Synergy of WEE1 and mTOR Inhibition in Mutant KRAS-Driven Lung Cancers

Lung cancer is the leading cause of cancer-related deaths worldwide and non–small cell lung cancer (NSCLC) accounts for more than 80% of those cases (1, 2). Kirsten RAS viral oncogene homolog (KRAS) is mutated in 25% to 30% of lung adenocarcinomas (3). Unlike NSCLC patients harboring EGFR-activating mutations or ALK fusions, for which targeted inhibitors have achieved objective responses in up to 80% of cases, direct molecular inhibition of mutant RAS has proved difficult owing to its structure as well as picomolar affinity to GDP/GTP (3–6). Patients with tumors harboring KRAS-mutations are among the most difficult to treat and individual inhibitors targeting mutant RAS downstream signaling pathways such as PI3K/AKT/mTOR and RAF/MEK/ERK have yielded limited response rates of less than 20% in clinical trials (7). This suggests that RAS is associated with crosstalk of highly complex and redundant signaling cascades leading to bypass pathways and negative feedback loops (8). Complicating this, we previously showed that KRAS-mutant lung tumors bearing TP53 or LKB1 co-mutations differ in their response to docetaxel with or without selumetinib, suggesting that co-mutations may also impact treatment response (9, 10).

Given greater molecular diversity in KRAS-mutant tumors compared with other actionable oncogenic targets in lung adenocarcinoma, one approach to improving the clinical efficacy of inhibitors is to identify drug combinations that either target multiple RAS-driven pathways or circumvent resistance. Previously, Weisberg and colleagues (11) used a chemical screen in a mutant NRAS-transformed Ba/F3 cell line to show that an inhibitor of WEE1, AZD1775, synergized with the mTOR inhibitor Torin2 in acute leukemia. mTOR is a critical downstream effector of RAS in lung cancer; however, clinical testing of mTOR inhibitors alone have demonstrated limited efficacy (12). Preclinical studies have shown that mTOR inhibitors suppress homologous recombination-mediated DNA repair (HDR) and synergize with PARP inhibitors in BRCA1-proficient triple-negative breast cancers (13). AZD2014 is a novel small-molecule ATP-competitive dual inhibitor of both mTORC1 and mTORC2 kinase that is well tolerated in preclinical studies and is currently in phase II clinical trials (14, 15).

WEE1 is a protein kinase that negatively regulates the G₂–M checkpoint by inhibiting cyclin-dependent kinase (CDK) 1 via tyrosine 15 phosphorylation (16). AZD1775 is a first-in-class, pyrazolo-pyrimidine derivative and potent small-molecule WEE1 kinase inhibitor (17). CDK1 is also necessary for BRCA1-mediated S phase checkpoint activation and HDR (18). Although previous studies have shown that WEE1 inhibitors such as AZD1775 augment the effects of chemotherapy by driving transformed cells to mitotic catastrophe, it is unclear how mTOR inhibition potentiates the effects of AZD1775 (19).

Here, we investigated the mechanism underlying KRAS-mutant cell sensitivity to dual WEE1 and mTOR inhibition. We show that WEE1 inhibition in KRAS-mutant cells induces DNA damage accumulation, which is further compounded by simultaneous compromise of DNA repair, mediated by mTOR inhibition. Importantly, oral administration of mTOR and WEE1 inhibitors caused significant reduction in human xenografted tumor growth as well as tumor regression in autochthonous lung adenocarcinoma murine models. Our results support further clinical investigation of combined WEE1/mTOR inhibition as a potential KRAS-driven NSCLC therapy.

Human NSCLC cell lines (A427, NCI-H23, NCI-H1355, NCI-H441, NCI-H2009, NCI-H358, Calu1, NCI-H460, HCC827, HCC4006, NCI-H1975, NCI-H3255, and NCI-H1650) were obtained from the ATCC and cultured in RPMI-1650 media supplemented with 10% FBS (Thermo Fisher) and antibiotics. Patient-derived xenograft (PDX) cell lines (PDX239, 267, 277, and 462) were previously derived using protocols approved by the University Health Network Human Research Ethics and Animal Care Committee (20). KRAS mutation status for PDX cell lines were detected using the Sequenom OncoCarta panel v1.0 as previously described (20, 21). Human embryonic kidney 293T (HEK293T) cells were cultured in DMEM media supplemented with 10% FBS and antibiotics. All cells were cultivated at 37°C and 5% CO₂. Authentication of ATCC human cell lines and PDX cells were done by short tandem repeat DNA profiling analysis.

LentiCRISPRv2 vector containing SpCas9 (Addgene, #52961) was digested with BsmB1 and ligated with annealed oligos. Guide RNAs (gRNA) against human CCND1 were designed using CHOPCHOP (https://chopchop.rc.fas.harvard.edu; Supplementary Table S1). Retroviral constructs containing human full-length LKB1 (#8592) and kinase-dead LKB1 (K78I; #8593) were obtained from Addgene and sequence verified. Transient transfections and virus preparation in HEK293T cells were performed using Fugene reagents (Promega) as per the manufacturer's protocol. Retro- and lentivirus were prepared by transfecting two packaging plasmids into 293T cells using protocols from The RNAi Consortium (TRC; Broad Institute, Cambridge, MA; ref. 22). Stable cell lines were isolated following viral transduction and selection with puromycin antibiotics (1–2 μg/mL).

Nude mice were obtained from Charles River Laboratories (Wilmington, MA). All manipulations were performed under sterile conditions in a laminar flow hood, in accordance with procedures approved by the DFCI Animal Care and Use Committee. A427 (5 × 10⁶) and H1355 (5 × 10⁶) cells were injected subcutaneously in the flanks of 6-week-old nude mice (n = 8–10/group). For treatment, mice were randomized into groups with similar mean tumor volumes of 100 to 150 mm^3. Treatment began at Day 11 after implantation for A427 lines and Day 32 for H1355 lines. AZD2014 and AZD1775 was dissolved in 0.5% hydroxypropyl methylcellulose (HPMC) and administered by oral gavage once a day at 15 mg/kg (AZD2014) and 40 mg/kg (AZD1775).

Mice were examined every 2 to 3 days, and tumor length and width were measured using calipers. Tumor volume was calculated using the following formula: (length × width²)π/6. At sacrifice, portions of tumors were snap-frozen and stored in liquid nitrogen or were fixed in 10% buffered formalin for routine histopathologic processing.

Mouse strains harboring a conditional activating mutation (G12D) at the endogenous KRAS locus were induced intranasally with 5 × 10⁷ p.f.u. adeno-Cre recombinase (University of Iowa adenoviral core). All experimental mice were maintained on a mixed genetic background (C57BL/6, BALB/c, and S129). Upon detection of tumor burden at approximately 14 to 16 weeks after adeno-Cre recombinase induction by analyzing MRI scans using 3D Slicer software, animals were randomly assigned to treatment groups (23). AZD2014 and AZD1775 was dissolved in 0.5% HPMC and administered by oral gavage once a day at 10 mg/kg (AZD2014) and 20 mg/kg (AZD1775).

Whole cell extracts were lysed with lysis buffer [10 mmol/L Tris [pH 8.0], 1% NP-40, 2 mmol/L EDTA, 150 mmol/L, 0.1 mmol/L Na₃VO₄ and protease inhibitors (Roche)], resolved by SDS-PAGE and transferred to polyvinyliden fluoride membranes. Primary antibodies included: p-70S6K (Cell Signaling Technology, #9205), p70S6 (#2708), pCDK1^Y15 (#4539), CCND1 (#2978), cPARP (#9541), LKB1 (#3080), phospho-histone H2AX (Ser 139; 20E3, #9718), ß-actin (Sigma, clone AC-15), CDK1 (Santa Cruz Biotechnology sc-54). After blocking, membranes were incubated with relevant antibodies and probed with corresponding HRP-conjugated secondary antibodies (Cell Signaling Technology). All blots were developed on Amersham Imager 600 (GE, Pittsburg, PA).

Tumors were fixed with 10% buffered formalin overnight and embedded in paraffin (FFPE). FFPE were cut at 4-μm thickness, dried in a 60°C oven overnight and stained with the following antibodies: TUNEL (Millipore, #S7100) and γH2AX (#9718). Ten fields per tumor section were quantified for TUNEL or γH2AX-positive cell staining with a minimum sample size of 3 to 8 animals per cohort.

Cells were fixed with 4% paraformaldehyde in PBS (10 minutes), permeabilized with 0.1% Triton X-100 in PBS (10 minutes), and incubated with 0.05% SDS diluted in PBS for 5 minutes. After blocking in green antibody dilution buffer (Ventana Medical Systems), cells were incubated overnight at 4° with followed by secondary antibodies conjugated to Alexa Fluor-488 or Alexa Fluor-647 (Invitrogen). The Yokogawa spinning disk confocal microscope (Zeiss USA) was used to image the fluorescently stained slides. We acquired a range of 10 to 20 fields per treatment using Andor iQ software (Andor USA) at high-power oil ×100 objective.

Cell proliferation/viability was evaluated by the tetrazolium dye (MTS) assay (Promega). Each cell line was plated at a seeding density to give logarithmic growth over the course of the assay in a 96-well tissue culture plate. The combination indices were calculated by CalcuSyn software using the Chou–Talalay method (Biosoft). For colony formation assays, 0.5 × 10³, 1 × 10³, and 2 × 10³ cells were seeded on 6-well dishes and after 4 to 6 weeks, observed colonies were stained with crystal violet and were enumerated using ImageJ software (NIH, Bethesda). ImageJ filters scored colonies that were 50 to 100 μm (A427) or ≥100 μm (H1355) in size.

Caspase 3 and 7 activities were performed using Caspase-Glo 3/7 assay (Promega) according to the manufacturer's protocol. Live time-course imaging of cells was performed and measured using the IncuCyte analysis system (Essen BioSciences).

All numerical data are presented as mean ± SEM. Statistical significance was determined using one-way ANOVA with post hoc testing for comparisons of groups, including colony formation, confocal, and IHC experiments. Difference in tumor growth rates and in vitro growth dose curves were assessed by mixed-model two-way ANOVA. Tests that produced P ≤ 0.05 were considered to be significant. All statistical analyses were performed with GraphPad Prism 7.0.

We first evaluated the ability of dual inhibition of mTOR and WEE1 to inhibit cell proliferation of a panel of KRAS-mutant (A427, H1355, H23, H441, H2009, H358, Calu1, and H460) and KRAS-wild type (HCC827, HCC4006, H1975, H3255, and H1650) NSCLC cell lines (Fig. 1A–C; Supplementary Tables S2 and S3 and Supplementary Fig. S1). Because PDXs closely mimic the molecular characteristics and heterogeneity of the original patient tumors, we further tested four cell lines cultured from early passage PDXs (PDX239, PDX277, PDX462, and PDX267; ref. 24). Nine of the 11 cell lines achieved combination indices <1.0, consistent with a synergistic response at nanomolar drug concentrations (Fig. 1B; Supplementary Table S4). In contrast, no synergistic effects (CI > 1.0) were observed in any of the KRAS-wild type NSCLC cell lines treated with the drug combination, but rather single-agent activity (Fig. 1C).

Further assessment of the colony growth ability of H1355 and A427 cells under long-term treatment demonstrated significantly reduced colony formation in response to dual inhibition treatments compared with monotherapy or vehicle controls (Fig. 1D–F). Of note, we did not observe significant emergence of resistant clones after 6-weeks of chronic treatment. Consistent with our cell viability assays, the combination of AZD1775 and AZD2014 led to a significant increase in apoptotic caspase 3/7 activity compared with either agent alone, confirming the observed synergistic anti-tumor effects (Fig. 1G and Supplementary Fig. S2).

Given the biological implications of our in vitro data, we next tested the anti-tumor activity of AZD1775 and AZD2014 in human tumor xenograft mouse models. Daily oral administration of combined AZD2014 and AZD1775 treatment at clinically relevant doses significantly inhibited the rate of A427 xenograft tumor growth without toxicity (P = 0.0066; Fig. 2A and Supplementary Fig. S3). At time of sacrifice, tumors treated with the combination weighed nearly 60% less than the vehicle-treated group (Fig. 2B). Combination treatment for 21 days suppressed tumor progression (baseline 130 ± 15; post-treatment 132 ± 12 mm³), whereas the vehicle-treated tumors progressed from 140 ± 22 to 330 ± 115 mm³ (Fig. 2A). We also observed significant growth suppression upon chronic administration of the combination therapy for 33 days in mice bearing H1355 xenografts (Fig. 2C; P < 0.0001). At endpoint, tumors treated with the combination weighed significantly lower on average than vehicle-treated tumors (P = 0.0087; Fig. 2D). For both models, single treatment arms showed no significant tumor reduction, suggesting lack of efficacy for the monotherapies against human KRAS-driven tumors (Fig. 2B and D).

To evaluate the ability of combined therapy to induce apoptosis and DNA damage in vivo, cells staining positively for the DNA fragmentation marker TUNEL (Terminal deoxynucleotidyl transferase dUTP nick end labeling) and the DNA damage marker γH2AX were quantified in A427 tumors 24 hours after combined treatment and compared to vehicle controls (Fig. 2E–G). We detected significantly increased numbers of TUNEL and γH2AX-positive tumor cells (2.7-fold and 2.6-fold, respectively) in the combination-treated tumors compared to controls (P < 0.0001; Fig. 2F and G). To further verify the ability of combined treatment to induce both apoptotic and DNA damage response in vivo, xenograft-bearing mice were treated for 6, 12, or 24 hours. Western blot analysis showed a time-dependent increase in γH2AX and levels of cleaved PARP (Fig. 2H). Of note, monotherapy with either AZD2014 and AZD1775 alone failed to induce detectable accumulation of γH2AX or cleaved PARP (Fig. 2H).

To establish the translational significance of the observed synergy of WEE1 and mTOR inhibition in human NSCLC cell lines, we extended the combination treatment in well-established genetically engineered mouse models of lung adenocarcinoma having an active KRAS^G12D mutation (25). This autochthonous, immunocompetent murine model closely recapitulates the genetic and histopathological features of human lung KRAS-driven adenocarcinoma disease and serves as a suitable clinical predictor (25). Disease course in KRAS^G12D mice was monitored by magnetic resonance imaging (MRI). Upon detection of lung tumor burden, mice were treated with vehicle, single agents, or combined AZD1775 and AZD2014 (Fig. 3A). All vehicle-treated mice showed progressive disease by the 2-week time point (PD: more than 20% increase in tumor volume compared to baseline) with tumor volume doubling at 3-weeks (Fig. 3B and C). We detected significant tumor regression when mice were treated with the AZD2014 and AZD1775 combination compared to vehicle-treated mice at the two- and three-week time points (P = 0.0433 and P = 0.004, respectively; Fig. 3B, Supplementary Fig. S4). The combination therapy achieved a nearly 83% disease control rate (33% partial responses and 50% stable disease; Fig. 3C) at the 3 week-time point as defined by RECIST 1.0; Partial response was defined as more than 30% decrease in tumor volume compared to baseline, while stable disease was defined as neither partial or progressive response. In contrast, 100% of tumors singly treated with either AZD1775 or AZD2014 alone progressed at nearly the same rate as vehicle-treated mice (Fig. 3B–D). In fact, a few single agent-treated mice succumbed to their tumor burden before the three-week time point, emphasizing the lack of efficacy of single agent treatment alone as well as the aggressive disease course in KRAS GEM models. Furthermore, we observed no significant body weight differences between mice treated with the drug combination, suggesting that this combination did not result in treatment-related toxicity in mice (Supplementary Fig. S4). Taken together, these results suggest that the combination of WEE1 and mTOR inhibition may be of potential clinical value in reducing progression of mutant KRAS-driven lung adenocarcinomas.

Given the potent in vitro and in vivo treatment response, we next interrogated downstream effectors of WEE1 and mTOR inhibition. Combined treatment for two hours led to enhanced suppression of phosphorylation of major signaling molecules downstream of mTOR and WEE1, including p70^S6K and CDK1 at tyrosine 15, respectively (Fig. 4A). Moreover, combined treatment reduced phosphorylation of CDK1 at low nanomolar range in a dose-dependent fashion (Fig. 4B).

Treatment of exponentially proliferating A427, H1355, and H23 cells for 48-hours with AZD1775 and AZD2014 led to the accumulation of γH2AX, a marker for DNA damage (Fig. 4C). Consistent with this finding, we showed significantly elevated γH2AX foci in cells treated with both AZD2014 and AZD1775 (P < 0.0001; Fig. 4D and E). In addition to regulating the cell cycle machinery, CDK1-mediated phosphorylation of BRCA1 is also necessary for homology directed DNA repair (HR; ref. 18). We therefore hypothesized that in addition to CDK1 activation, WEE1 inhibition would also lead to increased repair by HR in response to DNA damage from untimely mitosis. We confirmed that H23, H1355 and A427 were HR-proficient cells by assessing RAD51 focus formation upon treatment with DNA-damaging agents such as gemcitabine and etoposide (Supplementary Fig. S5). Compared to vehicle-treated cells, there was significantly increased co-localization of RAD51 and γH2AX foci in response to AZD1775 treatment in NSCLC cell lines (H1355, P = 0.018; A427, P < 0.0001; H23, P = 0.0164; Fig. 4F–H). However, this enhanced level of RAD51 and γH2AX foci formation was abolished when cells were co-treated with AZD2014, suggesting disrupted repair of DNA damage may underlie potentiation of AZD1775 by AZD2014 (Fig. 4F–H).

Activation of mTOR signaling leads to upregulation of components of the protein synthetic machinery and cap-dependent translation of crucial mRNAs for cell cycle transit, such as cyclin D1 and c-MYC (26). Cyclin D1 facilitates RAD51 recruitment to DNA repair foci (27). Therefore, we reasoned that cyclin D1-depletion may similarly sensitize cells to WEE1 inhibition. We used clustered regularly interspaced short palindromic repeats (CRISPR)-based gene editing to generate CCND1-deficient isogenic cell lines and lacZ-targeted sgRNA controls (Fig. 5A). Knockout of CCND1 led to significantly reduced recruitment of RAD51 to sites of DNA damage after AZD1775 treatment of H1355 and H23 cells, suggesting inefficient HR-mediated DNA repair (P = 0.0128 and P = 0.0017, respectively; Fig. 5B and C). Additionally, we observed marked reduction of the number of co-localized RAD51 and γH2AX foci in cyclin D1-deficient cells treated with AZD2014. Importantly, AZD1775 treatment of cyclin D1-deficient cells decreased cell viability in a dose-dependent fashion compared to isogenic control cells (P = 0.0254; Fig. 5D). These results suggest that reduced cyclin D1 translation contributes to, but does not fully account for, the inhibition of HR repair mediated by AZD2014. Taken together, our results suggest that inhibition of mTOR potentiates WEE1 inhibition by abrogating compensatory activation of DNA repair, thus contributing to further DNA damage accumulation and eventual cell death (Fig. 5E).

KRAS activating mutations commonly co-occur with LKB1 inactivating mutations in NSCLCs (9). We and others have demonstrated that KRAS/LKB1-mutant NSCLCs represent a genetically and functionally distinct subset of NSCLC (8). Interestingly, our dose-response data suggest that dual therapy was highly potent in LKB1/KRAS-mutant cell lines (A427, H23, H1355 and H460 cells) and previous studies have implicated LKB1 in DNA damage response (28). Hence, we investigated whether LKB1 depletion sensitizes cells to dual inhibition. We first assessed whether wild type and kinase-dead mutant forms of LKB1(K78I) could render LKB1-deficient RAS-driven cells resistant to dual therapy. H23 and H1355 containing empty vector (EV) constructs displayed increased sensitivity to dual therapy as evidenced by the accumulation of γH2AX protein levels and significantly increased caspase 3/7 activity compared to single-agent treatments (Fig. 6A and B). When cells expressed wild type LKB1, the induction of DNA damage and apoptotic caspase activity were significantly abrogated in response to combination therapy (P < 0.0001, Fig. 6A–D, Supplementary Fig. S6). In contrast, ectopic expression of dominant-negative kinase-dead mutant LKB1(K78I) in H23 and H1355 cells phenocopied the response of parental cells to dual therapy. Consistent with reports that LKB1 negatively regulates mTOR signaling, we found that expression of the wild type LKB1 construct in KRAS/LKB1-mutant cells reduced phospho-S6 activity (Supplementary Fig. S7). These results confirm that LKB1 activity was indeed restored upon transfection and that KRAS/LKB1-mutant cells confer hyperactive mTOR signaling. Despite the reduction in mTOR signaling afforded by expression of wild type LKB1, these cells displayed effective HR repair as observed by restored RAD51 and γH2AX foci upon dual treatment (Fig. 6C–F, Supplementary Fig. S8). Indeed, LKB1 can stimulate DNA repair by modulating BRCA1 levels (28). In contrast, LKB1-deficient cells may be more reliant on mTOR activity to simulate repair and therefore may be vulnerable to AZD2014 treatment after AZ1775-mediated DNA damage.

Despite the development of agents targeting specific genetic subsets of NSCLCs, KRAS-driven lung cancer presently remains “untargetable” and is therefore a highly lethal disease. Exploiting synthetic lethal interactions to selectively target KRAS-mutant lung cancers is warranted. Here, we provide mechanistic evidence that simultaneous inhibition of WEE1 and mTOR signaling allows for more complete disruption of compensatory pathways, resulting in cytotoxic synergy in KRAS-driven NSCLC cell lines and PDX-derived cell lines that more closely mimic the heterogeneity of patient tumors. Importantly, we demonstrated anti-tumor efficacy upon combined treatment with inhibitors at clinically relevant doses in KRAS-GEM models that closely recapitulate human disease.

Activation of KRAS or inhibition of WEE1 deregulates CDK activity and leads to replication stress and subsequent DNA damage (29, 30). Initial preclinical studies on AZD1775 focused on chemo-potentiation based on the rationale that disrupted cell-cycle checkpoints would allow for untimely entry into cell division with unrepaired DNA lesions, resulting in cell death (19, 31, 32). However, more recent observations suggest that the mechanism of AZD1775 cytotoxicity is primarily through DNA damage rather than premature mitosis, consistent with our observations of increased γH2AX-positive cells and apoptotic activity upon single-agent treatment in KRAS-mutant cells (33, 34). Earlier studies have focused on p53-deficient tumors for WEE1 inhibition, based on the reasoning that cells defective in the G₁ checkpoint as a result of p53 loss-of-function would be dependent on the G2 checkpoint survival. Here we observed similar cytotoxic profiles to dual therapy among A427 and H460 cells bearing wild type TP53 and cells containing mutant TP53 (H1355 and H23), suggesting that the cytotoxic synergy is independent of p53 status (31, 34).

In agreement with previous studies implicating CDK1 in DNA repair, we show here that WEE1 inhibition not only increased CDK1 activity in a dose-dependent manner, but also significantly induced HR repair to compensate for the resultant DNA damage (18). Johnson and colleagues previously showed that CDK1-mediated phosphorylation of BRCA1 is required for HR repair and CDK1-depleted cancer cells are HR-defective and sensitized to PARP inhibition both in vitro and in vivo (18). This finding partly explains why single-agent treatment may lack long-term efficiency as cancer cells acquire a CDK1-mediated bypass response. Based on our data, coupling AZD1775 to agents that compromise DNA repair would exacerbate cancer cell killing. Indeed, combined inhibition of checkpoint kinase 1 (CHK1), a kinase implicated in the DNA damage response, and WEE1 has shown therapeutic efficacy in neuroblastoma (35). However, a prior study demonstrated that forced activation of CDK1 via WEE1 inhibition can also impair HR repair in breast cancer cell lines, suggesting that the effects of AZD1775 may also be genotype and cell-type dependent (36). Furthermore, recent studies have revealed a novel role of WEE1 in the maintenance of nucleotide (dNTP) pools through regulation of ribonucleotide reductase subunit 2, RRM2 (30). Pfister and colleagues identified that loss of methyltransferase SETD2 is synthetically lethal with loss of WEE1 in cancer cells due to dNTP starvation via RRM2 deregulation (30). Taken together, these data reveal alternative mechanisms by which WEE1 inhibition can lead to DNA damage. It is therefore possible that TORC1/2 inhibitor-mediated reduction in DNA repair may lead to a synthetic lethal interaction in other cell types besides NSCLC harboring KRAS mutation, constituting a direction for future studies.

Several reports have indicated that cell cycle regulator proteins directly control proteins in DNA repair pathways (27, 37). Given that cyclin D1 facilities RAD51 recruitment to DNA repair foci and is a downstream target of mTOR signaling, we asked whether depletion of cyclin D1 expression was sufficient to override AZD2014-mediated potentiation to AZD1775 (27, 38). We showed that AZD1775 treatment was insufficient to induce HR repair in CRISPR CCND1-deficient cells compared to sgLacZ-control cells. Moreover, AZD2014 treatment further reduced levels of RAD51 and γH2AX foci formation in CCND1-deficient cells compared to control cells, confirming that mTOR signaling is necessary for proper HR repair in KRAS-mutant cells. Although cell viability assays showed that CCND1 depletion was significantly more sensitive to AZD1775 monotherapy than control cells, this response was less potent than the dual inhibition response. Previous studies have shown that mTOR inhibitors suppress HR repair by deregulating expression of a key histone methyltransferase, SUV39H1, in BRCA1-proficient breast cancer cells (39). Shen and colleagues demonstrated that mTOR positively regulates Fanconi anemia group D2 protein, FANCD2, which is recruited to DNA interstrand crosslinks required for proper DNA repair (40). These studies suggest that mTOR elicits multiple cascades linked to cellular DNA repair machinery (41, 42). Therefore, we postulate that other mTOR downstream effectors may also contribute to HR repair. Taken together, our data show that DNA repair is a major target of the synthetic lethal interaction between WEE1 and mTOR inhibition.

LKB1 is a multifaceted tumor suppressor implicated in a variety of cellular processes including signal transduction, energy sensing, cell polarity and dNTP metabolism (43–46). Recently, Gupta and colleagues showed a role for LKB1 in preserving genome integrity by stimulating the expression of BRCA1 to mediate DNA damage response pathways (28). LKB1 has been found to post-transcriptionally stabilize BRCA1 mRNA by inhibiting the cytoplasmic localization of the RNA-binding protein, Hu antigen R (HuR). Here, we demonstrated that ectopic expression of wild type LKB1 in KRAS-mutant cells bearing LKB1 inactivating mutations prevented accumulation of DNA damage after AZD1775 monotherapy treatment. Additionally, wild type LKB1 restoration also prevented reduced HR repair after AZD2014 exposure and therefore also prevented the accumulation of DNA damage after dual therapy. Conversely, expression of a kinase-dead LKB1 mutant phenocopied the response of LKB-mutant cells to dual therapy, confirming that LKB1 activity is necessary for proper DNA repair. Corroborating this observation, a recent Phase I study of single-agent AZD1775 in adult patients with refractory solid tumors demonstrated objective responses in two patients carrying BRCA1 mutations, supporting the notion that LKB1 plays a role in the response to DNA damage induced by WEE1 inhibition (17).

Consistent with prior studies that reported increased sensitivity to MAPK and mTOR signaling inhibition in KRAS/LKB1-mutant subsets of lung cancers, we observed that KRAS/LKB1-mutants conferred hyperactivation of mTOR signaling (ref. 47; Supplementary Fig. S4). LKB1-deficient cells would be unable to stimulate BRCA1 expression after WEE1 inhibitor-mediated DNA damage and may be very reliant on hyperactivated mTOR activity to stimulate DNA repair, explaining the particularly high degree of sensitivity of KRAS/LKB1-mutant cells to dual therapy. Such cells would be expected to be hypersensitive to mTOR inhibition. Additionally, HuR localization of which is affected by LKB1, also regulates levels of WEE1 (28, 48). We speculate that loss of LKB1 may elicit elevated WEE1 activity, which may depend on AZD1775 to override signals in RAS-mutant cells (28, 48). Based on these studies, it would be interesting to comprehensively examine the role of other HuR downstream targets to understand their role in mediating LKB1-loss-induced sensitization to DNA damage via dual therapy.

Our in vitro observations were translated in xenograft models, where dual therapy resulted in the accumulation of both γH2AX and TUNEL-positive staining in tumors. We also evaluated treatment response in mice with lung-specific conditional activation of KRAS^G12D mutations that developed aggressive lung adenocarcinomas (25). The WEE1 and mTOR inhibitor combination induced regression and disease stabilization over 2 to 3 weeks of treatment in established tumors. Interestingly, the combinatorial treatment was successful in vivo in an LKB1-proficient setting, suggesting that induction of HR mediated by LKB1 in vivo may not be as profound as that observed in in vitro models studied here or in other cell systems (40). It will be of interest to determine whether combinatorial effects are even more profound in LKB1-deficient mouse models. Notably, we did not detect any drug-related toxicity during the course of 3-week treatment on mice, consistent with a recent Phase I study showing AZD1775 tolerability and safety as monotherapy and in combination with standard chemotherapy in 202 patients (49).

In summary, the present study is the first to show that AZD2014 potentiates AZD1775 by abrogating compensatory activation of HR repair, resulting in cytotoxic synergism. Considering the absence of targeted therapies available for KRAS-driven NSCLC, dual inhibition of mTOR and WEE1 should be investigated as a potential strategy. Additionally, since tumors harboring concomitant KRAS activation and LKB1 loss are among the most highly aggressive, combined mTOR and WEE1 inhibition may be particularly compelling for this lung cancer subset.

G.I. Shapiro is a consultant/advisory board member for G1 Therapeutics, Lilly, Pfizer, Roche, and Vertex, and reports receiving commercial research grants from Lilly and Pfizer. K.K. Wong reports receiving commercial research grants from AstraZeneca, Janssen Pharmaceuticals and Lilly Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

All animal experiments were performed in accordance with procedures approved by the DFCI Animal Care and Use Committee.

Conception and design: J. Hai, K. Do, T. Chen, G.I. Shapiro, K.-K. Wong

Development of methodology: J. Hai, X. Wang, K.-K. Wong

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Hai, S. Liu, X. Wang, M.-S. Tsao, K.-K. Wong

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Hai, S. Liu, L. Bufe, K. Do, X. Wang, G.I. Shapiro, K.-K. Wong

Writing, review, and/or revision of the manuscript: J. Hai, S. Liu, K. Do, X. Wang, S. Li, M.-S. Tsao, G.I. Shapiro, K.-K. Wong

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Hai, T. Chen, C. Ng, K.-K. Wong

Study supervision: J. Hai, K.-K. Wong

The authors thank Mei Zhang for help with immunohistochemistry work, Yanxi Zhang for technical animal assistance and Dr. Shunsuke Kitajima, Dr. Nhu-An Pham, and Belfer Institute for providing PDX and NSCLC cell lines. We also thank AstraZeneca for making AZD2014 and AZD1775 available for this study and Belfer Institute and Pinar Eser for making the IncuCyte system available to us.

This work was supported by the National Cancer Institute (R01CA195740, CA163896, CA166480, CA122794, and CA140594), the Thoracic Foundation (to K.K. Wong), the Gross-Loh Family Fund for Lung Cancer Research, and the Susan Spooner Family Lung Cancer Research Fund at Dana-Farber Cancer Institute (to K.K. Wong). K.K. Wong was supported by a Stand Up to Cancer—American Cancer Society Lung Cancer Dream Team Translational Research Grant (Grant Number: SU2C-AACR-DT1715). Stand Up To Cancer is a program of the Entertainment Industry Foundation. Research grants are administered by the American Association for Cancer Research, the Scientific Partner of SU2C.

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.