Inhibition of the T790M Gatekeeper Mutant of the Epidermal Growth Factor Receptor by EXEL-7647

Receptor tyrosine kinases (RTK) are transmembrane proteins with an extracellular ligand binding domain and an intracellular tyrosine kinase catalytic domain. On binding to their cognate ligands, most RTKs dimerize and become activated through autophosphorylation of intracellular tyrosine residues. Activation of RTKs results in up-regulation of multiple cellular signaling pathways that promote cell growth, survival, and angiogenesis and normally occurs only in response to specific developmental or environmental stimuli. Inappropriate activation of RTKs via mutation, overexpression, or ectopic ligand production is a frequent feature of human tumor development and progression and is thought to be a major mechanism by which cancer cells subvert normal growth controls. Consequently, in recent years, modulation of RTK signal transduction has been an active area in oncology drug discovery (1, 2).

The epidermal growth factor receptor (EGFR; also known as ErbB1) and other ErbB family RTKs (ErbB2/HER-2/neu, ErbB3/HER-3, and ErbB4/HER-4) encoded by the c-erbB proto-oncogenes have been strongly implicated in cancer development and progression (reviewed in refs. 1, 3, 4). Overexpression of ErbB receptors and ligands has been documented in multiple tumor types, including lung, breast, colorectal, prostate, head and neck, and glioblastoma (3). Several mechanisms can cause aberrant receptor activation, resulting in tyrosine kinase activity, which is observed in cancer, including receptor overexpression, mutation, ligand-dependent receptor dimerization, and ligand-independent activation. For ErbB2, where a specific ligand has not been identified, activation occurs by homodimerization or heterodimerization alone, whereas ErbB3 does not have significant kinase activity (5). However, on activation, all four receptors are capable of signal transduction (6), causing activation of the ras/mitogen-activated protein kinase pathway, the phosphatidylinositol 3-kinase/AKT pathway, Src family kinases, and signal transducer and activator of transcription proteins. Activation of these pathways promotes cell proliferation, survival, and angiogenesis (reviewed in refs. 7, 8).

Activating somatic mutations in the EGFR gene are found in several human tumor types (9, 10). A deletion mutation in the extracellular domain of the receptor known as EGFR variant III (EGFRvIII) is overexpressed in breast, glioma, and lung cancers (11) and exhibits ligand-independent, constitutive kinase activity (12). Small deletions or point mutations in the tyrosine kinase domain of EGFR are present in the tumors of non–small cell lung cancer (NSCLC) patients that experienced significant clinical responses following treatment with erlotinib or gefitinib (9, 10, 13). These mutations confer ligand independence or hypersensitivity to EGF and are more sensitive to inhibition by erlotinib or gefitinib than the wild-type (WT) receptor. However, some patients who initially experienced substantial objective tumor responses when treated with EGFR inhibitors have subsequently relapsed. In some cases, this resistance is associated with the acquisition of a “second site” point mutation (T790M) in the EGFR kinase domain (14). This residue is referred to as the “gatekeeper” threonine in EGFR and resides deep within the ATP-binding pocket. Analogous mutations in Bcr-Abl and c-Kit have been implicated in imatinib-resistant chronic myelogenous leukemia and gastrointestinal stromal tumors. In vitro experiments show that this gatekeeper mutation renders the activated EGFR insensitive to the inhibitory effects of erlotinib and gefitinib.

In addition to the ErbB family members that drive proliferation within the tumor, RTKs expressed in the supporting vasculature are important during tumor development. Vascular endothelial growth factors (VEGF) and their receptors (VEGFR1/Flt-1, VEGFR2/KDR, and VEGFR3/Flt-4) play critical roles in tumor angiogenesis and lymphangiogenesis (15). Like EGFR, activation of VEGF receptors (VEGFR) initiates the ras/mitogen-activated protein kinase cascade within endothelial cells that is responsible, in part, for the proangiogenic phenotype of VEGF (16). The Eph receptors comprise the largest subfamily of RTKs, members of which are also required for normal development of the vasculature and nervous system. Eph receptors and their cell-presented ligands, the ephrins, are frequently overexpressed in human tumors (17). In particular, ephrin type B subclass 4 receptor (EphB4) is up-regulated in prostate, breast, endometrial, small-cell lung, and gastrointestinal cancers (18–22). Overexpression of EphB4 in transgenic mice promotes tumor formation, whereas reduction of EphB4 in tumor cell lines inhibits growth both in vitro and in vivo (23).

Agents that selectively target EGFR, ErbB2, or VEGF have shown promising activity in clinical trials and several are now approved for use in select cancer indications. These include monoclonal antibodies targeting EGFR (cetuximab), ErbB2 (trastuzumab), VEGF (bevacizumab), and small-molecule EGFR inhibitors (erlotinib and gefitinib). Several lines of evidence suggest that simultaneous targeting of multiple members of the EGF and VEGF RTK families could provide improved and/or broader spectrum efficacy versus highly selective agents. For example, combinations of small-molecule EGFR and VEGFR inhibitors show improved efficacy in preclinical models compared with either agent alone (24). In addition, a small molecule that inhibits both EGFR and ErbB2 has shown activity in trastuzumab-resistant cells (25). We have identified a novel small-molecule inhibitor that targets multiple EGF and VEGF RTK family members, with the aim of affecting tumor cell growth and survival through direct effects on tumor cells and through inhibition of angiogenesis. This compound, EXEL-7647(XL647), is a potent inhibitor of EGFR, ErbB2, KDR, and EphB4 both enzymatically and in cellular assays of receptor phosphorylation. In addition to potent and long-lived inhibition of WT EGFR in vivo, EXEL-7647 retains significant EGFR inhibitory activity in the bronchoalveolar cell line, H1975, which harbors both an activating mutation, L858R, as well as the T790M mutation that confers resistance to erlotinib and gefitinib. When H1975 tumor xenografts were established in female severe combined immunodeficient mice, once-daily oral administration of EXEL-7647 significantly inhibited tumor growth in a dose-dependent manner. In contrast, erlotinib showed less efficacy and no dose dependence.

Immunohistochemical analysis of the tumors at the end of the dosing period showed a dose-dependent decrease in activated EGFR and its downstream effectors phosphorylated ERK (pERK^{Thr202/Tyr204}) and phosphorylated AKT (pAKT^Ser473), as well as decreases in tumor vasculature (CD31) and cell proliferation (Ki67), and an increase in apoptosis (terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling).

EXEL-7647 also shows antiangiogenic activity in vivo, consistent with the potent inhibition of EXEL-7647 against the VEGFR family and EphB4 in vitro. When examined in MDA-MB-231 breast adenocarcinoma xenografts, which highly express VEGF (26, 27) and are not as reliant on EGFR signaling for tumor growth (28), EXEL-7647, but not erlotinib, showed potent and dose-dependent tumor growth inhibition and decreases in tumor vascularization (CD31). Taken together, these data show that EXEL-7647 is a potent inhibitor of RTKs that promote tumor epithelial cell growth and endothelial cell function that may have broad clinical utility particularly in tumors that are resistant to existing EGFR-targeted therapies.

EXEL-7647 (29) was synthesized and purified in the Medicinal Chemistry Department at Exelixis, Inc. Gefitinib (30) and erlotinib (31) for use in biochemical and cellular assays were synthesized at Exelixis. For in vitro assays, a 10 mmol/L stock solution of each compound was prepared in DMSO and diluted in optimal assay buffers or culture medium. The final DMSO assay concentration did not exceed 0.3% (v/v). For in vivo studies, EXEL-7647 was formulated for oral administration by dissolution of the dry powder in sterile filtered (0.45 μm; Nalge Nunc International) saline (0.9% USP, Baxter Corp.) or in sterile water for injection (Baxter). For the A431 pharmacodynamic and efficacy studies, erlotinib and gefitinib were formulated as suspensions in 0.5% carboxymethylcellulose/5% ethanol. For the H1975 and MDA-MB-231 efficacy studies, erlotinib was formulated as a suspension of finely crushed clinical tablets (100 mg Tarceva tablets, Genentech, Inc.) suspended in water. All compounds were mixed by vortexing and sonicated in a water bath to disrupt large particles. All dosing solutions/suspensions were prepared fresh daily.

Kinase activity and compound inhibition were investigated using one of three assay formats described below. The ATP concentration for each assay was equivalent to the K_m for each kinase. The kinase assay buffer contained 20 mmol/L Tris-HCl (pH 7.5), 10 mmol/L MgCl₂, 3 mmol/L MnCl₂, 1 mmol/L DTT, and 0.01% Triton X-100. Dose-response experiments were done using 10 different inhibitor concentrations in 384-well microtiter plates. IC₅₀ values were calculated by nonlinear regression analysis using the four-variable equation: Y = min + (max − min) / [1 + ([I] / IC₅₀)^N], where Y is the observed signal, [I] is the inhibitor concentration, min is the background signal in the absence of enzyme (0% enzyme activity), max is the signal in the absence of inhibitor (100% enzyme activity), IC₅₀ is the inhibitor concentration required at 50% enzyme inhibition, and N represents the empirical Hill slope as a measure of cooperativity.

³³P-Phosphoryl transfer kinase assay. Reactions were done in 384-well white, clear bottom, high-binding microtiter plates (Greiner). Plates were coated with 2 μg/well peptide substrate in a 50 μL volume. The coating buffer contained 40 μg/mL EphB4 substrate poly(Ala-Glu-Lys-Tyr) 6:2:5:1 (Perkin-Elmer), 22.5 mmol/L Na₂CO₃, 27.5 mmol/L NaHCO₃, 150 mmol/L NaCl, and 3 mmol/L NaN₃. The coated plates were washed once with 50 μL assay buffer following overnight incubation at room temperature. Test compound and either 5 nmol/L EphB4 (residues E605-E890 of human EphB4 containing a six histidine NH₂-terminal tag, which was expressed in a baculovirus expression system and purified using metal chelate chromatography), 4 nmol/L insulin-like growth factor-I receptor (residues M954-C1367 of human insulin-like growth factor-I receptor, Proqinase GmbH), or 15 nmol/L insulin receptor 1 (residues P948-S1343 of human insulin receptor 1, Proqinase) were combined with [³³P]γ-ATP (5 μmol/L, 3.3 μCi/nmol) in a total volume of 20 μL. The reaction mixture was incubated at room temperature for 1.5 to 2.5 h and terminated by aspiration. The microtiter plates were subsequently washed six times with 0.05% Tween-PBS buffer. Scintillation fluid (50 μL/well) was added and incorporated ³³P was measured by liquid scintillation spectrometry using a MicroBeta scintillation counter (Perkin-Elmer).

Luciferase-coupled chemiluminescence assay. Kinase activity was measured as the percentage of ATP consumed following the kinase reaction using luciferase-luciferin–coupled chemiluminescence. Reactions were conducted in 384-well white, medium-binding microtiter plates (Greiner). Kinase reactions were initiated by combining test compound, 3 μmol/L ATP, 1.6 μmol/L substrate [poly(Glu,Tyr) 4:1; Perkin-Elmer], and either EGFR (7 nmol/L, residues H672-A1210 of human EGFR, Proqinase) or KDR (5 nmol/L, residues D807-V1356 of human KDR, Proqinase) in a 20 μL volume. The reaction mixture was incubated at room temperature for 4 h. Following the kinase reaction, a 20 μL aliquot of Kinase Glo (Promega) was added and luminescence signal was measured using a Victor2 plate reader (Perkin-Elmer). Total ATP consumption was limited to 50%.

AlphaScreen tyrosine kinase assay. Donor beads coated with streptavidin and acceptor beads coated with PY100 anti-phosphotyrosine antibody (Perkin-Elmer) were used. Biotinylated poly(Glu,Tyr) 4:1 was used as the substrate. Substrate phosphorylation was measured by luminescence following donor-acceptor bead addition followed by complex formation. Test compound, 3 μmol/L ATP, 3 nmol/L biotinylated poly(Glu,Tyr) 4:1, and 1 nmol/L ErbB2 (residues Q679-V1255 of human ErbB2, Proqinase) or Flt-4 (residues D725-R1298 of human Flt-4, Proqinase) were combined in a volume of 20 μL in a 384-well white, medium-binding microtiter plate (Greiner). Reaction mixtures were incubated for 1 h at room temperature. Reactions were quenched by addition of 10 μL of 15 to 30 μg/mL AlphaScreen bead suspension containing 75 mmol/L HEPES (pH 7.4), 300 mmol/L NaCl, 120 mmol/L EDTA, 0.3% bovine serum albumin, and 0.03% Tween 20. After 2 to 16 h of incubation at room temperature, plates were read using an AlphaQuest reader (Perkin-Elmer).

The A431 human epidermoid carcinoma cell line was purchased from the American Type Culture Collection, and MDA-MB-231 human adenocarcinoma cells were generously provided by Georgetown University. Both cell lines were maintained and propagated as monolayer cultures in DMEM (Mediatech) containing l-glutamine supplemented with 10% heat-inactivated fetal bovine serum (Hyclone), 100 units/mL penicillin G, 100 μg/mL streptomycin (1% penicillin/streptomycin, Mediatech), and 1% nonessential amino acids (Mediatech) at 37°C in a humidified 5% CO₂ incubator. The NSCLC adenocarcinoma H1975 (American Type Culture Collection) and the Lx-1 squamous cell carcinoma (Department of Oncology Drug Discovery, Bristol-Myers Squibb) cell lines were maintained in complete RPMI 1640 (30-2,001; American Type Culture Collection; containing l-glutamine supplemented with 10% heat-inactivated fetal bovine serum, 1% penicillin/streptomycin, and 1% nonessential amino acids) at 37°C in a humidified 5% CO₂ incubator.

WT EGFR autophosphorylation. A431 cells were seeded at 5 × 10⁴ per well in 96-well microtiter plates (3904 Costar, VWR) and incubated in fully supplemented DMEM for 16 h after which growth medium was replaced with serum-free DMEM and the cells were incubated for an additional 24 h. Serial dilutions of EXEL-7647 (in triplicate) in serum-free medium were added to the quiescent cells and incubated for 1 h before stimulation with 100 ng/mL recombinant human EGF (R&D Systems) for 10 min. Negative control wells did not receive EGF. After treatment, cell monolayers were washed with cold PBS and immediately lysed with cold lysis buffer [50 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 10% glycerol, 1% NP40, 0.1% SDS, 0.5% sodium deoxycholate, 1 mmol/L EDTA, 50 mmol/L NaF, 1 mmol/L sodium pyrophosphate, 1 mmol/L sodium orthovanadate, 2 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin, 5 μg/mL leupeptin, 5 μg/mL pepstatin]. Lysates were centrifuged, transferred to 96-well streptavidin-coated plates (Pierce) containing biotin-conjugated, mouse monoclonal anti-human EGFR (2 μg/mL; Research Diagnostics), and incubated for 2 h. Plates were washed thrice with TBST [25 mmol/L Tris, 150 mmol/L NaCl (pH 7.2), 0.1% bovine serum albumin, and 0.05% Tween 20] and incubated with horseradish peroxidase–conjugated anti-phosphotyrosine antibody (1:10,000; Zymed Laboratories). Horseradish peroxidase activity was determined by reading the plates in a Victor2 plate reader following addition of the ELISA Femto substrate (Pierce). IC₅₀ values were determined based on total EGFR tyrosine phosphorylation with EXEL-7647 treatment versus total EGFR tyrosine phosphorylation with growth factor treatment alone, normalized to receptor levels.

Expression vector cloning. A clone corresponding to the longest EGFR isoform (Genbank accession no. NM_005228.3/NP_005219.2 #21-176, Upstate Biotechnology) was used as a template to produce two mutant EGFR genes (coding for L858R and L861Q) by site-directed mutagenesis. The WT and the two sequence-verified mutants were transferred to a COOH-terminal Flag-tagged retroviral cytomegalovirus promoter-driven mammalian expression vector. The two Tet-On expression vectors, EGFR WT (Tet-On) and EGFRvIII (Tet-On), which were COOH-terminally Flag tagged, were generously provided by Dr. Abhijit Guha (University of Toronto, Toronto, Ontario, Canada).

Transient transfections. Transient transfections of Lx-1 cells were done using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's protocol. For transfections of the WT, L858R, and L861Q constructs, 1 μg plasmid DNA was used for each transfection (each well of a 12-well plate). For transfections of Tet-controlled EGFR WT and variant III constructs, 0.5 μg of either construct was combined with 0.5 μg of the pTet-On plasmid (BD Biosciences) for each transfection. Cells were harvested 24 h after transfection and replated in either 96-well plates (4 × 10⁴ cells per well) for compound treatments or 12-well plates (2 × 10⁵ cells) for immunoblot assays. Expression of the EGFRvIII transgene was induced 24 h after transfection by adding 1 μg/mL doxycycline to the medium. These cells were maintained in the presence of doxycycline for the remainder of the experiment. After a 12-h incubation, the cells were serum starved (in fetal bovine serum–free medium) and immediately treated with the indicated compounds in triplicate for 24 h followed by a 10-min treatment with recombinant human EGF (100 ng/mL). Whole-cell lysates were made by adding 125 μL radioimmunoprecipitation assay buffer (Boston Bioproducts) containing protease inhibitors (Protease Inhibitor Cocktail Tablets, Roche) in addition to 50 mmol/L NaF, 1 mmol/L sodium pyrophosphate, 1 mmol/L sodium orthovanadate, 2 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin, and 5 μg/mL leupeptin in each well for either EGFR phosphorylation ELISA or immunoblot.

EGFR ELISA. Reacti-Bind streptavidin-coated plates (Pierce) were coated with 2 μg/mL biotin-conjugated anti-Flag antibody (Sigma). Whole-cell lysates (10 μg) were then added to the anti-Flag–coated wells in a final volume of 100 μL for 2 h at room temperature and then washed thrice with TBST. The anti-phosphotyrosine horseradish peroxidase–coupled secondary antibody (1:10,000; Zymed) was used to detect phosphorylated EGFR (pEGFR; 1 h at room temperature followed by three washes with TBST). Horseradish peroxidase activity was determined by reading the plates in a Victor2 plate reader following addition of the ELISA Femto substrate.

Cell viability assay. Growth inhibition of H1975 and A431 cells by increasing concentrations of EXEL-7647, gefitinib, or erlotinib was determined by seeding 5 × 10³ cells per well in 96-well plates. The following day, cells were washed once with low-serum RPMI 1640 (0.1% fetal bovine serum, 1% nonessential amino acids, and 1% penicillin/streptomycin), after which 90 μL of the low-serum RPMI 1640 were added. Test compounds were diluted to 10 times the test concentrations and 10 μL were added to triplicate wells for a 72-h incubation. Cell viability was determined using Alamar blue solution (Biosource).

Analysis of H1975 cell lysates. For the H1975 immunoblot studies, 3 × 10⁵ cells were plated in each well (12-well plate) and incubated for 16 h in complete RPMI 1640, rinsed with fetal bovine serum–free RPMI 1640, and incubated with serial dilutions of test compounds in fetal bovine serum–free medium for 2 h followed by stimulation with 100 ng/mL human recombinant EGF for 10 min. Whole-cell protein lysates were prepared as described above and centrifuged for 10 min at 13,000 × g at 4°C to remove any insoluble material. Total protein was determined using bicinchoninic acid reagent and an equal amount of protein was combined with LDS loading buffer (Invitrogen) according to the manufacturer's instructions. Proteins were separated by gel electrophoresis on 4% to 15% polyacrylamide gels, transferred to nitrocellulose membranes, and detected by immunoblotting. Antibody-antigen complexes were detected using chemiluminescence. The following antibodies from Cell Signaling Technology were used at a 1:1,000 dilution: anti-EGFR, anti-pEGFR^Tyr1068, anti-AKT, anti-pAKT^Ser473, anti-ERK, and anti-pERK^{Thr202/Tyr204}. The anti-β-actin primary antibody (Accurate Chemical and Scientific) was used at 1:10,000 and the horseradish peroxidase–coupled secondary antibodies were purchased from Jackson ImmunoResearch and used at 1:5,000.

Female severe combined immunodeficient mice and female athymic nude mice (NCr), 5 to 8 weeks of age and weighing ∼20 to 25 g, were purchased from The Jackson Laboratory and Taconic, respectively. The animals were housed at the Exelixis vivarium facilities according to guidelines outlined by the Exelixis Institutional Animal Care and Use Committee. During all studies, animals were provided food and water ad libitum and housed in a room conditioned at 70°F to 75°F and 60% relative humidity.

H1975, A431, or MDA-MB-231 cells were harvested from exponentially growing cultures, detached by brief trypsinization, washed twice in cold HBSS, resuspended in ice-cold HBSS, and implanted either s.c. (H1975, 3 × 10⁶ cells per mouse) or i.d. (A431, 1 × 10⁶ cells per mouse) into the dorsal hind flank or s.c. into the mammary fat pad (MDA-MB-231, 1 × 10⁶ cells per mouse). Palpable tumors were measured by caliper twice weekly until the mean tumor weight was in the range of ∼80 to 120 mg. Tumor weight was determined by measuring perpendicular diameters with a caliper and multiplying the measurements of diameters in two dimensions: tumor volume (mm³) / 2 = length (mm) × width² (mm²) / 2. Tumor weight (mg) was extrapolated from tumor volume (mm³) by assuming a conversion factor of 1. On the appropriate day after implantation, mice were grouped (10 mice per group) such that the group mean tumor weight was ∼100 ± 15 mg. The mean tumor weight of each animal in the respective control and treatment groups was determined twice weekly during the dosing periods. The response of tumors to treatment was determined by comparing the mean tumor weight of the treatment group with the appropriate control group. Percentage inhibition of tumor growth was determined with the following formula: Percentage inhibition = 100 × [1 − (X_f − X_o) / (Y_f − Y_o)], where X_f and Y_f are the mean tumor weights of the treatment and control groups, respectively, on day f, and X_o and Y_o are the mean tumor weights of treatment and control groups respectively, on day zero (staged tumor weights after grouping).

EGFR autophosphorylation was evaluated in tumor xenografts derived from A431 cells that overexpress WT EGFRs (29). Tumor-bearing mice were given either EXEL-7647, erlotinib, or gefitinib at 100 mg/kg and tumors were harvested 1 to 72 h later. Half an hour before respective time point, EGF (50 μg/mouse; R&D Systems) was given via i.v. bolus injection with tumors dissected 30 min later and tumor extracts were prepared by homogenization (Polytron PT 10/35, Brinkmann) in 10 volumes of ice-cold lysis buffer [50 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 1% NP40, 0.1% SDS, 0.5% deoxycholate, 1 mmol/L EDTA, 50 mmol/L NaF, 1 mmol/L sodium pyrophosphate, 1 mmol/L orthovanadate, 2 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin, 5 μg/mL leupeptin, 5 μg/mL pepstatin A]. Lysates were clarified by centrifugation (20,000 × g for 10 min at 4°C) and EGFR tyrosine phosphorylation levels were determined by ELISA (as described above).

For determination of compound levels in plasma following oral administration of EXEL-7647, gefitinib, or erlotinib, whole blood was placed in heparinized Eppendorf tubes on ice and centrifuged at 20,000 × g for 4 min. The plasma supernatant (50 μL) was added to 100 μL internal standard solution (250 ng/mL internal standard in acetonitrile), mixed by vortexing, and centrifuged. Sample extract (20 μL) was assayed for EXEL-7647, gefitinib, or erlotinib by LC/MS/MS analysis. Plasma levels were calculated using an authentic standard curve for each compound. The limit of quantification was 0.004 μmol/L (2 ng/mL) for EXEL-7647, gefitinib, and erlotinib. Mean values and SD were calculated for each time point and dose concentration was assessed.

For immunohistochemical analysis of H1975 and MDA-MB-231 xenografts, tumors were excised after euthanasia and fixed in zinc fixative (BD Biosciences) for 48 h before being processed into paraffin blocks. Serial sections at 5 μm were obtained from the area of largest possible surface for each tumor and stained using standard methods. The following antibodies were used: Ki67 (SP6; Labvision), CD31 (MECA.32; BD Biosciences), pERK^{Thr202/Tyr204} (phospho-p44/42 mitogen-activated protein kinase; Cell Signaling Technology), pAKT^Ser473 (Cell Signaling Technology), and pEGFR^Tyr1068 (Cell Signaling Technology). For immunofluorescent staining, sections were then incubated with Alexa 594–conjugted goat anti-rabbit secondary antibody (Invitrogen) and mounted in Fluorescent Mounting Medium (DAKO) containing 4′,6-diamidino-2-phenylindole (Molecular Probes) as a nuclear counterstain. Fluorescent staining was visualized using a Zeiss AxioImager and digitally captured using a Zeiss high-resolution camera coupled to AxioVision image analysis software. Two to three nonoverlapping representative fields were captured at ×200 or ×400 magnification depending on histologic readout and quantified using the integrated morphometric analysis functions in Metamorph software (Universal Imaging Corp.). Apoptotic cells were detected using terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling in situ cell death detection kit according to the manufacturer's instructions (Roche Diagnostics GmbH).

CD31-positive tumor vessels, Ki67-positive proliferating cells, and pERK staining in each tumor section were identified and quantified using the integrated morphometric analysis functions in the ACIS automatic cellular imaging system (Clarient, Inc.) and reviewed by a blinded observer. Number of CD31-positive vessels were identified across 5 to 10 randomly chosen fields of equal size at ×100 magnification in viable tumor tissue and calculated as number of vessels per square millimeter for each tumor, averaged for each treatment group, and compared with vehicle-treated controls. Percentage Ki67-positive cells were calculated as the ratio between Ki67-positive cells divided by the total number of cells identified across 5 to 10 randomly chosen fields of equal size in viable tumor tissue. The results for each tumor and treatment group were averaged and compared with vehicle-treated controls. The level of pERK staining was determined as described above and calculated as the ratio of antibody staining divided by the total number of cells identified, averaged for each treatment group, and compared with vehicle-treated controls.

Results are presented as mean ± SD or SE as indicated for each graph or table. For IC₅₀ comparison in the A431 cell viability experiment, two sample Student's t tests were applied to determine P values for each IC₅₀ pair assuming that the random fluctuations of replicates around the dose-response curve are distributed [log]normally with the individual replicates used as the ‘sample size’ for the t test (nine point dose response done in triplicate). For statistical analysis of immunohistochemical results from in vivo studies, two-tailed Student's t test analysis and Bonferroni correction were done to identify significant differences compared with vehicle control group (multiple use of a single vehicle control group) with a cumulative minimal requirement of P < 0.05. Final tumor weight measurements at the end of the H1975 efficacy study and percentage pEGFR, pAKT, Ki67 index, CD31, and terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling in H1975 xenografts were analyzed with one-way ANOVA followed by post hoc Student-Newman-Keul analysis for determination of statistical differences between EXEL-7647 and erlotinib.

In vitro target inhibition profile of EXEL-7647. EXEL-7647 was optimized as an inhibitor of a spectrum of growth-promoting and angiogenic RTKs to simultaneously block tumor growth and vascularization. In particular, EXEL-7647 potently inhibits the EGF/ErbB2, VEGF, and ephrin RTK families (Table 1). Mechanism of action studies for EGFR, ErbB2, KDR, and EphB4 confirmed that EXEL-7647 is a reversible and ATP competitive inhibitor (data not shown). EXEL-7647 was inactive against a panel of 10 tyrosine kinases (including the insulin and the insulin-like growth factor-1 receptor) and 55 serine-threonine kinases (including cyclin-dependent kinases, stress-activated protein kinases, and protein kinase C isoforms).

Potent biochemical inhibition of WT EGFR was confirmed in a cell-based assay using the A431 cell line that endogenously expresses elevated levels of EGFR (31). As shown in Table 2, EXEL-7647 shows potent inhibition of EGFR with a IC₅₀ value of 1 nmol/L. Activating somatic mutations resulting in either deletions or amino acid substitutions in the kinase domain of the EGFR gene have been found to correlate with clinical response to the EGFR RTK inhibitors, gefitinib and erlotinib, in NSCLC patients (9, 10). To assess the activity of EXEL-7647 against these EGFR mutants, WT, L861Q, and L858R mutant EGFRs were overexpressed by transient transfection in the squamous cell carcinoma cell line, Lx-1, and EGFR activation was measured. Lx-1 cells were chosen for these studies because EGFR expression is undetectable in this cell line. In this assay system, EXEL-7647 inhibited the WT EGFR with a IC₅₀ value of 12 nmol/L. Consistent with previous reports (9, 10), the mutant EGFRs seem to be more sensitive to EGFR inhibition than the WT receptor (Table 2). EGFRvIII is frequently present and amplified in malignant gliomas and exhibits ligand-independent kinase activity (11, 12). EXEL-7647 was assessed for inhibition of the activity of EGFRvIII in EGFR phosphorylation assays. A tetracycline-inducible (Tet-On) system was used to express both the WT and the EGFRvIII construct in Lx-1 cells. In this assay system, EXEL-7647 inhibited the WT and mutant EGFRvIII with IC₅₀ values of 5 and 74 nmol/L, respectively (Table 2).

EXEL-7647 inhibits mutant EGFR resistant to gefitinib and erlotinib. An EGFR mutation (T790M) has been identified in NSCLC patients who have activating mutations in the EGFR gene, who initially responded to gefitinib or erlotinib but later developed drug resistance (14). To assess the ability of EXEL-7647 and other EGFR inhibitors to inhibit the activity of an EGFR harboring the T790M mutation, cell viability assays were done using the NSCLC cell line H1975 that contains both an activating (L858R) and resistance-associated mutation (T790M). For comparison, cell viability assays were done in parallel using A431 cells, which express WT EGFR (29). In A431 cells, EXEL-7647, gefitinib, and erlotinib reduced cell viability with IC₅₀ values of 13, 31, and 84 nmol/L, with 95% confidence intervals of 10 to 15, 23 to 39, and 67 to 100 nmol/L, respectively (Fig. 1A). As expected, H1975 cells are more resistant to EGFR inhibitors than cells harboring the WT receptor. However, H1975 cells are 11 and 17 times more sensitive to EXEL-7647 (IC₅₀, 0.92 μmol/L) than to gefitinib (IC₅₀, 10.41 μmol/L) and erlotinib (IC₅₀, 16.14 μmol/L), with 95% confidence intervals of 0.84 to 1.01, 8.80 to 12.02, and 9.34 to 22.94 μmol/L, respectively (Fig. 1A). Student's t test analysis confirmed that EXEL-7647 is significantly more potent in the H1975 cell viability assay than either gefitinib or erlotinib (P < 0.001). Analysis of EGFR-mediated signaling in H1975 cells following treatment with the different inhibitors supports the observations made using the viability assay. As expected, EGFR in H1975 cells is constitutively phosphorylated (Fig. 1B). Downstream signaling pathways are also activated as judged by the constitutive phosphorylation of both AKT and ERK. Addition of exogenous EGF further increases EGFR, AKT, and ERK phosphorylation. Treatment with gefitinib or erlotinib has minimal effect on the phosphorylation of EGFR, AKT, or ERK when tested at concentrations up to 30 μmol/L. In contrast, treatment with 10 μmol/L EXEL-7647 results in an almost complete inhibition of EGFR, AKT, and ERK phosphorylation (Fig. 1B). These results are consistent with the effects observed in the viability assay and show that EXEL-7647 retains significant potency against the T790M EGFR mutant.

EXEL-7647 inhibits the WT and the gatekeeper mutant EGFR in vivo. Consistent with in vitro data, EXEL-7647 shows long-lasting inhibition of WT EGFR autophosphorylation in the A431 xenograft model, which overexpresses WT EGFR (32). Single oral doses of 100 mg/kg EXEL-7647 were given to A431 tumor-bearing athymic nude mice and tumors were harvested at various times between 1 and 72 h later, with administration of EGF 0.5 h before each tumor harvest. EGFR phosphorylation was substantially inhibited (>90%) throughout the entire time course of this study (Fig. 2A). At the 72-h time point, phosphorylation of EGFR in response to EGF is still maximally decreased, despite the plasma concentration of EXEL-7647 being <1 μmol/L. A second time course study was conducted with gefitinib and erlotinib, to compare the duration of action of EXEL-7647 with clinically validated EGFR inhibitors. Gefitinib and erlotinib showed substantial inhibition of EGFR phosphorylation in this model at the 4- and 8-h time points, but little or no inhibition was evident at subsequent time points (Fig. 2A).

Consistent with the pharmacodynamic assays described above, EXEL-7647 shows potent and dose-dependent tumor growth inhibition in the A431 xenograft model. Substantial tumor growth inhibition of 77% was achieved by once-daily oral administration of 10 mg/kg EXEL-7647, whereas the highest dose of 100 mg/kg causes tumor regression to 65% below the starting tumor size (Fig. 2B). This result is consistent with the potent antiproliferative activity of EXEL-7647 in A431 cells (Fig. 1A).

In cellular assays of EGFR activation, EXEL-7647 has reduced potency against the T790M EGFR mutant compared with WT EGFR; however, the concentrations required to inhibit mutant EGFR in H1975 cells are in the same range as the plasma levels of EXEL-7647 achieved following oral dosing (Fig. 2A). To examine the in vivo effects of EXEL-7647 on the T790M mutant EGFR, H1975 human tumor xenografts were established s.c. in female severe combined immunodeficient mice and allowed to reach ∼100 mg before once-daily oral administration of EXEL-7647 at 10, 30, and 100 mg/kg or erlotinib at 30 and 100 mg/kg for 14 days. Figure 3 shows that EXEL-7647 substantially inhibited tumor growth in a dose-dependent manner when compared with the vehicle-treated control group. Tumor growth inhibition reached 33%, 66%, and 92% for the 10, 30, and 100 mg/kg dose groups (P < 0.005 when compared with the vehicle-treated control group), respectively, with a calculated ED₅₀ of 17 mg/kg. Consistent with these data, the plasma concentrations of EXEL-7647 following the final dose increased in a dose-dependent manner (C_max was 1, 3, and 15 μmol/L for the 10, 30, and 100 mg/kg dose groups, respectively; T_max was 4 h). These end of study plasma levels determined after 14 consecutive days of dosing indicate a 1.5-fold accumulation of EXEL-7647 in the plasma on repeat dosing when compared with a single dose (i.e., at 4 h after a single dose of 100 mg/kg, the plasma concentration was 10.6 μmol/L, whereas at 4 h following the 14th daily dose, the plasma level was 15 μmol/L). EXEL-7647 was well tolerated at all dosing levels throughout the course of the experiment as evidenced by the lack of body weight loss in all groups (Fig. 3, inset). In contrast to EXEL-7647, erlotinib only modestly inhibited H1975 xenograft tumor growth in this study (Fig. 3). Tumor growth inhibition was statistically significant (P < 0.005) when compared with the vehicle-treated control group, but modest and not dose independent (33-37% inhibition for the 30 and 100 mg/kg dose groups). For erlotinib, efficacy was likely limited for mechanistic, rather than pharmacokinetic, reasons because last dose pharmacokinetic profiling showed that the erlotinib plasma exposure was substantial and increased with dose (C_max was 12 and 27 μmol/L, respectively, for the 30 and 100 mg/kg dose groups; T_max was 4 h). Statistical analysis by one-way ANOVA with post hoc Student-Newman-Keul revealed that EXEL-7647 was significantly more potent in preventing tumor growth in this xenograft model than erlotinib (P = 0.01, dose to dose comparison).

Immunohistochemical analyses of the tumors at the end of the dosing period showed that, in contrast to erlotinib, EXEL-7647 treatment was associated with a significant (P < 0.001) and dose-dependent reduction in the levels of pEGFR when compared with vehicle-treated tumors (Fig. 4; Table 3). In addition, EXEL-7647 showed significant and dose-dependent decreases in phosphorylation of the EGFR effectors AKT and ERK (Table 3), whereas erlotinib showed moderate decreases (Table 3). Further immunohistochemical analyses showed that EXEL-7647 increased the percentage of terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling–positive cells with a 7- to 16-fold increase in apoptosis compared with vehicle-treated tumors (Table 4). The percentage of CD31-positive vessels in viable tumor tissue was also significantly (P < 0.005) decreased following EXEL-7647 treatment, indicating a dose-dependent loss of microvessel density (Fig. 4; Table 4). Similar decreases in the percentage of Ki67-expressing cells (Ki67 index) in the H1975 tumors were also observed at all dose levels, indicating a significant reduction in proliferating cells in the tumors at the end of the study. Tumors from the groups that received erlotinib at 30 and 100 mg/kg showed moderate effects on apoptosis and Ki67 index and no reduction in number of tumor vessels. Thus, treatment of H1975 xenografts with EXEL-7647 results in significant inhibition of EGFR and its downstream effectors in vivo, resulting in inhibition of tumor cell proliferation and vascularization and an induction of tumor cell apoptosis. The observed decreases in pEGFR and pAKT in the EXEL-7647–treated groups as well as the effects on tumor cell proliferation, vascularization, and apoptosis were significantly more potent than those observed in the erlotinib-treated groups (P < 0.01; one site ANOVA with post hoc Student-Newman-Keul analysis).

The A431 and H1975 xenograft studies described here examined the in vivo effects of EXEL-7647 on WT and mutant EGFR; however, as mentioned previously (Table 1), EXEL-7647 is a RTK inhibitor that inhibits a spectrum of growth-promoting and proangiogenic RTKs, including VEGFRs. Therefore, to examine the effects of EXEL-7647 and erlotinib in a xenograft model that is less dependent on EGFR signaling for tumor growth (28) and expressing high levels of VEGF (26, 27), MDA-MB-231 human breast adenocarcinoma xenograft tumors were established in the mammary fat pad of athymic nude mice. When tumors reached an approximate weight of 100 mg, once-daily oral administration of EXEL-7647 or erlotinib at 30 or 100 mg/kg was initiated and continued for 14 days. Figure 5A shows that EXEL-7647 at 30 and 100 mg/kg significantly inhibited tumor growth by 35% and 81%, respectively, when compared with the vehicle-treated control group (P < 0.005). In contrast, erlotinib did not significantly inhibit tumor growth in this model (Fig. 5B) exhibiting tumor growth inhibition of 19% and 17% in the 30 and 100 mg/kg groups, respectively, which did not reach statistical significance. Immunohistochemical analysis of the tumors at the end of the study (Table 5) showed that, consistent with the inhibitory activity against VEGFRs, there was a significant decrease in tumor vessel density (CD31) in the EXEL-7647–treated groups but not in the erlotinib-treated groups compared with respective vehicle control groups. In addition, EXEL-7647 treatment resulted in a significant decrease in cell proliferation (Ki67; Table 5). These data indicate that EXEL-7647 significantly inhibited tumor growth in a xenograft model that does not exhibit a strong dependency on EGFR signaling and supports the hypothesis that simultaneous targeting of a spectrum of growth-promoting and proangiogenic RTKs translates into broader efficacy than agents that exhibit a more selective RTK inhibition profile.

Targeted therapies have recently attracted significant attention as anticancer agents. The most encouraging clinical results to date have been achieved in malignancies driven by mutationally activated kinases, including chronic myelogenous leukemia (Bcr-Abl), gastrointestinal stromal tumors (c-Kit), and a subset of NSCLC patients with mutationally activated EGFRs (9, 10, 33, 34). However, the currently approved therapies targeting cancers driven by such activating mutations are only impactful for a relatively small fraction of the cancer patient population. It has also become apparent that individuals who initially benefit from treatment with a selective kinase inhibitor (e.g., imatinib) may later become unresponsive due to the emergence of drug resistance mutations in the gene encoding the target kinase (34). The majority of cancer patients are therefore in need of more broadly active anticancer agents that will combine targeted potency against both WT and drug resistance mutations with tolerable side effect profiles. Here, we describe a novel spectrum-selective kinase inhibitor, EXEL-7647, that is a potent inhibitor of both EGFR and VEGFR family members and additionally inhibits EphB4. EXEL-7647 inhibits EGFR signaling in cells and retains significant activity against mutant EGFRs associated with resistance to erlotinib and gefitinib.

Members of the EGF RTK family (EGFR, ErbB2, ErbB3, and ErbB4) are overexpressed in many solid tumors. Several anti-EGFR therapies (gefitinib, erlotinib, and cetuximab) have been approved for the treatment of NSCLC or colorectal cancer, and the anti-ErbB2 antibody, trastuzumab, is approved for the treatment of ErbB2-positive breast cancers. EXEL-7647 inhibits the kinase activities and activation of both EGFR and ErbB2. Although patients who have EGFR kinase domain mutations respond to gefitinib and erlotinib, many develop resistance to these drugs over time. A recent study by Pao et al. (14) has defined the mechanism that underlies the acquired resistance observed in some of these patients. A “second site” mutation in EGFR results in a substitution of a methionine for a conserved threonine at residue 790 (T790M). This change is presumed to inhibit the ability of gefitinib or erlotinib to form a critical hydrogen bond within the ATP-binding pocket of the catalytic region (35). The acquisition of clinical resistance observed following treatment with small-molecule EGFR inhibitors in NSCLC is similar to that observed in chronic myelogenous leukemia and gastrointestinal stromal tumor subjects who relapse following treatment with imatinib (36, 37). The observation that inhibition of drug action can be accomplished by a similar mechanism in three different disease/treatment scenarios emphasizes the need for therapeutics that show efficacy against kinases harboring the gatekeeper mutation.

Our studies using the bronchoalveolar cell line H1975, which harbors the EGFR double mutation (L858R and T790M), provide a means to evaluate the ability of small molecules to inhibit activation of the gatekeeper EGFR mutant. EXEL-7647 inhibits cellular proliferation of H1975 with a IC₅₀ value of 820 nmol/L, which is ∼12 and 15 times more potent than gefitinib and erlotinib, respectively. Additionally, EXEL-7647 inhibits the phosphorylation of the double-mutant EGFR and the subsequent activation of downstream survival and proliferation pathways as measured by inhibition of phosphorylation of AKT and ERK, respectively. When H1975 cells were established as s.c. xenografts in severe combined immunodeficient mice, once-daily oral administration of EXEL-7647 resulted in strong, dose-dependent inhibition of tumor growth. In contrast, erlotinib showed only moderate efficacy. Immunohistochemical analysis of tumors at the end of the dosing period showed that EXEL-7647 inhibited the phosphorylation of EGFR and the downstream effectors AKT and ERK, consistent with inhibition of EGFR T790M in vivo. EXEL-7647 treatment also caused significant reductions in tumor vessel density, which is consistent with inhibition of VEGFRs. Tumors from EXEL-7647–treated animals also showed marked inhibition of cell proliferation and increases in apoptosis when compared with tumors from vehicle-treated animals. In contrast, tumors from erlotinib-treated animals showed no changes in tumor vessel density and modest changes in proliferation, apoptosis, pEGFR, pAKT, and pERK.

In addition to inhibition of EGFR, EXEL-7647 is also a potent inhibitor of VEGFRs; therefore, the effect of EXEL-7647 and erlotinib was examined in the MDA-MB-231 xenograft model that highly expresses VEGF (26, 27) and is less reliant on signaling through EGFR (28). Consistent with inhibition of both tumor growth and proangiogenic RTKs, EXEL-7647 was highly efficacious in this model, whereas erlotinib did not significantly inhibit tumor growth. Immunohistochemical analyses of tumors at the end of the study showed that EXEL-7647 caused a significant decrease in tumor vessel density (CD31) and tumor cell proliferation (Ki67) showing inhibitory effects of EXEL-7647 on both tumor cell growth and the tumor vasculature. In contrast, the selective EGFR inhibitor, erlotinib, had no effect on number of CD31-positive tumor vessels and showed a moderate but significant decrease in tumor cell proliferation. Given that human malignancy is a multifaceted disease, these data show the potential advantage of a small molecule that targets multiple RTKs associated with critical components of tumor pathobiology. This spectrum of potent inhibitory activities translates to increased and broader efficacy when compared with a selective agent.

EXEL-7647 is an ATP competitive and reversible inhibitor of EGFR, ErbB2, EphB4, and KDR. A class of irreversible EGFR inhibitors has been described that also shows in vitro activity against the T790M mutation in cell lines (38, 39). However, because no in vivo studies have been described using these compounds, their effect on tumor growth in animal studies is currently unknown.

In summary, EXEL-7647 is a highly potent spectrum-selective kinase inhibitor of the clinically validated RTK EGFR and importantly retains inhibitory activity against the gatekeeper EGFR mutation that is resistant to EGFR inhibitors that are currently approved. In addition to inhibition of EGFR, EXEL-7647 also potently inhibits two other clinically validated RTKs ErbB2 and KDR as well as the proangiogenic EphB4. By simultaneously targeting multiple signal transduction pathways that are involved in tumor growth and angiogenesis, EXEL-7647 may have broader clinical utility than agents with more restricted specificity profiles.

We thank Dr. Abhijit Guha for generously providing the Tet-On inducible EGFR and EGFRvIII expression constructs, Bristol-Myers Squibb Oncology for Lx-1 cells, and numerous members of the Exelixis Drug Discovery and Translational Medicine Departments for their support.