ATM Kinase Inhibition Preferentially Sensitizes p53-Mutant Glioma to Ionizing Radiation

This article is featured in Highlights of This Issue, p. 3109

Each year in the United States about 7 of 100,000 people are diagnosed with high-grade glioma, classified by the World Health Organization (WHO) as grade 3 and 4 gliomas. Grade 4 is also referred to as glioblastoma multiforme (GBM). Hallmark characteristics of GBM include uncontrolled cell proliferation, diffuse infiltration, and resistance to apoptosis (1, 2). These characteristics, at least in part, account for GBM's poor prognosis and resistance toward radio- and chemotherapy, and a median survival of just 12 to 15 months (3). Primary GBM (type II) arises de novo with no prior symptoms or evidence of lower grade tumor and occurs most frequently in individual 60 years or older. Secondary GBM (type I) occurs in younger patients from lower grade brain tumors. Type II constitutes the majority (>90%) of all GBMs, whereas the remaining are type I. It has been suggested that genetic mutations and amplifications affecting signaling pathways may account, in part, for GBM's aggressive nature (4, 5). GBMs are also classified into 4 different subgroups based on gene expression profiling: (i) classical, (ii) mesenchymal, (iii) neural, and (iv) proneural (6, 7). Type II GBM is mostly found in the classical subgroup with EGF receptor (EGFR) amplifications and PTEN deletions, as well as some with mutations in p53 (8, 9). On the other hand, type I is usually found in the proneural subgroup with frequent mutations in platelet-derived growth factor receptor (PDGFR), IDH1, and p53 (6). The frequency of p53 mutations in type I GBM is 65% or greater, whereas classical GBM harbors p53 mutations 30% of the time (10).

Mutation of the ataxia–telangiectasia (A–T) mutated (ATM) gene is the underlying cause of the rare autosomal recessive disease A–T with an associated extreme radiosensitivity (11). In addition to ATR (A–T and Rad3-related kinase) and DNA-PK (DNA-dependent protein kinase), ATM plays a major role in regulating the DNA damage response (DDR), in particular after ionizing radiation (12). Importantly, radiation activates DNA damage checkpoints and increase DNA repair capacity of cancer stem cells leading to glioma radioresistance and tumor recurrence (13). Therefore, inhibitors of the DDR represent very attractive therapeutic agents worthy of exploration as potential adjuvants to standard GBM therapy (4, 14, 15).

Currently, standard treatment of GBM is surgical resection followed by temozolomide and radiation (16). This therapy results in a median survival of just over 1 year; thus, it is critical that more effective treatments are developed and explored (17). The last few years have seen the advancement of small-molecule inhibitors that disrupt ATM function and sensitize tumor cells to radiation and chemotherapeutic agents (14). One of the first specific ATM inhibitors developed was KU-55933, which binds competitively to the ATP-binding pocket of the ATM kinase (18). Another ATM inhibitor, CP466722, was also described recently (19). KU-60019 is an improved, second-generation analog of KU-55933 >30-fold more effective at blocking radiation-induced phosphorylation of key ATM targets and possesses an impressive ability to radiosensitize human glioma cells in vitro (20). In addition, we also showed that KU-60019 inhibited glioma cell migration and invasion (20), perhaps providing a second level of tumor growth control in between radiation fractions (21). However, the ability of KU-60019, or any other ATM inhibitor to radiosensitize tumors in vivo, has not yet been reported.

We show herein that orthotopic gliomas are radiosensitized with the ATM kinase-specific inhibitor, KU-60019, resulting in significant increases in the survival of mice. Importantly, p53-mutant glioma is much more sensitive to the combination of KU-60019 and radiation resulting in impressive extended survival, and in some cases, the apparent curing of treated mice.

Antibodies used in both Western blotting and/or immunohistochemistry were anti-p-(S1981)-ATM (Millipore); anti-ATM (GeneTex); anti-Ki67 (Novus Biologicals); 53BP1 (BD Biosciences); p53 (Cell Signaling Technology); glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Millipore); mouse anti-γ-H2AX (Millipore); anti-p-(S824)-KAP1 (Abcam); Alexa Fluor 594 goat anti-mouse immunoglobulin G (IgG), and Alexa Fluor 488 goat anti-rabbit IgG (Invitrogen). Matrigel (BD Biosciences) was used for tumor implantation. KU-60019 was obtained from AstraZeneca.

Human glioblastoma U1242/luc-GFP cells have been described previously (20). U87/luc-DsRed and U87/luc-DsRed-p53(281G) cells were generated by infection of U87 cells with Fluc-DsRed2 lentivirus (22) followed by cell sorting, and subsequent infection with LNCX-p53 (281G) retrovirus and selection with G418. U1242/luc-GFP-p53(281G) cells were generated the same way. Cell characterization included the ability to form intracranial tumors in athymic mice, and quantitative reverse transcriptase PCR expression profiling (U1242). Cells were not tested and authenticated by an external service provider. Pooled populations of cells were used in all instances. Irradiations were conducted using an MDS Nordion Gammacell 40 research irradiator with a Cs-137 source delivering at a dose rate of 1.05 Gy/min.

For the continuous KU-60019 exposure growth assay, cells were counted on a hemocytometer and seeded in a 12-well plate. After cells were allowed to grow for 48 hours, culture medium was removed, and medium containing KU-60019 (3 μmol/L) or dimethyl sulfoxide (DMSO) vehicle was added to quadruplicate wells and returned to the incubator. Seven days after initial exposure to drug, each well was trypsinized and counted on a Beckman Coulter counter, and total number of cells for each well was calculated. Counts from quadruplicate wells were averaged, and plotted.

Clonogenic radiosurvival experiments were carried out as described previously (18, 20, 21). Briefly, U87 and U1242 cells and derivatives were diluted, seeded on 10- or 6-cm dishes, and after the cells attached, treated with KU-60019 (3 μmol/L) or not in the medium for 1 hour before irradiation. Cells were incubated overnight, and the medium changed to nondrug containing medium 16 hours postirradiation. Cells were incubated for further 14 days, stained with crystal violet, and colonies consisting of 50 or more cells were counted by eye. Radiosensitization by KU-60019 was determined by fitting radiation survival curves using the linear quadratic equation, and dose enhancement ratios (DER) were determined by dividing the D₃₇ dose of control cells by the D₃₇ dose of treated cells. A DER more than 1 indicates radiosensitization.

Luciferase assay were conducted using a PerkinElmer Precisely (2103 Envision Multilabel plate reader) and luciferase kit from PerkinElmer and used as recommended by the manufacturer. For the radiosurvival assay, U87/luc-DsRed+/-p53-281G cells were serially diluted in 96-well microplates and irradiated with the indicated doses. Three days later, luciferase assay was conducted to determine viable cell numbers. The mouse glioma cells were derived from GBM of “Mut6” (hGFAP-Cre; p53 f/⁻; NF1 f/⁺; PTEN f/⁺) mice (23). Tumors were harvested and enriched for stem-like cells by growing them as neurospheres in serum-free media supplemented with EGF and basic fibroblast growth factor (bFGF; ref. 24). Cells were plated in 96-well plates and treated with KU-60019 at the indicated concentrations and irradiated 3 hours later. After 6 days, cell viability was determined by CellTiter-Glo (Promega). Bioluminescence imaging (BLI) was conducted on a Caliper-IVIS-200 after subcutaneous injection of d-luciferin (Gold Biotechnology) at 30 mg/mL in PBS.

Flow cytometry was carried out as described previously (25). Cells were fixed in 100% methanol and permeabilized in 1% Triton X-100/casein. Cell-cycle distribution was analyzed by 7-amino-actinomycin D (7-AAD) and propidium iodide staining (5 mg/mL, 0.1% Triton X-100/PBS). Bromodeoxyuridine (BrdUrd) flow cytometry was carried out as recommended by the manufacturer (BD Biosciences) as described previously (26). Briefly, U87 and U87-281G cells were irradiated with 10 Gy or left untreated. After 16 hours, BrdUrd was added to the medium (10 μmol/L final), and 5 hours later cells isolated for further processing. Flow cytometry was conducted on a BD Biosciences FACS Canto flow cytometer at the Virginia Commonwealth University (VCU) Flow Cytometry Core Facility. Data were analyzed using the FACSDiva software.

Confocal fluorescence microscopy was conducted as previously described (25). Cells were grown on Lab-Tek glass slides coated with Matrigel. After treatment, cells were fixed with 3% paraformaldehyde, permeabilized with 0.5% Triton-X 100 in PBS, and blocked with casein/0.1% goat serum before exposure to primary antibodies followed by secondary antibodies Alexa 488 goat anti-rabbit or goat anti-mouse 546 Fab fragment (Invitrogen) at 1:500 dilution, and nuclei counterstained with 4′,6-diamidino-2-phenylindole (DAPI; 1 mg/mL). Cells were imaged and analyzed using a Zeiss LSM 510 Meta imaging system in the VCU Microscopy Facility.

Following treatment with irradiation and KU-60019, animals were transcardially perfused with 3% paraformaldehyde followed by postfixing of brains and treatment with 30% sucrose until no longer buoyant. Brains were cut coronally through the injection site, and the hindbrain embedded in Tissue-Tek OCT Compound (Sakura), frozen on dry ice, and stored at −20°C. Cryosectioning was carried out on a Leica CM1850 cryostat with 6-μm thickness and slides were stored frozen until staining. Sections were rinsed in PBS, followed by pretreatment in sodium citrate antigen unmasking solution (0.01 mol/L sodium citrate; 0.05% Tween-20; pH 6.0) and boiled followed by aldehyde reduction with NaBH₄. Sections were permeabilized with 1% Triton-X-100/PBS and blocked with 10% normal goat serum in Triton-X-100/PBS before the exposure to primary antibodies followed by secondary antibodies [Alexa Fluor-594 goat anti-mouse IgG or Alexa Fluor-488 goat anti-rabbit IgG (Invitrogen)] and DAPI. Hematoxylin and eosin (H&E) staining was carried out on frozen sections fixed in 3% formaldehyde and stained with modified Mayer's H&E Y. After dehydration with ethanol, sections were cleared with Histoclear (National Diagnostics #HS-200), and coverslips mounted on sections with Permount (Fisher #SP15-100). Sections were imaged and analyzed using a Zeiss LSM 510 Meta imaging system at ×63 magnification in the Massey Cancer Center Flow Cytometry and Imaging Facility or using an Applied Imaging Ariol SL-50 system (Applied Imaging Corp.).

To ensure tumor take and equal tumor size distribution throughout the treatment groups, all mice injected with U1242/luc-GFP or U87/luc-DsRed cells were imaged using BLI typically at 5, 16, and 32 days postcell injection using a Caliper IVIS-200. If necessary, later images were also acquired. Fixed and unfixed brains were imaged on a Zeiss SV11 stereomicroscope (Zeiss) with a Hamamatsu digital CCD camera C4742-95-12NRB (Hamamatsu) and AttoArc2/HB 100 Arc variable intensity microscopy illuminator system (Zeiss) using excitation/emission filters (GFP; 470/525, DsRed; 546/575), and a Caliper IVIS-200 system using GFP filters. Images were captured using AxioVision 3.1 or Living Image 4.0 software, and processed with Adobe PhotoShop 5.0.

All procedures were carried out in accordance with protocol AM10197 approved by the VCU (Richmond, VA) Institutional Animal Care and Use Committee (IACUC).

All mice were subjected to a light/dark cycle of 12 hours with water and food available ad libitum. Athymic nude female mice (5–6-weeks old) were purchased from National Cancer Institute (NCI)-Frederick (Gaithersburg, MD). Mice weighing between 15 and 25 g were anesthetized intraperitoneally with a mixture of ketamine and xylazine (10 mg/kg; 1 mg/kg) in saline, and placed on a heating pad. A 1-cm sagital incision was made, and the skull exposed. A 1-mm burr hole was made 2 mm left lateral and 1 mm posterior to the bregma. A small pocket was created in the brain by inserting a 26-gauge needle with a beveled edge to a depth of 3 mm. This location was found optimal for the insertion of the Brain Infusion Kit 3 (ALZET) cannula connected to an osmotic pump and was therefore used throughout these experiments. U1242/luc-GFP, U87/luc-DsRed, or derivatives thereof were trypsinized, washed, and resuspended in PBS and Matrigel (BD Biosciences), and loaded into a Hamilton syringe fitted with a 26-gauge blunt tip needle. The needle was inserted to a depth of 3 mm and 1 μL of cells was injected over a 5-minute period. After injection, the needle was removed over a 5-minute period.

Osmotic pumps (ALZET M2002, M1002, M1007D), containing KU-60019, were implanted 7 days postcell injection to allow continuous drug delivery to the tumor site for up to 19 days (5.28 μL/d; 100 μL total volume) followed by radiation. An incision was made to expose the burr hole previously made for cell injection and cleaned to remove all bone wax. The pump was inserted and the cannula tip was positioned into the burr hole and glued into place. All radiations were carried out using a MDS Nordion Gammacell 40 research irradiator with a 137-Cs source delivering a dose rate of 1.05 Gy/min to the head only.

Convection-enhanced delivery (CED) of KU-60019 was done on day 6, 9, and 12 postcell injection. A guide screw with a stylet (PlasticsOne, Inc.) was inserted into the burr hole to facilitate cell placement and subsequent drug infusions and the incision closed using vet-glue (GLUture Topical Tissue Adhesive; Abbott Laboratories). Next, 12.5 μL of 250 μmol/L of KU-60019 in PBS was directly infused into the tumor via a cannula inserted into the guide-screw by CED using a BeeHive pump (BeeHive; Bioanalytical Systems) set at rate of 0.4 μL/min. Vehicle alone (PBS) was infused into mice without KU-60019 treatment. Immediately following surgery, mice either received 3 Gy to the head or left untreated. For subsequent treatments, the skin was opened and the stylet removed before repeating the infusion followed by irradiation.

Mouse survival studies were modeled using Kaplan–Meier curves. Statistical analyses were conducted using log-rank test and mean survival in days was reported with SD for all treatment groups. Clonogenic radiosurvival experiments were analyzed by two-way ANOVA with radiation survival fraction as the response to cell type and radiation dose along with interaction. Least-squared means of the treatment difference were obtained from the model and tested with ANOVA F test. All statistical analyses were done in SAS v9.2.

To confirm and validate the target for KU-60019 as the ATM kinase, U1242 and U87 cells were exposed to KU-60019 and radiation and extracts prepared for Western blotting with p-(S1981) ATM antibody. S1981 is a well-established radiation-induced auto-phosphorylation site (27). Both cell lines increased p-(S1981) ATM levels after exposure to radiation and KU-60019 suppressed this increase showing that ATM kinase was inhibited (Fig. 1A). Furthermore, U1242 and U87 cells were similarly radiosensitized in vitro in the presence of low levels of KU-60019 (Fig. 1B). U1242 cells show high DNA repair capacity (curved slope with low α/β ratio), whereas U87 cells have a straighter curve and thus lower repair capacity. Nevertheless, KU-60019 was able to radiosensitize both cell lines to a similar extent. When p53 null glioma cells isolated from a spontaneous mouse GBM model were grown as neurospheres and tested they were also found to be radiosensitized by KU-60019. Altogether, KU-60019 is an effective and specific radiosensitizer of established human glioma cells and primary mouse glioma stem cells in vitro in agreement with earlier studies (20, 21, 28, 29).

One characteristic phenotype of human glioma is its invasiveness, which is rarely seen when human xenografts established from tissue culture cell lines are propagated as orthotopic tumors in mice. The U1242 human glioma cell line is unusual in that it has a stable, invasive, and very aggressive phenotype when grown in nude mice (30, 31). We expressed the reporter genes luciferase and GFP, or DsRed, from lentivirus to follow tumor growth. Although U1242 cells spread invasively in the brain with high mitotic index, similar to human disease, U87 cells form encapsulated tumors (Fig. 2 and Supplementary Fig. S1). We chose to test KU-60019 on U1242 tumors in our initial experiments because these cells recapitulate the pathology of the human disease to some extent. ATM phosphorylates many proteins in response to radiation and DNA damage (32). Among the best characterized targets are H2AX, KAP1, and 53BP1, all linked to processes associated with DNA double-strand break repair. When mice growing intracranial tumors were irradiated, we were able to document dose-dependent increases in γ-H2AX, p-(S824)-KAP1, and 53BP1 repair foci immediately following radiation (Fig. 3A–C), and in a time course (Fig. 3D). These results show that invasive human glioma U1242 cells show evidence of radiation-induced DDR, permitting the analysis of KU-60019–dependent ATM inhibition in mice.

When KU-60019 is systemically administered either intraperitoneally or orally, plasma levels only reach the low micromolar range, which is barely able to radiosensitize flank tumors (21). The blood–brain barrier (BBB) and blood–tumor barrier (BTB) represent major obstacles in the effective treatment of brain tumors (33). To reach meaningful drug concentrations of KU-60019 within the tumor, the BBB/BTB need to be bypassed or drugs administered locally. Both osmotic pumps, as well as clinically used CED, partially bypass the BBB/BTB and deliver drugs directly to the tumor to improve efficacy and reduce potential systemic toxicity (33, 34). We first explored the use of commercially available osmotic pumps to deliver KU-60019 at effective levels to the brain. Preliminary results with a dye acting as a surrogate for KU-60019 suggested that efficient spread occurred (Supplementary Fig. S2). When mice harboring U1242/luc-GFP tumors were treated with KU-60019 administered from an osmotic pump followed by 2 Gy of irradiation, we were able to show a complete elimination of tumor cells on day 20 by BLI 7 days after radiation (Fig. 4A, top), a finding that was confirmed by GFP imaging of the extracted brain (Fig. 4A, bottom). This result suggested that osmotic pumps administer KU-60019 to sufficient levels to radiosensitize U1242 gliomas.

Encouraged by this marked result, we then designed an experiment to determine the effect of KU-60019 and radiation on the survival of mice. Mice with U1242/luc-GFP intracranial tumors were implanted with osmotic pumps loaded with KU-60019 7 days after cell implantation, followed by a single dose of radiation on day 13. Mice treated with KU-60019 and radiation survived for 109 days, significantly longer (P = 0.004 vs. all other treatments) than mice treated with KU-60019 alone (35 days), radiation alone (34 days), or after no treatment (27 days; Fig. 4B). However, we noticed a considerable variation in the survival of the mice in the KU-60019 and radiation group. Some mice barely showed a survival benefit, whereas others (e.g., A51 and A53) showed complete responses and survived for more than 200 days. Despite initially having tumors, at the time euthanasia these mice showed no signs of the disease (Supplementary Fig. S3). We suspected these differences in survival were a result of variable drug delivery throughout the brain because of insufficient ability of this type of pump to apply positive pressure. Therefore, we decided to administer KU-60019 in a manner used clinically, that is, by CED, immediately before irradiation. We also decided to repeat this procedure 3 times in line with clinical practice and possibly achieve a better therapeutic response, whereas using a low dose of radiation to maximize detection of the KU-60019 effect.

To compensate for the relatively small volume that can be infused by CED over a reasonable time, the concentration of KU-60019 was increased to 250 μmol/L. Using these conditions, drug distribution in the mouse brain was determined by liquid chromatography/mass spectrometry (LC/MS) analysis and shown to be concentrated around the infusion site spreading 1 to 2 mm longitudinally (Supplementary Table S1). KU-60019 reached peak concentrations of approximately 5 μmol/L, above those needed to inhibit ATM (21). Mice with U1242/luc-GFP intracranial tumors were infused with KU-60019 by CED on day 6, 9, and 12, immediately followed each time with 3 Gy of radiation (Fig. 5A). Mice treated with KU-60019 and radiation survived for 60 days, significantly longer (P = 0.00238 vs. all other treatments) than mice treated with KU-60019 alone (38 days), radiation alone (38 days), or after no treatment (35 days). Increased survival after KU-60019 and radiation was reflected in the reduction of tumor growth determined by BLI on day 5, 16, and 30 (Supplementary Fig. S4). To ensure that the levels of drug delivered into the brain were sufficient to block ATM kinase activity, we examined the ability of the drug to inhibit γ-H2AX foci formation (25, 35). We found that KU-60019 was able to inhibit foci formation in both tumor (elongated nuclei) as well as mouse brain (round nuclei with characteristic, large nucleoli; Fig. 5B and C). A similar reduction was seen when p-(S824)-KAP1 foci was examined (Supplementary Fig. S5). These results suggest that CED is able to effectively administer KU-60019 and penetrate diseased brain to radiosensitize the tumor. In addition, normal brain showed a similar response to radiation and KU-60019 in what seemed to be a subpopulation of cells, but presently, we do not know what these cells are or the impact on the brain. Nevertheless, a significant prolongation of survival was seen with KU-60019 and radiation at a radiation dose that produced negligible effects on survival when used as a single agent.

To explore the ability of KU-60019 to radiosensitize other human gliomas, we turned to U87 cells, which have been used extensively in the past in similar studies. Furthermore, U87 cells express wild-type p53 making it possible to manipulate p53 status and show possible effects on radiosensitization and the response to KU-60019. To generate U87/luc-DsRed cells with mutant p53, we infected cells with a retrovirus expressing the p53-281G allele (36), and selected for G-418 resistance resulting in constitutive expression of the p53-281G protein in untreated cells (Fig. 6A). These cells were then characterized in response to KU-60019 and radiation. As expected, the U87-281G cells have an impaired G₁–S checkpoint after radiation, are significantly more sensitive to KU-60019 and radiation alone (20, 21), and to KU-60019 and radiation combined relative to the parental U87 cells (Fig. 6B–E). Radiation followed by BrdUrd labeling showed that while entry into S-phase was reduced by almost 50% (28.2% to 14.4% BrdUrd⁺ cells) in U87 cells, the levels in U87-281G cells were barely reduced (25.1% to 23.2% BrdUrd⁺; Fig. 6B). Similar reductions were seen when cells were examined by fluorescence microscopy (Supplementary Fig. S6). Thus, the G₁–S checkpoint is intact in U87 cells and impaired in U87-281G cells as expected (37). In addition, both U87 and U87-281G cells showed distinct G₂–M arrest after radiation, which was more clearly seen with U87-281G cells because of the disrupted G₁–S checkpoint.

Next, mice were implanted with either U87 or U87/p53-281G cells followed by treatment with KU-60019 and radiation to determine the effect on survival. Mice with U87 gliomas treated with KU-60019 delivered by CED on day 6, 9, and 12 followed by radiation survived for 65 days, whereas mice treated with KU-60019 alone survived for 70 days, radiation alone for 49 days, and untreated mice for 46 days (Fig. 6F, bottom). Although mice in the KU-60019 and radiation combination group survived significantly longer than radiation alone mice (P = 0.0448), there was no difference in survival between this group and the KU-60019 alone group. There was a statistical difference in survival between KU-60019 alone and the no treatment groups (P = 0.00019). Thus, U87 glioma is relatively refractory to KU-60019 radiosensitization. On the other hand, mice with U87-281G tumors (Fig. 6F, top), treated with the KU-60019 and radiation combination survived more than 160 days, significantly longer (P = 0.00011 vs. all other treatments) than mice treated with KU-60019 alone (51 days), radiation alone (44 days), or after no treatment (42 days). Similar to the results with mice harboring parental U87 tumors, mice treated with KU-60019 alone showed a trend toward longer survival but this difference was not significant (P = 0.13). Thus, genetically matched p53-mutant U87 tumors dramatically responded to KU-60019 radiosensitization resulting in a subset of mice in this group apparently cured of disease. In addition, there was a trend of increased survival of mice treated with KU-60019 alone in both the p53 wild-type and 281G-mutant sets lending support to the idea that inhibiting ATM might reduce tumor dispersal in between radiation fractions (21).

Before we had verified the p53 status of the U1242 cells as having the R175H mutation, in agreement with previous reports (38, 39), we had already infected these cells with the retrovirus expressing the p53-281G allele and initiated a survival experiment. In support of the data obtained with the U87 p53-matched experiment, we noticed a significant improvement in the response to KU-60019 and radiation with U1242/p53-281G tumors compared with parental U1242 tumors (Supplementary Fig. S7). Mice with U1242/p53-281G tumors treated with KU-60019 followed by radiation, survived for 145 days, significantly longer (P = 0.0004) than mice treated with KU-60019 alone (66 days), radiation alone (58 days), or no treatment (52 days). Also, mice with U1242/p53-281G-mutant tumors responded significantly better to treatment than mice with parental U1242 tumors across all treatments (P = 0.0004), corroborating the results with the U87 set that glioma with a p53 mutation is more sensitive to KU-60019 radiosensitization. The mechanism underlying the enhanced response as a result of manipulating p53 status in the U1242 cells is not known but currently under investigation. It is possible that 281G and 175H cooperate in some fashion to make these cells more sensitive to radiosensitization by KU-60019 (40), or, alternatively, 175H might gain some function when 281G is also expressed resulting in a more responsive phenotype.

Inhibiting ATM for cancer therapy has been a long-sought goal for radiobiologists and oncologists alike (14). With the advent of highly specific inhibitors for the ATM kinase, this has recently become a reality (18–20). We have shown herein that the ATM kinase inhibitor, KU-60019, was able to radiosensitize gliomas grown orthotopically in mice. Radiosensitization was most striking in U87 gliomas expressing mutant p53, where a large proportion of the mice seemed to have been cured of disease.

There are no known unique ATM phosphorylation sites (32). However, H2AX and KAP1 are considered more specific targets for the ATM kinase at early times after radiation, whereas the related DNA-PK and ATR kinases serve as complementary backups for ATM at later times (20, 41–43). We were able to validate ATM inhibition by KU-60019 in vivo by showing reduced γ-H2AX, as well as p-(S824)-KAP1, foci formation immediately following irradiation in both the tumor and mouse brain. These results suggest that KU-60019 is able to inhibit the ATM kinase in tissue regardless of whether it is cancerous or not. Because increased survival correlated well with the ability of KU-60019 to suppress foci formation, this suggests that interfering with ATM signaling caused the reduction in tumor growth and prolonged the survival of mice.

We have shown in this study that glioma with mutant p53 is much more sensitive to KU-60019–dependent radiosensitization. This result was not unexpected as several recent studies have shown that human lymphomas with double mutations in p53 and ATM, or Chk2, were much more sensitive to chemotherapy than tumors with a single mutation (44–46). Several excellent reviews have discussed therapeutic strategies based on the DDR synthetic lethality principle (47, 48). In addition, caffeine, a broad-spectrum phosphatidyl inositol 3′ kinase-related kinase inhibitor, was shown to radiosensitize p53 null but not matched p53 wild-type cells in vitro (49, 50). The results from our study underscore these previous findings and extend this concept by showing enhanced radiosensitization of p53-mutant glioma with an ATM kinase inhibitor in vivo. Importantly, tumors with normal p53 were found to be protected against chemotherapy (44). Thus, one would assume that normal tissues and cells would be protected as well. It was recently shown that under conditions in which the cell-cycle kinase Cdk1 was inhibited or depleted, normal, nontransformed cells were protected against DNA-damaging drugs while tumor cells were efficiently killed (51). We are also encouraged by the result that p53 null mouse glioma stem cells were radiosensitized by KU-60019, a result in agreement with a previous study (29). Therefore, targeting the ATM kinase with a small-molecule inhibitor in p53-mutant glioma would be very attractive as it is expected that the glioma would be sensitive to radiation, whereas normal tissue would be protected or less affected.

Even though U1242 tumors harbor the p53 R175H allele, they were not as sensitive to KU-60019 and radiation as the U87/p53-281G tumors based on survival. We speculate that this is due to the ability of U1242 tumors to spread throughout the brain (see Supplementary Fig. S1), and perhaps evade the drug. On the other hand, U87 tumors are encapsulated and more contained, thus providing a more accessible, physical target for KU-60019. Therefore, it is likely that mice harboring U1242 tumors die from tumor dispersal throughout the brain parenchyma, whereas mice with U87 tumors die from increased intracranial pressure produced by the tumor. The differential response to KU-60019 and enhanced radiosensitization seen in vivo with p53-mutant gliomas is not reflected in vitro as both U87 (wild-type) and U1242 (p53 mutant) are similarly radiosensitized (see Fig. 1). However, these 2 cell lines are derived from different individuals with many genetic differences in addition to p53 that would likely influence the response to KU-60019 in vitro. In sharp contrast, the p53 effect can be seen clearly in vitro when isogenic cells differing only in p53 status are compared as we did with the U87 cells (see Fig. 6C–E). We attribute differences in results between experiments using osmotic pump and CED a reflection of different experimental conditions and duration of drug exposure. Ongoing studies using panels of human and mouse gliomas and primary glioma stem cells with varying p53 status are exploring the use of higher doses of radiation to maximize survival, and identifying potential side effects of KU-60019 on normal brain. Whether KU-60019 radiosensitization seen at higher doses of radiation in vitro would also be observed in vivo is currently not known. However, at higher radiation doses one would not be able to take full advantage of the radiosensitizer and spare normal tissue as higher radiation doses will likely damage the brain despite having normal p53. Furthermore, whether treatment with KU-60019 alone provides a survival benefit is currently under investigation.

Local drug delivery increases the therapeutic effect and at the same time reduces general toxicity to treated subjects. CED is becoming a more frequent experimental treatment option in the management of brain tumors but adds complexity and cost to the procedure. In addition, technical difficulties associated with cannula placement verification and the prevention of drug backflow are issues that still need to be resolved (33). Although the effect of KU-60019 radiosensitization on tumors is clear, the effect on mouse brain is not known. This issue has not been a primary concern when patients survive for just a little more than a year, at best. We have not observed gross abnormalities on normal mouse brain after treatment with KU-60019 and radiation. Further studies on the response of normal tissue to KU-60019, and the brain in particular, with and without radiation, are therefore needed as prolonging survival of subjects would likely expose any detrimental late effects on brain function and cognition due to treatment. However, it is important to point out that all treatments induce damage in normal tissues, a side effect of cancer therapy and fact that needs to be balanced against therapeutic gain and quality of life. Delivery of an ATM inhibitor locally by CED should in part alleviate such concern.

Altogether, our results suggest that ATM kinase inhibition may be an effective adjuvant therapy for patients with mutant p53 brain cancers. Because a third of all GBM cases have p53 mutations, this population of patients might be a particularly attractive target group.

No potential conflicts of interest were disclosed.

Conception and design: L. Biddlestone-Thorpe, M. Sajjad, M.J. O'Connor, K. Valerie

Development of methodology: L. Biddlestone-Thorpe, M. Sajjad, E. Rosenberg, A. Khalil, D.G. Temesi, N.D. Mukhopadhyay, K. Valerie

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Biddlestone-Thorpe, M. Sajjad, E. Rosenberg, J.M. Beckta, N.C.K. Valerie, M. Tokarz, A.F. Wagner, D. Gilfor, S.E. Golding, D.G. Temesi, A. Lau, K.S. Choe, L.F. Parada, S.K. Lim

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Biddlestone-Thorpe, M. Sajjad, E. Rosenberg, J.M. Beckta, B.R. Adams, S.E. Golding, S. Deb, A. Lau, K.S. Choe, N.D. Mukhopadhyay, K. Valerie

Writing, review, and/or revision of the manuscript: L. Biddlestone-Thorpe, M. Sajjad, J.M. Beckta, N.C.K. Valerie, A.F. Wagner, A. Khalil, S.E. Golding, S. Deb, A. Lau, M.J. O'Connor, K.S. Choe, S.K. Lim, N.D. Mukhopadhyay, K. Valerie

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Sajjad, E. Rosenberg, M. Tokarz, N.D. Mukhopadhyay

Study supervision: K. Valerie

The authors thank Khalid Shah (Harvard University, Cambridge, MA) for virus reagent and Michael Kastan (Duke University, Durham, NC) for helpful discussion.

This work was supported by NIH Grants R01NS064593 and R21CA156995 (to K. Valerie). L. Biddlestone-Thorpe was supported by T32CA113277. S.E. Golding was supported by American Brain Tumor Association and T32CA085159, A.F. Wagner by T32CA085159, and J.M. Beckta by F30CA171893. Shared Resources used in this project were supported by VCU, the Massey Cancer Center (P30CA016059), and the Department of Neurobiology & Anatomy Microscopy Facility (supported in part by P30NS047463).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.