Cancer cell line profiling to identify previously unrecognized kinase dependencies revealed a novel nonmutational dependency on the DNA damage response checkpoint kinase Chk1. Although Chk1 is a promising therapeutic target in p53-deficient cancers, we found that Ras–MEK signaling engages Chk1 in a subset of osteosarcoma, ovarian, and breast cancer cells to enable their survival upon DNA damage, irrespective of p53 mutation status. Mechanistically, Ras–MEK signaling drives Chk1 expression and promotes cancer cell growth that produces genotoxic stress that requires Chk1 to mediate a response to the consequent DNA damage. Reciprocally, Chk1 engages a negative feedback loop to prevent hyperactivation of Ras–MEK signaling, thereby limiting DNA damage. Furthermore, exogenous DNA damage promotes Chk1 dependency, and pharmacologic Chk1 inhibition combined with genotoxic chemotherapy potentiates a DNA damage response and tumor cell killing. These findings reveal a mechanism-based diagnostic strategy to identify cancer patients that may benefit from Chk1-targeted therapy. Mol Cancer Ther; 16(4); 694–704. ©2017 AACR.

This article is featured in Highlights of This Issue, p. 553

The successful clinical development of several “rationally targeted” drug treatments for a variety of human cancers has been largely enabled by the discovery of recurrent tumor mutations that are associated with pathway dependency, often described as “oncogene addiction.” In many cases, these dependencies are associated with the activation of kinase-mediated signaling pathways, and direct targeting of the mutationally activated kinase, or a proximal signaling pathway component, has proven to be an effective clinical strategy in the context of personalized cancer treatment (1, 2). Among the more than 500 kinases encoded by the human genome, although many have been implicated in cancer in various preclinical studies, only a small number of them have been found to be recurrently activated by mutation in tumors, raising the question as to whether the paradigm of targeting “kinase addiction” is somewhat limited in scope as a cancer therapeutic strategy.

Drug sensitivity profiling of cultured tumor-derived cell lines has proven to be an efficient approach to the identification of potentially treatment-sensitive cancer subsets, and in many cases, these subsets are associated with specific genomic features that can inform a diagnostic strategy for patient stratification. Such profiling has demonstrated a strong genotype-associated sensitivity to various targeted inhibitors in cancer cells addicted to mutationally activated kinases, including BCR-ABL, PDGFR, BRAF, EGFR, HER2, PI-3K, MET, FGFR, and ALK (3–5).

Notably, drug sensitivity profiling has also revealed subsets of cancer cells that can be grouped according to their sensitivity to a particular kinase inhibitor, but without evidence of a common genomic lesion that explains their seemingly related dependency. Such findings may reflect the phenomenon of “nononcogene addiction,” wherein a variety of genes and pathways are not oncogenic themselves, but are essential for supporting maintenance of the malignant state, thereby possibly expanding opportunities for targeted therapeutics to drive mutation-independent vulnerabilities (6). In this context, large-scale profiling of tumor-derived cell lines could provide an opportunity to discover targets for candidate anticancer therapeutics in a variety of nononcogene-addicted tumors. Here, using high-throughput tumor cell line profiling of sensitivity to several kinase inhibitors, further informed by proteomic and transcriptomic approaches, we observed an unexpected sensitivity profile among subtypes of bone, ovarian, and triple-negative breast cancers to an inhibitor of the DNA damage response (DDR) checkpoint kinase Chk1 and established an associated nonmutational mechanism that may be useful in guiding patient stratification for treatment.

Cancer cell lines, reagents, and high-throughput tumor cell line screening

All cancer cell lines were from the ATCC. Cell line authentication was routinely conducted by SNP-based genotyping at the Genentech cell line core facility. Cell lines were maintained in RPMI1640 or DMEM/F12 (GIBCO) supplemented with 10% FBS (Sigma), 50 U/mL penicillin, 50 U/mL streptomycin, and 2 mmol/L l-glutamine (Gibco). Drug treatment was performed in 5% FBS. Gemcitabine was from Selleckchem. GDC-0425 and cobimetinib were synthesized at Genentech. All other reagents were from LC Laboratories. For the high-throughput tumor cell line screening, drug sensitivity of cancer cell lines was assessed using an automated platform as described previously (3). Cells were treated with 1 μmol/L Gö6976 or various concentrations of GDC-0425 for 72 hours and then assayed for cell viability using either SYTO60 (Invitrogen) or Cell Titer-Glo (Promega).

Immunoblot analysis

Cells were harvested in RIPA lysis buffer (Sigma) containing a protease inhibitor cocktail (Roche). Lysates were then analyzed for immunoblotting. Antibodies against pChk1, c-PARP, pp53, γH2A.X, pERK, ERK, pan-Ras, pRAD50, pNBS1, pATM/ATR substrate, and pDNA-PKcs were from Cell Signaling Technology. Chk1, Chk2, and cdc25 antibodies were from Santa Cruz Biotechnology. Isoform-specific antibodies used for Ras were anti-HRAS (MAB3291) from Millipore, KRAS (sc-30), and NRAS (sc-31) from Santa Cruz Biotechnology.

Kinase inhibitor target profiling

Cell extracts from U-2 OS cells were subjected to affinity chromatography using KinAffinity beads (Evotec) representing a set of various broad-spectrum kinase inhibitors immobilized on sepharose beads and treated with various concentrations of Gö6976 and sotrastaurin. Target identification was performed by LC/MS-MS.

Cell-cycle analysis

Cell-cycle analysis was performed using flow cytometry. BrdU Flow Kits (Becton Dickinson) were used according to the manufacturer's instructions. Briefly, cells were cultured with 10 μmol/L bromodeoxyuridine (BrdUrd) for 30 minutes. Cells were then washed with cold PBS, fixed with Cytofix/Cytoperm buffer, treated with DNase, and labeled with FITC-BrdUrd antibody. Finally, cells were suspended with 7-aminoactinomycin (7-AAD) to stain total DNA and analyzed by flow cytometry.

RNA sequencing

RNA sequencing (RNA-seq) was performed as described previously (7). Briefly, RNA-seq was performed by the Illumina HiSeq2000 system using the standard paired-end protocol. Approximately 20 to 30 million 75-bp paired reads were obtained and mapped to the human genome by GSNAP (8). Differentially expressed gene analysis was performed using the DEseq package from Bioconductor (9). These data have been deposited in European Genome-phenome Archive and are accessible through EGA with accession number of EGAS00001000610.

Phospho-antibody array analysis

Phospho-antibody array analysis was performed using the Proteome Profiler Kit ARY003B (R&D Systems) according to the manufacturer's instructions. Briefly, cells were treated with kinase inhibitors for various time points. Cells were then lysed and 500 μg of cellular extract was subjected to a protein array after centrifugation at 14,000 × g for 10 minutes. Phosphorylated kinases were detected by incubating arrays with biotinylated detection antibodies, streptavidin–HRP antibodies, and chemiluminescent detection reagents.

Xenograft studies

All procedures involving animals were reviewed and approved by the Institutional Animal Care and Use Committee at Genentech and carried out in an AAALAC (Association for the Assessment and Accreditation of Laboratory Animal Care) accredited facility. Cells (1 × 107) from the indicated cell lines were implanted subcutaneously into the right flanks of NCr nude mice. When tumors reached approximately 200 mm3, mice were randomized into 4 arms of 5 mice for 143B PML BK TK cells, or 6 arms of 5 mice each for HCC1806 and HCC70 cells. For the 4-arm study, mice were treated with vehicle, gemcitabine 120 mg/kg, GDC-0425 75 mg/kg alone, or gemcitabine and GDC-0425 combination for 15 days. For 6-arm studies of HCC1806 and HCC70 models, mice were treated with vehicle, gemcitabine 120 mg/kg, GDC-0425 50 mg/kg, GDC-0425 75 mg/kg alone, or gemcitabine and GDC-0425 combination until tumor sizes in the vehicle control group reached 1,200 mm3. For cotreatment studies, GDC-0425 was orally administrated at 24, 48, and 72 hours after gemcitabine administration by intraperitoneal injection. Tumor volumes were determined using digital calipers using the formula (L × W × W)/2.

Cancer cell line profiling reveals selective sensitivity of osteosarcoma cells to Chk1 kinase inhibition

In a high-throughput screen of 705 human cancer cell lines derived from a variety of tumor types for sensitivity to a panel of kinase inhibitors, we observed that 5 of 30 bone cancer–derived cell lines were more than 70% growth inhibited and 10 cell lines were more than 50% growth inhibited by 1 μmol/L Gö6976, a multitargeted kinase inhibitor that is also widely used as a pan-PKC inhibitor. Notably, the proportion of sensitive cell lines from all other tissues that were growth inhibited greater than 70% is significantly lower than that from bone tissue, suggesting a significant and specific vulnerability of bone cancers to Gö6976 (Fig. 1A). As Gö6976 is primarily used for research purposes as a potent PKC inhibitor, we determined whether PKC signaling is required for growth of the Gö6976-sensitive cells. Sotrastaurin, a structurally similar potent pan-PKC inhibitor (Supplementary Fig. S1A), failed to suppress the growth of Gö6976 responders, such as U-2 OS and 143B PML BK TK cells, excluding PKC as a relevant target of Gö6976 in this context (Supplementary Fig. S1B).

To identify the relevant target of Gö6976, we performed kinome profiling in a binding competition screen in U-2 OS cell lysates using a mixture of immobilized kinase inhibitors to a matrix that enriches kinases. Kinases that bind to the master matrix were subjected to competitive binding with either Gö6976 or sotrastaurin addition and were then identified by LC/MS-MS. Despite the structural similarity between the two compounds, there were only three kinases among 24 hits detected in the submicromolar range of Gö6976 whose release was promoted by both compounds, including kinases involved in PKC signaling. Notably, Gö6976 inhibits kinases downstream of PKCs, such as, PRKD1, 2, and 3, while sotrastaurin directly inhibits PKCs including PRKDA, D, E, Q, and H (Fig. 1B; Supplementary Fig. S1C).

We then sought to determine which kinase of 21 hits from the Gö6976-only list was responsible for the observed Gö6976-induced growth inhibition. Thus, RNAi was used to deplete each of the identified kinase hits, followed by cell viability measurement in three Gö6976-sensitive osteosarcoma lines. This analysis demonstrated that only Chk1 knockdown significantly suppressed the growth of all three cell lines, consistent with previous studies indicating that Gö6976 can inhibit Chk1 (Fig. 1C; ref. 10). Chk1 expression was higher in the Gö6976-sensitive cell lines, indicating a correlation between Chk1 expression and Chk1 dependency (Fig. 1D; Supplementary Fig. S1D). To determine whether Chk1 inhibition is cytotoxic or cytostatic, three Gö6976-sensitive and two refractory cell lines were transfected with Chk1 siRNA, and PARP cleavage was measured. All of the Gö6976-sensitive cell lines showed increased PARP cleavage, indicative of apoptotic cell death, while PARP remained intact in the two Gö6976-refractory cell lines (Fig. 1E). We then confirmed the requirement for Chk1 using AZD-7762, a selective Chk1/2 kinase inhibitor, which similarly promoted growth inhibition and caspase activation in Gö6976-sensitive cell lines (Fig. 1F; Supplementary Fig. S1E; Supplementary Methods).

Hyperphosphorylation of Chk1 upon Chk1 inhibition has previously been reported as a consequence of a negative feedback loop involving regulation by protein phosphatase 2A of ATR-mediated Chk1 phosphorylation on serine 345 (11). To examine the potency of Gö6976 on Chk1 in comparison with GDC-0425, another investigational Chk1 inhibitor with greater Chk1 selectivity than AZD-7762 (12), we first performed a dose–response study with Gö6976 and GDC-0425. Six hours of drug treatment at 3 μmol/L of both inhibitors was sufficient to inhibit Chk1, as measured by hyperphosphorylation of Chk1. Notably, both inhibitors significantly increased phosphorylation of p53 and H2AX, indicating DNA damage (Supplementary Fig. S1F). In addition, we determined whether Chk1 inhibition–mediated hyperphosphorylation of Chk1 affected cell death. A selective ATR or ATM inhibitor significantly suppressed hyperphosphorylation of Chk1 by GDC-0425 treatment. However, both inhibitors, either alone or together, failed to block PARP cleavage, whereas a pan-caspase inhibitor sufficiently blocked PARP cleavage (Fig. 1G), suggesting that hyperphosphorylation of Chk1 only serves as an indicator of Chk1 inhibition. Together, these findings indicate that Chk1 is the relevant target of Gö6976 in Gö6976-sensitive osteosarcoma cells where Chk1 inhibition induces apoptotic cell death possibly via a DDR.

Sensitivity to Chk1 inhibition is p53-independent in osteosarcoma cells

Previous studies have implicated p53 deficiency in the response to combination treatment with Chk1 inhibitors and DNA-damaging agents (13–15). Significantly, osteosarcoma lines demonstrating sensitivity to Chk1 inhibition harbor wild-type p53, and we observed synergistic cell killing activity by assessing the fractional response of a combination of two drugs (Bliss Independence; see Supplementary Methods) following cotreatment with a Chk1 kinase inhibitor and the DNA-damaging chemotherapy drug gemcitabine in U-2 OS cells (Supplementary Fig. S1G). This prompted us to examine the status of the Chk1 and p53 pathways in treatment-sensitive cells.

On the basis of the observed induction of p53 phosphorylation by Chk1 inhibition (Supplementary Fig. S1F), we speculated that p53 might be activated by a DDR caused by Chk1 inhibition. We first determined whether p53 signaling affects the sensitivity of cells to gemcitabine alone or in combination with GDC-0425. Using CRISPR gene knockout technology, p53 knockout cell lines were generated from U-2 OS and 143B PML BK TK cells (Fig. 1H). Surprisingly, p53 ablation did not confer sensitivity on these cells to either gemcitabine or GDC-0425 alone or in combination (Supplementary Fig. S1H; Fig. 1I). This suggests that p53 activation by Chk1 inhibition may be a consequence of the DDR and that an unknown p53-independent mechanism underlies the observed growth inhibition of Chk1-dependent osteosarcoma cells.

Ovarian and basal breast cancer cells with elevated Chk1 are sensitive to Chk1 inhibition

To expand the Chk1 dependency observations in bone cancer to other tissues, we performed a second high-throughput screen of 723 cancer cell lines derived from diverse tissue types for sensitivity to the more selective Chk1 inhibitor GDC-0425 (Supplementary Fig. S2A). Analysis of the association between Chk1 expression and sensitivity to GDC-0425 demonstrated that ovarian cancers exhibited significant differential sensitivity between Chk1-high and low expressers (Fig. 2A; Supplementary Fig. S2G). This was further confirmed by immunoblotting for Chk1 expression in a panel of ovarian cancers (Supplementary Fig. S2B). A similar trend was observed among breast cancers (not statistically significant, Supplementary Fig. S2C), while other tissues did not show such differences (Supplementary Fig. S2D). Correlation scatter plots for Chk1 expression and sensitivity in ovarian and breast tissues also demonstrated that the correlation is much higher in ovarian cancers (Supplementary Fig. S2E and S2F). An apoptosis assay showed that drug-sensitive ovarian cancer cells underwent PARP cleavage upon GDC-0425 treatment, as well as induction of pChk1 and γH2AX, suggesting an apoptotic death caused by DNA damage upon Chk1 inhibition (Fig. 2B).

As breast cancer cell lines showed a similar trend, we determined whether stratifying these cell lines on the basis of well-established subtypes could reveal a disease-specific association between Chk1 expression and sensitivity to Chk1 inhibition (16, 17). Indeed, subtype analysis showed that among the subclasses of breast cancers, Chk1 expression in basal-type cancer cells was specifically elevated (Fig. 2C; Supplementary Fig. S2H and S2I). Consistent with this observation, Chk1 RNA levels were found to be significantly higher within the basal subtype of human breast tumors within TCGA data (Fig. 2D). Further analysis of GDC-0425 sensitivity demonstrated that basal cells were significantly more sensitive to GDC-0425 (54.5%) than luminal (18%) or Her-2+ cells (25%; Fig. 2E; Supplementary Fig. S2J). A broad comparison of the efficacy of GDC-0425 in Chk1-high versus low cell lines in three tissue types suggested that Chk1 expression was well correlated with GDC-0425–induced apoptosis (Fig. 2F). Further cell-cycle analysis using BrdUrd incorporation and 7-AAD revealed a GDC-0425–induced reduction in the S-phase and increase in the G0–G1 phase population in both drug-sensitive (Fig. 2G) and resistant cell lines (Fig. 2H), except for SW1353 cells. However, GDC-0425 induced a large increase in the sub-G0–G1 population, together with PARP cleavage (Fig. 2F), only in sensitive cell lines, indicating apoptotic cell death (Fig. 2I).

Feed-forward Chk1 activation by Ras–MEK signaling in Chk1-dependent cancer cells

Although we had observed a correlation between Chk1 expression and dependency, there were “outliers” among the Chk1-high expressers that were refractory to GDC-0425 across the subtypes in breast cancers (Supplementary Fig. S2H and S2J). This prompted us to further investigate additional signatures of Chk1 dependency. As we previously demonstrated a p53-independent sensitivity to Chk1 inhibition in osteosarcoma, we selected p53-mutant breast cancers to rule out a role for p53, and to explore the mechanism underlying GDC-0425 sensitivity in this context. Thus, we tested 6 each of GDC-0425–sensitive and refractory cell lines, which all harbor p53 mutations and exhibit relatively high Chk1 expression (Supplementary Fig. S3A; Fig. 3A). Gene set enrichment analysis (GSEA) comparing the sensitive versus refractory groups, followed by ingenuity upstream regulator analysis, suggested that Ras/ERK signaling is differentially activated in the sensitive group. Consistent with these findings, hierarchical unsupervised gene clustering of genes previously associated with the functional output of the MAPK pathway demonstrated a marked separation between the sensitive and refractory groups (Supplementary Fig. S3B; ref. 18). In addition, pERK expression in the GDC-0425–sensitive group was significantly higher than in the resistant group (Supplementary Fig. S3C). Furthermore, a number of receptor tyrosine kinases (RTK) were identified as upstream regulators, implicating RTK-mediated Ras activation (Fig. 3B). Notably, the same approach in ovarian cancers also demonstrated that the Ras/ERK pathway is significantly activated in a GDC-0425–sensitive group, regardless of p53 mutation status, suggesting that these two models share mechanisms of Chk1 dependency related to Ras/ERK (Supplementary Fig. S3D and S3E).

As GDC-0425 sensitivity in breast and ovarian cell lines tested by GSEA correlates with Ras signaling, we further explored the possibility of cross-talk between the Chk1 and Ras signaling pathways. First, to determine whether Ras signaling regulates Chk1, cells were treated with cobimetinib, a selective MEK kinase inhibitor. Notably, Chk1-dependent cells exhibited a dramatic decrease in Chk1 expression that was not significantly changed in Chk1-independent MB-453 cells (Fig. 3C). To determine which Ras isoform is responsible for Chk1 expression, RNAi was used to knockdown K-, N-, or H-Ras. Knockdown of each Ras isoform partially suppressed Chk1 expression in JIMT-1 cells, suggesting that signaling from multiple wild-type Ras isoforms converges on Chk1 (Supplementary Fig. S3F). Then, we determined whether Ras signaling regulates Chk1 gene transcription or protein expression. qPCR analysis indicated that Chk1 mRNA was significantly suppressed by cobimetinib (Fig. 3D). Further analysis with cycloheximide, a protein synthesis inhibitor, in the presence or absence of cobimetinib showed additive suppression of Chk1 expression by cobimetinib/cycloheximide, suggesting that Ras-mediated Chk1 expression is regulated at the transcriptional level (Supplementary Fig. S3G).

We next determined whether inhibition of Ras–MEK signaling affects the sensitivity of cells to Chk1 inhibition. Treatment with single-agent cobimetinib or AZD6244, clinical MEK kinase inhibitors, suppressed cell growth without inducing apoptosis. However, MEK inhibition either by pharmacologic inhibitors or RNAi-mediated gene silencing significantly protected cells from reduced viability upon GDC-0425 treatment (Fig. 3E and F; and Supplementary Fig. S3H). This prompted us to determine whether pretreatment with cobimetinib could render cells more resistant to Chk1 inhibition. To address this question, we first asked whether suppression of Chk1 expression by cobimetinib is reversible. Cells were treated with cobimetinib for 6 days and either maintained in cobimetinib or maintained drug free for an additional 3 days. Indeed, drug wash-out transiently restored Chk1 expression, implicating a feed-forward regulation of Ras on Chk1 (Fig. 3G). As expected, a cell viability assay showed that suppression of Ras–MEK signaling increased resistance to GDC-0425, while reactivation of Ras-MEK restored sensitivity (Fig. 3H and I). Collectively, these observations suggest that Ras–MEK signaling is required for Chk1 dependency in a subset of cancer cells.

Chk1 promotes negative feedback to Ras–MEK signaling

To explore the possibility of reciprocal regulation of Chk1 on the Ras signaling pathway, multiple cell lines from each group were treated with GDC-0425, and pERK levels were measured. Surprisingly, pERK was strongly increased in sensitive cells as early as 1 hour after treatment and either rebounded at the 8-hour time point or was maintained throughout the time course, while pERK was not modulated until 24 hours after treatment of the GDC-0425–refractory cell lines CAL-148 or MDA-MB-453 (MB-453; Fig. 4A and B; Supplementary Fig. S4A). Phosphorylation profiling of 43 arrayed kinases in JIMT-1 cells treated with either GDC-0425 or AZD-7762 for 1 hour or U-2 OS cells treated with Gö6976 for 0.5 to 3 hours revealed significant pERK elevation, while phosphorylation of other kinases remained unchanged, confirming a transient and specific feedback activation of ERK by Chk1 inhibition (Fig. 4C; Supplementary Fig. S4B). We then asked whether the observed pERK increase is mediated by Ras activation. “Pull-down” assays to detect active Ras demonstrated a significant increase in Ras binding to recombinant Raf-1 protein in the presence of GDC-0425 (Fig. 4D). Together with the feed-forward regulation, these findings suggest that a feedback loop between Ras–MEK and Chk1 is selectively activated in Chk1-dependent cells.

Ras–MEK promotes a DDR that requires Chk1

As described above, we observed feedback inhibition of Ras–MEK signaling by Chk1 and that reciprocal Chk1 inhibition promoted Ras activation and a DDR (Fig. 4D; Supplementary Fig. S1F). Therefore, we asked whether inhibition of Ras signaling in the presence of GDC-0425 would suppress the DDR. Indeed, cobimetinib reduced activation of the DDR pathway by Chk1 inhibition, suggesting that feedback Ras activation caused by Chk1 inhibition augments DNA damage (Fig. 5A; Supplementary Fig. S5).

As Chk1 inhibitor–sensitive osteosarcoma and ovarian cancer cells exhibited increased γH2AX, we asked whether Chk1-dependent breast lines accumulate significant DNA damage upon Chk1 inhibition. Indeed, GDC-0425 treatment led to hyperphosphorylation of Chk1, as expected, and induced significant DNA damage, indicated by activation of ATM and DNA–PK signaling, specifically in GDC-0425–sensitive cells (Fig. 5B). This suggests a critical role of Chk1 in regulating the DDR in Chk1-dependent cells. We then explored whether Chk1-dependent cells exhibit increased DNA repair activity through Chk1. Thus, two dependent cell lines and one independent line were exposed to gemcitabine for one day, after which time the drug was removed for the following 3 days. A time course assessment of DNA damage revealed more efficient damage recovery, with hyperactivation of Chk1, in MDA-MB-231 and BT-549 cells relative to MDA-MB-453 cells, suggesting that Chk1-dependent cells are less prone to DNA damage (Fig. 5C).

Combining Chk1 inhibition with DNA damage potentiates antitumor activity

As previous studies had shown that Chk1 inhibitors sensitize cancer cells to chemotherapy (15), we tested the hypothesis that Chk1 inhibition could promote chemopotentiation, especially in the context of Chk1-dependent cells. Chk1-dependent cell lines from osteosarcoma, ovarian, and breast cancers, and a Chk1-independent cell line, were treated with gemcitabine and GDC-0425. Strikingly, combinatorial treatment of gemcitabine and GDC-0425 potentiated DNA damage and apoptotic cell death in all Chk1-dependent cell lines, while the Chk1-independent MDA-MB-453 cells were much less affected by the combination (Fig. 6A; Supplementary Fig. S6A). Additional viability assays of Chk1-dependent breast cancer cells and Chk1-independent MDA-MB-453 cells confirmed that IC50 shifts of GDC-0425 for gemcitabine were greater in Chk1-dependent cells than in Chk1-independent cells, suggesting that Chk1-dependent cells respond synergistically to gemcitabine (Fig. 6B; Supplementary Fig. S6B and S6C).

We then extended these cell culture findings to in vivo efficacy studies. Xenografts of both osteosarcoma and triple-negative breast cancer models (143B PML BK TK, HCC1806, and HCC70 cell lines) exhibited partial suppression of tumor growth upon treatment with either gemcitabine or GDC-0425 alone. Notably, the gemcitabine/GDC-0425 combination resulted in significant tumor regression in all tested models (Fig. 6C–E). The combination of gemcitabine and GDC-0425 did not cause significant body weight loss in both 143B PML BK TK and HCC1806 models, demonstrating the tolerability of this combination in vivo (Supplementary Fig. S6D and S6E).

We have observed that a subset of cancer cells derived from bone, ovarian, and breast tumors require Ras–MEK signaling to maintain cell growth and require Chk1 to manage the repair of DNA damage that is generated in these cells. Ras–MEK signaling transcriptionally activates Chk1, which appears to sustain cancer cell growth by maintaining DNA damage levels below a threshold that would otherwise drive apoptosis. Consequently, inhibition of Ras–MEK signaling significantly retards cell growth, leading to a substantial loss of Chk1 dependency (Fig. 6F). To manage these cancers therapeutically, one would therefore expect that inhibiting the Ras pathway would be sufficient to disrupt tumor growth. However, Ras pathway inhibition alone was found to only reduce cell proliferation, without promoting apoptosis. Considering that Chk1 dependency is closely associated with DNA damage levels produced by Ras activation, we therefore focused on exacerbating Chk1 dependency by introducing exogenous DNA damage, for example, with gemcitabine, to enhance the cytotoxic effect of Chk1 inhibition (Fig. 6G).

DNA damage–induced stress is a hallmark of cancer, which is widely observed in solid tumors experiencing genomic instability associated with constitutive endogenous DNA damage, resulting in activation of the DDR (6). When DDR occurs, DNA double-strand breaks are detected by ATM/ATR and DNA-PK, which belong to the phosphatidylinositol 3-kinase like protein kinase (PIKK) family (19, 20). Subsequently, Chk1 is phosphorylated by ATR and activates S- and G2 checkpoints, whereas p53 is required for the G1 checkpoint to arrest the cell cycle and repair-damaged DNA. p53-deficient cancer cells lack a G1–S checkpoint, which causes cells to rely on S or G2–M checkpoints to prevent death by mitotic catastrophe. Chk1 is a major S and G2–M checkpoint protein that can compensate for insufficient DDR in rapidly proliferating cancer cells in the absence of functional p53. Therefore, p53 deficiency has emerged as a candidate biomarker for Chk1-targeted therapy (13–15). Notably, we observed that the sensitivity of Chk1-dependent osteosarcoma cancers to Chk1 inhibition was not affected by p53 depletion. Furthermore, a differential sensitivity to Chk1 inhibition among p53-mutant breast cancer cells suggests that Chk1 dependency of cancer cells is not merely determined by the functionality of p53. Through differential gene expression and pathway analyses, we unexpectedly observed that Chk1 dependency is closely associated with Ras activity in each of the three tested cancer models.

The oncogenic roles of mutant Ras in cancer have been widely demonstrated, while the contribution of wild-type Ras to tumorigenesis remains less clear. Recent studies demonstrated that wild-type Ras proteins contribute to tumorigenesis in Ras-mutant cancer cells (21–24). However, the oncogenic roles of Ras in Ras wild-type cancers have yet to be established. Here, we observed that some of the Chk1 inhibitor–sensitive cells harbor wild-type Ras. In this context, all three Ras isoforms can contribute to Chk1 activation. Our mechanistic studies implicate a feedback-inhibitory pathway from Chk1 to Ras. Such feedback may function to sense the level of DNA damage that accumulates as a consequence of replicative stress associated with Ras-mediated cell proliferation. Thus, these feed-forward and feedback-regulatory links between Ras and Chk1 enable cancers to maximize growth without surpassing a critical threshold of DNA damage.

All authors were employees of Genentech at the time the studies were conducted and may be shareholders of Roche Pharmaceuticals.

Conception and design: H.-J. Lee, J. Settleman

Development of methodology: H.-J. Lee, E. Blackwood, M. Evangelista

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H.-J. Lee, V. Pham, E. Blackwood

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H.-J. Lee, Y. Cao, V. Pham, C. Wilson, M. Evangelista, C. Klijn, J. Settleman

Writing, review, and/or revision of the manuscript: H.-J. Lee, Y. Cao, E. Blackwood, M. Evangelista, D. Stokoe, J. Settleman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H.-J. Lee, J. Settleman

Study supervision: H.-J. Lee, D. Stokoe, J. Settleman

Other (mass spectrometric data acquisition and data analysis): V. Pham

We are grateful to members of the Settleman Laboratory for helpful discussions. We thank the Genentech gCSI facility for drug sensitivity profiling of cancer cell lines and gCell facility for providing cancer cell lines and conducting cell line authentication.

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.

1.
Sharma
SV
,
Settleman
J
. 
Oncogene addiction: setting the stage for molecularly targeted cancer therapy
.
Genes Dev
2007
;
21
:
3214
31
.
2.
de Bono
JS
,
Ashworth
A
. 
Translating cancer research into targeted therapeutics
.
Nature
2010
;
467
:
543
9
.
3.
McDermott
U
,
Sharma
SV
,
Dowell
L
,
Greninger
P
,
Montagut
C
,
Lamb
J
, et al
Identification of genotype-correlated sensitivity to selective kinase inhibitors by using high-throughput tumor cell line profiling
.
Proc Natl Acad Sci U S A
2007
;
104
:
19936
41
.
4.
Barretina
J
,
Caponigro
G
,
Stransky
N
,
Venkatesan
K
,
Margolin
AA
,
Kim
S
, et al
The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity
.
Nature
2012
;
483
:
603
7
.
5.
Garnett
MJ
,
Edelman
EJ
,
Heidorn
SJ
,
Greenman
CD
,
Dastur
A
,
Lau
KW
, et al
Systematic identification of genomic markers of drug sensitivity in cancer cells
.
Nature
2012
;
483
:
570
5
.
6.
Luo
J
,
Solimini
NL
,
Elledge
SJ
. 
Principles of cancer therapy: oncogene and non-oncogene addiction
.
Cell
2009
;
136
:
823
37
.
7.
Lee
HJ
,
Zhuang
G
,
Cao
Y
,
Du
P
,
Kim
HJ
,
Settleman
J
. 
Drug resistance via feedback activation of Stat3 in oncogene-addicted cancer cells
.
Cancer Cell
2014
;
26
:
207
21
.
8.
Wu
TD
,
Nacu
S
. 
Fast and SNP-tolerant detection of complex variants and splicing in short reads
.
Bioinformatics
2010
;
26
:
873
81
.
9.
Anders
S
,
Huber
W
. 
Differential expression analysis for sequence count data
.
Genome Biol
2010
;
11
:
R106
.
10.
Kohn
EA
,
Yoo
CJ
,
Eastman
A
. 
The protein kinase C inhibitor Gö6976 is a potent inhibitor of DNA damage-induced S and G2 cell cycle checkpoints
.
Cancer Res
2003
;
63
:
31
5
.
11.
Leung-Pineda
V
,
Ryan
CE
,
Piwnica-Worms
H
. 
Phosphorylation of Chk1 by ATR is antagonized by a Chk1-regulated protein phosphatase 2A circuit
.
Mol Cell Biol
2006
;
26
:
7529
38
.
12.
Stumpf
A
,
Cheng
ZK
,
Wong
B
,
Reynolds
M
,
Angelaud
R
,
Girotti
J
, et al
Development of an expedient process for the multi-kilogram synthesis of Chk1 inhibitor GDC-0425
.
Org Process Res Dev
2015
;
9
:
661
72
.
13.
Wang
Q
,
Fan
S
,
Eastman
A
,
Worland
PJ
,
Sausville
EA
,
O'Connor
PM
. 
UCN-01: a potent abrogator of G2 checkpoint function in cancer cells with disrupted p53
.
J Natl Cancer Inst
1996
;
88
:
956
65
.
14.
Toledo
LI
,
Murga
M
,
Fernandez-Capetillo
O
. 
Targeting ATR and Chk1 kinases for cancer treatment: a new model for new (and old) drugs
.
Mol Oncol
2011
;
5
:
368
73
.
15.
Ma
CX
,
Janetka
JW
,
Piwnica-Worms
H
. 
Death by releasing the breaks: CHK1 inhibitors as cancer therapeutics
.
Trends Mol Med
2011
;
17
:
88
96
.
16.
Nielsen
TO
,
Parker
JS
,
Leung
S
,
Voduc
D
,
Ebbert
M
,
Vickery
T
, et al
A comparison of PAM50 intrinsic subtyping with immunohistochemistry and clinical prognostic factors in tamoxifen-treated estrogen receptor-positive breast cancer
.
Clin Cancer Res
2010
;
16
:
5222
32
.
17.
Neve
RM
,
Chin
K
,
Fridlyand
J
,
Yeh
J
,
Baehner
FL
,
Fevr
T
, et al
A collection of breast cancer cell lines for the study of functionally distinct cancer subtypes
.
Cancer Cell
2006
;
10
:
515
27
.
18.
Dry
JR
,
Pavey
S
,
Pratilas
CA
,
Harbron
C
,
Runswick
S
,
Hodgson
D
, et al
Transcriptional pathway signatures predict MEK addition and response to selumetinib (AZD6244)
.
Cancer Res
2010
;
70
:
2264
73
.
19.
Zhou
BB
,
Elledge
SJ
. 
The DNA damage response: putting checkpoints in perspective
.
Nature
2000
;
408
:
433
9
.
20.
Jackson
SP
,
Bartek
J
. 
The DNA-damage response in human biology and disease
.
Nature
2009
;
461
:
1071
8
.
21.
Lim
KH
,
Ancrile
BB
,
Kashatus
DF
,
Counter
CM
. 
Tumour maintenance is mediated by eNOS
.
Nature
2008
;
452
:
646
9
.
22.
Young
A
,
Lou
D
,
McCormick
F
. 
Oncogenic and wild-type Ras play divergent roles in the regulation of mitogen-activated protein kinase signaling
.
Cancer Discov
2013
;
3
:
112
23
.
23.
Bentley
C
,
Jurinka
SS
,
Kljavin
NM
,
Vartanian
S
,
Ramani
SR
,
Gonzalez
LC
, et al
A requirement for wild-type Ras isoforms in mutant KRas-driven signalling and transformation
.
Biochem J
2013
;
452
:
313
20
.
24.
Grabocka
E
,
Pylayeva-Gupta
Y
,
Jones
MJ
,
Lubkov
V
,
Yemanaberhan
E
,
Taylor
L
, et al
Wild-type H- and N-Ras promote mutant K-Ras-driven tumorigenesis by modulating the DNA damage response
.
Cancer Cell
2014
;
25
:
243
56
.