Research ArticleCancer

Therapeutic Targeting of a Robust Non-Oncogene Addiction to PRKDC in ATM-Defective Tumors

See allHide authors and affiliations

Science Translational Medicine  12 Jun 2013:
Vol. 5, Issue 189, pp. 189ra78
DOI: 10.1126/scitranslmed.3005814

Abstract

When the integrity of the genome is threatened, cells activate a complex, kinase-based signaling network to arrest the cell cycle, initiate DNA repair, or, if the extent of damage is beyond repair capacity, induce apoptotic cell death. The ATM protein lies at the heart of this signaling network, which is collectively referred to as the DNA damage response (DDR). ATM is involved in numerous DDR-regulated cellular responses—cell cycle arrest, DNA repair, and apoptosis. Disabling mutations in the gene encoding ATM occur frequently in various human tumors, including lung cancer and hematological malignancies. We report that ATM deficiency prevents apoptosis in human and murine cancer cells exposed to genotoxic chemotherapy. Using genetic and pharmacological approaches, we demonstrate in vitro and in vivo that ATM-defective cells display strong non-oncogene addiction to DNA-PKcs (DNA-dependent protein kinase catalytic subunit). Further, this dependence of ATM-defective cells on DNA-PKcs offers a window of opportunity for therapeutic intervention: We show that pharmacological or genetic abrogation of DNA-PKcs in ATM-defective cells leads to the accumulation of DNA double-strand breaks and the subsequent CtBP-interacting protein (CtIP)–dependent generation of large single-stranded DNA (ssDNA) repair intermediates. These ssDNA structures trigger proapoptotic signaling through the RPA/ATRIP/ATR/Chk1/p53/Puma axis, ultimately leading to the apoptotic demise of ATM-defective cells exposed to DNA-PKcs inhibitors. Finally, we demonstrate that DNA-PKcs inhibitors are effective as single agents against ATM-defective lymphomas in vivo. Together, our data implicate DNA-PKcs as a drug target for the treatment of ATM-defective malignancies.

Introduction

In response to DNA damage, cells activate a signaling cascade to prevent further cell cycle progression. Activation of this DNA damage response (DDR) network allows time for DNA repair or, if the lesions are beyond repair capacity, leads to the induction of apoptosis (1). The proximal DDR kinase ATM, which is mutated in the human cancer–prone disorder ataxia-telangiectasia (A-T), is a master regulator of three essential DDR processes—cell cycle regulation, DNA repair, and apoptosis. ATM affects the different cellular outcomes through the phosphorylation of numerous substrates, including H2AX, MDC1, Nbs1, Chk2, p53, and MDM2 (2, 3). ATM is frequently mutated in various sporadic human cancers, and biallelic loss of ATM is associated with chemotherapy resistance and poor survival (410). It has been recently shown that ATM is required for the induction of p53-driven apoptosis after genotoxic chemotherapy (10). Thus, ATM deficiency is likely a selected genomic aberration in cancer because it protects from p53-driven apoptosis. Beyond mediating apoptosis, ATM also plays a critical role in DNA double-strand break (DSB) repair. Mammalian cells use two distinct DSB repair pathways. Nonhomologous end joining (NHEJ) is an error-prone DSB repair pathway that is preferentially used during early phases of the cell cycle, when no sister chromatid is available (11). During NHEJ, the noncatalytic subunits Ku70 and Ku80 form a heterodimer that binds to the free DNA ends and subsequently recruits DNA-PKcs (DNA-dependent protein kinase catalytic subunit). DNA-PKcs kinase activity is essential for XRCC4- and Lig4-mediated rejoining of the broken ends during NHEJ (12). The second major DSB repair pathway, homologous recombination (HR), is largely restricted to the S and G2 phases of the cell cycle, when a sister chromatid is available as a template for DSB repair (13). One of the early events necessary for completion of the HR process is DSB end resection to create a 3′ single-stranded overhang, which becomes rapidly coated with RPA and provides a substrate for activation of the proximal DDR kinase ATR (14). During the ensuing steps of the HR process, RPA is replaced by Rad51, which mediates the core reactions of HR, namely, homology searching, strand exchange, and Holliday junction formation (13). There is strong evidence for a role of ATM in HR-mediated DSB repair, with less pronounced effects on NHEJ (1518). Cells derived from A-T patients show a DSB repair defect, which is due to impaired assembly and functioning of the RAD51-associated protein complexes in the HR arm of DSB repair (16, 18, 19). Recruitment of Rad51 to DSBs requires resection of DNA ends to generate RPA-coated 3′ single-stranded overhangs. This resection process and the resulting Rad51 focus formation are ATM-dependent (2022). ATM is required for the HR-dependent DSB repair component in G2, as supported by the observation that ionizing radiation–induced sister chromatid exchanges in G2 require ATM (2325). Thus, the apoptosis-evading effect of ATM deficiency in human neoplasias likely comes at the cost of a reduced ability to repair chemotherapy-induced DSB lesions via error-free HR.

Because ATM-deficient human tumors frequently display chemotherapy resistance (4, 610), one might speculate that chemotherapy-induced DSBs are repaired in ATM-deficient cancer cells to ensure long-term survival. It is likely that alternative, error-prone DSB repair pathways, such as NHEJ, compensate for the HR defect in ATM-deficient cells. Consistent with the idea that NHEJ might serve as a backup mechanism for failed HR-mediated DSB repair, mice deficient for both ATM and PRKDC (encoding DNA-PKcs) display early embryonic lethality (26), whereas animals lacking either ATM or PRKDC are viable (27, 28). Here, we aimed to characterize DNA-PKcs as a drug target for the treatment of inherently chemotherapy-resistant ATM-defective neoplastic disease.

Results

ATM-defective cancer cells display DNA-PKcs addiction

We have recently shown that ATM depletion renders murine cells and tumors resistant to genotoxic chemotherapy, mimicking the effects of disabling ATM mutations in human patients (10). We further showed that DNA-PKcs repression in ATM-depleted murine embryonic fibroblasts increases their sensitivity to DSB-inducing chemotherapy (10). However, it remains unclear whether ATM-defective human cancer cells display a similar addiction to DNA-PKcs and whether DNA-PKcs is indeed a druggable target for the treatment of ATM-defective cancers. Finally, the molecular details of the apparent synthetic lethal interaction between ATM and PRKDC remain elusive.

To test whether ATM-defective cancer cells display DNA-PKcs addiction, we investigated HT144 and H1395 cells. The human melanoma cell line HT144 carries a homozygous GG to AA substitution at codon 2845 in ATM, resulting in a premature stop codon (29). In addition, this cell line carries a homozygous BRAFV600E mutation. As an ATM-proficient control for this cell line, we used BRAFV600E-driven A375 melanoma cells. The human non–small cell lung cancer (NSCLC) cell line H1395 carries an A to G substitution at codon 2666 of ATM, resulting in a Thr to Ala mutation in the ATM Ser/Thr kinase domain (30). As an ATM-proficient control for this cell line, we used A549 NSCLC cells. All four cell lines are p53-proficient. We assessed the effects of DNA-PKcs inhibition using the DNA-PKcs inhibitor KU-0060648. Cells were treated for 24 hours with KU-0060648 (0.5 μM), the DSB-inducing topoisomerase II inhibitor etoposide (10 μM), or a combination of both compounds (Fig. 1A). Apoptosis was assessed by flow cytometry after cells had been stained with antibodies to cleaved caspase-3. Etoposide induced widespread apoptosis in the ATM-proficient cell lines A375 (90.6 ± 6.8%) and A549 (54.6 ± 5.4%) (Fig. 1A), whereas ATM-defective HT144 and H1395 cells were resistant against etoposide with only 14.2 ± 2.5% and 15.0 ± 2.6% of apoptotic cells after 24 hours, respectively. When these cell lines were exposed to KU-0060648, we observed a clear segregation on the basis of their ATM status. ATM-proficient cells showed only marginally increased apoptosis, compared to the nontreated controls (Fig. 1A), whereas the ATM-defective cell lines displayed a marked apoptotic response after 24 hours of exposure to KU-0060648. Specifically, HT144 cells showed 44.3 ± 10.3% and H1395 cells displayed 49.6 ± 10.2% apoptotic cells compared to less than 3% of apoptotic cells in the respective untreated controls (Fig. 1A). Combination treatment with etoposide plus KU-0060648 had no additional significant effect on the apoptotic response of ATM-defective cells compared to either drug alone (Fig. 1A). Similar effects were observed when we repeated these experiments and replaced KU-0060648 with NU7441 (0.5 μM) as an alternative DNA-PKcs inhibitor (Fig. 1B).

Fig. 1 Non-oncogene addiction to DNA-PKcs in ATM-defective human cancer cells.

(A) ATM-proficient (A375 and A549) and ATM-defective (HT144 and H1395) cells were grown to assess their response to etoposide (10 μM), the DNA-PKcs inhibitor KU-0060648 (KU) (0.5 μM), or a combination treatment. After 24 hours, cells were harvested, and cleaved caspase-3 staining was analyzed by flow cytometry to assess the percentage of apoptotic cells (bars indicate means ± SEM, n = 12). (B) To exclude off-target effects of KU-0060648, we treated cells as in (A) with the exception that KU-0060648 was replaced by NU7441 (NU) as an alternative DNA-PKcs inhibitor (bars indicate means ± SEM, n = 12). (C) Clonogenic survival assay. ATM-proficient (A375 and A549) and ATM-defective (HT144 and H1395) cells were treated with 10 μl of phosphate-buffered saline (PBS) (vehicle control) or exposed to etoposide (10 μM), KU-0060648 (0.5 μM), or a combination treatment for 12 hours, washed, and replated at 5000 cells per 10-cm dish. Fourteen days later, colonies were stained and counted. (D) Quantification of the experiments described in (C) [bars indicate means ± SEM (n = 3), normalized to untreated control]. *P < 0.05, two-tailed Student’s t test.

To validate these experiments, we used colony survival assays. Cells were left untreated and exposed to etoposide, KU-0060648, or a combination treatment consisting of etoposide plus KU-0060648 for 12 hours (Fig. 1C). Surviving colonies were assayed 14 days after reseeding subsequent to completion of the different treatments. ATM-defective cells displayed etoposide resistance compared to their ATM-proficient counterparts, which is reflected in significantly more surviving colonies in HT144 and H1395 cells (Fig. 1, C and D). In contrast, ATM-defective cells appeared to critically depend on DNA-PKcs signaling for their survival, even in the absence of exogenous genotoxic stress. Treatment with KU-0060648 as a single agent resulted in a significant decrease in the number of surviving colonies in HT144 and H1395 cells compared to untreated controls. This is in contrast to their ATM-proficient counterparts, which show essentially no difference in the number of surviving colonies when comparing control cells to those exposed to KU-0060648 (Fig. 1D). These data suggest that ATM deficiency is associated with marked etoposide resistance in human cancer cells. Furthermore, the observation that DNA-PKcs inhibition promotes apoptosis in ATM-defective cells suggests that these cells are DNA-PKcs–dependent.

Because studies with adenosine triphosphate–competitive inhibitors are frequently hampered by off-target effects, we next performed genetic studies to further assess the DNA-PKcs dependence of ATM-defective cancer cells. To examine whether DNA-PKcs was required for survival in ATM-proficient and ATM-defective human cancer cells, we used RNA interference (RNAi) to deplete DNA-PKcs and examined population doubling rates upon knockdown (Fig. 2, A to D). Cells were infected with lentiviruses delivering short hairpin RNAs (shRNAs) against luciferase (control) or DNA-PKcs. We used three DNA-PKcs–targeting shRNAs with different degrees of knockdown efficiency, essentially allowing us to analyze an allelic series of DNA-PKcs expression levels (fig. S1). DNA-PKcs depletion completely prevented further proliferation of ATM-defective H1395 and HT144 cells when we used the two most potent shRNAs (#1 and #2) (Fig. 2, A and B). When we tested the effects of an shRNA with a less strong knockdown efficiency (#3), proliferation of both H1395 and HT144 cells was markedly reduced, but not completely abolished, compared to cells expressing control shRNA. In contrast, DNA-PKcs repression in ATM-proficient cells did not significantly reduce the population doubling rates in these cells, indicating that DNA-PKcs is not essential in ATM-proficient settings (Fig. 2, C and D).

Fig. 2 Genetic validation of the synthetic lethal interaction between ATM and PRKDC.

(A to D) Effect of a series of DNA-PKcs–targeting shRNAs with varying degrees of knockdown efficiency. ATM-defective H1395 (A) and HT144 (B) cells, as well as ATM-proficient A549 (C) and A375 (D) cells, were transduced with control shRNAs or three distinct DNA-PKcs–targeting shRNAs, and population doublings were recorded. Experiments shown in (A) to (D) were performed at n = 8 (bars indicate means ± SEM). (E and F) ATM-defective HT144 (E) and H1395 (F) cells were transiently transfected with a plasmid encoding Flag.ATM and GFP. Twenty-four hours later, cells were treated with etoposide (10 μM) for 24 hours and harvested, and apoptosis was assessed by flow cytometry. GFP coexpression was used to separate ATM-complemented cells from the parental cells [top left panels in (E) and (F)]. Gate M1 indicates GFP-negative cells, and gate M2 indicates GFP-expressing cells. The bottom panels in (E) and (F) show histogram plots of the parental (left) and ATM-complemented (right) HT144 and H1395 cells. Gates M3 and M4 indicate the fraction of cleaved caspase-3–positive parental and ATM-complemented cells, respectively. Quantification of the individual experiments is shown at the top right of (E) and (F). Bars indicate means ± SEM (n = 8). *P < 0.05, two-tailed Student’s t test. (G) A375 and A549 cells were infected with lentiviruses encoding ATM-targeting or luciferase control shRNAs. Cells were left untreated or exposed to etoposide (10 μM), KU-0060648 (KU) (0.5 μM), or a combination treatment for 24 hours before apoptosis was quantified using flow cytometry. The experiments shown in (G) were performed at n = 8 (bars indicate means ± SEM). (H) Eμ:MYC;ARF−/−-driven cells were infected with retroviruses encoding ATM-targeting or luciferase control shRNAs. Upon puromycin selection, cells were left untreated or exposed to etoposide (10 μM), KU-0060648 (0.5 μM), or a combination treatment for 24 hours before apoptosis was quantified using flow cytometry (bars indicate means ± SEM, n = 9).

We next performed a set of complementation experiments in which we compared the effect of KU-0060648 in the ATM-defective parental cell lines H1395 and HT144 and their ATM-complemented counterparts. Cells were transfected with a plasmid driving the expression of ATM and green fluorescent protein (GFP) (fig. S2). We chose conditions under which 40 to 50% transfection efficiency was reached. Transfected cultures were exposed to etoposide for 12 hours, and apoptosis was analyzed by flow cytometry. Coexpression of GFP and ATM allowed the separate gating of ATM-defective (GFP-negative) and ATM-complemented (GFP-positive) cells (Fig. 2, E and F). In keeping with our hypothesis, we found that complementation of ATM-defective and inherently chemotherapy-resistant HT144 and H1395 cells significantly enhanced their apoptotic response to etoposide (Fig. 2, E and F, bottom panels).

If ATM deficiency was indeed responsible for the etoposide resistance and DNA-PKcs addiction that we observed in HT144 and H1395 cells, one might expect ATM depletion to produce a similar phenotype in initially ATM-proficient cells. To test this, we infected A375 and A549 control cells with lentiviruses expressing ATM-targeting shRNAs and analyzed the apoptotic response of these cells to KU-0060648, etoposide, or a combination treatment (Fig. 2G and fig. S3). Consistent with our hypothesis, we found that ATM depletion rendered A375 and A549 cells resistant to etoposide and exquisitely sensitive to DNA-PKcs inhibition (Fig. 2G). To further prove our hypothesis that ATM deficiency is associated not only with resistance against genotoxic chemotherapy but also with DNA-PKcs dependence, we turned to murine Eμ:MYC;ARF−/−-driven B–non-Hodgkin’s lymphoma (B-NHL) cells (Fig. 2H and fig. S4). To address the effect of ATM depletion in an otherwise isogenic system, we infected Eμ:MYC;ARF−/−-driven lymphoma cells with retroviruses expressing ATM-targeting shRNA and compared the effects of KU-0060648, etoposide, or a combination treatment. In agreement with our initial experiments, we found that ATM-proficient control cells were highly sensitive to etoposide, although this sensitivity was completely abolished when ATM was depleted (Fig. 2H). Furthermore, ATM-depleted Eμ:MYC;ARF−/− lymphoma cells became DNA-PKcs–addicted because KU-0060648 exposure induced massive apoptosis in these cells (Fig. 2H). This DNA-PKcs addiction of ATM-depleted cells was likely not due to DNA-PKcs overexpression in ATM-depleted cells. When we performed immunoblotting to analyze DNA-PKcs expression levels and activation (as monitored by an antibody detecting phospho-Thr2647), we found that knockdown of ATM did not result in increased DNA-PKcs expression, but in increased DNA-PKcs activity, even in the absence of genotoxic stress (fig. S5). Together, these data indicate that ATM deficiency is associated with resistance against genotoxic chemotherapy, likely through an abrogation of p53-driven apoptosis. On the other hand, this apoptosis evasion appears to come at the cost of a non-oncogene addiction to DNA-PKcs, which could be targeted with DNA-PKcs inhibitors.

ATM-defective cells fail to repair DSBs when DNA-PKcs is inhibited

We next aimed to mechanistically characterize the DNA-PKcs addiction of ATM-defective cells. Beyond mediating apoptosis (19), ATM has also been shown to be an important driver of HR-mediated DSB repair (1518). However, because ATM-defective cancers appear to be largely resistant against genotoxic chemotherapy, these HR-defective malignancies may use alternative DSB repair pathways, such as DNA-PKcs–dependent NHEJ, to repair DSBs. If this was the case, one might expect to observe the prolonged persistence of unrepaired DSBs in ATM-defective cells that had been exposed to DNA-PKcs inhibitors. To test this, we used immunofluorescence to monitor the persistence of etoposide-induced γH2AX and 53BP1 nuclear foci in ATM-proficient and ATM-deficient cells exposed to KU-0060648 or vehicle control. We chose to monitor γH2AX and 53BP1 foci because both are established markers of DSBs (31, 32).

We treated ATM-proficient (A375 and A549) and ATM-deficient (HT144 and H1395) cells for 20 min with a low-dose etoposide pulse (0.1 μM) to induce DSBs. In a parallel experiment, cells were pretreated with KU-0060648 for 1 hour before addition of etoposide. KU-0060648 remained present in the medium after etoposide removal. In both experiments, cells were protected from premature apoptosis by addition of the irreversible pan-caspase inhibitor Z-VAD (10 μM), which was applied together with etoposide. Consistent with the induction of DSBs by etoposide, we detected similar numbers of γH2AX and 53BP1 foci in all cell lines 1 hour after etoposide removal (Fig. 3, A to D). Furthermore, there were no detectable differences in the repair kinetics of all cell lines, when the number of γH2AX/53BP1 foci–positive cells was assessed at 72 and 96 hours. However, marked differences in the DSB repair kinetics could be observed in ATM-proficient and ATM-deficient cells that were treated with KU-0060648. ATM-proficient cells showed no evidence of increased γH2AX or 53BP1 foci at 72 and 96 hours compared to vehicle-treated controls. This picture was different in ATM-defective cells treated with KU-0060648. Both HT144 and H1395 cells displayed persistent γH2AX and 53BP1 foci even 96 hours after etoposide removal, when DNA-PKcs was inhibited (Fig. 3, A to D). These observations are in line with a DSB repair defect being present in cells lacking both ATM and DNA-PKcs activity.

Fig. 3 DSB repair defect resulting from DNA-PKcs inhibition in ATM-defective cells.

(A and B) Treatment of ATM-defective cells with KU-0060648 (KU) results in persistent DSBs as indicated by persistent nuclear γH2AX staining. ATM-proficient (A375) and ATM-defective (HT144) human melanoma cells (A), as well as ATM-proficient (A549) and ATM-defective (H1395) human NSCLC cells (B), were exposed to a low-dose etoposide pulse (0.1 μM, 20 min) and harvested 1, 72, and 96 hours later. Control cells were left untreated. In a parallel experiment, cells were pretreated with KU-0060648 (0.5 μM) for 1 hour before addition of etoposide. Top panels show a quantification of these experiments (n = 9; bars indicate means ± SEM). Bottom panels depict representative original immunofluorescence data. (C and D) Exposure of ATM-defective HT144 and H1395 cells to KU-0060648 results in persistent DSBs as indicated by persistent 53BP1 nuclear foci. A375, A549, HT144, and H1395 cells were treated as in (A) and (B) and stained with antibodies detecting 53BP1. Top panels show a quantification of these experiments (n = 7, bars indicate means ± SEM). Bottom panels depict representative original immunofluorescence data.

ATM- and DNA-PKcs–defective cancer cells generate RPA-coated single-stranded DNA intermediates

We next aimed to further characterize the DSB repair defect in ATM-defective cells. Recruitment of Rad51, the core component of the HR machinery, to DSBs requires resection of DNA ends to generate RPA-coated 3′ single-stranded DNA (ssDNA) overhangs. To assess whether ATM-defective cells that were exposed to KU-0060648 initiated early steps of the HR process, we submitted ATM-proficient and ATM-deficient cells to the same treatment regimen as in Fig. 3 to monitor the occurrence and persistence of nuclear RPA foci, markers of ssDNA repair intermediates (33). All four cell lines displayed prominent RPA foci 1 hour after removal of etoposide, regardless of ATM status and independent of the presence or absence of a DNA-PKcs inhibitor (Fig. 4, A and B). At 72 and 96 hours after etoposide exposure, the ATM-proficient cells (A375 and A549) remained largely RPA foci–negative, paralleling their lack of γH2AX and 53BP1 foci and indicative of completed DSB repair at these late time points. There was no difference among the ATM-competent cells that were exposed to either KU-0060648 or vehicle. Furthermore, RPA foci were largely undetectable in ATM-defective HT144 and H1395 cells at 72 and 96 hours, when no DNA-PKcs inhibitor was present. In contrast, when ATM-defective cells were treated with an etoposide pulse and KU-0060648, large RPA foci were detectable in these cells 72 and 96 hours after etoposide removal. These data indicate that DSBs not only persist for extended periods in ATM-defective cells that are exposed to DNA-PKcs inhibitors but also are extensively modified in these cells to yield RPA-coated ssDNA structures.

Fig. 4 RPA-coated ssDNA intermediates in ATM-defective cells without functional DNA-PKcs.

(A and B) ATM-proficient (A375) and ATM-defective (HT144) melanoma cells, as well as ATM-proficient (A549) and ATM-defective (H1395) NSCLC cells, were exposed to a low-dose etoposide pulse (0.1 μM, 20 min) and harvested 1, 72, and 96 hours later. Control cells were left untreated. In a parallel experiment, cells were pretreated with KU-0060648 (KU) (0.5 μM) for 1 hour before addition of etoposide. RPA foci were visualized using indirect immunofluorescence. Top panels show a quantification of these experiments (n = 12; bars indicate means ± SEM). Bottom panels depict representative original immunofluorescence data.

RPA-coated ssDNA triggers activation of the ATR/Chk1/p53/Puma axis

We next asked whether ssDNA repair intermediates might induce proapoptotic signaling in ATM-defective cells. Because both the HT144 and H1395 cell lines are p53-proficient, we reasoned that the KU-0060648–induced apoptosis of these cells might be p53-dependent. Numerous kinases can activate p53, including ATM, DNA-PKcs, and ATR (34). Intriguingly, RPA-coated ssDNA recruits ATR through binding of its regulatory subunit ATRIP to RPA (14). We reasoned that the RPA-coated ssDNA that we had observed in ATM-defective cells treated with KU-0060648 might trigger an ATR-dependent p53 activation, ultimately promoting p53-driven apoptosis. To test this, we used immunoblotting to assess the activation status of the ATR/Chk1/p53 signaling axis in ATM-proficient and ATM-deficient cells that were treated with etoposide, KU-0060648, or vehicle control. ATR activation was monitored with an antibody detecting a phospho-epitope on Thr1989. Chk1 activation was assessed with an antibody to phospho-Ser317. p53 activation was assessed with an antibody detecting total (stabilized) p53 and an antibody directed against phospho-Ser20, the residue targeted by Chk1. Using these assays, we found the ATR/Chk1/p53 axis to be activated in the ATM-defective cell lines 24 hours after addition of KU-0060648 (Fig. 5A). Consistent with the primary resistance of these ATM-defective cells, we failed to detect any activation of the ATR/Chk1/p53 axis after etoposide treatment. A strikingly different activation pattern of the ATR/Chk1/p53 axis emerged in ATM-proficient cell lines. Twenty four hours after etoposide, these cells displayed prominent activation of ATR, Chk1, and p53, whereas KU-0060648 treatment did not result in any substantial activation of the ATR/Chk1/p53 axis (Fig. 5A). These data suggest that DNA-PK inhibition leads to ATR/Chk1-dependent p53 activation in ATM-defective cells.

Fig. 5 Activation of apoptosis through ATR in ATM-defective cells treated with DNA-PKcs inhibitors.

(A) ATM-proficient (A375 and A549) and ATM-deficient (HT144 and H1395) cells were left untreated, exposed to etoposide (10 μM), or treated with KU-0060648 (KU) (0.5 μM). Cells were harvested after 24 hours and analyzed by immunoblotting using the indicated antibodies. β-Actin staining served as a loading control. (B and C) To assess the functional consequence of the p53 accumulation depicted in (A), we used qPCR to monitor the expression level of the p53 target genes PUMA, BAX, BAK, GADD45A, and RPRM in ATM-defective HT144 (B) and H1395 (C) cells after exposure to either etoposide (10 μM) or KU-0060648 (0.5 μM). Cells were treated as in (A), harvested, and analyzed by qPCR. Expression levels of the indicated p53 target genes were normalized against glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression (bars indicate means ± SEM, n = 4).

To assess the outcome of this p53 accumulation, we used quantitative polymerase chain reaction (qPCR) to monitor the expression of the p53 target genes PUMA, BAX, BAK, GADD45A, and RPRM. KU-0060648 treatment markedly increased mRNA expression of the proapoptotic p53 target gene PUMA in ATM-defective cells (Fig. 5, B and C). In contrast, etoposide treatment did not result in a significant change in the expression level of any of the investigated p53 target genes. These observations agree with a primary resistance of ATM-defective cells to DNA damage–induced apoptosis through the p53 pathway. However, this resistance appears to be overcome by exposure of ATM-defective cells to KU-0060648, suggesting that repression of DNA-PK activity in ATM-defective cells leads to an ATR/Chk1-dependent activation of the proapoptotic p53/Puma axis, likely as a result of aberrant ATR activation downstream of RPA-coated ssDNA repair intermediates.

We next asked whether interception of signaling through the ATR/Chk1/p53/Puma axis would abolish KU-0060648–induced apoptosis in ATM-defective cells. To this end, we exposed ATM-proficient and ATM-deficient cells to either etoposide, etoposide plus KU-0060648, or etoposide plus KU-0060648 plus the Chk1 inhibitor AZD-7762 (200 nM) for 12 hours. Addition of either KU-0060648 or KU-0060648 plus AZD-7762 did not significantly alter the degree of apoptosis in ATM-proficient cells when compared to the effects of etoposide alone (fig. S6, A and B). In contrast, exposure of ATM-defective cells to a combination treatment consisting of etoposide plus KU-0060648 led to the robust induction of apoptosis. Addition of AZD-7762 to this regimen led to a precipitous drop in the percentage of apoptotic cells, suggesting that signaling through the ATR/Chk1 axis is involved in mediating apoptosis in DNA-PKcs inhibitor–treated ATM-defective cells (fig. S6, C and D).

To further interrogate the contribution of the ATR/Chk1/p53/Puma axis in mediating KU-0060648–induced apoptosis in ATM-defective cells, we next assessed the effects of Puma repression in this setting. ATM-proficient and ATM-deficient cells were infected with lentiviruses encoding either control or Puma-targeting shRNAs. Cells were then treated with KU-0060648, etoposide, or a combination of both compounds for 24 hours before apoptosis was assessed. As expected, depletion of Puma resulted in marked resistance of ATM-proficient cells to etoposide or combination treatment with etoposide plus KU-0060648 (fig. S7, A and B). Loss of Puma also repressed the apoptotic response of ATM-deficient HT144 and H1395 cells treated with KU-0060648 or a combination of etoposide plus KU-0060648 (fig. S7, C and D). These data lend further support to our hypothesis that signaling through the ATR/Chk1/p53/Puma axis is involved in mediating KU-0060648–dependent apoptosis in ATM-defective cells.

Repression of CtBP-interacting protein abolishes 3′-ssDNA repair intermediates and prevents apoptosis

CtBP-interacting protein (CtIP) is required for the resection of DNA ends to generate RPA-coated 3′-ssDNA overhangs (35, 36). There is accumulating evidence suggesting a two-step model for DSB resection where CtIP and the Mre11/Rad50/Nbs1 complex cooperate to initiate resection before the exonuclease Exo1 continues the resection process to generate the 3′-ssDNA tails required for the HR process (37). Our data suggested that the generation of RPA-coated ssDNA repair intermediates triggered ATR/Chk1 activation, ultimately promoting p53-dependent PUMA induction and subsequent apoptosis of ATM-defective cells treated with KU-0060648. We hence speculated that repression of CtIP might prevent the generation of ssDNA intermediates and thus might preclude activation of the proapoptotic ATR/Chk1/p53/Puma axis in ATM-defective cells treated with KU-0060648. To test this, we compared the dynamics of nuclear RPA foci formation in ATM-defective HT144 and H1395 cells expressing either control or CtIP-targeting shRNAs (fig. S8). As shown in fig. S9A, CtIP depletion significantly reduced the number of RPA foci–positive HT144 and H1395 cells 1 hour after etoposide removal and almost completely blocked the generation of RPA foci at 72 and 96 hours. As described by others, we found that shRNA-mediated depletion of CtIP caused hypersensitivity toward etoposide (fig. S9B) (38). CtIP depletion also resulted in a significant (P < 0.05) reduction of KU-0060648 toxicity when applied alone or in combination with etoposide (fig. S9B). These results point toward a critical role for CtIP not only in mediating the resection of DSBs in DNA-PK inhibitor–treated ATM-defective cells but also in promoting subsequent activation of apoptosis.

ATM-defective chronic lymphocytic leukemia cells display DNA-PKcs addiction

Deletions on the long arm of chromosome 11 (harboring the ATM gene located at 11q22.3–11q23.1) are found in about 20% of patients with chronic lymphocytic leukemia (CLL) and identify a subgroup with poor outcome (9). CLL with del(11q) can be further divided into two subgroups based on the integrity of the residual ATM allele. Patients with biallelic ATM alterations display defective responses to cytotoxic chemotherapeutics in vitro and a poorer clinical outcome (4). Hence, we next aimed at validating our findings in primary CLL cells derived from patients with either del(11q) or a wild-type configuration on 11q. Patients were stratified as being wild type or del(11q) by clinical-grade fluorescence in situ hybridization (FISH) analysis (Fig. 6, A and B). Primary CLL cells were seeded onto a feeder layer of CD40 ligand–expressing NIH 3T3 cells before treatment with KU-0060648, etoposide, or a combination of both compounds. Wild-type CLL cells were highly sensitive to etoposide but resistant against KU-0060648 (Fig. 6). Addition of KU-0060648 to the etoposide regimen did not significantly enhance the response of the ATM-proficient CLL cells. In contrast, del(11q) CLL cells were exclusively sensitive to KU-0060648 but largely resistant to etoposide (Fig. 6). These data suggest that DNA-PK inhibition might be a useful strategy to treat chemotherapy-resistant, ATM-defective CLLs.

Fig. 6 Etoposide resistance and DNA-PKcs addiction in CLL cells carrying del(11q).

(A and B) CLL patients were stratified as wild type on 11q22 (n = 13) or del(11q) (n = 13) through FISH analysis (left panels). Primary CLL cells freshly isolated from 26 patients were seeded onto a feeder layer of CD40 ligand–expressing NIH 3T3 cells before treatment with KU-0060648 (KU) (0.5 μM), etoposide (10 μM), or a combination of both compounds. After 24 hours, cells were harvested and incubated with 7-aminoactinomycin D (7AAD) and fluorescein isothiocyanate–labeled annexin V and then analyzed by flow cytometry. Survival was quantified as the percentage of 7AADlow/annexin Vlow cells (bars indicate means ± SEM, n = 9). *P < 0.05, two-tailed Student’s t test.

DNA-PKcs is a valid target for the therapy of ATM-defective lymphoma

To validate our observations in vivo, we used the Eμ:MYC;ARF−/−-driven B-NHL model (10). Lymphoma cells derived from Eμ:MYC;ARF−/− mice were infected with lentiviruses encoding either luciferase control or ATM-specific shRNAs. C57BL/6J recipient mice were transplanted with 1.5 × 106 transduced lymphoma cells. Upon lymphoma manifestation, treatment with either KU-0060648, etoposide, or a combination of KU-0060648 plus etoposide was initiated. Untreated control animals carrying either luciferase shRNA– or ATM shRNA–expressing lymphomas were used to monitor the natural course of the disease. The entire cohort of animals bearing untreated control tumors succumbed to their disease within 25 days after initial manifestation (Fig. 7A). The overall survival of animals carrying control shRNA–expressing lymphomas could be significantly enhanced when these animals were treated with etoposide (Fig. 7A). KU-0060648 treatment did not produce a significant extension in overall survival of these animals. Furthermore, addition of KU-0060648 to the etoposide regimen did not result in significant additional survival gains beyond those that were achievable with etoposide. In contrast, survival of untreated animals bearing ATM-depleted lymphomas was slightly reduced compared to untreated control lymphomas, likely reflecting a more aggressive phenotype. Furthermore, ATM-depleted lymphomas were resistant to etoposide, with no significant survival gains compared to the untreated cohort. ATM-depleted lymphomas were highly sensitive to KU-0060648, leading to significant (P < 0.05) survival gains when compared with untreated and etoposide-treated animals. Combination treatment with KU-0060648 and etoposide prolonged the survival of ATM-depleted lymphoma-bearing animals even further, and a plateau was reached at about 20% (Fig. 7B). This plateau persisted for up to 140 days after initiation of treatment, suggesting that a cure rate of 20% might be achievable with a single course of KU-0060648 plus etoposide in this model. These data strongly suggest that DNA-PKcs inhibitors either alone or in combination with DSB-inducing genotoxic agents might be a valuable strategy to target ATM-defective human cancers.

Fig. 7 Validation of DNA-PKcs as a drug target for the treatment of ATM-defective B-NHL in vivo.

(A and B) RNAi-mediated suppression of ATM in Eμ:MYC;ARF−/−-driven lymphomas confers etoposide resistance and DNA-PKcs dependence in vivo. Lymphoma cells were transduced with luciferase shRNA– or ATM shRNA–expressing retroviruses and injected into isogenic C57BL/6 recipient mice. Upon lymphoma manifestation, animals were treated with one course of KU-0060648 (KU) [blue lines in (A) and (B)], etoposide [red lines in (A) and (B)], or a combination of both compounds [purple lines in (A) and (B)] or left untreated [black lines in (A) and (B)]. Overall survival is shown in Kaplan-Meier format. Recording of survival was initiated on day 1 of each treatment regimen. In total, 23 mice in each treatment cohort carrying luciferase shRNA–expressing lymphomas and 34 animals in each treatment cohort carrying ATM shRNA–expressing lymphomas were included. Statistically significant survival differences are indicated (two-tailed Student’s t test). (C) Proposed mechanism of DNA-PKcs addiction of ATM-defective cancer cells and the therapeutic targeting of the synthetic lethal interaction between ATM and PRKDC.

Discussion

Loss of ATM is associated with addiction to DNA-PKcs

Loss of ATM in neoplastic disease is associated with resistance against genotoxic therapies (4, 710, 39, 40). This resistance has been attributed to a functional interception of the ATM/Chk2/p53 DDR signaling axis, which relays the presence of genotoxic lesions to an apoptotic cellular outcome (10, 40) (Fig. 7C). However, ATM not only mediates apoptosis but also plays a critical role in HR-mediated DSB repair (16, 18, 19). Thus, apoptosis resistance is associated with reduced HR-driven DSB repair capacity in ATM-defective neoplastic disease. The observation that ATM-defective cells not only proliferate but also are resistant against genotoxic chemotherapy suggests that these cells can repair DSBs. Mammalian cells exploit two major DSB repair pathways—the ATM-dependent HR pathway and DNA-PKcs–mediated NHEJ. We show that ATM-defective cells rely on functional DNA-PKcs signaling for their survival, even in the absence of exogenously induced DNA damage. Our data suggest that the NHEJ pathway is a backup pathway for DSB repair in ATM-defective HR-impaired cells. Thus, although isolated loss of ATM appears to protect cancer cells from genotoxic stress by blunting the proapoptotic p53 response, it renders these cells exquisitely susceptible to DNA-PKcs inhibition. Our data are reminiscent of those seen in a synthetic lethal interaction between BRCA1/2 and PARP1 (41, 42). Cells exposed to a PARP1 inhibitor accumulate DSBs, likely as a result of impaired base excision repair. Although these lesions are typically resolved through HR-dependent DSB repair, BRCA1/2-defective cells, as a result of their inherent HR defect, fail to repair PARP1 inhibitor–induced DSBs, ultimately resulting in cell death (41, 42).

Targeting the synthetic lethal interaction between ATM and PRKDC

We focused on the exploitability of DNA-PKcs inhibitors for the treatment of ATM-defective cancers. We showed that both cancer-associated ATM mutations and ATM depletion resulted in DNA-PKcs dependence. To validate DNA-PKcs as a drug target for the treatment of ATM-defective human cancers, we used two distinct DNA-PKcs inhibitors, namely, KU-0060648 [a dual DNA-PKcs and phosphatidylinositol 3-kinase (PI3K) inhibitor (43)] and NU7441 [a DNA-PKcs inhibitor with only weak activity against PI3K (44)]. Both compounds displayed cytotoxic activity specifically in ATM-defective cells. Given that both compounds display at least some degree of activity against PI3K as an off-target effect, it is conceivable that the effects we observed were at least partially due to PI3K inhibition. However, we believe that the contribution of PI3K inhibition to the cytotoxic effects of KU-0060648 and NU7441 is only marginal because RNAi-mediated DNA-PKcs depletion in ATM-defective human cancer cells mimicked the cytotoxic effects of KU-0060648 and NU7441. Furthermore, DNA-PKcs depletion produced dose-dependent effects, with potent shRNAs resulting in complete prevention of cell proliferation, whereas a less efficient shRNA allowed minimal residual growth of ATM-defective cells. Together, these data indicate that activity against DNA-PKcs mediates the cytotoxic effects of KU-0060648 and NU7441 in ATM-defective cells.

Clinical perspective

Disabling ATM mutations occur in about 10% of human tumors (5, 10). Recently, two large CLL genome resequencing projects analyzing distinct patient cohorts have been published (45, 46). One cohort only included treatment-naïve patients (45), whereas the second cohort also included pretreated patients (46). Consistent with our hypothesis that ATM deficiency is associated with resistance against frontline genotoxic chemotherapy, the number of ATM mutations was lower in the untreated cohort [4 of 105 patients (45)] than in the cohort that included pretreated patients [8 of 91 patients, 5 of whom were in the pretreated group (46)]. These independent observations suggest that ATM mutations accumulate in therapy-refractory CLL patients. Therapeutic options are currently very limited for these patients because they typically do not qualify for allogeneic transplantation. Thus, it will be interesting to test DNA-PKcs inhibitors in CLL patients who have been stratified on the basis of their ATM status. One such molecule might be CC-115, a dual mammalian target of rapamycin (mTOR)/DNA-PKcs inhibitor, currently in phase 1 clinical trials (47). Additional tumors in which ATM is frequently inactivated include head and neck squamous cell carcinoma and mantle cell lymphoma (4850). Given the availability of DNA-PKcs inhibitors that are in clinical testing, as well as the various human malignancies with high rates of ATM inactivation, there will be ample opportunity to validate our findings in human patients with ATM-defective neoplastic disease.

Materials and Methods

Lymphoma model

C57BL/6J recipient mice were anesthetized with isoflurane, and 1.5 × 106 Eμ:MYC;ARF−/− lymphoma cells were injected intravenously. Lymphoma cells had been isolated from the spleen of Eμ:MYC;ARF−/− lymphoma-bearing animals. Lymphoma burden was monitored by palpation of the axillary and brachial lymph nodes. Upon the appearance of substantial tumor burden (palpable lesions with a diameter of >0.5 cm, usually 11 to 13 days after injection), mice were exposed to the indicated treatments. KU-0060648 was administered at 10 mg/kg, twice daily, on days 1 to 4, and etoposide was given at 20 mg/kg, once daily, on days 1 to 4. Overall survival was measured as an end point of the current study. Experiments were approved by the local animal care committee of the University of Cologne.

Statistics

Values reported represent means ± SEM. P values were calculated with GraphPad Prism, with P < 0.05 considered significant. Experiments were done 3 to 12 times, and the particular statistical analyses used in the experiments are noted in the figure captions. Statistics were performed to illustrate significance between groups where n ≥ 3.

Cell culture methods, virus production, immunoblotting, immunofluorescence, clonogenic survival assay, fluorescence-activated cell sorting and FISH analyses, and all reagents are described in detail in the Supplementary Materials and Methods.

Supplementary Materials

www.sciencetranslationalmedicine.org/cgi/content/full/5/189/189ra78/DC1

Materials and Methods

Fig. S1. shRNA-mediated knockdown of DNA-PKcs.

Fig. S2. ATM complementation of HT144 and H1395 cells.

Fig. S3. shRNA-mediated knockdown of ATM in A375 and A549 cells.

Fig. S4. shRNA-mediated knockdown of ATM in Eμ:MYC;ARF−/− lymphoma cells.

Fig. S5. ATM depletion in Eμ:MYC;ARF−/− lymphoma cells leading to DNA-PKcs hyperactivation.

Fig. S6. Induction of apoptosis in DNA-PKcs inhibitor–treated ATM-defective cells rescued through Chk1 inhibition.

Fig. S7. Induction of apoptosis in DNA-PKcs inhibitor–treated ATM-defective cells rescued by suppressing Puma.

Fig. S8. shRNA-mediated knockdown of CtIP in ATM-defective HT144 and H1395 cells.

Fig. S9. Prevention of apoptosis by CtIP repression in DNA-PKcs inhibitor–treated ATM-defective cells.

References and Notes

  1. Funding: This work was supported by the Volkswagenstiftung (Lichtenberg Program to H.C.R.), the Deutsche Forschungsgemeinschaft (KFO-286, RE2246/2-1 to H.C.R.), the Helmholtz-Gemeinschaft (Preclinical Comprehensive Cancer Center to H.C.R.), the Max Planck Society (M.I.F.A.NEUR8061 to R.K.T.), the Ministry for Science and Technology, NRW (MIWT, 313-005-0910-0102 to H.C.R.), and Deutsche Jose Carreras Stiftung (DJCLS-R12/26 to L.P.F. and H.C.R.). Author contributions: H.C.R. designed and supervised the experiments and wrote the manuscript. S.C., M.D., and C.F. conducted the in vitro experiments and wrote the manuscript. A.R., P.M., G.S.H.-S., and J.M.-B. performed the in vivo experiments. M.H., L.P.F., and R.K.T. provided cell lines, patient material, and equipment. K.-A.K. performed FISH analyses. J.W. contributed to the writing of the manuscript, designed qPCR experiments, and performed statistical analyses. Competing interests: R.K.T. is a founder and shareholder of Blackfield AG, a company focused on cancer genome diagnostics and cancer genomics–based drug discovery. R.K.T. received consulting and lecture fees (Sanofi-Aventis, Merck, Roche, Lilly, Boehringer Ingelheim, AstraZeneca, Atlas-Biolabs, Daiichi-Sankyo, and Blackfield AG) as well as research support (Merck, EOS, and AstraZeneca). All other authors declare that they have no competing interests. Data and materials availability: HT144, H1395, A375, and A549 cell lines are available from commercial suppliers. Sharing of cell lines is restricted due to agreements with the cell line providers. ATM shRNA was from Y. Shiloh, CtIP shRNA was from B. P. Sleckman, and DNA-PKcs shRNA was from J. B. Lazaro. Full-length ATM complementary DNA in pHyRING was provided by X. O. Breakefield. All of these plasmids are subject to material transfer agreements, and sharing of these vectors is thus restricted.
View Abstract

Navigate This Article