WEE1 Kinase Inhibitor AZD1775 Has Preclinical Efficacy in LKB1-Deficient Non–Small Cell Lung Cancer

Lung cancer remains the leading cause of cancer-related mortality, and is expected to kill more than one million people worldwide in 2016 (1). Despite clinical successes with targeted therapeutic agents against EGFR or ELM4–ALK fusion protein, the majority of lung adenocarcinomas (the most prevalent histologic subtype of lung cancer) demonstrate no genomic alterations in these oncogenes. Instead, most genomic alterations occur in TP53, KRAS, STK11/LKB1, and KEAP1—genes with no effective targeted therapeutic avenues (2). Therefore, development of novel therapeutic approaches for lung adenocarcinoma with oncogenic KRAS or loss of tumor suppressors such as LKB1 is of the utmost clinical importance.

Mutations in LKB1 tumor suppressor were initially identified to be responsible for Peutz–Jeghers syndrome, a hereditary condition that carries increased cancer risk (3). Although the incidence of LKB1 somatic mutations is low in most tumor types, inactivating LKB1 alterations in lung adenocarcinomas are the third most common event (2). Deletion of Lkb1 combined with expression of mutant Kras (G12 mutation) in genetically engineered mouse models leads to highly aggressive, metastatic tumors. Researchers have linked loss of Lkb1 to significant changes in cellular metabolism, polarity, differentiation, and the tumor microenvironment (4–7). These changes collectively provide opportunities for therapeutic interventions, as targeting the altered cellular functions associated with LKB1 loss have shown preclinical efficacy. Whether these preclinical approaches can be translated to the clinical arena remains an area of active work.

The G₂–M cell cycle checkpoint is a critical mechanism for DNA repair. Activation of the G₂–M checkpoint delays mitosis after DNA damage, which allows for DNA repair (8). Tumor cells preferentially engage this checkpoint as a mechanism to survive genotoxic events and resist DNA-damaging therapies (9). Studies indicate that inhibition of the cyclin B-CDK1 (CDC2) complex is pivotal for engagement of the G₂–M DNA damage checkpoint (10). The WEE1 kinase directly phosphorylates CDC2 at tyrosine 15, thereby inactivating the CDC2/cyclin B complex and halting the cell cycle (11, 12). Because of the regulatory role of WEE1 in the G₂–M checkpoint, pharmacologic suppression of WEE1 activity by the small-molecule AZD1775 has been shown to have antitumor effects across a number of cancer types (13).Early studies reported that tumor cells harboring mutations in TP53 were more susceptible to WEE1 inhibition in combination with DNA-damaging agents, but more recent investigations suggest that the antitumor effects of AZD1775 and WEE1 inhibition may affect function in tumor cells with genetic alterations other than TP53 (14). A phase I clinical trial of single-agent AZD1775 demonstrated tolerability in patients, suppression of CDC2 phosphorylation, and multiple partial responses (15). Interestingly, no response was observed in five patients with tumors harboring TP53 mutations; however, responses were detected in tumors with BRCA mutations, a known tumor suppressor gene associated with DNA damage response. AZD1775 is currently being used in a number of phase II clinical trials in lung cancers and other tumor types (www.ClinicalTrials.gov).

As highlighted by the role of BRCA mutations in a phase I study of AZD1775 (15), there is a great push for understanding the genomic contexts in which WEE1 inhibition would be most efficacious. In the current study, we investigated the effects of AZD1775, alone and in combination with DNA agents in LKB1-deficient non–small cell lung cancer (NSCLC). Relevant in vitro and in vivo models of LKB1-deficient NSCLC were used to understand the role of LKB1 in AZD1775 effectiveness. We further explored the role of the DNA damage response pathway to elucidate the mechanism of AZD1775 efficacy in LKB1-deficient NSCLC.

A549, A427, and H2030 human lung cancer cell lines were obtained from the ATCC within the last 3 years, and maintained in RPMI1640 media (Invitrogen) with 10% FBS and penicillin/streptomycin (Thermo Fisher Scientific) under standard tissue culture conditions (37°/5% CO₂ atmosphere). Cells were used within three passages of thawing for experiments. Mycoplasma testing was performed according to the MycoAlert Kit (Lonza). Short-term murine 3381A NSCLC cell line was isolated from Kras/Lkb1 transgenic mice by grossly dissecting lung tumor nodules 8 weeks postinfection with AdenoCre by intranasal inhalation. Tumor nodules were washed in warm media, diced into approximately 2-mm chunks, and allowed to settle to allow for outgrowth of tumor cells. Confluent plates of tumor cells were collected and maintained in RPMI1640 media and standard tissue culture conditions. Experiments were performed at earliest passage available. Isogenic cell lines A549-LKB1, A549-KDLKB1, H2030-LKB1 and H2030-KDLKB1, H2030-pBabe, H2030-TP53, 3381A-pBabe and 3381A-LKB1 were constructed from early passage parental lines (within 5 passages of purchase/derivation) using previously described methods (16). pBabe retroviral constructs were obtained from Addgene. The constructs, which contained FLAG-tagged full-length LKB1 and kinase-dead LKB1 (KDLKB1), were originally developed in the laboratory of Dr. Lewis Cantley (Harvard Medical School, Boston, MA; ref. 17). The pBabe-puro-TP53 vector was purchased from Cell Biolabs, Inc. Mutation of threonine-366 to alanine in FLAG-tagged LKB1 (T336A) was performed using the Q5 site-directed mutagenesis according to the manufacturer instructions (New England Biolabs Inc.). Mutagenesis of the 366 site was confirmed by DNA sequencing of the plasmid. Cell lines transfected with genetic constructs were maintained using the selective antibiotic puromycin (Thermo Fisher Scientific). AZD1775 (MK-1775) was purchased from MedChem Express and diluted in dimethyl sulfoxide (DMSO). Cisplatin was purchased from Tocris Bioscience and diluted in sterile H₂O.

Antibody to phosphorylated-CDC2 (pCDC2) were obtained from Abcam. Phosphorylated (serine 139)-histone H2A.X (γH2AX), LKB1, phosphorylated AMPK (pAMPK), cleaved PARP, cleaved caspase-3, CDKN1A (p21), GAPDH, and α-tubulin antibodies were purchased from Cell Signaling Technology. Immunoblot analysis was performed following standard protocols for adherent cell lines. Cells were washed twice with wash buffer containing 1× PBS, 0.1 mol/L PMSF and protease inhibitor sodium orthovanadate (NaOV). Scraped cells were transferred to prechilled Eppendorf tubes and pelleted by centrifugation following lysis with RIPA lysis buffer (0.1 mol/L PMSF, NaOV and ultra-pure distilled H20). Samples were then sonicated and centrifuged, and the supernatant was transferred and to a new prechilled tube. Total protein lysates were separated by SDS-PAGE and transferred to nitrocellulose membranes. Blots were blocked with either 5% milk in Tris-buffered saline/0.1% Tween 20 (TBST) or in 5% BSA (Sigma-Aldrich) in TBST. Primary antibodies were diluted in 5% BSA and incubated overnight on a shaker at 4°C. Blots were developed with Pierce ECL western blotting substrate (Thermo Fisher Scientific) and visualized on a Carestream Image Station 4000 MM (Carestream Health Inc.). Densitometry was performed using Carestream Molecular Imaging software. Proteins of interest were normalized against the loading control for the immunoblot, and fold change was compared against the untreated control sample.

Cell viability assay was performed using the CellTiter-Glo Luminescent Cell Viability Assay Kit (Promega) to determine the number of viable cells based on ATP quantification. Cells were plated in 96-well plates and allowed to adhere overnight before being treated with the indicated drugs for 48 hours. Plates were treated with CellTiter-Glo substrate following the manufacturer's protocol and luminescence was read using the PerkinElmer EnVision 2102 multi-label plate reader (PerkinElmer). Cell viability was calculated relative to the nontreated vehicle control. Differences between groups were determined by the Student t test, with a P < 0.05 considered statistically significant.

For the clonogenic assay, the indicated cell lines were plated at 250 cells per well in 6-well dishes and allowed to adhere overnight. Cells were then treated for 24 hours with cisplatin, AZD1775, combined AZD1775 + cisplatin, or vehicle (i.e., DMSO) at indicated doses in triplicate. After 24 hours of drug incubation, the media were refreshed with fresh growth media (no drugs) and plates were incubated at 37°C until colonies achieved approximately 50 cells per colony (roughly 7–10 days). Cells were fixed with 10% acetic acid/10% methanol solution, stained with 0.5% crystal violet solution, and washed with deionized water. Once dry, colonies containing more than 50 cells were manually counted. The viable fraction was then calculated following the procedure described by Franken and colleagues (18). Differences between groups were determined by the Student t test, with a P value of <0.05 considered statistically significant.

Mutant Kras (Kras) with and without concomitant loss of Lkb1 (Kras/Lkb1) transgenic mice have been described previously (5, 19, 20). Kras and Kras/Lkb1 mice harboring the conditional (lox-stop-lox) firefly luciferase gene (Kras^luc and Kras^luc/Lkb1; ref. 20) were generated by selective breeding with Gt(ROSA)26Sor^tm1(Luc)Kael/J mice (The Jackson Laboratory). Mice were backcrossed for six generations onto the FVB/Nj background. Lung tumors were generated in 8-week-old Kras^luc and Kras^luc/Lkb1 following previously described methods (5, 20). The baseline tumor burden was determined at 7 weeks postinfection with AdenoCRE using the Xenogen IVIS system (PerkinElmer). Briefly, Kras^luc and Kras^luc/Lkb1 were injected intraperitoneally with d-luciferin (150 mg/kg in sterile PBS) and scanned 10 minutes postinjection. For drug treatment studies, Kras^luc/Lkb1 mice positive for tumor burden were randomized for 4 weeks of treatment with either cisplatin (2.5 mg/kg, once per week, intraperitoneally) or combined AZD1775 (30 mg/kg, 3 days per week, oral gavage) + cisplatin. Mice received a follow-up scan at the conclusion of treatment and were subsequently monitored. Moribund mice were sacrificed and necropsied. Overall survival was assessed by log-rank (Mantel–Cox) testing, with a P < 0.05 considered statistically significant.

For short-term dosing studies, Kras^luc and Kras^luc/Lkb1 mice underwent bioluminescent imaging to confirm tumor burden 8 weeks postinfection with AdenoCre. Mice with disease were randomized for treatment with either vehicle (0.5% methylcellulose), AZD1775 (30 mg/kg for 3 days), cisplatin (2.5 mg/kg, one dose) or combined AZD1775 + cisplatin. Mice were sacrificed 2 hours after the final dose of AZD1775 and necropsied. A sample of tumor was inflated and fixed in 10% neutral buffered formalin overnight. Fixed tissue was prepared as formalin-fixed, paraffin-embedded (FFPE) tissue using routine procedures. All treatments were done in accordance with a preapproved protocol authorized by the Institutional Animal Care and Use Committee at St. Joseph's Hospital and Medical Center in Phoenix, Arizona.

Tumors from Kras^luc or Kras^luc/Lkb1 mice treated with a single dose of 0.5% methylcellulose, cisplatin (2.5 mg/kg), AZD1775 (30 mg/kg), or combined AZD1775 + cisplatin were fixed overnight in 10% neutral-buffered formalin, and paraffin-embedded by routine procedure. FFPE samples were sectioned, floated onto charged slides, and baked for one hour. Five-micron thick sections underwent heat-induced epitope retrieval in Tris-EDTA buffer (pH 9), and IHC staining was performed using published methods (19). Antibodies to LKB1 (Cell Signaling Technology), γH2AX (Cell Signaling Technology), and phosphorylated CDC2 (Abcam) were used. Slides were scanned using an Aperio system (Leica Biosystems), and images were collected at a magnification of ×20. Quantification of γH2AX-positive nuclei was performed by manual counting from 10 random fields of view by a blinded reviewer. Two mice per treatment group were counted. Differences were assessed by the Student t test, with a P < 0.05 considered statistically significant.

LKB1 has been suggested to function in the DNA damage repair pathway (21–23). To test whether LKB1 inactivation conferred sensitivity of NSCLC to WEE1 inhibition by AZD1775, we treated isogeneic pairs of three human NSCLC cell lines (A427, A549, and H2030), all with oncogenic mutation in KRAS and inactivation of LKB1 with AZD1775. In each cell line, wild-type LKB1 (LKB1) or a kinase-dead LKB1 (KDLKB1) was stably introduced. In all three isogenic pairs, LKB1 expression reduced AZD1775 impact on cell viability compared with cells expressing KDLKB1 (Fig. 1A–C). This result was validated in a murine tumor cell line (3381A) derived from a NSCLC tumor of a Kras/Lkb1 transgenic mouse with stable expression of wild-type LKB1 or empty vector (pBabe). As shown in Fig. 1D, stable expression of LKB1 suppressed AZD1775-induced loss of viability compared with vector-alone (pBabe) 3381A cells. Consistent with the mechanism of action of AZD1775, both A549 and H2030 both showed inhibition of WEE1-dependent phosphorylation of CDC2 (pCDC2), regardless of LKB1 or KDLKB1 expression (Fig. 1E). However, in both A549 and H2030 NSCLC cell lines, the stable expression of LKB1 reduced the protein expression of γH2AX, a marker for DNA damage, compared with kinase-dead LKB1 expression (55% reduction in A549 and 49% in H2030, as determined by densitometry; Fig. 1E). Consistent reductions in NSCLC cell viability and induction of DNA damage response markers were observed in LKB1-deficient cells at concentrations below those achievable in serum (1.3–1.6 μmol/L) in clinical trials employing AZD1775 (15, 24). Ample evidence has pointed to LKB1-dependent activation of AMPK as the primary mediator of LKB1-dependent functions and importantly in promoting cell viability in response to extrinsic stressors (16, 17, 19). Consistent with these findings, reintroduction of LKB1 into LKB1-deficient A549 and H2030 cells resulted in increased phosphorylated AMPK (pAMPK) indicative of a functional LKB1–AMPK signaling pathway, whereas expression of a kinase-dead LKB1(KDLKB1) showed reduced pAMPK (Fig. 1E). AZD1775 exposure had no effect on pAMPK protein levels in LKB1-expressing NSCLC cells, suggesting that the effects of AZD1775 were independent of AMPK activity.

In both clinical and preclinical studies, AZD1775 is commonly paired with DNA-damaging agents (e.g., cisplatin, radiation), as these combinations are found to enhance cytotoxicity relative to DNA damage alone (13). In both human (H2030) and murine (3381A) cell lines, exposure to combined AZD1775 (200 nmol/L) and cisplatin (5 μmol/L) induced higher protein levels of γH2AX (5-fold and 22-fold over untreated in 3381A and H2030 cells, respectively) and cleaved PARP (15-fold and 3.4-fold over untreated in 3381A and H2030 cells, respectively), a marker of apoptosis, compared with either drug alone (Fig. 2A and B). Furthermore, ectopic expression of LKB1 blunted the induction of γH2AX (85% and 63% compared with LKB1-null 3381A and H2030 cells, respectively) and cleaved PARP (79% and 47% compared with LKB1-null 3381A and H2030 cells, respectively) by combined AZD1775 + cisplatin. Likewise, measurement of cell viability by the CellTiter-Glo assay in both isogenic human and mouse lung cancer cell lines showed that the addition of AZD1775 reduced cell viability across escalating doses of cisplatin. However, ectopic expression of wild-type LKB1 in both H2030 and 3381A cell lines resulted in improved cell viability compared with empty vector (3381A) or KDLKB1 (H2030; Fig. 2C and D). These observations were further validated in clonogenic cell survival assays, with H2030-LKB1 cells displaying improved survival to AZD1775 + cisplatin treatment compared to H2030-KDLKB1 (Fig. 2E). Notably, in the isogenic H2030 cell lines, exposure to combined AZD1775 + cisplatin reduced clonogenic cell survival compared with no treatment or to either treatment alone.

To test whether the combinational effects of AZD1775 + cisplatin in LKB1-deficient NSCLC extended to other DNA-damaging agents, we examined the effects of AZD1775 in combination with radiation exposure. To control for previous exposure to DNA-damaging treatment (a characteristic of many long-term human cancer cell lines), we used the isogeneic 3381A murine NSCLC cell line. Exposure to radiation (2 Gy) and AZD1775 increased both γH2AX (5-fold over untreated) and cleaved caspase-3 protein (8-fold over untreated) levels in the empty vector cells. In the 3381A NSCLC cells expressing wild-type LKB1, γH2AX, and cleaved caspase-3 were suppressed (28% and 73%, respectively) compared with the treated empty vector cells (Fig. 2F). Likewise, 3381A cells expressing the empty vector were more sensitive to radiation (4 Gy) in combination with 500 nmol/L AZD1775, compared with 3381A NSCLC cells expressing wild-type LKB1 (Fig. 2G).

Mutation of TP53 has been shown to confer sensitivity to AZD1775 in combination with DNA-damaging agents across a number of cancer types (13). Notably, comutations of LKB1 and TP53 have been reported, although parallel evidence shows that LKB1 and TP53 mutations exist within two distinctly separate populations of NSCLC, especially in the context of mutant KRAS (2, 25). As several investigations into the effectiveness of WEE1 inhibition have focused on loss of TP53, it is worth noting that the effects of LKB1 expression on AZD1775 efficacy were independent of TP53 mutation, as A427 and A549 have wild-type TP53, whereas H2030 contains mutant TP53 (Fig. 1A–C). As H2030 cells harbor mutant KRAS with concomitant loss of TP53 and LKB1, we generated isogenic cell lines by reintroducing either wild-type TP53 or wild-type LKB1 through retroviral infection and puromycin selection. H2030 isogenic NSCLC cells expressing vector-alone (pBabe), LKB1, or TP53 were treated with AZD1775, cisplatin, or combined AZD1775 + cisplatin and analyzed by both immunoblot and clonogenic cell survival assay. As shown in Figs. 3A and B, restoration of LKB1 and TP53 were capable of reducing protein levels of cleaved PARP and γH2AX, and improved cell survival in response to combined AZD1775 + cisplatin. However, wild-type expression of LKB1 had a more pronounced effect on these markers relative to TP53. Interestingly, p21, which was reported to be downstream of both LKB1 and TP53, was only found to increase with ectopic expression of TP53 in H2030 cells.

Based upon the effects of AZD1775 in combination with DNA-damaging agents, we postulated that this combination could function as a potential treatment for NSCLC tumors harboring mutant KRAS with concomitant loss of LKB1. To test this hypothesis in an in vivo system, we used a well-characterized transgenic NSCLC model (Kras/Lkb1) that has been further refined by introduction of a conditional firefly luciferase gene (luc) via selective breeding to generate Kras^luc/Lkb1 mice (20). Figure 4A demonstrates that LKB1 protein is present in tumors from Kras^luc mice, but is absent in tumor cells from Kras^luc/Lkb1 mice. LKB1 protein staining was restricted to stromal and normal lung alveolar epithelial cells, confirming specificity of Cre-mediated biallelic deletion of LKB1 in tumor cells. In concordance with an autochthonous lung tumor model of Kras/Lkb1, conditional biallelic deletion of the floxed Lkb1 gene and lox-stop-lox cassettes proximal to both mutant Kras and luc genes resulted in bioluminescent NSCLC tumors that were visible using noninvasive imaging (Supplementary Fig. S1A and S1B). Eight-week-old Kras^luc/Lkb1 mice were infected with adenoCre (5 × 10⁶ PFU) via intranasal inhalation and imaged 7 weeks postinfection. As with other autochthonous Kras/Lkb1 lung tumor models, mice at this time point display invasive, multifocal disease, thus allowing preclinical assessment of AZD1775 in mice with “late-stage” disease. We chose to test combined AZD1775 + cisplatin, as cisplatin is a commonly used chemotherapeutic agent for NSCLC. Animals with identified tumors were randomized for four weeks of treatment with cisplatin alone (2.5 mg/kg once per week) or combined AZD1775 (30 mg/kg three times per week) + cisplatin (2.5 mg/kg once per week; Supplementary Fig. S1C) and received a follow-up scan at the end of treatment. Animals were euthanized upon reaching a moribund state to determine overall survival. Analysis of tumor volumes [change in relative light units (RLU)] suggested a nonstatistically significant trend toward reduced tumor burden with AZD1775 + cisplatin, compared with cisplatin only (Supplementary Fig. S1D). However, median survival of mice treated with combined AZD1775 + cisplatin was significantly longer than mice exposed to cisplatin alone (14.14 vs. 12.28 weeks, respectively; P = 0.0285; Fig. 4B). Notably, cisplatin alone also improved survival compared with previously published 9-week median survival for the Kras/Lkb1 NSCLC model, and is depicted in the figure by a dashed line (5).

To understand the mechanistic underpinnings of combined AZD1775 + cisplatin treatment in NSCLC tumors, we performed a short-term dosing study in Kras^luc/Lkb1 and Kras^luc mice. Kras^luc/Lkb1 and Kras^luc mice with tumor burden were randomized for a single treatment course of cisplatin, AZD1775, or combined AZD1775 + cisplatin, and sacrificed for IHC analysis of DNA damage markers. As shown in Figs. 4C and D, cisplatin and combined AZD1775 + cisplatin demonstrated elevated protein staining of γH2AX, a marker for DNA damage, in Lkb1-deficient NSCLC tumors compared with Lkb1-expressing tumors. Furthermore, the combination of AZD1775 + cisplatin induced higher γH2AX protein levels than cisplatin alone in Lkb1-deficient tumors, whereas tumors of the Kras^luc mice showed comparable γH2AX protein levels regardless of treatment (Fig. 4C). Protein staining of phosphorylated CDC2 (pCDC2) was determined in Kras^luc/Lkb1 and Kras^luc mice by IHC to ensure AZD1775 activity. Exposure to cisplatin alone in Kras^luc or Kras^luc/Lkb1 (Fig. 4E) showed positive staining for pCDC2, suggesting activation of the G₂–M checkpoint. Exposure to combined AZD1775 + cisplatin demonstrated reduced pCDC2 protein staining compared with cisplatin alone in both Kras^luc and Kras^luc/Lkb1 tumors (Fig. 4E).

We next investigated potential mechanisms for the increased sensitivity of LKB1-deficient NSCLC to combined AZD1775 + cisplatin therapy. The ATM Serine/Threonine Kinase (ATM) is one of the kinases responsible for sensing and activation repair following DNA damage. Studies have shown that ATM directly phosphorylates LKB1 at threonine 366 (T366) in response to radiation (21, 23). To test whether ATM-mediated phosphorylation of LKB1 had a role in reducing the effects of combined AZD1775 + cisplatin treatment, we reintroduced a variant of LKB1 wherein threonine 366 was mutated to alanine to prevent ATM phosphorylation. As shown in Fig. 5A, expression of wild-type LKB1 improved survival of LKB1-deficient H2030 NSCLC cells to combined AZD1775 + cisplatin treatment, whereas expression of LKB1^T366A resulted in partial restoration of sensitivity to combined AZD1775 + cisplatin treatment. In addition, analysis of γH2AX protein levels after removal of combined AZD1775 + cisplatin found that although wild-type LKB1 displayed reduction of γH2AX levels within 6 hours of withdrawal (44% reduced compared with CA treated), both H2030-pBabe and H2030-LKB1^T366A NSCLC cells retained increased levels of DNA damage as determined by γH2AX (75% and 46% increase at 6 hours compared with wild-type LKB1, respectively; Fig. 5B). Notably, exposure to the AMPK activator 2-d-deoxyglucose resulted in phosphorylation of AMPK in both LKB1 and LKB1^T366A, but not in pBabe only, indicating that mutation of threonine 366 did not alter the ability of LKB1 to activate AMPK in response to energetic stress (Supplementary Fig. S2).

In this study, we show that in vitro exposure to AZD1775 enhances cytotoxicity and increases DNA damage in mutant KRAS/LKB1–deficient NSCLC cell lines. We further show that AZD1775 in combination with DNA-damaging agents has significant additive effects in mutant KRAS/LKB1–deficient NSCLC cells, even at low doses. Critically, combined AZD1775 + cisplatin displayed greater in vivo efficacy compared with cisplatin alone in a transgenic model of advanced Kras/Lkb1 NSCLC. We find that lack of ATM–LKB1 signaling in LKB1-deficient NSCLC cells may play a role in these mechanisms. Collectively, our studies suggest that taking therapeutic advantage of defects in the DNA damage response related to LKB1 inactivation may have potential in treating NSCLC.

Initial studies of AZD1775 (previously known as MK1775) demonstrated preferential antitumor effects in TP53-deficient tumor cells across a variety of tumor types (13, 26). Results from in vitro and in vivo studies suggested that the loss of TP53 caused dependence on the G₂–M checkpoint of the cell cycle and disruption of WEE1 signaling–induced tumor cell death both as a single agent and in combination with cytotoxic chemotherapies. Hirai and colleagues (26) initially demonstrated that the combination of MK1775 and DNA-damaging agents (i.e., gemcitabine, carboplatin, or cisplatin) was more effective in suppressing tumor growth and inducing apoptosis both in vitro and in vivo. Antitumor efficacies of the coadministration of AZD1775 and numerous other chemotherapies (i.e., 5-FU, capecitabine, pemetrexed, camptothecin, doxorubicin, or mitomycin C) have been subsequently elucidated in colon, cervical, pancreatic, ovarian, breast, and lung tumors (13, 27). The administration of WEE1 inhibitors also sensitized tumor cells to radiation exposure. Most of these studies have focused on the loss of TP53 as a molecular context for AZD1775 effectiveness. However, a number of studies have shown antitumor effects of AZD1775, regardless of TP53 status (14, 15, 28, 29). Similarly, results of a phase I clinical trial using single-agent AZD1775 elucidated multiple partial responses, although none in cases with TP53 mutations (15). Interestingly, partial responses were observed in patients with tumors harboring BRCA1/2 mutations, known players in DNA damage response.

Other preclinical data suggest that tumors with G₁–S checkpoint aberrations could sensitize to AZD1775 treatment (30), while a recent study showed that H3K36me3-deficient cancers were sensitive to AZD1775 treatment, associated with dNTP starvation (31). In this study, we show preferential responses to AZD1775 in mutant KRAS with concomitant LKB1-deficiency both alone and in combination with DNA-damaging therapeutics in vitro and in vivo. As previously established, the combination of AZD1775 with a DNA-damaging agent produced a more robust response on tumor cell viability than AZD1775 alone in vitro, although loss of LKB1 consistently impacted the effect of AZD1775, alone or in combination, compared with wild-type LKB1. It should be noted that the murine lines developed from a mouse model driven only by Kras/Lkb1 tended to have more robust responses than the human cell lines that harbor KRAS/LKB1. However, these differences may be due to the presence of other genomic alterations inherent in human tumor cell lines that might impact AZD1775 effect, a phenomenon to be further investigated. Of note, the improvement of overall survival in the Kras/Lkb1 mice who received combined AZD1775 + cisplatin was in an advanced-stage NSCLC tumor, based on timing of treatment. Testing this combination in tumors at earlier stages may further extend the survival time. We also demonstrate that the loss of LKB1 in NSCLC facilitates the antitumor effects of coadministration of AZD1775 + cisplatin, as the ectopic expression of wild-type LKB1 abrogated the combinational effect. Collectively, these studies suggest that the effects of AZD1775 and in particular inhibition of WEE1-CDC2 signaling may be mediated by factors other than TP53 inactivation alone.

Although loss or mutation of LKB1 is a common event in lung adenocarcinoma, several therapeutic strategies have been shown to be effective in this molecular context. As LKB1 was discovered as a regulator of cellular metabolism through AMPK, therapeutic strategies linked to enhanced energetic stress have shown promise in preclinical settings. Energetic stress induced by metformin or phenformin preferentially induced apoptosis in LKB1-deficient tumor cells (32). Our group demonstrated both in vitro and in vivo that 2-deoxyglucose could induce tumor cell death in LKB1-deficient tumor models (16, 19). LKB1 loss also regulates mTOR signaling/activation through AMPK, and thus mTOR inhibitors such as rapamycin, everolimus, and BEZ235 have shown preclinical in vitro efficacy in LKB1-deficient contexts (32); however, rapamycin failed to show in vivo efficacy in the Kras/Lkb1 NSCLC mouse model (33). Clinical trials are currently investigating the effects of everolimus or metformin in tumors harboring LKB1 mutation or NSCLC, respectively (www.clinicaltrials.gov). We sought to take advantage of LKB1 deficiency in the context of DNA damage/repair, and our data suggest that patient tumors harboring mutant KRAS with loss of LKB1 may respond more effectively to WEE1 inhibition, either alone or in combination with DNA-damaging therapies. Recent work by Wang and colleagues (34) demonstrated that LKB1-deficient cells were more sensitive to PARP inhibitors, and that LKB1 could interact with ATM and BRCA1 as part of the DNA damage response pathway. Our data also suggest that T366, a phosphorylation site for ATM on LKB1 is necessary for the resistance conferred by wild-type LKB1 to AZD1775 exposure. Furthermore, results from a phase I clinical trial with AZD1775 showed partial responses in patients with BRCA mutations, known alterations associated with DNA damage response (15). These data suggest that aberrant LKB1 expression or signaling might influence sensitivities to different classes of therapeutics intended to target DNA repair mechanisms.

Although the regulatory function of the cellular metabolism of LKB1 has been well studied, the role of LKB1 in DNA damage and repair are still unclear. Early work showed that activation of ATM by radiation led to LKB1 phosphorylation (21, 23), while further analysis within a transgenic model of skin cancer found that Lkb1 loss and subsequent lack of ATM–LKB1 signaling potentiated defects in DNA repair after UV-induced DNA damage (35). Other analyses have found that LKB1 localizes to the sites of nonhomologous end joining repair (36) and defects in pyrimidine metabolism due to LKB1 inactivation have been proposed to alter DNA repair in NSCLC cells (7). In this study, we show that phosphorylation of LKB1 by ATM in the setting of a WEE1 inhibitor may play a role in reducing the DNA damage and cytotoxicity of DNA-damaging agents.

In addition to the clinical implications of our work, our findings further substantiate a role for LKB1 in regulation of the DNA damage response. In line with previous studies, we found that prevention of ATM-dependent phosphorylation of LKB1 altered DNA repair and survival. Although the bulk of available data regarding ATM–LKB1 signaling has focused on UV- or radiation-induced DNA damage, our work shows activation of this pathway in response to chemotherapeutic agents. Notably, recent genetic analysis of NSCLC tumors carrying KRAS mutations revealed that genetic inactivation of ATM replicated the gene expression profile of mutant KRAS with loss of LKB1 (25). Furthermore, comutation of ATM and LKB1 are noted in The Cancer Genome Atlas (TCGA) dataset for lung adenocarcinoma (2). These findings prompt the question: does the loss of ATM–LKB1 signaling play a critical role in driving NSCLC tumorigenesis in mutant KRAS? Investigating this query is outside the scope of the current study, but should be addressed in the future.

A large proportion of patients with NSCLC demonstrate LKB1 inactivation and KRAS mutations. Despite the growing understanding of the characteristics of these tumors, clinical treatment has not yet been developed for this subset of patients. By using in vitro and in vivo models, we have shown that combined AZD1775 + cisplatin may have value for treating patients with NSCLC who have LKB1 inactivation and mutant KRAS. Despite these positive findings, it will be necessary to perform thorough investigations to determine the dose and schedule of AZD1775 with a DNA-damaging agent necessary for maximal clinical benefit, as others have done (26). However, these studies lie outside the focus of this study and are underway within our laboratories. AZD1775 is currently being studied in clinical trials for solid tumors, with particular emphasis on tumors harboring TP53 mutations. As indicated by our work, an understanding of the contributions of other genomic alterations will be critical for AZD1775 to achieve clinical application and efficacy.

No potential conflicts of interest were disclosed.

Conception and design: L.J. Inge, T.G. Whitsett

Development of methodology: L.J. Inge

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.L. Richer, J.M. Cala, K. O'Brien, V.M. Carson, L.J. Inge

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.L. Richer, J.M. Cala, L.J. Inge, T.G. Whitsett

Writing, review, and/or revision of the manuscript: V.M. Carson, L.J. Inge, T.G. Whitsett

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. O'Brien, L.J. Inge

Study supervision: L.J. Inge, T.G. Whitsett

The authors would like to acknowledge that part of this work was supported by a generous gift from the St. Joseph's Foundation.

L.J. Inge was supported by funds from the St. Joseph's Foundation. T.G. Whitsett was support by TGen start-up funding.

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.