Loss of ataxia telangiectasia mutated– and Rad3-related function potentiates the effects of chemotherapeutic drugs on cancer cell survival

The most common strategy for anticancer therapy is the generation of DNA damage in proliferating cells. Both cancer cells and normal proliferating cells are challenged by DNA damage, but cancer cells seem to be particularly sensitive. Cancer cells intrinsically have defective responses to DNA damage, and it is believed that these alterations, acquired during tumorigenesis, underlie the efficacy of current treatment modalities (1).

Whether caused by ionizing radiation or by chemotherapeutic agents, diverse types of DNA damage trigger the activation of the DNA damage signaling network. When activated by DNA lesions, these complex and interconnected pathways trigger a variety of cellular responses, including DNA repair, cell cycle arrest via the activation of checkpoints, and apoptosis (1, 2). Some of these responses are thought to be protective and serve to enhance cell survival. For example, transient arrest of the cell cycle may facilitate the repair of damaged chromosomes. Other responses such as apoptosis eliminate the clonogenic potential of damaged cells, thus preventing the propagation of acquired mutations. Cumulatively, DNA damage responses have potent effects on the long-term clonogenic survival of proliferating cell populations. Accordingly, there has been considerable interest in identifying targets in the DNA damage signaling network that might modulate these effects for therapeutic benefit (1, 3, 4).

There are many potential therapeutic targets, few of which have been validated in a rigorous genetic fashion. At the core of the DNA damage signaling network is a two-tiered protein kinase cascade, in which upstream signals from the phosphoinositide 3-kinase–like kinases (PIKK) are transduced to and amplified by the checkpoint kinases, Chk1 and Chk2 (5). Chk1 has been a major focus of interest in the attempt to abrogate DNA damage signaling in cancer cells. The compound 7-hydroxystaurosporine (also known as UCN-01) was originally isolated as an inhibitor of protein kinase C, but is a potent Chk1 inhibitor as well (6, 7). UCN-01 sensitizes tumor cells to several chemotherapeutics including mitomycin C, 5-fluorouracil (5-FU), cisplatin, and camptothecin (8–11). UCN-01 is currently being used in combination with DNA-damaging agents for treatment of several types of cancer in clinical trials (12).

Encouraging progress with Chk1 inhibitors has stimulated studies of additional targets. One promising drug target is ATR. ATR is upstream of Chk1 and signals directly to Chk1 in response to diverse forms of DNA damage (5). Recent studies have revealed that Chk1 activation requires ATR (13–15). Like Chk1, ATR is essential for viability in mouse (16) and human cells (17), seeming to function as both a DNA damage response element and as an essential factor for cell growth. Recent studies using cancer cells with engineered ATR alleles have shown that ATR promotes cell cycle progression in the presence of double-strand DNA (dsDNA) breaks, and that a loss of this activity leads to a form of cell cycle arrest known as S-phase stasis (15). The induction of S-phase stasis was associated with diminished clonogenic survival. Other studies have used overexpression of an ATR mutant (18) or inhibitory RNA to examine whether PIKK activity contributes to therapeutic sensitivity (19–22). The extent to which ATR inhibition can potentiate therapeutic responses has varied considerably among these studies.

Isogenic cell lines derived by homologous recombination have been valuable tools in the evaluation of potential therapeutic targets (23–27). In this study, we used ATR-mutant knock-in cells that express very low levels of ATR to comparatively evaluate the roles of ATR in cell survival and cell cycle progression following treatment with a broad panel of widely used chemotherapeutic agents. Using a defined genetic system, rather than siRNA or nonspecific chemical inhibitors, is advantageous because ATR expression is not variable between experiments, making quantitation and comparison of drug effects possible. These variables are critical to establish the extent to which ATR inhibition is likely to enhance the effectiveness of other DNA damage–causing agents as a combination therapy. We determined that ATR-mutant cancer cells displayed a marked sensitivity to DNA-damaging agents and DNA synthesis inhibitors but not to agents that impede growth by other mechanisms. Additionally, the degree of sensitization depended on the type of drug used and the type of DNA lesion imparted. ATR-mutant cells were most sensitive to agents that cause DNA cross-links and were more moderately sensitized to agents that inhibit nucleotide synthesis or generate dsDNA breaks. We show that therapeutically relevant concentrations of these drugs triggered ATR signaling to Chk1 and that a loss of ATR activity resulted in the inhibition of cell cycle progression. These results suggest that the combination of ATR inhibition with conventional chemotherapy could be a broadly applicable strategy.

The human lung cancer cell line H1299 and the human colon cancer cell lines HCT116 and DLD1 and their derivatives were cultured in McCoy's 5A medium supplemented with 6% FCS and penicillin/streptomycin (Invitrogen, Carlsbad, CA). The Seckel-Rescue ATR-expressing cell line was obtained by transfection of the ATR expression plasmid wt-ATR-pcDNA3 (a kind gift from Dr. Robert Abraham, Wyeth Research, Pearl River, NY) with Lipofectamine (Invitrogen). Stable clones that integrated the construct were selected with geneticin (Invitrogen).

Cells growing in 24-well plates were treated with drugs at indicated concentrations for 24 to 48 h. Drug-treated cells and untreated controls were washed with HBSS (Invitrogen) and harvested following detachment in 0.5-mL trypsin-EDTA. Cells were transferred to 1.5-mL medium and vortexed. Approximately 10⁴ cells were plated in triplicate onto 10-cm tissue culture dishes and incubated at 37°C for 14 days. Colonies were stained with 0.2% crystal violet in 50% methanol. Colonies containing >50 cells were scored. Clonogenic survival is expressed as a proportion of the colony number on control dishes.

Wortmannin (Sigma, St. Louis, MO) was dissolved in DMSO. Hydroxyurea, cyclophospamide, cisplatin, and mitomycin C were obtained from Sigma and dissolved in PBS. 5-FU (American Pharmaceuticals, Inc., Los Angeles, CA), methotrexate (Bedford Laboratories, Bedford, OH), raltitrexed (Tomudex; AstraZeneca, Wilmington, DE), doxorubicin (Adriamycin; Giensia Laboratories, Irvine, CA), gemcitabine (Gemzar; Eli Lilly and Co., Indianapolis, IN), irinotecan (Camptosar; Pharmacia-Upjohn Co., Kalamazoo, MI), and paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) were of pharmaceutical grade. Tumor necrosis factor–related apoptosis-inducing ligand (TRAIL) was obtained from Alexis Biochemicals (Montreal, Canada). According to the manufacturer's recommendations, TRAIL enhancer was added to facilitate TRAIL receptor cross-linking.

Cells were treated with each chemotherapeutic agent as indicated. Following treatment, drug-containing medium was aspirated and cells were washed thrice with HBSS (Invitrogen). Cells were then incubated with cell culture medium containing nocodazole at a final concentration of 0.2 μg/mL for the times indicated. Cells were detached with trypsin-EDTA and fixed in PBS containing 3.7% formaldehyde, 0.5% NP40, and stained with 10 μg/mL Hoechst 33258. Ten thousand cells were analyzed per sample on a flow cytometer.

Whole-cell lysates were denatured, sonicated, and fractionated on NuPAGE gels (Invitrogen). Proteins were transferred to polyvinylidene difluoride membranes, which were then incubated with antibodies directed against ATR, Chk1, Chk2, α-tubulin (Santa Cruz Biotechnology, Santa Cruz, CA), phospho-Chk1 (Ser³¹⁷), phospho-Chk2 (Thr⁶⁸), and phospho-Chk1 (Ser³⁴⁵; Cell Signaling Technologies, Beverly, MA) under conditions recommended by the manufacturers. Blots were developed by enhanced chemiluminescence (Amersham, Piscataway, NJ).

To examine the effects of combined PIKK inhibition and DNA damage, we examined the effects of wortmannin, a well-known kinase inhibitor that blocks the activity of PIKKs (28), including ataxia telangiectasia mutated (ATM) and ataxia telangiectasia mutated– and Rad3-related (ATR). Treatment of HCT116 cells with 12-Gy ionizing radiation stimulated robust phosphorylation of both checkpoint kinases, Chk1 and Chk2 (Fig. 1A). Chk2 is a canonical substrate of ATM (29) whereas Chk1 phosphorylation on Ser³⁴⁵ after ionizing radiation treatment has recently been shown to be dependent on ATR activity (13–15). Phosphorylation of both Chk1 and Chk2 was inhibited by wortmannin (Fig. 1A). Because the activation of ATR by double-strand breaks has recently been shown to be dependent on prior activation of ATM (13, 14, 30), it is not possible to infer whether both ATR and ATM were directly inhibited by the drug at the concentrations used. Previous studies have indicated that ATR is relatively more resistant to inhibition by wortmannin, with an IC₅₀ of 1.8 μmol/L (31).

Using a concentration of wortmannin (10 μmol/L) that moderately inhibited the phosphorylation of Chk1 and Chk2, we examined whether PIKK inhibition would affect clonogenic survival following treatment with 5-FU. 5-FU is an antimetabolite that is widely used in cancer therapy and is a mainstay for adjuvant therapy for colorectal cancer. We tested the effect of combination therapy on two colorectal cell lines, HCT116 and DLD1, and the lung cancer cell line H1299. Wortmannin was mildly toxic at the concentration used, as reflected by a decreased level of clonogenic survival (Fig. 1B and C). 5-FU treatment caused a marked decrease in the clonogenic survival of all three cell lines tested. Interestingly, combined treatment with 5-FU and wortmannin resulted in a consistent increase in survival as compared with 5-FU alone. This unexpected result suggested that inhibition of the PIKK family might not potentiate and may in fact mitigate the effects of at least some modes of DNA damage–related therapy. Whereas wortmannin is a potent inhibitor of PIKKs, the effects of this drug on other signaling molecules cannot be ruled out. This potential for off-target effects prompted us to use a genetic approach.

We sought to establish whether the activity of a single PIKK family member, ATR, might function to alter cell survival in response to chemotherapeutics. ATR was chosen as a target protein for two reasons. First, a downstream substrate of ATR, Chk1, has been successfully targeted by kinase inhibitors and these drugs have shown synergy with conventional therapies. Second, studies using RNA interference have shown ATR-dependent differences in survival and exposure to several DNA-damaging agents, but a thorough comparison of the effects of ATR inhibition on distinct classes of drugs has not been reported. We sought to identify which class of drugs ATR inhibition could potentiate most effectively.

Using a gene targeting approach, our group has derived a cancer cell line in which both ATR alleles are a hypomorphic variant (15). This derivative cell line, designated DLD-ATR-Seckel, expresses very low levels of ATR protein and is radiosensitive as compared with parental DLD1 (also known as HCT15), which has wild-type ATR alleles. Both DLD1 and its ATR-mutant derivative have inactivating mutations in p53 (32) and, like the vast majority of cancer cells, are defective in their checkpoint responses. Isogenic cell systems derived by homologous recombination have proved to be useful in studying the relationships between specific genes and drug sensitivity (23–27).

We used the DLD1/DLD-ATR-Seckel cell system to examine the contribution of ATR activity to cell survival following treatment with conventional chemotherapeutics. A diverse panel of clinically useful drugs were tested, including antimetabolites that inhibit DNA synthesis by various mechanisms (Fig. 2A), alkylating agents that create DNA adducts and cross-links (Fig. 2B), and a radiomimetic that induces dsDNA breaks (Fig. 2C). The analysis of clonogenic survival revealed that the ATR-mutant, DLD-ATR-Seckel cells were significantly more sensitive to these classes of drugs. Interestingly, the degree of sensitization was largely consistent among drugs of the same category. The alkylating agents cisplatin and mitomycin C showed the largest effect, with a difference in survival that was nearly 3 orders of magnitude. In comparison, cyclophosphamide was much less affected by ATR inhibition, even at millimolar concentrations. Cyclophosphamide is a prodrug that is activated in vivo by liver enzymes, which are limiting in cultured cancer cells. The relatively low levels of both cell killing and potentiation in ATR-mutant cells are likely a result of the low activity of the prodrug form.

The relative sensitivity of ATR-mutant cells to the effects of the various antimetabolites tested was less dramatic, but a roughly 10- to 20-fold reduction in survival could be seen after exposure to these compounds in this broad category. The relative sensitivity of DLD-ATR-Seckel cells to doxorubicin (Fig. 2C), a radiomimetic, was relatively modest and consistent with previously reported radiosensitivity, an ∼6-fold increase (15). Thus, dsDNA breaks caused similar differences in survival regardless of the mode of their generation. In contrast, treatment of cells with Taxol, a drug that interferes with microtubule depolymerization, and TRAIL, a cell death-inducing ligand, did not elicit any specificity and inhibited cell survival to a similar extent, regardless of genotype (Fig. 2D). Targeting of the XIAP gene in DLD1 cells has previously been shown to lead to TRAIL sensitivity (25), underscoring that the results obtained here were highly gene specific.

Given the broad sensitivity of the ATR-mutant cells, we sought to confirm that the concentrations of drugs used in this study were sufficient to activate ATR kinase activity. Using concentrations of drugs that gave at least a 10-fold reduction in the survival of ATR-mutant cells, we observed robust phosphorylation of the canonical substrate of ATR, Chk1 (Fig. 3). Phosphorylation of both Ser³¹⁷ and Ser³⁴⁵ of Chk1 was markedly lower in DLD-ATR-Seckel, regardless of the stimulus. As was previously observed after treatment with ionizing radiation (15), phosphorylation of Chk1 Ser³⁴⁵ triggered by chemotherapeutic compounds was ATR dependent; very low levels of signal could be seen in the DLD-ATR-Seckel cells.

In general, the compounds that stimulated robust Chk1 phosphorylation were the same ones that had elicited a genotype-specific difference in survival. Taxol, for example, did not trigger phosphorylation of Chk1 (Fig. 3). Interestingly, the singular exception to this pattern was cisplatin. Among all drugs tested, cisplatin showed the most dramatic ATR-dependent survival difference and yet seemed to trigger little detectable Chk1 phosphorylation (Fig. 3). No decrease in Chk1 protein levels was observed in cisplatin-treated cells. Whereas these results show that ATR was generally activated by the stimuli tested, they do not necessarily show that Chk1 was enzymatically activated nor that the Chk1 pathway was the mediator of the strong phenotypic differences observed.

Cancer cells are often checkpoint deficient and thus aberrantly proceed through cell cycle transitions despite the presence of DNA damage. Previous results have shown that ATR is required for cancer cells to progress from G₁ into S phase in the presence of dsDNA breaks (15). We sought to determine whether other types of DNA lesions and inhibitors of DNA synthesis similarly induce a requirement for ATR activity.

Cells were treated with representative compounds from each group that had elicited strong differences in survival that correlated with ATR status. After treatment, cells were released into fresh medium that contained nocodazole, which facilitated the analysis of cell cycle progression. The presence of this microtubule inhibitor blocked cell division and thus prevented the appearance of nascent cells in the 2N peak.

The most thoroughly characterized drug with respect to cell cycle progression is hydroxyurea. Hydroxyurea inhibits ribonucleotide reductase and thereby blocks synthesis of cytosine. In a wide range of yeast and mammalian cells, hydroxyurea has been shown to reversibly inhibit DNA synthesis (33–35). Before drug treatment, asynchronous DLD1 and DLD-ATR-Seckel cells had an indistinguishable cell cycle profile (Fig. 4A). Following treatment with hydroxyurea (time 0), cells of both types accumulated in a peak with 2N DNA content, consistent with a failure to synthesize DNA (Fig. 4B). Following release from hydroxyurea block, DLD1 cells with wild-type ATR progressed through S phase within 9 h and accumulated before mitosis due to the effect of nocodazole. In contrast, DLD-ATR-Seckel cells did not progress and remained in the 2N peak.

Radiomimetic drugs, such as doxorubicin, subject cells to chronic levels of dsDNA breaks rather than the acute breaks caused by ionizing radiation treatment. By the end of a 24-h doxorubicin treatment period, many of the DLD1 cells had entered S phase, consistent with known checkpoint defects in this cancer cell line (Fig. 4C). After an additional 24-h recovery period, the majority of cells attained 4N DNA content. A substantial proportion of DLD-ATR-Seckel cells remained in the 2N peak following doxorubicin treatment, and progression of these cells through S-phase was impaired throughout the recovery period as well. Treatment with cisplatin (Fig. 4D) also revealed a requirement for ATR in the progression of the cell cycle. In the presence of cisplatin-induced DNA damage, the entire population of DLD1 cells moved into the 4N peak whereas a substantial proportion of DLD-ATR-Seckel cells failed to complete DNA replication. No notable differences in the numbers of apoptotic cell fragments were observed even at later time points (data not shown).

To conclusively confirm that the survival defects in the ATR-mutant cells were due to the lack of ATR expression, we transfected DLD-ATR-Seckel cells with an ATR expression construct and selected a stable, ATR-expressing cell line. The level of expression of ATR in the resulting Seckel-Rescue cell line was ∼50% of that detected in the parental cell line, DLD1 (Fig. 5). Resistance of Seckel-Rescue cells to cisplatin, the compound that elicited the greatest extent of differential sensitivity among the panel of drugs tested, was significantly higher than that of DLD-ATR-Seckel.

Previous reports focusing on inhibition of the DNA damage response in cancer cells have suggested ATR as a potential target for novel anticancer therapies (18–22). Whereas the results have varied considerably, these exploratory studies have collectively indicated a role for ATR in at least some relevant therapeutic responses. With the use of a novel genetic system generated by homologous recombination, we have established the overall magnitude and generality of these effects in the most commonly used chemotherapeutics and thereby identified one class of chemotherapeutics that is most strongly potentiated by ATR inhibition. For the first time, we have related clonogenic survival with ATR signaling and with progression of the cell cycle. These experiments have revealed an unambiguous role for ATR activity in cancer cell growth and survival following DNA damage.

We propose that the survival defects observed in ATR-mutant cell populations are directly related to the failure of a large proportion of these cells to progress through S phase and complete DNA replication. Previously, we have shown that dsDNA breaks cause ATR-deficient cells to stall at the beginning of S phase (15), a type of growth arrest known as S-phase stasis (36). The data presented here suggest that other forms of DNA damage caused by chemotherapeutic drugs may similarly cause this distinctive effect. Presumably, a failure of a significant proportion of cells to efficiently progress through all phases of the cell cycle would impair overall survival.

Clearly, other ATR-dependent factors may also contribute to long-term survival. In addition to defects in cell cycle progression, ATR-deficient cells are known to have a defect in the G₂-M checkpoint (15, 37), and such defects have been shown to adversely affect long-term clonogenic survival. The effects on survival do not seem to be related to an increased level of apoptosis of ATR-mutant cells because increased numbers of apoptotic cells were not observed.

Nonspecific inhibition of the PIKKs did not recapitulate the survival defect caused by the genetically mediated inhibition of ATR activity. Rather, a nonspecific PIKK inhibitor actually increased clonogenic survival (Fig. 1B and C). This interesting result underscores the complex nature of the DNA damage signaling network and the many responses that can be triggered by PIKK activation. Because ATR signaling actively promotes the progression of S phase when DNA replication forks stall (38), ATR may be unique among the PIKKs in its ability to antagonize the effects of therapeutic agents that inhibit DNA replication. In contrast, the role of the other nonessential PIKKs seems to be the halting of the cell cycle by the activation of checkpoints.

Our comparative study revealed distinct classes of drugs that could elicit ATR-dependent differences in cell survival (Table 1). The greatest potentiation ratios were observed in the alkylating agents cisplatin and mitomycin C. Both of these agents form several types of DNA complexes, including intrastrand and interstrand adducts (39), which are thought to significantly impede active DNA replication forks. Stalled replication forks, in turn, are known to be the most potent inducers of ATR signaling (38). ATR also potentiated the effects of the antimetabolites tested in this panel, but to a lesser extent. In contrast to the alkylating agents, antimetabolites inhibit DNA replication indirectly by altering the biosynthesis or uptake of nucleotide precursors. It would thus seem that the potentiating effect of ATR inhibition on cell survival is directly related to the extent to which ATR is activated.

It was surprising that cisplatin did not seem to trigger Chk1 phosphorylation under the experimental conditions used. Phosphorylation of Chk1 has been shown to be dynamically affected by phosphatases that regulate cell cycle progression (40, 41). It seems possible that Chk1 was very robustly phosphorylated by ATR in cisplatin-treated cells, but then dephosphorylated shortly thereafter. Alternatively, it is possible that cisplatin triggers a distinct ATR pathway that does not involve Chk1. Cisplatin is a widely used drug that is used for the treatment of testicular, ovarian, cervical, head and neck, and non–small-cell lung cancers (39). Other cancers seem to be inherently resistant to this form of treatment, including colorectal cancer. Toxic side effects also limit the clinical utility of this drug. A means of potentiating the effects of cisplatin, thereby lowering the doses used and broadening its potential application, would be a significant step forward. Our study suggests that inhibition of ATR is a valid means of achieving this goal.

This study illuminates several avenues for future research. It will be important to assess the contributions of cancer-associated mutations that alter DNA damage responses, particularly those that affect p53, to the ATR-specific effects shown here. Compound knockout/knock-in cell lines would be particularly useful in this regard. Further validation of ATR as a therapeutic target will be greatly facilitated by the isolation of highly specific small-molecule inhibitors of ATR kinase activity. Whether ATR-mediated potentiation of standard chemotherapy is specific to checkpoint-deficient cancer cells will be ultimately addressed by testing ATR inhibitors in vivo on xenografts and orthotopic tumors and by the derivation and characterization of ATR-mutant strains of mice. Our studies suggest that high-throughput approaches for the discovery of ATR-specific inhibitors are clearly warranted.

We thank the Cell Imaging Core Facility of the Kimmel Cancer Center at Johns Hopkins University School of Medicine for the assistance with flow cytometry.