Discovery of a mutant-selective covalent inhibitor of EGFR that overcomes T790M-mediated resistance in NSCLC

Abstract

Patients with non-small cell lung cancer (NSCLC) with activating EGF receptor (EGFR) mutations initially respond to first-generation reversible EGFR tyrosine kinase inhibitors. However, clinical efficacy is limited by acquired resistance, frequently driven by the EGFR(T790M) mutation. CO-1686 is a novel, irreversible, and orally delivered kinase inhibitor that specifically targets the mutant forms of EGFR, including T790M, while exhibiting minimal activity toward the wild-type (WT) receptor. Oral administration of CO-1686 as single agent induces tumor regression in EGFR-mutated NSCLC tumor xenograft and transgenic models. Minimal activity of CO-1686 against the WT EGFR receptor was observed. In NSCLC cells with acquired resistance to CO-1686 in vitro, there was no evidence of additional mutations or amplification of the EGFR gene, but resistant cells exhibited signs of epithelial-mesenchymal transition and demonstrated increased sensitivity to AKT inhibitors. These results suggest that CO-1686 may offer a novel therapeutic option for patients with mutant EGFR NSCLC.

Significance: We report the preclinical development of a novel covalent inhibitor, CO-1686, that irreversibly and selectively inhibits mutant EGFR, in particular the T790M drug-resistance mutation, in NSCLC models. CO-1686 is the fi rst drug of its class in clinical development for the treatment of T790M-positive NSCLC, potentially offering potent inhibition of mutant EGFR while avoiding the on-target toxicity observed with inhibition of the WT EGFR.

Figure 1

(A) Chemical structure of CO-1686. Reactive acrylamide group is highlighted by dashed circle. (B) Structural modeling of CO-1686 binding to T790M EGFR. The EGFR T790M kinase is shown in a ribbon representation (green) with the bound CO-1686 in orange. The aminopyrimidine binds to the hinge residues Met793 through hydrogen bonding (yellow dashed lines). The C5-CF3 substitution points to the gatekeeper residue Met790. Both C2 and C4 substitutions adapt a U-shaped binding mode. The piperazine ring is facing an open space in the solvent exposure area. The meta acrylamide points to Cys797 and forms the covalent bond.

Figure 2

(A) CO-1686 inhibits viability in EGFR-mutant cell lines in vitro. The CO-1686 GI₅₀ was determined in cells harboring mutant or WT EGFR. Cell viability was analyzed using the CellTiter-Glo viability assay. GI₅₀ values shown on Log10 scale in nM as an average of 3 or more independent experiments ± SEM. (B) CO-1686 induces apoptosis in parental HCC827 (EGFR del19) as well as erlotinib-resistant HCC827-EPR cells (EGFR del19/T790M). Levels of cleaved PARP (CI-PARP), Bim-EL, phosphorylated EGFR (pEGFR), phosphorylated MAPK (pMAPK) and phosphorylated ribosomal protein S6 (pS6RP) were determined after 72 hours treatment with CO-1686 or erlotinib in HCC827 cells expressing EGFRdel19 +/− T790M.

Figure 3

In vivo antitumor efficacy of CO-1686 in lung (A) NCI-H1975, (B) LUM1868, (C) HCC827 and (D) squamous epidermoid A431 xenograft models. CO-1686 was administered orally (PO), daily (QD) or twice daily (BID) at concentrations ranging from 3, 10, 30 and 100 mg/kg/day. N=10 animals/gp. Data plotted as mean ± SEM.

Figure 4

CO-1686 does not inhibit WT EGFR signaling in vivo and is active in EGFR-mutant transgenic mouse lung cancer models. (A, B) EGFR signaling was examined in A431 (WT EGFR) tumors following CO-1686 (50 mg/kg BID, PO), erlotinib (75 mg/kg QD, PO) or afatinib (20 mg/kg QD, IP) administration (N=4 animals/gp). Established (80–120 mm³) A431 tumors following five days of drug administration were harvested at 4 hours post-last dose for tumor WT EGFR signaling analysis. (A) Western blot of tumor lysates analyzed for phosphorylated EGFR (pEGFR) and downstream signaling analysis. (B) Quantification of pEGFR levels in A431 tumors. pEGFR levels from (A) were quantified by densitometry and plotted relative to the vehicle treated group (vehicle = 100%). Data plotted as mean ± SEM. * indicates P < 0.05 and ** P < 0.005 comparing vehicle to treated groups. (C) WT EGFR signaling in mouse skin is not affected by CO-1686. Representative immunohistochemical pMAPK staining in normal skin from mice administered with vehicle, CO-1686 (100 mg/kg QD, PO), afatinib (20 mg/kg QD, IP) or erlotinib (100 mg/kg QD, PO). Arrow indicates pMAPK positive nuclear staining in vehicle and CO-1686 groups (20X magnification). (D and E) In vivo antitumor activity of CO-1686 in EGFR^L858R and EGFR^L858R/T790M GEM models. Representative lung MRI images of mice treated with the indicated compound are shown (* in MRI image indicates a region of tumor formation). The bar graph shows lung tumor volume measurements (mean ± SEM) at baseline and following 21-days of dosing with the indicated compounds. * indicates P < 0.05 and ** indicates P < 0.001 comparing vehicle and indicated groups 21-days post-dosing.

Figure 5

(A, B) CO-1686 resistant NCI-H1975 cell clones, COR1-1 and COR10-1 display a reduced dependence on EGFR signaling for survival. (A) Effect of EGFR siRNA transfection on EGFR protein expression in CO-1686 resistant clones (COR 1-1 and COR 10-1) and NCI-H1975 parental as determined Western blotting. (B) Impact of EGFR siRNA knock-down on cell viability in NCI-H1975, COR1-1 and COR10-1 clones. Cell viability was determined 72-hours post-siRNA transfection and plotted relative to the siRNA control treated group (representing 100%). Data plotted as mean ± SEM. * indicates P < 0.05 and ** P < 0.005 comparing siCTRL and siEGFR groups. (C) qRT-PCR analysis of EMT-related and EGFR-related genes in COR1-1 and COR10-1 cell clones. Data plotted as fold change relative to parental NCI-H1975 cells ± SEM. All data shown P < 0.05 comparing COR clone to parental. (D) Inhibition of AKT restores partial sensitivity of COR cell clones to CO-1686. COR10-1 was exposed to CO-1686, the AKT-inhibitors MK-2206 (left panel) or GDC-0068 (right panel), or an equimolar combination of AKT inhibitor + CO-1686. 72 hours post-drug addition cell viability was determined by CellTiterGlo. Data plotted as % viability relative to DMSO (no drug) control.