AZD6738, A Novel Oral Inhibitor of ATR, Induces Synthetic Lethality with ATM Deficiency in Gastric Cancer Cells

DNA-damaging agents represent the cornerstone of cancer treatment. Rapid advances in cancer biology have identified key pathways involved in the repair of DNA damage. Although there are various types of DNA repair response pathways, repair of DNA double-strand breaks (DSB) by homologous recombination (HR) and nonhomologous end joining (NHEH) primarily influence the therapeutic efficacies of DNA-damaging agents (1–3). HR deficiency (HRD) is frequently observed in cancer cells, and unlike normal cells, DSBs are dealt with by NHEJ pathways that result in an intolerable level of genomic instability and cancer cell death.

Cancer cells with HRD have been shown to be particularly sensitive to DNA-damaging agents and PARP inhibitors (4–6), for example, responses to olaparib (a PARP inhibitor) were observed across different tumor types associated with germline BRCA1/2 mutations (7). Olaparib has been approved in several countries for advanced ovarian cancer with germline BRCA mutation (7). However, cancer with a BRCA germ line mutation represents only a small proportion of all cancer cases, and in many countries, gastric cancer is still the leading cause of cancer-related death. Low ATM protein expressions, which contribute to HRD, predict response to PARP inhibitor in gastric cancer cells. A randomized phase II study in gastric cancer showed olaparib/paclitaxel is active in patients with metastatic gastric cancer that have failed first-line fluorouracil- and platinum-based therapy, and that its use was associated with greater overall survival in ATM_low patients (8). Therefore, targeting the DNA damage response (DDR) pathway might be a promising strategy for treating gastric cancer with a DNA damage repair pathway defect.

ATM and ATR play essential roles in DDR by facilitating connections between DNA damage sensing and DDR effectors. In addition, these factors regulate cell-cycle progression by controlling the activations of Chk1 and Chk2 (9–11). Interestingly, ATM and ATR are differentially activated by distinct types of DNA damage despite the fact that they function as compensatory partners by sharing substrates. ATM is primarily activated by DSBs, whereas ATR responds to a much broader spectrum of DNA damage, especially during DNA replication (9, 12). In addition, ATR plays a role in the G₂–M checkpoint, and thus, a p53 mutation leading to checkpoint-defective cells is lethal in the presence of ATR depletion (13, 14).

Because ATR inhibition is likely to have greater deleterious effects on cancer cells, ATR pathway components are considered promising therapeutic targets. Previous studies have demonstrated ATR inhibition is effective for treating cancers with HRD (15–17). Most importantly, ATR participates in functional interactions between repair proteins, especially ATM, during DDR (9, 12). Although a previous study indicated that an ATR inhibitor had a synergistic antitumor effect on ATM- or p53-deficient cancer cells when administered in combination with cisplatin in colorectal and lung cancer cell lines (14), the antitumor activity and underlying mechanisms of ATR inhibitor monotherapy on ATM status remain unclear. For this reason, we considered study of the antitumor effects of targeting ATR and of the underlying mechanisms involved would help elucidate the therapeutic role of AZD6738 in cancer and interactions between DDR-associated molecules. In addition, such studies could result in novel treatment strategies that increase the scope of ATR inhibitors to the broader HRD cancer patient population.

Although gastric cancer is rarely associated with BRCA mutation, about 14% of patients show low ATM expression and about 10% show high microsatellite instability levels, which are associated with defective DNA damage repair (8, 18). In addition, the genomes of several gastric cancer cells are modulated by epigenetic alterations that can regulate DNA damage signaling (19–23). In the present study, we investigated the antitumor effects of an ATR inhibitor, AZD6738, in vitro using human gastric cancer cell lines, and an in vivo xenograft model. AZD6738 was found to inhibit the proliferation of gastric cancer cells with dysfunctional ATM by suppressing proliferative signaling and blocking cell-cycle progression in the S phase. Furthermore, AZD6738 disrupted HR-mediated DNA repair in sensitive cells, whereas ATR inhibition activated the ATM-Chk2 pathway to promote the repair of DNA damage induced by AZD6738 in insensitive cells with functional ATM. Although earlier this year Kwok and colleagues reported that ATR inhibition is selectively sensitive to TP53- or ATM deficiency in chronic lymphocytic leukemia (CLL) cells (24), it is difficult to understand the action mechanism of ATR inhibitor and the synthetic lethal interaction between ATR and ATM in solid tumors, especially in gastric cancer. This article reveals the mechanisms underlying the action of AZD6738 in gastric cancer cells and suggests a synthetic lethal interaction between ATR inhibition and ATM deficiency. In addition, our report meaningfully evaluated that ATR to ATM switch using ATR inhibitor, that is, ATR suppression, led to compensatory ATM activation. Furthermore, we also found ATM inactivation in sensitive cells was mediated by histone deacetylase 1 (HDAC1) deficiency. These results help to understand the mechanism underlying the antitumor effect of AZD6738 and provide a rationale for a clinical trial that has been initiated for treating solid tumors, including gastric cancer.

The ATR inhibitor AZD6738 and ATM inhibitor were kindly provided by AstraZeneca, Ltd. Cisplatin and paclitaxel were obtained from Choongwoe Co., Ltd., and Samyang Genex Co., Ltd. AZD6738 was dissolved in dimethyl sulfoxide, and cisplatin and paclitaxel were dissolved in normal saline as 10 mmol/L, and subsequently serial dilution was performed for specific concentration. The structure of AZD6738 has been previously published (25).

Human gastric cancer cells (SNU-1, -5, -16, -216, -484, -601, -620, -638, -668, -719, AGS, NCI-N87, MKN-45, and KATO-III) were purchased from the Korean Cell Line Bank. Identities of the cell lines were confirmed by DNA fingerprinting analysis (26). All cells were passaged for less than 6 months before use and maintained at 37°C with a 5% CO₂ atmosphere in RPMI 1640 supplemented with 10% FBS and 10 μg/mL gentamicin.

Antiproliferative effects of AZD6738 were assessed with an MTT assay as previously described (27). Cells were seeded in 96-well plates and exposed to increasing doses of AZD6738 (ranging from 0–1 μmol/L) for 5 days. Cell viability was evaluated by measuring the absorbance at 540 nm, and IC₅₀ values were analyzed using SigmaPlot software [Statistical Package for the Social Sciences, Inc. (SPSS)]. The combined effects of AZD6738 and chemotherapeutic agents were assessed using the methods previously described by Chou and Talalay (28). Any synergistic effects were defined by combination index (CI) values less than 1, whereas antagonism was identified by values more than 1.

The effect of AZD6738 on cell-cycle progression was evaluated using a FACS Calibur flow cytometer (BD PharMingen) by quantifying the DNA contents of the cells stained with propidium Iodide (PI) according to previously described protocols (29). The degree of apoptosis was measured using an Annexin V–binding assay in accordance with the manufacturer's instructions (BD PharMingen). For the BrdU incorporation experiments, 10 μmol/L of BrdU were added to the cells for 3 hours before harvest. The cells were then incubated with anti-BrdU antibody for 30 minutes, stained with 7-AAD, and then subjected to FACScan cytometry according to the manufacturer's protocol (BD PharMingen). The percentage of cells in the S phase was calculated by using BD FACSDIVA software (BD PharMingen).

Total cell proteins were extracted, and 20 μg proteins were separated on 5% to 15% SDS-PAGE gels as previously described (30). Primary antibodies against phosphorylated (p)-ATM (S1981), ATM, p-ATR, ATR, p-STAT3, STAT3, p-AKT, AKT, p-ERK, ERK, p-Chk1, Chk1, p-Chk2, Chk2, p-p53, p53, p21, Mre11, XRCC1, TS, caspase-3, and cyclin E were acquired from Cell Signaling Technology. Antibodies against HDAC1, p-histone H2A.X (Millipore), PARP (BD Biosciences), and actin (Sigma Aldrich) were also purchased.

HDAC1–ATM interaction was examined by immunoprecipitation. Cells were transfected with HDAC-specific siRNA or nonspecific control siRNA. After 3 days, total protein was extracted from the cells and then incubated with anti-HDAC1 or anti-IgG antibodies (negative control) and protein A/G plus agarose (Santa Cruz Biotechnology).

siRNA specific for ATM or HDAC1 as well as a nonspecific control was purchased from Genolution. Transfection was performed using G-fectin (Genolution) according to the manufacturer's protocol. The sequence of the ATM-specific siRNA was 5′-AACATACTACTCAAAGACATT-3′, and the sequence of the HDAC1-specific siRNA was 5′-GGCAAGUAUUAUGCUGUUATT-3′. The sequence of the nonspecific control siRNA was 5′-AATTCTCCGAACGTGTCACG-3′.

An immunofluorescence assay was performed as previously described (27). The primary antibodies used for this experiment were rabbit polyclonal anti-RAD51 (H-92; Santa Cruz Biotechnology), mouse monoclonal anti–p-ATM (Cell Signaling Technology), rabbit polyclonal anti-HDAC1, and mouse monoclonal anti–p-H2A.X (Millipore) at a dilution of 1:50. The coverslips were rinsed 3 times for 10 minutes in PBS followed by incubation with the appropriate fluorophore-conjugated secondary antibody at a dilution of 1:50 (Invitrogen). The cells were counterstained with DAPI (300 nmol/L; Invitrogen), and the coverslips were mounted on slides using Faramount aqueous mounting medium (Dako). Immunofluorescence was visualized using a Zeiss LSM 510 laser scanning microscope (Carl Zeiss). At least 100 cells were analyzed for each experiment, and the ones containing more than five foci of each molecule were counted.

The degree of DNA damage was determined with an alkaline comet assay using a commercial kit (Trevigen) according to the manufacturer's protocol. Tail lengths were measured using the Comet assay IV program (Andor Technology).

Animal experiments were carried out at the animal facility of Seoul National University (Seoul, Korea) in accordance with institutional guidelines and prior approval from the Institutional Animal Care and Use Committee. A total of 10 female Balb/c athymic nude mice of 5 weeks old were purchased from Central Lab Animal, Inc., and 1 × 10⁸ SNU-601 cells in 100 μL of PBS were injected subcutaneously into the right flank. After implantation of the tumor cells, sizes of the resulting tumors and body weight of each mouse were measured. When the tumor volume reached 200 mm³, the mice were randomly divided into two groups (5 mice per group). One group of mice was given 50 mg/kg AZD6738 every day for 20 consecutive days via oral gavage. The control group was treated with a 10% 2-hydroxyl-propyl-β-cyclodextrine/PBS solution alone. Tumor volume was calculated using the following formula: [(width)² × (height)]/2. At the end of the measurement period, the mice were euthanized with CO₂, and the tumors were excised for further analysis.

Immunohistochemistry and a TUNEL assay using paraffin-embedded xenograft tumor tissues were conducted as previously described (27).

All results are expressed as the mean ± SE and analyzed using SigmaPlot version 9.0 (Systat Software Inc.). A two-sided Student t test was performed when appropriate. P values <0.05 were considered statistically significant.

To assess the antiproliferative activity of AZD6738 (a novel ATR inhibitor), its growth-inhibitory effects were investigated in 14 gastric cancer cell lines using an MTT assay. The various gastric cancer cell lines showed different responses to AZD6738 (Supplementary Table S1 and Supplementary Fig. S1). SNU-601 was chosen as a sensitive cell line and SNU-484 as a resistant cell line for further investigation. Because ATR responds to various types of DNA damage that interferes with DNA replication and plays roles in intracellular signal pathways involved in cell proliferation, we investigated the downregulation of proliferative signaling after treating cells AZD6738 (11, 31). In sensitive SNU-601 cells, ATR inhibition dose-dependently induced the downregulations of proliferative signaling molecules, including AKT, STAT3, and ERK. In contrast, AKT, STAT3, and ERK were not influenced by ATR inhibition in insensitive SNU-484 cells (Fig. 1A).

ATR signaling activates downstream pathways that control cell-cycle arrest during the S to G₂ phase transition (32). Therefore, we investigated how ATR inhibition affects the cell-cycle progression of gastric cancer cells by FACS analysis. The S and sub-G₁ populations of SNU-601 cells were dramatically and dose-dependently increased by AZD6738 (Fig. 1B), and increased levels of cleaved PARP and caspase-3 along with γ-H2AX expression were consistent with FACS data. Inhibition of cell-cycle progression by AZD6738 also led to the downregulation of thymidylate synthase (TS), cyclin E, and p21 expression in SNU-601 cells (Fig. 1C). To confirm these observations, we evaluated percentages of apoptotic and BrdU-positive cells. The percentage of Annexin V–positive SNU-601 cells (indicating apoptotic death) was significantly increased by AZD6738 treatment (Fig. 1D) as was the percentage of cells in the S phase, whereas these increases were not observed in SNU-484 cells (Fig. 1E). These results indicated cell-cycle arrest, and the downregulations of proliferation signal pathways underlie the increased cell death shown by AZD6738-sensitive cells.

Because ATR is an essential component of HR repair, we hypothesized that ATR inhibition leads to reduced HR repair capacity and an accumulation of DNA damage. The overall expression of DDR-associated proteins was downregulated by ATR inhibition in SNU-601 cells (Supplementary Fig. S2A; Fig. 2A). Furthermore, DNA damage accumulation was also observed in SNU-601 cells treated with AZD6738, whereas no changes of DNA damage accumulation were observed in AZD6738-treated SNU-484 cells (Supplementary Fig. S2B; Fig. 2B). To verify whether DNA damage accumulation was caused by defective HR repair due to the inhibition of ATR activation, an immunofluorescence assay was used to assess RAD51 foci formation, which is indicative of functional HR repair activity. Numbers of RAD51 foci in SNU-601 cells were significantly lower than in insensitive SNU-484 cells, although degrees of DNA damage caused by hydroxyurea (HU) were comparable (Supplementary Fig. S2C; Fig. 2C). These observations support our hypothesis that ATR inhibition leads to the accumulation of DNA damage resulting from HR inactivation.

To understand why SNU-601 and SNU-484 cells differ in terms of AZD6738 sensitivity, we focused on the expression of p-Chk1 (Supplementary Fig. S2A; Fig. 2A). Although Chk1 was not activated, DNA damage did not accumulate in AZD6738-treated SNU-484 cells. Because ATM and ATR share a number of substrates, including p53, and function in a complementary manner (10, 11, 13, 33), we examined the protein expressions of ATM-Chk2 pathway factors. ATM-Chk2 axis protein expressions were downregulated in SNU-601 cells, but the axis was activated to repair the DNA damage induced by AZD6738 in SNU-484 cells (Supplementary Fig. S3; Fig. 3A). Conversely, ATM could not be activated in SNU-601 cells, even when DNA damage was induced by irradiation (Supplementary Fig. S4).

To determine AZD6738 sensitivity was a direct result of ATM inactivation, we measured the viability of ATM-depleted SNU-484 cells treated with AZD6738. ATM knockdown using siRNA and ATM inhibition using an ATM inhibitor enhanced AZD6738 sensitivity in SNU-484 cells (Fig. 3B and C, and Supplementary Fig. S5). AZD6738-induced apoptosis and S phase cell-cycle arrest were increased in cells with siRNA-mediated or chemically downregulated ATM compared with cells transfected with control siRNA (Fig. 3D). Furthermore, the number of RAD51 foci in ATM-defective cells was significantly reduced at sites of DNA damage (Fig. 3E). Our results show ATM deficiency or inactivation might be a predictive marker for AZD6738 sensitivity in gastric cancer cells.

Previous studies have suggested HDAC1 plays a major role in ATM activation and expression (22, 23), and HDAC1 depletion is known to suppress HR repair by inducing ATM inactivation (23, 34). To explain why ATM is inactivated in SNU-601 cells, despite unaffected total ATM levels, we examined HDAC1 expression levels in SNU-601 and SNU-484 cells. HDAC1 expression was much lower in SNU-601 cells compared with SNU-484 cells (Fig. 4A). To determine whether HDAC1 silencing enhanced AZD6738 sensitivity by modulating ATM activation, we measured the viability of siRNA-transfected HDAC1-depleted SNU-484 cells treated with AZD6738 using an MTT assay. HDAC1 knockdown was found to restore AZD6738 sensitivity in SNU-484 cells (Fig. 4B). In addition, we found that HDAC1 expression was inversely correlated with sensitivity to AZD6738 in gastric cancer cells (Supplementary Fig. S6). Furthermore, IFA assays showed HDAC1 colocalized with p-ATM in SNU-484 cells, and that HDAC1 knockdown significantly reduced p-ATM expression (Fig. 4C). An immunoprecipitation assay confirmed interaction between HDAC1 and ATM (Fig. 4D). These results indicate HDAC1 deficiency leads to ATM inactivation and AZD6738 sensitivity.

To determine whether inhibition of ATR effectively inhibits in vivo tumor growth, we utilized a SNU-601 xenograft model. The tumor volumes in mice administered AZD6738 (50 mg/kg daily) were significantly smaller than in vehicle control (Fig. 5A), and at this dose, AZD6738 was well tolerated (Fig. 5B). Furthermore, Ki-67 expression (an indicator of proliferation) was lower in AZD6738-treated mice than in nontreated controls, indicating lower proliferative ability in AZD6738 treated mice, and TUNEL assay showed AZD6738 also increased numbers of apoptotic cells (Fig. 5C). These results were confirmed by observations of reduced expressions of proteins related to proliferation and of increased PARP cleavage following AZD6738 treatment (Fig. 5D). These findings reinforce the notion that ATR inhibition significantly suppresses cell proliferation and promotes apoptosis in vivo.

Because platinum and 5-fluorouracil are commonly used as a first-line chemotherapy and weekly paclitaxel as a second-line chemotherapy for gastric cancer, we hypothesized that the inhibition of ATR by AZD6738 might enhance the efficacies of cytotoxic chemotherapeutics. Possible synergistic interactions between AZD6738 and paclitaxel or cisplatin were examined according to the Chou–Talalay median effect principle (28, 35). As was expected, AZD6738 sensitized some gastric cancer cells to cisplatin and/or paclitaxel (Tables 1A and 1B), which suggests AZD6738 has therapeutic potential alone and in combination with established chemotherapeutics.

Because ATM and ATR are essential regulators of responses to DNA damage including DSBs and replication stress, techniques for targeting ATM and ATR using highly selective small-molecule inhibitors are currently under preclinical and clinical development. In preclinical studies, many efforts have been made to identify the marker, exhibits synthetic lethal interaction with ATR inhibitors (15, 17, 24, 36). Recent reports have suggested that ATR inhibitors are effective in cells with impaired HR activities due to p53 or ERCC1 deficiency (15, 36). Although ATR inhibition is selectively effective in HR-defective cancers, malignancies lacking specific markers of HR deficiency represent only a small proportion of cancers. Furthermore, the effects of ATR inhibitors in solid tumors, including gastric cancer, and underlying mechanisms are not fully understood, and thus, there is an urgent need to evaluate the effects of ATR inhibitors on gastric cancer. In the current study, we assessed the antitumor effects of an ATR inhibitor, AZD6738, on gastric cancer cells and in a xenograft mouse model. We also explored whether ATR inhibition could increase the effectiveness of the chemotherapeutic agents used to treat gastric cancer. The results of this study might aid the selection of combinatorial regimens containing an ATR inhibitor, and present a strong rationale for conducting clinical trials in gastric cancer.

In a previous study, we found that gastric cancer cells exhibited heterogeneous responses to a PARP inhibitor and exhibited different DDR abilities (27). Although both ATR and PARP inhibitors target HR-defective cancer, the gastric cancer cell lines evaluated in both present and previous studies responded differently to these inhibitors, suggesting ATR and PARP are inhibited by different mechanisms. The antitumor effects of PARP inhibitors are based on the increased levels of genomic instability genomic instability due to DSBs induced by SSB accumulations, whereas those of ATR inhibitors are due to DNA damage accumulation due to replication stress. Because high replication stress is a characteristic of cancer, it would appear ATR inhibitor monotherapy is likely to be more effective than PARP inhibitor monotherapy. According to our data, gastric cancer cell lines exposed to ATR inhibitor in short term showed a greater antiproliferative effect than those exposed to PARP inhibitor. Moreover, unlike PARP inhibitor, ATR inhibitor directly downregulates DDR molecules, such as, Chk1, Mre11, and ERCC1, which eventually leads to the creation of cells mimicking the HR-deficient phenotype and resulting in accumulation of genomic instability.

Interestingly, in the present study, we observed that the ATM-Chk2 signaling pathway was activated when ATR activity was blocked in insensitive cells. ATM and ATR have been reported to play critical roles in DDR with overlapping functions in a partnership- and time-dependent manner (11), but their cross-regulation of the ATM-Chk2 and ATR-Chk1 pathways is very rare. However, a switch from ATM to ATR signaling and of Chk2 phosphorylation by ATR have been reported (12, 37, 38), and suggest that attenuated ATM activation potentiates ATR activation. Based on these reports, it is evident ATM and ATR function in a reciprocal manner, which is in-line with our result that ATR suppression promotes repair of DNA damage via ATM activation. Therefore, when ATM activation is impaired, ATM cannot adequately repair the DNA damage induced by ATR inhibition and results in cell death. This study also shows that the correlation between ATM activity, which was evaluated in gastric cancer cell lines by Kubota and colleagues (39), and the sensitivity of ATR inhibitor is statically significant and biologically meaningful (Supplementary Fig. S7). Our data indicate ATR inhibition has a synthetic lethal interaction with ATM deficiency, and that the presence of dysfunctional ATM might predict gastric cancer cell sensitivity to AZD6738. Interestingly, ATM is one of the most frequently mutated kinases in human cancers (40), and genetic alterations in ATM have been reported 12.2% of 287 gastric tumor samples (21), and low ATM expression has been reported in 14% of gastric cancer patients (8). The lethal effect of AZD6738 in the presence of ATM deficiency indicates ATR inhibition offers the possibility of highly attractive, effective therapeutic strategy for gastric cancer with ATM deficiency.

Recent studies have reported HDACs are involved in DDR by regulating the expressions of HR repair–associated genes (41, 42). In particular, HDACs play a critical role in mitigating ATM pathway response to DNA damage, because ATM is a substrate of HDAC (22, 23, 34). Furthermore, functional ATM activation was detected in SNU-484 cells, and not in SNU-601 cells, after IR-induced DNA damage, and it has been demonstrated HDAC1 regulates ATM activity and that HDAC1 depletion is sufficient to modulate ATM activation in response to DNA damage (22, 23, 41). These reports support our result that HDAC1 deficiency led to ATM inactivation and lethality when ATR was inhibited. HDAC1 is a well-known target of enhancer of zeste homolog 2 (Ezh2), and Ezh2 downregulation increases HDAC1 expression levels in gastric cancer. In addition, high levels of Ezh2 are frequently observed in gastric cancer tissues, and SNU-601 cells produce high levels of Ezh2 (43). These observations suggest HDAC1 depletion is caused by epigenetic modulation associated with high levels of Ezh2 expression. These results hint at a new cancer treatment strategy involving the administration of ATR inhibitor in combination with HDAC inhibitors. During our studies, we also observed the antitumor effect of AZD6738 in combination with SAHA in gastric cancer cells (data not shown). However, HDAC inhibition using small-molecule inhibitors affects numerous DNA repair factors involved at multiple levels of DNA repair pathways, and thus, side effects on normal tissues should be considered before using this combination. Taken together, our results show HDAC1 deficiency modulates ATM activity and confers sensitivity to AZD6738-induced DNA damage.

This is first study to evaluate the antitumor effects of AZD6738 on human gastric cancer cells and in a mouse model. Our findings suggest ATM activation is the main mechanism of resistance to AZD6738 and show ATM and ATR act in a compensatory manner. The study also demonstrates that AZD6738 attenuates ATR activity and induces ATM activation, and thus, promotes an ATR to ATM switch in the presence of DNA damage. These findings show the interaction between AZD6738 has a synthetic lethal interaction with ATM defect in gastric cancer cells and that ATM inactivation in ATM dysfunctional SNU-601 cells is due to HDAC1 deficiency. We believe that our findings have potential clinical implications for the treatment of ATM-defective gastric cancer, increase understanding of the mechanisms governing the action of AZD6738 alone and in combination with chemotherapeutics, and provide a rationale for present and future clinical trials.

S.-A. Im is a consultant/advisory board member for AstraZeneca, Novartis, Hanmi, and Spectrum. Y.-J. Bang is a consultant/advisory board member for AstraZeneca. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Min, S.-A. Im, K.-H. Lee, D.-Y. Oh, J. Brown, A. Lau, M.J. O'Connor, Y.-J. Bang

Development of methodology: A. Min, S.-A. Im, K.-H. Lee

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Min, S.-A. Im, H. Jang, H.-J. Kim

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Min, S.-A. Im, Y. Yang, K.-H. Lee, D.-Y. Oh, A. Lau, Y.-J. Bang

Writing, review, and/or revision of the manuscript: A. Min, S.-A. Im, D.K. Kim, Y. Yang, K.-H. Lee, J.W. Kim, T.-Y. Kim, J. Brown, A. Lau, Y.-J. Bang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Min, H. Jang, S. Kim, D.K. Kim, H.-J. Kim, J. Brown

Study supervision: S.-A. Im, T.-Y. Kim, Y.-J. Bang

Other (assisted the experiment, especially performed in vivo experiment and also wrote and reviewed the article): M. Lee

The ATR inhibitor AZD6738 and ATM inhibitor were kindly provided by AstraZeneca, Ltd.

This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT, and Future Planning (2015R1A2A2A01004655; to S.-A. Im). This work was also funded by Basic Science Research Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Science, ICT & Future Planning(2014R1A1A3052365; to H.-J. Kim). We thank to AstraZeneca for supporting this study through AZ-KHIDI Oncology Research Program.

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.