway. Similar to recent studies in breast cancer (14), combined inhibition of both pathways potently suppressed release of negative feedback loops; thereby, resulting in enhanced induction of apoptosis in tumor cells and tumor shrinkage in vivo. Thus, patients whose tumors harbor genomic aberrations in RTKs or any of the RAS-/RAF-oncogenes might benefit from treatment with a combination of a PI3K inhibitor and a MEK inhibitor.

It has been generally assumed that the engagement of both the MAPK and PI3K pathways by mutant RTKs is essential (1, 34). Our results, by contrast, suggest that the primary downstream dependency of such tumors is on the PI3K pathway, whereas activation of the MAPK pathway may primarily be the result of inhibition of the PI3K pathway. This finding might be of particular interest for treatment of patients whose RTK-driven cancer has acquired secondary resistance after an initial response (e.g., EGFR-mutant lung cancer treated with EGFR inhibitors). In these tumors, secondary resistance mechanisms arise that either abrogate the binding of the kinase inhibitor (35, 36), or that substitute the primary signaling input by activation of additional kinases that reactivate the same downstream pathway (12, 37, 38). Notably, we found H1975 cells that express the T790M resistance mutation of EGFR to be sensitive to PI3K inhibition. Similarly, HCC827 GR cells (12) that acquired EGFR inhibitor resistance by amplification of MET retained the high sensitivity to PI3K inhibition of the parental HCC827 cell line. Thus, the primary signaling dependency encoded by an activated mutation in a receptor tyrosine kinase remains exploitable by PI3K inhibition alone, or better, in combination with a MAPK pathway inhibitor.

BRAF-mutant tumors were found in previous studies to be addicted to downstream activation of MEK (9). Our findings corroborate these observations; however, they further indicate that these and RAS-mutant tumors exhibit the highest degree of susceptibility to combined PI3K/MAPK pathway inhibition. Thus, as is the case with RTK-driven tumors, suppression of feedback-mediated PI3K activation is still synergistic, even in the case of a direct downstream dependency. We further found NRAS-mutant NSCLC tumors to be exquisitely sensitive to MEK inhibition; thereby, adding these genotypes to the growing list of genetic lesions that might be amenable to specific therapeutic intervention (13).

Of note, we found that intermittent administration of a combination of a MEK inhibitor and a PI3K inhibitor at their respective MTD was similarly effective as the daily administration at lower doses. This observation suggests that noncontinuous, but potent inhibition of these pathways might be sufficient for tumor growth inhibition, a finding reminiscent of BCR-ABL inhibition in chronic myeloid leukemia (39). Thus, rather than administering continuously lower doses of inhibitors targeting downstream signaling pathways, intermittent high dosing of such drug combinations might be better tolerated and allow for more potent target inhibition, induction of apoptosis, and tumor control.

In summary, we have defined the role of PI3K and MAPK signaling in genetically defined cancers, and provide strong evidence that combined inhibition of both pathways might be clinically beneficial. More broadly, our chemical-genomics approach may be useful for the study of novel therapeutics and might help to direct future drug development and patient stratification in clinical trials.

Previous SectionNext Section

Materials and Methods

The cell line panel described previously (19) was used for cell-based screening against various inhibitors using CellTiterGlo as a growth inhibition assay or using Annexin-V and propidium iodide staining of cells as a measure of apoptosis. The accuracy of the measurement of apoptosis was assessed in nine representative cell lines (Fig. S10). Cell-based screening was performed as described (19). This cell line panel has been shown to represent the distribution of genetic aberrations present in primary lung tumors (19). Other cell lines representing additional non-NSCLC tumor types were included to test whether observations were general features of tumors with the respective genotypes. Calculation of the P values was performed using a two-tailed t test implemented in “R” and, where appropriate, corrected for testing of multiple hypotheses. Pharmacodynamic response of signaling was measured by immunoblotting of cellular lysates of treated cells using phospho-specific antibodies. Mouse experiments were performed under approval of the respective animal care review board. Mice were treated with inhibitors by oral gavage using the indicated doses and schedules. Tumor size was determined by magnetic-resonance imaging in the case of transgenic mice and by measuring diameters using a caliper in the case of xenografts. After treatment, mice were killed and tumors were explanted and, in some instances, subjected to immunohistochemical staining of markers of response. For more details, see SI Methods.

Previous SectionNext Section

Acknowledgments

This work was supported by Deutsche Krebshilfe Grant 107954, Fritz-Thyssen-Stiftung Grant 10.08.2.175, and German Ministry of Science and Education NGFNplus-program BMBF Grant 01GS08100 (to R.K.T.), by the Deutsche Forschungsgemeinschaft through SFB832 (to R.K.T. and L.C.H.), and National Cancer Institute Grant R01 CA121210 (to W.P.). S.F. holds a scholarship of the Cologne Fortune Program/Faculty of Medicine, University of Cologne.

Previous SectionNext Section

Footnotes

  • 2To whom correspondence should be addressed. E-mail: nini{at}nf.mpg.de

  • Author contributions: M.L.S., S.F., R.U., M.P., M.K., S.H., I.S., J.W., K.M., A.G., L.R., K.A.P., S.P., L.C.H., T.Z., R.B., H.K., C.O., K.P.H., L.F., K.-K.W., W.P., and R.K.T. designed research; M.L.S., S.F., R.U., J.M.H., M.K., S.H., I.S., J.W., F.F., K.M., A.G., L.R., K.A.P., S.P., H.K., C.O., and K.P.H. performed research; M.G., K.M.S., D.R., and L.F. contributed new reagents/analytic tools; M.L.S., S.F., R.U., M.P., J.M.H., M.K., S.H., J.W., F.F., K.M., A.G., L.R., K.A.P., S.P., M.G., L.C.H., S.A., T.Z., R.B., H.K., K.M.S., W.R.S., D.R., C.O., K.P.H., L.F., K.-K.W., W.P., and R.K.T. analyzed data; and M.L.S., S.F., M.P., S.A., W.R.S., D.R., K.-K.W., W.P., and R.K.T. wrote the paper.

  • Conflict of interest statement: R.K.T. has received research support from AstraZeneca and Novartis. W.R.S. is an employee of Novartis. C.O., K.P.H., and L.F. are employees of Genentech.

  • This article is a PNAS Direct Submission.

  • This article contains supporting information online at www.pnas.org/cgi/content/full/0907325106/DCSupplemental.

  • Freely available online through the PNAS open access option.

Previous Section
 

References

  1. ↵
    1. Sharma SV,
    2. Bell DW,
    3. Settleman J,
    4. Haber DA
    (2007) Epidermal growth factor receptor mutations in lung cancer. Nat Rev Cancer 7:169–181.
    CrossRefMedlineWeb of Science
  2. ↵
    1. Kantarjian H,
    2. et al.
    (2002) Hematologic and cytogenetic responses to imatinib mesylate in chronic myelogenous leukemia. N Engl J Med 346:645–652.
    Abstract/FREE Full Text
  3. ↵
    1. Shah NP,
    2. et al.
    (2004) Overriding imatinib resistance with a novel ABL kinase inhibitor. Science 305:399–401.
    Abstract/FREE Full Text
  4. ↵
    1. Wood LD,
    2. et al.
    (2007) The genomic landscapes of human breast and colorectal cancers. Science 318:1108–1113.
    Abstract/FREE Full Text
  5. ↵
    1. Parsons DW,
    2. et al.
    (2008) An integrated genomic analysis of human glioblastoma multiforme. Science 321:1807–1812.
    Abstract/FREE Full Text
  6. ↵
    1. Jones S,
    2. et al.
    (2008) Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science 321:1801–1806.
    Abstract/FREE Full Text
  7. ↵
    1. Ding L,
    2. et al.
    (2008) Somatic mutations affect key pathways in lung adenocarcinoma. Nature 455:1069–1075.
    CrossRefMedlineWeb of Science
  8. ↵
    1. McLendon R,
    2. et al.
    (2008) Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455:1061–1068.
    CrossRefMedlineWeb of Science
  9. ↵
    1. Solit DB,
    2. et al.
    (2006) BRAF mutation predicts sensitivity to MEK inhibition. Nature 439:358–362.
    CrossRefMedline
  10. ↵
    1. Engelman JA,
    2. et al.
    (2008) Effective use of PI3K and MEK inhibitors to treat mutant Kras G12D and PIK3CA H1047R murine lung cancers. Nat Med 14:1351–1356.
    CrossRefMedlineWeb of Science
  11. ↵
    1. Engelman JA,
    2. et al.
    (2005) ErbB-3 mediates phosphoinositide 3-kinase activity in gefitinib-sensitive non-small cell lung cancer cell lines. Proc Natl Acad Sci USA 102:3788–3793.
    Abstract/FREE Full Text
  12. ↵
    1. Engelman JA,
    2. et al.
    (2007) MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling. Science 316:1039–1043.
    Abstract/FREE Full Text
  13. ↵
    1. Jaiswal BS,
    2. et al.
    (2009) Combined targeting of BRAF and CRAF or BRAF and PI3K effector pathways is required for efficacy in NRAS mutant tumors. PLoS One 4:e5717.
    CrossRefMedline
  14. ↵
    1. Hoeflich KP,
    2. et al.
    (2009) In vivo antitumor activity of MEK and phosphatidylinositol 3-kinase inhibitors in basal-like breast cancer models. Clin Cancer Res 15:4649–4664.
    Abstract/FREE Full Text
  15. ↵
    1. Pratilas CA,
    2. et al.
    (2009) (V600E)BRAF is associated with disabled feedback inhibition of RAF-MEK signaling and elevated transcriptional output of the pathway. Proc Natl Acad Sci USA 106:4519–4524.
    Abstract/FREE Full Text
  16. ↵
    1. O'Reilly KE,
    2. et al.
    (2006) mTOR inhibition induces upstream receptor tyrosine kinase signaling and activates Akt. Cancer Res 66:1500–1508.
    Abstract/FREE Full Text
  17. ↵
    1. Carracedo A,
    2. et al.
    (2008) Inhibition of mTORC1 leads to MAPK pathway activation through a PI3K-dependent feedback loop in human cancer. J Clin Invest 118:3065–3074.
    MedlineWeb of Science
  18. ↵
    1. Alessi DR,
    2. Cuenda A,
    3. Cohen P,
    4. Dudley DT,
    5. Saltiel AR
    (1995) PD 098059 is a specific inhibitor of the activation of mitogen-activated protein kinase kinase in vitro and in vivo. J Biol Chem 270:27489–27494.
    Abstract/FREE Full Text
  19. ↵
    1. Sos ML,
    2. et al.
    (2009) Predicting drug susceptibility of non-small cell lung cancers based on genetic lesions. J Clin Invest 119:1727–1740.
    CrossRefMedlineWeb of Science
  20. ↵
    1. Hara K,
    2. et al.
    (2002) Raptor, a binding partner of target of rapamycin (TOR), mediates TOR action. Cell 110:177–189.
    CrossRefMedlineWeb of Science
  21. ↵
    1. Kim DH,
    2. et al.
    (2002) mTOR interacts with raptor to form a nutrient-sensitive complex that signals to the cell growth machinery. Cell 110:163–175.
    CrossRefMedlineWeb of Science
  22. ↵
    1. Loewith R,
    2. et al.
    (2002) Two TOR complexes, only one of which is rapamycin sensitive, have distinct roles in cell growth control. Mol Cell 10:457–468.
    CrossRefMedlineWeb of Science
  23. ↵
    1. Zheng XF,
    2. Florentino D,
    3. Chen J,
    4. Crabtree GR,
    5. Schreiber SL
    (1995) TOR kinase domains are required for two distinct functions, only one of which is inhibited by rapamycin. Cell 82:121–130.
    CrossRefMedlineWeb of Science
  24. ↵
    1. Engelman JA,
    2. Luo J,
    3. Cantley LC
    (2006) The evolution of phosphatidylinositol 3-kinases as regulators of growth and metabolism. Nat Rev Genet 7:606–619.
    CrossRefMedlineWeb of Science
  25. ↵
    1. Zunder ER,
    2. Knight ZA,
    3. Houseman BT,
    4. Apsel B,
    5. Shokat KM
    (2008) Discovery of drug-resistant and drug-sensitizing mutations in the oncogenic PI3K isoform p110 alpha. Cancer Cell 14:180–192.
    CrossRefMedlineWeb of Science
  26. ↵
    1. Fan QW,
    2. et al.
    (2006) A dual PI3 kinase/mTOR inhibitor reveals emergent efficacy in glioma. Cancer Cell 9:341–349.
    CrossRefMedlineWeb of Science
  27. ↵
    1. Folkes AJ,
    2. et al.
    (2008) The identification of 2-(1H-indazol-4-yl)-6-(4-methanesulfonyl-piperazin-1-ylmethyl)-4-morpholin -4-yl-thieno[3,2-d]pyrimidine (GDC-0941) as a potent, selective, orally bioavailable inhibitor of class I PI3 kinase for the treatment of cancer. J Med Chem 51:5522–5532.
    CrossRefMedlineWeb of Science
  28. ↵
    1. Perera SA,
    2. et al.
    (2009) HER2YVMA drives rapid development of adenosquamous lung tumors in mice that are sensitive to BIBW2992 and rapamycin combination therapy. Proc Natl Acad Sci USA 106:474–479.
    Abstract/FREE Full Text
  29. ↵
    1. Regales L,
    2. et al.
    (2007) Development of new mouse lung tumor models expressing EGFR T790M mutants associated with clinical resistance to kinase inhibitors. PLoS ONE 2:e810.
    CrossRefMedline
  30. ↵
    1. Friday BB,
    2. Adjei AA
    (2008) Advances in targeting the Ras/Raf/MEK/Erk mitogen-activated protein kinase cascade with MEK inhibitors for cancer therapy. Clin Cancer Res 14:342–346.
    Abstract/FREE Full Text
  31. ↵
    1. Ji H,
    2. et al.
    (2007) Mutations in BRAF and KRAS converge on activation of the mitogen-activated protein kinase pathway in lung cancer mouse models. Cancer Res 67:4933–4939.
    Abstract/FREE Full Text
  32. ↵
    1. Chou TC
    (2006) Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug combination studies. Pharmacol Rev 58:621–681.
    Abstract/FREE Full Text
  33. ↵
    1. Chou TC,
    2. Talalay P
    (1984) Quantitative analysis of dose-effect relationships: The combined effects of multiple drugs or enzyme inhibitors. Adv Enzyme Regul 22:27–55.
    CrossRefMedlineWeb of Science
  34. ↵
    1. Sharma SV,
    2. et al.
    (2006) A common signaling cascade may underlie “addiction” to the Src, BCR-ABL, and EGF receptor oncogenes. Cancer Cell 10:425–435.
    CrossRefMedlineWeb of Science
  35. ↵
    1. Kobayashi S,
    2. et al.
    (2005) EGFR mutation and resistance of non-small-cell lung cancer to gefitinib. N Engl J Med 352:786–792.
    Abstract/FREE Full Text
  36. ↵
    1. Pao W,
    2. et al.
    (2005) Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is associated with a second mutation in the EGFR kinase domain. PLoS Med 2:e73.
    CrossRefMedline
  37. ↵
    1. Bean J,
    2. et al.
    (2007) MET amplification occurs with or without T790M mutations in EGFR mutant lung tumors with acquired resistance to gefitinib or erlotinib. Proc Natl Acad Sci USA 104:20932–20937.
    Abstract/FREE Full Text
  38. ↵
    1. Sos ML,
    2. et al.
    (2009) PTEN loss contributes to erlotinib resistance in EGFR-mutant lung cancer by activation of Akt and EGFR. Cancer Res 69:3256–3261.
    Abstract/FREE Full Text
  39. ↵
    1. Shah NP,
    2. et al.
    (2008) Transient potent BCR-ABL inhibition is sufficient to commit chronic myeloid leukemia cells irreversibly to apoptosis. Cancer Cell 14:485–493.
    CrossRefMedlineWeb of Science
« Previous | Next Article »Table of Contents
OPEN ACCESS ARTICLE

This Article

  1. Published online before print October 5, 2009, doi: 10.1073/pnas.0907325106 PNAS October 27, 2009 vol. 106 no. 43 18351-18356
  1. Free via Open Access: OA
  2. OA Abstract
  3. Figures Only
  4. » Full Text
  5. Full Text (PDF)
  6. Supporting Information

Readers Also Viewed

Classifications

Navigate This Article

Link: Advanced Search

This Week's Issue

  1. February 9, 2010, 107 (6)
  1. Current Issue
  1. Alert me to new issues of PNAS

Most