|Home | About | Journals | Submit | Contact Us | Français|
The protein kinase encoded by the ataxia-telangiectasia and Rad3-related (ATR) gene is activated by DNA damaging agents that are frequently employed as anticancer therapeutics. Inhibition of ATR expression in cultured cancer cells has been demonstrated to increase sensitivity to chemotherapeutic drugs, including the DNA crosslinking agent cisplatin. Cisplatin is a widely employed and effective drug, but its use is associated with significant toxicity. Here, we demonstrate that genetic inhibition of ATR expression selectively enhanced cisplatin sensitivity in human colorectal cancer cells with inactivated p53. A knockin strategy was employed to restore wild type p53 in cells harboring wild type or mutant ATR alleles. Knockin of functional p53 in ATR-deficient cells restored checkpoint function, suppressed apoptotic pathways, and dramatically increased clonogenic survival after cisplatin treatment. These results suggest that a strategy that combines specific inhibitors of ATR and conventional therapies might promote synthetic lethality in p53-deficient tumors while minimizing toxicity to normal tissues.
The most widely used anticancer agents prevent cell growth by damaging chromosomal DNA or inhibiting DNA replication (Kastan and Bartek, 2004). Diverse types of DNA lesions and DNA structures trigger the activation of the phosphatydlinositol kinase-like kinases ATM and ATR. ATM is primarily activated by the double strand DNA breaks caused by ionizing radiation and radiomimetic drugs. ATR functions downstream of ATM in response to DNA damage (Jazayeri et al., 2006), but is also activated independently of ATM by a wide range of agents that inhibit DNA replication and cause the accumulation of replication intermediates (Osborn et al., 2002; Hurley and Bunz, 2007; Cimprich and Cortez, 2008). Replication inhibitors that can robustly activate ATR include antimetabolites that alter nucleotide metabolism and alkylating agents that cause DNA lesions that physically impede DNA replication forks. ATM and ATR directly phosphorylate over 700 downstream substrates that collectively control cell growth and survival (Matsuoka et al., 2007).
Prominent among the regulatory proteins activated after DNA damage is the tumor suppressor p53, a transcription factor that is stabilized upon phosphorylation and thereby activated (Tibbetts et al., 1999; Shiloh, 2006). Genetic alterations that cause loss of p53 function - which occur in a large proportion of all human cancers - cause defective regulation of cell growth and death in response to DNA damage, and therefore present a potential obstacle to effective therapy (El-Deiry, 2003; Meek, 2009). Because the toxicity of most therapeutic agents to normal tissues limits the doses that can be safely administered to patients, strategies for selectively sensitizing p53-deficient cancer cells to existing anticancer drugs and radiation would have significant clinical impact.
Recent studies have demonstrated parallel interactions between upstream DNA damage signaling pathways and p53 that may be exploited to selectively impair coordinated cell cycle arrest (Chung and Bunz, 2010) and improve therapeutic responses (Jiang et al., 2009) in p53-mutant cells. For example, targeted inhibition of ATM and its substrate Chk2 has been shown to increase the sensitivity of p53-/- human cancer cells to the radiomimetic drug doxorubicin, while increasing the resistance of p53+/+ cells (Jiang et al., 2009). It is currently unknown whether specific targeting of ATR might similarly increase the sensitivity of p53-deficient cells.
ATR is a particularly attractive target for combination therapies as it is robustly activated by many different types of drugs. Inhibiting ATR activity, either by RNAi-mediated knockdown of ATR expression (Collis et al., 2003) or by overexpression of a dominant-negative mutant ATR protein (Cliby et al., 1998), has been shown to confer sensitivity to diverse anticancer agents, including ionizing radiation, methyl methanesulfonate and cisplatin.
To rigorously study the role of ATR in therapeutic responses, we generated a genetic model system wherein a human colorectal cancer cell line was engineered to harbor the hypomorphic mutation at the ATR locus that causes ATR-Seckel syndrome (Hurley et al., 2007). At the cellular level, the ATR-Seckel mutation causes aberrant splicing of the ATR transcript and markedly decreased ATR expression (Alderton et al., 2004; O'Driscoll et al., 2003). Cancer cells homozygous for ATR-Seckel alleles (ATRS/S) exhibit greatly reduced clonogenic survival in response to many commonly used anticancer agents, particularly to DNA crosslinking agents such as cisplatin (Wilsker and Bunz, 2007). Decreased ATR expression also causes differential and highly reproducible sensitization to antimetabolites, ionizing radiation and radiomimetic drugs in this in vitro system (Hurley et al., 2007; Wilsker and Bunz, 2007). An important question that arises from these studies concerns the potential efficacy of anti-ATR therapy. Would inhibiting ATR preferentially sensitize cancer cells with loss of p53 function?
Recent studies in mice suggest that the effects of ATR inhibition on cell survival are antagonized by p53. In a mouse model of ATR-Seckel syndrome, the deficiency of ATR causes impaired DNA replication during embryogenesis and accelerated aging in adult mice (Murga et al., 2009). Homozygous disruption of p53 in the ATR-Seckel background aggravates this aging phenotype. In a mosaic mouse model, the conditional disruption of ATR in a p53-mutant background causes the accumulation of DNA damage and tissue degradation (Ruzankina et al., 2009). Together, these studies demonstrate that p53 functions to protect against the detrimental effects of ATR deficiency. In this study, we examined whether p53-mutant human cancer cells might be preferentially chemosensitized by genetic ATR inhibition. We show that RNAi-mediated knockdown of ATR preferentially sensitized p53-/- cells to the effects of cisplatin, and that knockin of wild type p53 into the ATR-Seckel background suppressed apoptotic pathways, restored checkpoints and increased cisplatin resistance to the level exhibited by cells with wild type ATR. These data support specific ATR inhibition as a therapeutic strategy to target p53-deficient tumors.
Our previous studies have shown that ATR-mutant colorectal cancer cells are dramatically sensitized to the alkylating agent cisplatin, as well as to other drugs that similarly cause DNA crosslinks (Wilsker and Bunz, 2007). To explore whether the inhibition of ATR might selectively sensitize p53-deficient cancer cells to this important class of therapeutic agent, we used gene-specific siRNA to reduce ATR expression in p53+/+ and p53-/- derivatives of the colorectal cell line HCT116 (Fig. 1a). ATR is an essential protein required for efficient DNA replication. As expected, transient inhibition of ATR expression caused a reduction in clonogenic survival in both HCT116 p53+/+ and p53-/- cells (Fig. 1b). A low dose of cisplatin also suppressed clonogenic growth, irrespective of p53 genotype. The combination of ATR knockdown and treatment with cisplatin reduced the survival of p53-/- cells to a significantly greater extent than p53+/+ cells, which were relatively resistant.
To examine in detail the combined effects of ATR and p53 in a stable genetic system, we used a knockin strategy (Fig. 2a) to restore normal p53 function in cells with previously engineered ATR-Seckel alleles. The parental DLD-1 cell line from which the ATR-mutant (ATRS/S) cells were derived harbors a naturally-occurring inactivating point mutation in p53 exon 7 (S241F) and expresses only mutant p53 transcripts (Fig. 2b). The remaining p53 allele in this diploid colorectal cell line has a wild type coding sequence, but is not expressed (Sur et al., 2009). We used a recombinant adeno-associated virus (rAAV)-based gene targeting vector containing a synthetic exon promoter trap (SEPT) (Topaloglu et al., 2005) to replace the mutant exon 7 with wild type sequence, and thereby reverse the inactivating mutation. The SEPT selection cassette, flanked by LoxP sites, was excised by transient expression of the cre recombinase, thereby restoring the functional architecture of the p53 locus (Fig. 2a).
Following infection of target cells with the p53-specific rAAV (Sur et al., 2009), and stable selection of transgenic colonies, we identified three clones that had independently integrated the targeting construct into the p53 locus, replacing the expressed mutant p53 allele (Fig. 2b). One of these clones was chosen for detailed analysis; all effects described were reproduced in an independent clone. Clonal derivatives with the same genotype were phenotypically indistinguishable, as is typically the case for knockin/knockout cell derived by homologous recombination (Rago et al. 2007; Chung and Bunz, 2010).
Excision of the SEPT cassette resulted in the expression of wild type p53 transcripts (Fig. 2b), and increased expression of the p53 target proteins p53R2 and p21 (Fig. 2c). The majority of inactivating p53 mutations cause increased stability of the encoded protein. Accordingly, restoration of functional p53 resulted in a decrease in steady state p53 protein expression (Fig. 2c).
To further assess p53 function in knockin (p53+/Sil) cells, we examined well-characterized responses to ionizing radiation (IR). After upregulation by IR, p53 activates the G1/S checkpoint that controls entry into S-phase and stabilizes arrest at the G2/M checkpoint, thereby inhibiting the onset of mitosis (Waldman et al., 1995; Bunz et al., 1998). Cells were irradiated and then immediately treated with nocodazole to prevent cell division and trap G2/M checkpoint-defective cells in mitosis. Control treatment with nocodazole alone (no irradiation) revealed a modest increase in cells with a stable 2N DNA content in the p53-knockin population (Fig. 2d). Presumably, this failure of a proportion of cells to enter S-phase after 24 h was related to the increased levels of basal p21 induced by p53 pathway restoration (Fig. 2c). Asynchronous ATRS/S cells with mutant p53 (p53-/Sil) have been shown to be markedly checkpoint-defective (Hurley et al., 2007). Accordingly, after treatment with both nocodazole and IR, ATRS/Sp53-/Sil cells failed to accumulate at a 2N peak corresponding to G1/S (Fig. 2d) and entered mitosis in large numbers (Fig. 2e). In contrast, the ATRS/Sp53+/Sil cells exhibited restored function of both G1/S and G2/M checkpoints.
We next used our new isogenic cell panel to assess the combined effects of ATR and p53 status on cell survival after cisplatin treatment. Consistent with our previously published results (Wilsker and Bunz, 2007), ATRS/Sp53-/Sil cells exhibited markedly decreased survival after treatment with cisplatin across a broad dose range (Fig. 3a). The highest dose tested (1 μM) reduced survival of ATR-deficient cells nearly one hundred-fold, compared with isogenic DLD-1 cells that are ATR-proficient. Strikingly, the response of p53-knockin ATRS/Sp53+/Sil cells to cisplatin at all doses was similar to that of ATR-proficient parental DLD-1 cells, demonstrating that p53 could counteract the drug sensitization caused by ATR deficiency.
To evaluate the effect of p53 restoration on different drug responses, we separately compared the effects of p53 and ATR on survival after cisplatin and the DNA synthesis inhibitor hydroxyurea (HU). In a p53-deficient background, ATR is required for clonogenic survival in response to both of these drugs (Wilsker and Bunz, 2007). Restoration of p53 function in parental DLD-1 with wild type ATR alleles caused a significant sensitization of these cells to both cisplatin and HU (Fig. 3b), consistent with the well known role of p53 as an inhibitor of cell proliferation after DNA damage and DNA replication inhibition. Cells with the ATRS/S genotype were rendered more resistant to cisplatin and HU when wild type p53 function was restored. Interestingly, the desensitizing effect of restored p53 was significantly more pronounced in cisplatin-treated cells. While ATRS/Sp53+/Sil cells treated with HU exhibited similar survival to HU-treated ATR+/+p53+/Sil cells, the survival of ATRS/Sp53+/Sil cells after cisplatin most resembled cells with the ATR+/+p53+/Sil genotype (Fig. 3b). We conclude that p53 most potently modified ATR-mediated survival pathways in cisplatin-treated cells, and propose that anti-ATR therapy might uniquely sensitize p53-deficient tumor cells to cisplatin. These results complement previous studies demonstrating that ATM inhibition can differentially promote survival or sensitivity to DNA damage, depending on p53 status (Jiang et al., 2009).
It is interesting to note that the p53-mediated response to cisplatin was more striking in cells with restored p53 function (p53-knockin DLD-1; Fig. 3b) than in cells that retained wild type p53 alleles (HCT116); p53-knockout in HCT116 caused a minimal difference in survival (Fig. 1b). It has been reported that some p53-dependent responses may not be well maintained in p53-wild type cells such as HCT116 (Zhang et al., 2006). Like all human cancer cell lines that have been adapted to in vitro culture and extensively propagated following explantation, HCT116 and DLD-1 cells have been selected for robust growth. The growth-suppressive pathways downstream of p53 are inactive in p53-mutant DLD-1 cells (Sur et al., 2009), and presumably not subjected to negative selective pressure. One might therefore predict that growth suppressive pathways would be retained but inactive in p53-mutant DLD-1 and then robustly reactivated by wild type p53 knockin. In support of this view, the upregulation of growth inhibitory p53 target genes that encode PUMA and Ferrodoxin reductase were more strongly upregulated by DNA damage in the DLD-1 p53+/Sil derivative than in other colorectal cancer cells lines that naturally harbor wild type p53 alleles (Sur et al., 2009). We have observed the p53-dependent phenotypes in knockin DLD-1 cells to be highly stable and infer that, in p53-wild type cells, strong selection against p53 phenotypes might occur during the initial establishment of new cell lines, rather than during routine passaging. We therefore predict that human cell lines with restored p53 function will be valuable reagents for the study of the p53-dependent responses to anticancer agents.
We tested whether the decrease in clonogenic survival exhibited by ATRS/Sp53-/Sil cells in response to cisplatin (Fig. 3a,b) might be related to the induction of apoptosis. Indeed, morphologic evidence of apoptosis was apparent in a large fraction of ATRS/Sp53-/Sil cells after cisplatin treatment (Fig. 4a). Cisplatin-induced apoptosis occurred independently of functional p53, and was in fact suppressed by p53 restoration. To confirm the biochemical activation of apoptotic pathways, we assayed the cleavage of caspase-3 and the phosphorylation of p53 on S46 (p-p53S46), a marker of upstream apoptotic signaling (Oda et al., 2000). These biochemical markers of apoptosis were coordinately expressed in cisplatin-treated ATRS/Sp53-/Sil cells and suppressed in all isogenic cells with functional ATR or p53 (Fig. 4b). The small amount of p-p53S46 detected in cells of the other genotypes corresponded well with the relatively smaller proportion of apoptotic cells (1-3%) observed in these populations (Fig. 4a). Interestingly, phosphorylation of p53 S46 after cisplatin treatment was impaired in cells in which p53 had been functionally restored (Fig. 4b), suggesting the presence of a negative feedback loop. Overall, isogenic cells with proficient ATR and/or p53 pathways were resistant to the apoptotic pathways stimulated by cisplatin (Fig. 4a,b). These data suggest that increased apoptotic signaling in response to cisplatin may have contributed to the reduced clonogenic survival of ATRS/Sp53-/Sil cells after cisplatin treatment.
The chemoprotective effects of ATR have been largely attributed to the activation of its downstream effector kinase, Chk1 (Cimprich and Cortez, 2008). Recently, the stress activated kinase pathway composed of p38MAPK and MAPKAP2 (MK2) has emerged as another important responder to DNA damaging agents, including cisplatin (Reinhardt et al., 2007; Reinhardt et al. 2010). Like Chk1, p38MAPK is controlled by ATR (Jirmanova et al., 2005; Reinhardt and Yaffe, 2009). Yaffe and colleagues (Reinhardt et al., 2007) have reported that loss of p53 functionally “rewires” the cellular response to cisplatin, thereby increasing the requirement for MK2 in the control of downstream survival pathways. Because the effect of p53 restoration in our cell lines differed in ATR-wild type and –mutant cells (Fig. 3b), we examined whether p53 status affected relative activation of Chk1 and MK2 by cisplatin. Two isogenic cell pairs were tested: HCT116 cells and their p53-knockout derivative (Bunz et al., 1998), and DLD-1 cells and their derivative harboring the wild type p53 knockin (Sur et al., 2009). Of these two cell lines, DLD-1 exhibited a greater p53-dependent effect on cisplatin sensitivity (Figs. 1a, 3a,b). Irrespective of cisplatin treatment, DLD-1 cells with restored p53 exhibited a decrease in total Chk1 protein and an increase in MK2 protein and T334 phosphoprotein (Fig. 5a). The decreased Chk1 protein in the p53-restored DLD-1 cells correlated with decreased levels of Chk1 phosphoproteins after cisplatin treatment (Fig. 5a). The HCT116 lines, in which the effect of p53 status alone (in the absence of ATR manipulation) on cisplatin sensitivity was minimal (Fig. 1b) exhibited only a modest decrease in Chk1 phosphorylation in p53+/+ compared with the p53-/- derivative, but no apparent difference in total Chk1 or MK2 proteins.
We next examined the combined effects of ATR and p53 status on the p38MAPK-MK2 signaling pathway. Levels of MK2 were highest in cells with both wild type ATR and wild type p53 (Fig. 5b). While the levels of p38MAPK protein did not vary with genotype, increased p38MAPK activation by cisplatin could be observed in the cells with restored p53. The re-introduction of p53 into DLD-1 cells thus caused an apparent shift in ATR-dependent signaling from Chk1 to MK2 (Fig. 5a). This rewiring of ATR responses after restoration of p53 correlated with decreased survival after DNA damage (Fig. 3b).
ATR promotes cell survival after DNA damage and impeded DNA replication, and is therefore a logical target for sensitization to commonly employed anticancer therapeutics (Cimprich and Cortez, 2008; Wagner and Kaufmann, 2010). A critical question is whether the p53-deficient cells that compose a high proportion of human cancers would be selectively killed by such a combinatorial strategy. The answer to this question is not obvious. Depending on the context in which it is activated, p53 can trigger cell cycle arrest - and an overall increase in cell survival - or cell death by apoptosis (Vousden and Lu, 2002). We observed that knockdown of ATR preferentially sensitized p53-knockout colorectal cells to cisplatin (Fig. 1a,b) and that cells with inactivating mutations in both ATR and p53 exhibited increased apoptosis (Fig. 4a,b) and reduced survival (Fig. 3a, b) after cisplatin treatment. Restoration of functional p53 dramatically suppressed these phenotypes. These results suggest that specific targeting of ATR would be an effective means of increasing the sensitivity of p53-mutant tumor cells to cisplatin while preserving the resistance of normal tissues that retain p53 function, thereby minimizing toxicity. The ultimate test of this hypothesis in patients awaits the development of highly specific ATR inhibitors (Wagner and Kaufmann, 2010).
Recent efforts to enhance the effects of DNA damaging agents and DNA replication inhibitors on p53-deficient cancer cells have largely focused downstream of ATR, on Chk1 (Tse et al., 2007). While genetic inhibition of Chk1 and chemical Chk1 kinase inhibitors have been reported to be effective in sensitizing cancer cells to drugs such as gemcitabine, HU and 5-fluorouracil (Cho et al., 2005; Karnitz et al., 2005; Robinson et al., 2006; Blasina et al., 2008; Wilsker et al., 2008), targeting Chk1 has proven a relatively ineffective approach to reduce cell survival in combination with cisplatin (Wagner and Karnitz, 2009; Zenvirt et al. 2010). In response to cisplatin, ATR activates multiple downstream effectors (Reinhardt et al., 2007), including p53 (Pabla et al., 2008), that may cumulatively affect survival. The rewiring of these downstream pathways that occurs upon loss of p53 function (Reinhardt et al., 2007) may create an enhanced dependence on ATR. Studies arising from diverse experimental systems therefore suggest that targeting upstream ATR activity may represent a more effective means of sensitizing p53-deficient tumor cells to the distinct effects of cisplatin.
In cellular models and in actual human tumors, functional p53 can confer resistance to some types of DNA damage and sensitivity to others (Bunz et al., 1999; Pirollo et al., 2000). These distinct outcomes are probably related to the alternative cell cycle arrest or cell death pathways that can be activated by p53 (Vousden and Lu, 2002). Which of these downstream pathways predominates in a given cell appears to depend upon many factors, including the type of cell and the DNA damaging agent employed. One mechanism by which the choice between arrest and apoptosis is determined involves the kinase HIPK2, which phosphorylates p53 on S46 (Rui et al., 2004). Stimuli that promote cell cycle arrest over apoptosis trigger the suppression of HIPK2, thereby suppressing p53 S46 phosphorylation (Li et al., 2009). The p53 S46 site can also be phosphorylated directly by ATM (Kodama et al. 2010) and by DYRK2, an ATM-responsive kinase (Taira et al. 2007). Understanding how ATR might influence these upstream pathways should provide significant insight into the manner in which p53 mediates different cell fates in response to DNA damage.
Cisplatin (cis-Diammineplatinum(II) dichloride) and hydroxyurea were purchased from Sigma-Aldrich (St. Louis, MO). The human colorectal cell lines HCT116 and DLD-1and their isogenic derivatives with targeted alterations in p53 (Bunz et al., 1998; Sur et al., 2009) and ATR (Hurley et al., 2007) were cultured in McCoy's 5A medium (Invitrogen, Carlsbad CA) supplemented with 6% fetal calf serum (Hyclone) and penicillin/streptomycin (Invitrogen).
For the transient knockdown of ATR, 3 × 104 cells in 24-well plates were transfected with 8 μmol of ATR-targeted siRNA (AACCUCCGUGAUGUUGCUUGA, synthesized by Integrated DNA Technologies, Inc, Coralville, IA) or non-targeting control siRNA (Dharmacon, Lafayette, CO), using 2μl Lipofectamine 2000 (Invitrogen) for 72 h.
The restoration of wild type p53 in the ATRS/S DLD-1 cells was confirmed by sequencing amplified genomic DNA- and cDNA-derived PCR products from selected clones using the primers 100-P53 exon7 FOR: CTTGGGCCTGTGTTATCTCC and 101-p53 exon 7 rev: ATGGAAGAAATCGGTAAGAGG.
Cells in 24-well plates were treated with drug under the conditions described, washed, harvested and replated at low density in 100 mm dishes. Following 14 d of growth, surviving cells were stained with crystal violet (Sigma-Aldrich). Colonies containing more than 50 cells were scored and the total number of colonies per dish was normalized to untreated controls. Each data point represents the average of three separate dishes.
To assess IR-dependent cell cycle checkpoints, cells were treated with a single 12 Gy dose delivered by a 135Cs irradiator and then immediately incubated in medium containing 0.2 μg/ml nocodazole. After 24 h, cells were harvested, fixed, stained with Hoechst 33258, and analyzed by flow cytometry and fluorescence microscopy, as described (Jallepalli et al., 2003). After cisplatin treatment, the proportion of Hoechst 33258-stained cells with blebbing of the nuclear membrane, and fragmented and condensed nuclei was assessed by fluorescence microscopy. At least 300 nuclei from different fields were scored for each data point.
Whole-cell lysates were denatured and fractionated on NuPAGE gels (Invitrogen). Proteins were transferred to PVDF membranes (Millipore, Billerica, MA) which were probed with antibodies directed against ATR, α-tubulin, p53, p53R2, Chk1, p21 (Santa Cruz Biotechnology, Santa Cruz, CA), phoshpo-p53Ser46, phospho-Chk1S345, phospho-Chk1S317, MK2, phospho-MK2T334, p38MAPK, phospho-p38MAPKT180/Y182 and cleaved caspase-3 (Cell SignalingTechnologies, Danvers, MA) under conditions recommended by the manufacturers. Blots were developed using enhanced chemiluminescence (GE Healthcare, Piscataway, NJ).
This work was supported by grants from the Flight Attendant Medical Research Institute and the NCI (T32CA121937).
Conflict of Interest: The authors declare no conflict of interest.