MERTK Promotes Resistance to Irreversible EGFR Tyrosine Kinase Inhibitors in Non–small Cell Lung Cancers Expressing Wild-type EGFR Family Members

Lung cancer is the leading cause of cancer-related death world-wide, accounting for 26% of U.S. cancer-related deaths (https://seer.cancer.gov/statfacts/html/lungb.html). Non–small cell lung cancer (NSCLC) comprises 85% of lung tumors. Current therapy depends on the tumor characteristics. NSCLCs harboring an activating EGFR kinase domain mutation (mtEGFR) can be treated with EGFR tyrosine kinase inhibitors (TKIs), such as erlotinib and gefitinib; however, disease progression invariably occurs, usually within 9 to 14 months (1). In approximately 60% of cases treated with first-generation EGFR TKIs, resistance is due to a secondary EGFR-T790M mutation (1), which can be targeted with third-generation irreversible EGFR TKIs, including CO-1686 (rociletinib) and osimertinib (1–3). Although clinical development of rociletinib has been halted, the recent phase III FLAURA trial demonstrated superior outcomes in patients treated with osimertinib compared with first-generation EGFR TKIs, irrespective of T790M mutation status (4). Osimertinib was subsequently FDA-approved for first-line application and has become the standard-of-care for mtEGFR NSCLC (5). Although only 10% to 20% of NSCLCs have mtEGFR (6), 40% to 80% express EGFR without an activating mutation (wtEGFR; ref. 7) and recent data indicate a cooperative role for other EGFR family members (ERBB2 and ERBB3; refs. 2, 8). However, wtEGFR NSCLCs are relatively insensitive to EGFR TKIs and patients with tumors that express wtEGFR in the absence of other targetable mutations have limited treatment options. Immunotherapies are effective for some patients, but most are not cured, and toxic chemotherapies remain an important component of treatment (9). Thus, novel therapies are urgently needed.

One potential therapeutic target is MERTK, a member of the TAM family (TYRO3, AXL, and MERTK) receptor tyrosine kinases that is expressed in a variety of malignancies (10), including 70% of NSCLCs (11). Furthermore, MERTK inhibition using shRNA or a MERTK TKI decreased colony formation, irrespective of driver oncogene status, and delayed tumor progression in murine wtEGFR NSCLC models (11, 12). MRX-2843 is a potent ATP-competitive dual MERTK and FLT3 TKI that is predicted to have significant off-target activity (i.e., <8-fold selectivity) against only 3 other kinases (TRKA, AXL, and LOK) in cell-based assays and is currently in clinical trials (13, 14).

Intrinsic and adaptive resistance significantly limit clinical application of molecularly targeted single agents (15) and will likely limit the effectiveness of MERTK-targeted monotherapy. Combination therapies may overcome or prevent resistance and often have enhanced efficacy. Here, we screened >300 TKIs to identify inhibitors that synergize with MRX-2843 against wtEGFR NSCLC cells. Strikingly, the majority of third-generation EGFR TKIs had marked synergistic efficacy, while first-generation EGFR TKIs did not. Combined inhibition of MERTK and EGFR family members blocked redundant downstream pathways and provided robust therapeutic responses in wtEGFR NSCLCs.

Cells were cultured in RPMI 1640 medium with 10% FBS, penicillin (100 U/mL), and streptomycin (100 μg/mL). siRNAs were transfected using Lipofectamine 2000 (Thermo Fisher Scientific). Cell line identities were confirmed by short-tandem repeat microsatellite loci analysis. A549 cells were tested for Mycoplasma before injection in animals and were negative. Cells were used within 3 months of sequence verification and Mycoplasma testing. Free base CO-1686 and CO-1686 hydrobromide were used for cell culture and mouse studies, respectively. MRX-2843 was synthesized as described previously (13, 14). See Supplementary Table S1; Supplementary Table S2 for additional information regarding cell lines and reagents.

Cells were infected with 0.15 PFU/cell Nuclight^TMRed lentivirus (Essen BioScience) and incubated for 48 hours before selection in 1 μg/mL puromycin.

Nuclight^TMRed cells (3,000/96-well) were treated with drugs for 3 to 5 days. Cells were counted and images captured at 2-hour intervals using the Incucyte^TMZOOM system (Essen Bioscience).

Nuclight^TMRed A549 cells were treated with a library (1 μmol/L) of 378 kinase inhibitors (SelleckChem, Z118523) with/without concurrent MRX-2843 and cell expansion was assessed. Controls were treated with DMSO, MRX-2843, or cycloheximide (100 μg/mL). Z′ values were calculated as 1-[(3SDC+3SDB)/(meanC-meanB)], where SD is standard deviation, C is control, and B is background (16). Z scores were calculated as [S_i-median(S_all)]/SD(S_all), where S is the relative cell number for individual (S_i) or all (S_all) compounds on a plate (16).

Cells were treated with MRX-2843 and/or CO-1686 in serum-free DMEM with/without EGF ligand for 2 hours, then lysed and assessed by immunoblot (11). Phosphorylated and total MERTK were detected and quantitated after pervanadate treatment and immunoprecipitation (13). Antibodies are indicated in Supplementary Table S3.

A549 cells (800/12-well) were treated with DMSO, MRX-2843, and/or CO-1686 for 8 days. Colonies were stained with crystal violet (0.2% w/v in 25% methanol) and counted using a GelCount colony counter (Oxford Optronix).

A549 cells (5.5 million/mouse) were injected subcutaneously into the flank of athymic Nude-Foxn1^nu mice in 100 μL PBS with 50% Matrigel (Corning). Tumors were measured twice weekly with calipers and volume = πa²b/6, where “a” is the shortest diameter measured perpendicular to the longest diameter, “b.” When tumors reached 150 to 215 mm³, mice were randomized to groups and treated with vehicle, MRX-2843, CO-1686, osimertinib, or combination (2, 3). Starting tumor volumes were not statistically different between groups within each experiment. Mice with significant tumor ulceration were removed from study. All animal studies were approved by the Emory University Animal Care and Use Committee.

The fractional product method was used to assess interactions between drugs (17). The fractional reduction in cell number expected for an additive interaction was calculated using the equation E_additive = E_A+E_B-E_A × E_B, where E_A and E_B are the fractional reductions for drugs A and B alone. Experimental cell numbers that are significantly less or greater than E_additive define synergistic and antagonistic interactions, respectively.

Statistical significance was determined by one-way or two-way ANOVA with Bonferroni multiple comparison test using Prism v5 (GraphPad Software, Inc.). When <60% of tumor volume measurements were collected all values for that animal were censored. For CO-1686 studies, outliers with tumor volume more than 2 SDs different from the residual mean for that group were also censored. Unless otherwise indicated, cell culture data were derived from ≥3 independent experiments.

We previously reported antitumor effects in NSCLC cell culture and xenograft models in response to MERTK inhibition using shRNA or UNC2025, a MERTK-selective TKI (11, 12). To test whether MRX-2843 can similarly target wtEGFR NSCLC, Nuclight^TMRed-expressing cell lines were generated, then treated with MRX-2843 and cell numbers were monitored using live cell imaging. MRX-2843 had potent cellular activity, with 130–170 nmol/L EC₅₀ values (Fig. 1A). Inhibition of MERTK phosphorylation was dose-dependent in all four cell lines with statistically significant decreases evident at 50 to 100 nmol/L concentrations and phosphorylation nearly abrogated in cultures treated with 300 to 400 nmol/L MRX-2843 (Fig. 1B–E). Similar concentrations were required for inhibition of both MERTK and tumor cell expansion, suggesting MERTK kinase inhibition as a mechanism of reduced cell expansion.

A library of 378 kinase inhibitors was screened to identify MRX-2843 combination therapies that more effectively target wtEGFR NSCLC cells compared with single agents. Nuclight^TMRed–expressing A549 cells were treated with 1 μmol/L library compound alone and combined with 100 nmol/L MRX-2843 and cell numbers were determined at intervals (Supplementary Fig. S1A). A subtherapeutic dose of MRX-2843 was used for robust detection of synergistic interactions. Values were normalized to the median for all treated samples. Z′ values for single agent and combination therapies were 0.83 ± 0.05 and 0.74 ± 0.04, respectively, indicating that the assay is sufficiently robust for drug screening (16). Z scores were calculated to rank the library compounds, with a more negative Z score reflecting more potent inhibition (16) and ranged from −2.17 to 2.85 (Supplementary Fig. S1B). Of 25 EGFR TKIs in the library, 12 had negative Z scores when combined with MRX-2843 (Supplementary Fig. S1C). Of the 10 that were most effective in combination with MRX-2843, 8 were irreversible (Supplementary Table S4) and 9 of 10 irreversible EGFR TKIs had negative Z scores in combination with MRX-2843, compared with only 3 of 15 reversible EGFR TKIs (Supplementary Fig. S1C). CO-1686 was the most potent EGFR TKI in combination with MRX-2843 (Z = −1.53) and osimertinib also inhibited cell expansion when combined with MRX-2843 (Z = −0.77; Supplementary Table S4; Supplementary Fig. S1A–S1C). Erlotinib was not in the library but, like other first-generation EGFR TKIs, did not synergize with MRX-2843 (Supplementary Fig. S1D). In general, combined treatment with MRX-2843 and irreversible EGFR TKIs mediated robust antitumor activity (Supplementary Table S4; Supplementary Fig. S1C), implicating MERTK as a mediator of resistance to EGFR TKIs in wtEGFR NSCLC cells. CO-1686 and osimertinib were chosen for further study based on their optimal potency in combination with MRX-2843 and clinical relevance, respectively.

CO-1686 is not effective against wtEGFR NSCLCs (3). Indeed, treatment with 1 μmol/L CO-1686 did not affect expansion of A549 or H1299 cells compared with vehicle (Fig. 2A). In contrast, expansion of PC9 mtEGFR NSCLC cells was significantly inhibited by CO-1686. Similarly, CO-1686 inhibited EGFR-Y1068 phosphorylation and activation of downstream prosurvival signaling through ERK and AKT in PC9 cells, both with/without prior EGF stimulation, but had little or no impact on signaling in H1299 and A549 cells (Fig. 2B).

In subsequent studies, CO-1686 and MRX-2843 monotherapies mediated dose-dependent inhibition of A549 cell expansion and 100 nmol/L MRX-2843 and 1 μmol/L CO-1686 were identified as the best concentrations for assessment of interactions (Supplementary Fig. S2A). Treatment with CO-1686 and this subtherapeutic dose of MRX-2843 mediated near complete abrogation of tumor cell expansion in A549 cultures (Fig. 2C and D). Osimertinib and MRX-2843 also mediated potent antitumor activity against A549 cells (Fig. 2E). The combination effect was time-dependent. When CO-1686 and MRX-2843 were removed from cultures after 4 hours of combined treatment, A549 cells recovered and continued to expand (Supplementary Fig. S2B); however, expansion only partially recovered after 16 hours and the effect of treatment was irreversible after 24 hours.

Similar antitumor effects were observed in a panel of NSCLC cell lines with different driver mutations, including KRAS (A549, H157, H358, and H2009), NRAS (H1299), FGFR1 (COLO699 and H226) and ALK (H3122; Fig. 2F; Supplementary Fig. S3). Cell numbers were significantly reduced in all eight cell lines treated with CO-1686 and MRX-2843 (Fig. 2F) and in A549 cells treated with osimertinib and MRX-2843 compared with single agents (Fig. 2G). Application of the fractional product method to evaluate interactions between CO-1686 and MRX-2843 (17) revealed significantly reduced numbers of cells in cultures treated with the combination compared with the values expected for additive interactions, indicating synergistic antitumor activity mediated by MRX-2843 and CO-1686 in 7 of 8 wtEGFR cell lines (Fig. 2F). Similarly, treatment with MRX-2843 and osimertinib was significantly more effective than either single agent and the effect of the combination therapy was increased compared with the expected additive value, although the difference was not statistically significant (Fig. 2G). The degree of response to the combination therapy varied between cell lines and was independent of driver oncogene status (Fig. 2F).

To determine the biochemical mechanism by which MRX-2843 and CO-1686 combination therapy inhibits expansion of wtEGFR NSCLC cells, we evaluated components of the MERTK and EGFR family signaling pathways. MERTK, EGFR, and EGFR family members were coexpressed in all evaluated wtEGFR NSCLC cell lines (Fig. 3A and B; Supplementary Fig. S4A). The degree of response to the combination therapy did not correlate with MERTK or EGFR expression levels (Figs. 2F and 3A).

EGFR family members and MERTK mediate downstream signaling through numerous shared pathways, including PI3K-AKT and MAPK-ERK (10, 15). In both A549 and H1299 cells, treatment with MRX-2843 or CO-1686 alone had little or no effect on phosphorylated/activated AKT and ERK (Fig. 3C). Interestingly, both the subtherapeutic dose of MRX-2843 and single-agent CO-1686–mediated partial inhibition of MERTK phosphorylation. CO-1686 also partially inhibited EGFR phosphorylation. In contrast, treatment with MRX-2843 and CO-1686 together resulted in near complete abrogation of phosphorylated MERTK and EGFR and downstream AKT and ERK signaling. Similarly, inhibition of MERTK or EGFR alone using siRNA had no effect on AKT phosphorylation, whereas combined inhibition of MERTK and EGFR inhibited AKT, confirming that the biochemical effects of the combination are mediated by on-target inhibition of MERTK and EGFR (Supplementary Fig. S4B). Thus, synergistic antitumor activity mediated by MRX-2843 and CO-1686 coincides with synergistic inhibition of MERTK and EGFR signaling through AKT and ERK. This cross-talk between MERTK and EGFR may reflect redundancy between downstream signaling pathways and/or could be mediated by physical interactions between receptors.

Coimmunoprecipitation studies were conducted to further investigate the relationship between MERTK and EGFR in NSCLC cells. EGFR immunoprecipitated with MERTK from wtEGFR A549, COLO699, and H1299 cell lysates (Fig. 3D). Similarly, MERTK was detected in EGFR immunoprecipitates. These data suggest a physical interaction between MERTK and EGFR and the existence of this complex may explain the decreased MERTK phosphorylation mediated by CO-1686. However, given that first-generation EGFR TKIs (Supplementary Fig. S1C) and EGFR siRNA (Fig. 3E and F) failed to synergize with MRX-2843, it is likely that additional targets mediate synergy.

The selective synergy mediated by irreversible EGFR TKIs may also provide clues to the mechanism of interaction. EGFR can form heterodimers with ERBB2 and ERBB3 that mediate more robust signaling than EGFR homodimers (18) and, unlike earlier EGFR TKIs, newer-generation EGFR TKIs can target other ERBB family members (2, 8, 19). Indeed, CO-1686 and osimertinib inhibited ERBB2 and ERBB3 phosphorylation in mtEGFR PC9 cells (Supplementary Fig. S4C) and ERBB2 and ERBB3 co-immunoprecipitated with MERTK from wtEGFR cell lines (Supplementary Fig. S4D), suggesting a physical interaction. Moreover, the combination therapy synergistically inhibited ERBB2 and ERBB3 phosphorylation in A549 cells, whereas single agents had minimal effect (Supplementary Fig. S4E). The coincident changes in phosphorylation and enhanced antitumor activity in A549 cells implicate ERBB2 and ERBB3 as potential mediators of combinatorial effects. However, individual antibodies that selectively target EGFR-family members (EGFR/ERBB1, ERBB2, or ERBB3) did not synergize with MRX-2843, suggesting that coincident inhibition of multiple family members is required (Supplementary Fig. S4F–S4H). Inhibition of ERBB3 and AKT phosphorylation in A549 cells treated with anti-ERBB2 or anti-ERBB3 antibodies was confirmed by immunoblot (Supplementary Fig. S4I).

Cells containing more than two distinct nuclei are defined as polyploid (20). NSCLC cells treated with the combination therapy exhibited morphologic changes consistent with induction of polyploidy, including larger nuclear and cell size (Supplementary Fig. S5A; Supplementary Videos; ref. 21). Polyploidy was especially evident in H1299 cells, which became multi-nucleate (Supplementary Videos). AURKA and AURKB are cell cycle-regulated kinases involved in chromosome segregation and cytokinesis and AURK inhibition induces polyploidy (22). Nuclear EGFR is a co-transcription factor for AURKA (23) and phosphorylation of EGFR-Y1101 initiates EGFR nuclear trafficking (24). Here, treatment with MRX-2843 and CO-1686 synergistically reduced phosphorylation of EGFR-Y1101 and AURKs, implicating AURK inhibition as a mechanism of polyploidy, and perhaps other antitumor activities of the combination therapy (Supplementary Fig. S5B).

The frequent coexpression and physical interactions between MERTK and EGFR family members, synergistic abrogation of MERTK and EGFR family activities mediated by MRX-2843 and CO-1686, and coincident synergistic inhibition of downstream signaling through numerous shared oncogenic pathways, including PI3K-AKT and MAPK-ERK (11, 15), suggest that alternative activation of oncogenic pathways downstream of MERTK drives resistance of wtEGFR NSCLC cells to third-generation EGFR TKIs. To test this, A549 cells were treated with CO-1686 and PI3K inhibitor GDC-0941 (PI3Ki), or AKT inhibitor MK-2006 (AKTi) singly or in combination. Biochemical studies confirmed pAKT inhibition in response to treatment with PI3Ki or AKTi (Fig. 4A) and both PI3Ki and AKTi synergistically inhibited tumor cell expansion in combination with CO-1686 (Fig. 4B), implicating alternative activation of PI3K-AKT mediated by MERTK as a mechanism of resistance to EGFR TKIs in wtEGFR NSCLCs. In contrast, the MEK inhibitor SL-327 and mTOR inhibitor ridaforolimus did not synergize with CO-1686 (Supplementary Fig. S6A–S6B). Barasertib, an AURK-selective inhibitor (22), also mediated potent synergistic inhibition of A549 expansion in combination with CO-1686. The degree of AURK inhibition (Fig. 4C) and decreased cell expansion (Fig. 4D) in response to barasertib and CO-1686 was similar to combined treatment with MRX-2843 and CO-1686, even though treatment with barasertib and CO-1686 did not influence MERTK activity (Fig. 4C).

Together these data support a model whereby MERTK and EGFR-family members function redundantly to promote oncogenic signaling, at least partially through PI3K-AKT with additional effect on AURK (Fig. 4E). EGFR TKIs do not block MERTK and therefore do not efficiently inhibit downstream signaling through shared pathways. In contrast, dual inhibition of MERTK and EGFR family members blocks downstream signaling and thereby inhibits tumor cell expansion. Alternatively or in addition, EGFR family members and MERTK may form heterodimers that promote activation of downstream signaling and may not be effectively targeted by single inhibitors (Fig. 4F).

To determine the longer-term functional effects of treatment with CO-1686 and MRX-2843, A549 cells were cultured at low density with MRX-2843, CO-1686, or both and colonies were stained and counted after 8 days. Treatment with single agents had no significant effect, but combined treatment significantly and synergistically reduced colony number (Fig. 5A and B). Although treatment with a higher dose of 1 μmol/L CO-1686 mediated a 30% reduction in colony formation, combined treatment with MRX-2843 was significantly more effective and mediated a synergistic 86% decrease in colony formation (Fig. 5B). To assess the combination therapy in animal models, A549 xenografts were established in nude mice and then treated twice daily with 20 mg/kg MRX-2843, 30 mg/kg CO-1686, combination therapy, or vehicle (3, 14). Tumor growth was significantly reduced in mice treated with the combination relative to mice treated with vehicle or single agents (Fig. 5C and D). After 48 days of treatment, tumor volumes were not significantly different in mice treated with MRX-2843 (581.3 ± 48.4 mm³), CO-1686 (517.6 ± 46.2 mm³), or vehicle (589.3 ± 35.5 mm³), but tumor volume was reduced by 39.1% (358.8 ± 40.1 mm³) in mice treated with the combination therapy compared with vehicle (Fig. 5D). Moreover, this response was durable and mean tumor volume was not significantly different in mice treated with MRX-2843 and CO-1686 29 days after treatment was stopped (387.2 ± 44.4 mm³) compared with end of treatment (364.9 ± 42.3 mm³; Fig. 5C and E). Similar results were obtained in a smaller independent experiment. Although this study was not powered for robust statistical analysis, mean tumor volume was again significantly reduced at the end of treatment in mice that received 20 mg/kg MRX-2843 and 30 mg/kg CO-1686 (282.2 ± 91.3 mm³) compared with vehicle-treated mice (944.4 ± 328.5 mm³; Supplementary Fig. S7A–S7B). Increasing the MRX-2843 dose to 30 mg/kg provided no obvious additional benefit in combination with CO-1686 (Supplementary Fig. S7C–S7D).

Therapeutic effects were also assessed in mice with A549 xenografts treated with 20 mg/kg MRX-2843 twice daily, 10 mg/kg osimertinib once daily, combination therapy, or vehicle. Tumor volume was significantly reduced in mice treated with MRX-2843 and osimertinib relative to mice treated with vehicle or MRX-2843 alone (Fig. 6A). Although there was a trend toward decreased tumor volume in mice treated with 10 mg/kg osimertinib alone and a further decrease in mice treated with the combination therapy, these differences were not statistically significant. After treatment with osimertinib or the combination therapy was stopped, tumors continued to grow and mean tumor volumes were significantly greater 28 days later compared with end of treatment tumor volumes (1401.1 ± 197.3 mm³ vs. 658.0 ± 143.6 mm³ and 650.3 ± 115.7 mm³ vs. 407.6 ± 83.1 mm³, respectively). There was also a trend toward decreased tumor growth after end of treatment with the combination therapy compared with osimertinib alone (Fig. 6A). In a subsequent study, the dose of osimertinib was increased to 25 mg/kg once daily to improve therapeutic response. After 43 days of treatment, mean tumor volumes were significantly reduced in mice treated with the higher dose of osimertinib alone (449.9 ± 64.2 mm³) or with MRX-2843 and osimertinib combination therapy (297.4 ± 30.0 mm³) relative to mice treated with vehicle (972.8 ± 163.0 mm³) or MRX-2843 alone (727.9 ± 90.9 mm³; Fig. 6B and C). Again, there was a trend toward decreased tumor volume in mice treated with the combination compared with mice treated with 25 mg/kg osimertinib alone, but the difference was not statistically significant. However, in contrast with the lower dose of osimertinib, mean tumor volume was not significantly increased in mice treated with the combination after treatment was stopped for 84 days (284.1 ± 103.0 mm³) compared with end of treatment (223.4 ± 49.8 mm³; Fig. 6B and D). One of the five evaluable tumors disappeared and three additional tumors remained stable after treatment was stopped. To better understand how this durable response is mediated, tumors were collected from mice treated with MRX-2843 and osimertinib at end of study (Fig. 6D) or from freshly established untreated tumors (15 days) and sections were stained with hematoxylin and eosin and anti–KI-67 antibody to detect proliferating (KI-67⁺) tumor cells. Tumors treated with vehicle retained a high proliferative index, with dense tumor cellularity and KI-67 staining evident in the vast majority of cells (Fig. 6E). In contrast, tumors treated with the combination therapy had low cellularity with little KI-67 staining. Interestingly, one tumor (N351) had a higher incidence of proliferating cells and progressed upon removal of treatment, whereas the remaining tumors remained stable. Thus, although tumors did not regress in response to the combination therapy, in most cases the residual tumor consisted primarily of nonproliferative fibrous tissue.

Our unbiased screen with a potent MERTK TKI uncovered a preponderance of third-generation EGFR TKIs that synergistically inhibited expansion of NSCLC lines driven by different driver oncogenes (KRAS, NRAS, FGFR, and ALK), suggesting that these different drivers rely on survival mechanisms provided by common pathways downstream of MERTK and EGFR family members. MRX-2843 is selective for MERTK and FLT3 (13), which is not expressed in NSCLC (12), making MERTK the major target for MRX-2843 in this context. Here the use of low concentrations (50 nmol/L) further supports MERTK-selectivity. Analysis of downstream signaling revealed dual inhibition of MERTK and EGFR family kinases leading to decreased activation of AKT and AURK. The data from the screen implicating numerous third-generation, but few first-generation EGFR TKIs suggest that the combination effects are not solely mediated by EGFR inhibition. Indeed, EGFR knockdown with siRNA did not recapitulate the profound synergy, suggesting that “off-target” kinase inhibition likely plays a role in the synergistic interactions between EGFR TKIs and MRX-2843. The large number of third-generation EGFR TKIs with distinct off-target profiles that interact synergistically suggest that the relevant off-target mediators are closely related proteins, such as other members of the ERBB family. Third-generation TKIs target both ERBB2 and ERBB3, which both regulate the PI3K pathway (2, 19). However, inhibitory ERBB2 or ERBB3 antibodies did not synergize with MRX-2843, suggesting that inhibition of multiple EGFR family members is required for synergy. Similarly, EGFR expression alone was not a significant predictor of 5-year survival rate in a study of 119 patients with NSCLC, while patients with both ERBB2 and EGFR expression had a poorer prognosis (25). In addition, in at least the cells tested we were able to immunoprecipitate MERTK complexes that contained EGFR, ERBB2, and ERBB3. Whether these oligomers containing multiple kinases were true activating heterodimers remains to be determined; however, the combination of MRX-2843 and CO-1686 synergistically reduced EGFR phosphorylation implying heterodimer activation. Taken together, the evidence suggests that combined MERTK and EGFR family inhibition undermines the viability of NSCLC lines with multiple different drivers, presumably by abrogating survival signals through AKT and modulating AURK activity to disrupt cell-cycle progression.

What are the potential clinical implications of these findings? Although 62% of NSCLCs express wtEGFR and the first-generation EGFR TKI erlotinib is approved for patients with relapsed and refractory wtEGFR NSCLCs (26), treatment with EGFR TKIs has not been therapeutically effective in this setting for most patients (27). Immunotherapy has recently emerged as an effective option for some patients (28), but nonselective cytotoxic chemotherapies are still a therapeutic mainstay and many tumors are irresponsive to immunotherapy and/or chemotherapies. Thus, new, less toxic and more effective approaches are needed. Molecularly targeted agents such as EGFR TKIs have also been limited by development of resistance, including secondary EGFR-T790M mutation (1). Although third-generation EGFR TKIs such as osimertinib effectively target EGFR-T790M, conversion back to wild-type has emerged as a dominant source of resistance (29, 30). Thus, novel therapies are needed to target compensatory pathways to overcome and/or prevent primary and acquired resistance in mtEGFR NSCLCs. Although not the focus of this article, combined treatment with MRX-2843 and third-generation EGFR TKIs may also be effective in this context.

Previous data implicated TAM family kinases in resistance to EGFR TKIs. AXL upregulation induced resistance to EGFR TKIs erlotinib and lapatinib and the anti-EGFR mAb cetuximab and reactivated the AKT, ERK, and NFκB pathways in breast, lung, and head and neck cancers (31). Treatment with an AXL TKI or an AXL-neutralizing antibody overcame resistance and resensitized tumor cells to EGFR TKIs. Although overexpression of MERTK was not correlated with overall survival in patients with NSCLC, ectopic overexpression of MERTK in mtEGFR PC9 NSCLC cells attenuated sensitivity to erlotinib by decreasing induction of apoptosis and treatment with UNC569, an early-generation MERTK TKI, resensitized the cells, implicating a similar role for MERTK (32). However, the MERTK inhibitor used for these early studies was unsuitable for translational and clinical applications (33) and analysis was restricted to one cell line. In contrast to mtEGFR NSCLCs, wtEGFR NSCLC cells are inherently insensitive to EGFR TKIs. Our previous data implicate MERTK as a potential therapeutic target in NSCLC and reveal roles for MERTK in chemoresistance (11). KRAS mutation has also been correlated with resistance to EGFR TKIs and chemotherapy in NSCLC (34, 35). Both MERTK shRNA and a MERTK TKI mediated potent antitumor activity against A549 cells, which express both wtEGFR and mutated KRAS (11, 12).

Here, we used A549 cells to identify novel combination therapies targeting therapy-refractory NSCLCs and demonstrated synergistic antitumor activity mediated by MRX-2843 and irreversible EGFR TKIs such as CO-1686 and osimertinib. In a panel of wtEGFR NSCLC cell lines, treatment with MRX-2843 and CO-1686 was significantly more effective than either single-agent therapy. Furthermore, the combination mediated synergistic inhibition of colony formation in vitro and reduced tumor growth in vivo. Even one month after treatment was stopped, tumor volume remained stable. MRX-2843 also synergized with osimertinib in A549 cell culture and xenograft models and in this case, tumors treated with the combination did not grow even 84 days after treatment. These durable responses in the absence of continued treatment in a majority of mice are particularly remarkable and implicate combined treatment with MRX-2843 and an irreversible EGFR TKI as a potential therapeutic option for patients with relapsed or refractory wtEGFR NSCLC.

Biochemical analyses revealed the importance of inhibiting both MERTK and EGFR family receptors to provide robust inhibition of downstream signaling through AKT and AURK. In wtEGFR NSCLCs, MERTK and EGFR are coexpressed and interact. While treatment with CO-1686 (3) or a low dose of MRX-2843 alone had minimal impact on phosphorylation of wtEGFR and cell expansion, treatment with CO-1686 and MRX-2843 resulted in near complete inhibition of EGFR phosphorylation and tumor cell expansion. Interestingly, treatment with CO-1686 also sensitized cells to MRX-2843, such that MERTK phosphorylation was more effectively inhibited. Concurrent treatment with MRX-2843 and CO-1686 mediated more robust inhibition of downstream antiapoptotic and survival signaling through critical oncogenic pathways, such as PI3K-AKT, MAPK-ERK, and AURK, consistent with the known roles for these pathways downstream of EGFR family members and MERTK (15, 23, 36). Together, these data implicate MERTK as a mediator of resistance to EGFR TKIs in wtEGFR NSCLCs. In this context, MERTK sustains activation of key downstream pathways, and provides a mechanism for tumor cells to bypass the requirement for EGFR, such that combined inhibition of both EGFR and MERTK is necessary to effectively block downstream signaling and tumor cell expansion.

Previous studies demonstrated several biological mechanisms of acquired resistance that converge on activation of key downstream pathways. Reactivation of the PI3K pathway via activating mutations to bypass receptor tyrosine kinases is a common mechanism of acquired resistance and treatment with a PI3K inhibitor sensitized tumors to gefitinib (37). Here, treatment with a PI3K or AKT inhibitor synergized with CO-1686 to inhibit A549 cell expansion. The serine-threonine kinase mTOR is downstream of PI3K and is a critical mediator of PI3K oncogenic activity in tumor cells (38). In a genetically engineered lung tumor model driven by EGFR-T790M/L858R mutations, treatment with an mTOR inhibitor sensitized tumors to an irreversible EGFR TKI (39). However, response rates were lower than expected in phase Ib and II studies evaluating combined treatment with afatinib or gefitinib and an mTOR inhibitor in patients with advanced NSCLC (38, 40). In our studies, ridaforolimus, an mTOR inhibitor, and CO-1686 did not mediate synergistic antitumor activity, indicating that mTOR is not a critical downstream mediator of MERTK or PI3K activity in the context of MERTK- and EGFR-targeted combination therapy.

Treatment with an AURK inhibitor also provided synergistic antitumor effects. AURKA and AURKB regulate mitosis (41). AURKA inhibition delayed mitotic entry and progression, leading to accumulation of cells in G₂–M phase (42) and AURKB inhibition prevented proper alignment of multinucleated cells (41). Here, AURKA and AURKB were synergistically inhibited in response to treatment with MRX-2843 and CO-1686, resulting in formation of multinucleated/polyploid wtEGFR NSCLC cells, consistent with previous studies demonstrating induction of polyploidy in leukemia and glioblastoma cells treated with MERTK-selective TKIs (43) and enhanced therapeutic efficacy in response to combined inhibition of AURKA and EGFR (44). Coexpression of EGFR and AURKA was also associated with decreased disease-free and overall survival in patients with head and neck squamous cell carcinoma (44). Although MRX-2843 combination therapy clearly dampened AURK phosphorylation, the mechanism is not clear. FOX subclass M1 (FOXM1), an oncogenic transcription factor involved in drug resistance, is regulated downstream of PI3K-AKT and MAPK-ERK and activates AURKA expression. In turn, AURKA functions as a cofactor allowing kinase-independent transactivation of FOXM1 target genes (45).

Taken together, the evidence suggests roles for MERTK as a mediator of resistance to EGFR TKIs in wtEGFR NSCLC cells and points to combined inhibition of MERTK and EGFR-family members as an effective means to undermine the growth of NSCLCs driven by different oncogenes, in part by abrogating AKT and AURK activity. Additional studies are necessary to understand the complex signaling networks impacted by coinhibition of MERTK and EGFR. In conclusion, the findings reported here reinforce development of therapeutic strategies aimed at targets that mediate cross-talk between key oncogenic pathways, such as EGFR and MERTK. These studies identify a novel therapeutic strategy for cancers that are often refractory to existing therapies and support continued development of MERTK- and EGFR-targeted combination therapies for treatment of wtEGFR NSCLCs.

X. Wang holds ownership interest (including patents) in Meryx. S.V. Frye is an employee of and holds ownership interest (including patents) in Meryx. H.S. Earp III is an employee of, holds ownership interest (including patents) in, and is a consultant/advisory board member for Meryx. D. DeRyckere holds ownership interest (including patents) in Meryx. D.K. Graham holds ownership interest (including patents) in Meryx. No potential conflicts of interest were disclosed by the other authors.

Conception and design: D. Yan, X. Wang, H.S. Earp III, D. DeRyckere, D.K. Graham

Development of methodology: D. Yan, H.S. Earp III, D. DeRyckere, D.K. Graham

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D. Yan, R.E. Parker, X. Wang, S.V. Frye, D. DeRyckere, D.K. Graham

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D. Yan, R.E. Parker, D. DeRyckere, D.K. Graham

Writing, review, and/or revision of the manuscript: D. Yan, X. Wang, S.V. Frye, H.S. Earp III, D. DeRyckere, D.K. Graham

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D. Yan, R.E. Parker, X. Wang, D.K. Graham

Study supervision: D. Yan, D. DeRyckere, D.K. Graham

Other (contributed to design and discovery of MerTK inhibitors used): S.V. Frye

Development of MRX-2843 was supported by federal funds from the National Cancer Institute Experimental Therapeutics (NExT) Program, NIH (contract no. HHSN261200800001E).

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