Small-Molecule Drugs Mimicking DNA Damage: A New Strategy for Sensitizing Tumors to Radiotherapy Free

Cancer treatment primarily consists of surgery whenever possible, chemotherapy and radiotherapy. Chemotherapy and radiotherapy cytotoxicity is at least in part due to unrepaired DNA damage and the ability of cancer cells to recognize DNA damage and initiate repair is an important mechanism of resistance (1–6). Pharmacologic inhibition of DNA repair has the potential to make cancer cells more vulnerable to the damaging effects of therapy, therefore increasing the response to treatment (7). Over the last decade, numerous academic laboratories and pharmaceutical companies have developed molecules to modulate DNA repair by targeting proteins involved in different repair pathways. Briefly, two families of inhibitors were tested: (a) damage signaling inhibitors, targeting poly(ADP-ribose) polymerase (8–10) and ataxia telangiectasia mutated (ATM; ref. 11) and (b) base excision repair via O⁶-methylguanine DNA methyltransferase (12), homologous recombination repair via RAD51, or nonhomologous end joining via the DNA-dependent protein kinase (DNA-PK; refs. 13–18). All inhibition strategies were designed to target one key enzyme and therefore may be thwarted by mutations in the target or overactivation of an alternative repair pathway.

DNA double-strand breaks (DSB) are one of the most severe types of DNA damage, which if left unrepaired are lethal to the cell. Two main DNA repair pathways combat DSBs: homologous recombination and nonhomologous end joining, the latter being the most important in mammalian cells (1). Recruitment of DNA damage signaling and repair proteins to the sites of genomic damage constitutes a primary event required for repair. Many components of the DNA damage response, including ATM, BRCA1, p53-binding protein 1 (53BP1), MDC1, RAD51, and the MRE11/RAD50/Nijmegen breakage syndrome 1 (NBS1) complex, form ionizing radiation-induced foci (19–21). These foci colocalize with γ-H2AX, a phosphorylated form of the H2AX histone (22). The principal mediators of this signaling pathway are the phosphatidylinositol 3-kinase-like family of kinases. At least four phosphatidylinositol 3-kinase-like kinases are involved in the transduction of the signal that originates at broken DNA ends: ATX, ATR, ATM, and DNA-PK, the two later being the most important in the response to irradiation (23, 24). Although γ-H2AX facilitates faithful repair, the biochemical mechanism remains unclear. It is currently admitted that it “tags” the DSBs on chromosomes and facilitates the increase of local concentration of repair factors near the lesion (25–27).

In this study, we propose that introducing short DNA molecules that would be recognized as DSBs by the signaling and repair proteins would hijack them away from the chromosome DSBs to be repaired and consequently increase the sensitivity of the cell to irradiation. To test this hypothesis, we synthesized molecules that were screened for their ability to be recognized by the DNA-PK complex and trigger DNA-PKcs kinase activation. The effect of these molecules, called Dbait (for DSB bait), was first tested in cell culture and then on xenografted tumors.

Cell lines and Dbait. Studies of cultured cells were done using Hep2 (head and neck squamous cell carcinoma), HeLa S3 (epithelial carcinoma of the cervix), MO59K and MO59J (glioblastomas), and MRC5 and AT5BI (fibroblasts). Xenograft was done with human tumor cell lines Hep2 and LU1205 and SK28 (melanomas). Cells were grown at 37°C in monolayer cultures in complete DMEM containing 10% heat-inactivated fetal bovine serum (Invitrogen) and antibiotics (100 μg/mL streptomycin and 100 μg/mL penicillin) under conditions of 100% humidity, 95% air, and 5% CO₂. LU1205 (melanoma) was grown in MCDB containing 4% heat-inactivated fetal bovine serum, 1% glutamine, and antibiotics (100 μg/mL streptomycin and 100 μg/mL penicillin). In vitro, transfection of Dbait molecules was done with 2 μg Dbait and 20 μL Superfect reagent (Qiagen) in 1.2 mL medium (in 60 mm diameter plates). One nanomole of Dbait32H (molecular mass, 20,153) is ∼20 μg. All Dbait molecules were made by automated solid-phase oligonucleotide synthesis (Eurogentec) as described in Supplementary Fig. S1. Sequences are 5′-GCTGTGCCCACAACCCAGCAAACAAGCCTAGA-(H)-TCTAGGCTTGTTTGCTGGGTTGTGGGCACAGC-3′ Dbait32Hc, where H is a hexaethyleneglycol linker; 5′ACGCACGGGTGTTGGGTCGTTTGTTCGATCT-(H)-AGATCGAACAAACGACCCAACACCCGTGCGT-3′ Dbait32H; 5′ACGCACGG-(H)-CCGTGCGT-3′ Dbait8H; and 5′ACGCACGGGTGTTGGGTCGTTTGTTCGATCT-3′ Dbait32ss.

Kinase assay. DNA-PK activity was monitored using the kit SignaTECT DNA-PK Assay System (Promega). The biotinylated peptide substrate, various amounts of nuclear extract (cleaned of endogenous DNA by DEAE-Sepharose filtration), and 20 nmol/L Dbait molecules were incubated for 5 min at 30°C with [γ-³²P]ATP according to the manufacturer's instructions. The biotinylated substrate was captured on a streptavidin membrane, washed, and counted in a scintillation counter. Percentage of phosphorylation is calculated by dividing the bound radioactivity by the total count of [γ-³²P]ATP per sample.

Single-cell gel electrophoresis comet assay. MRC5 cells, mock transfected or transfected by 2 μg Dbait32H, were suspended in 0.5% low melting point agarose in DMEM and transferred onto a frosted glass microscope slide precoated with a layer of 0.5% normal melting point agarose. Slides were irradiated or not and incubated in DMEM under cell growth conditions for various times after irradiation. Slides were immersed in lysis solution [2.5 mol/L NaCl, 100 mmol/L EDTA, 10 mmol/L Tris, 1% sodium lauryl sarcosinate, 10% DMSO, 1% Triton X-100 (pH 10)] at 4°C for 1 h, placed in a electrophoresis tank containing 0.3 mol/L NaOH (pH 13) and 1 mmol/L EDTA for 40 min, electrophoresis for 25 min at 25 V (300 mA), washed with neutral buffer [400 mmol/L Tris-HCl (pH 7.5)], and stained with 20 μg/mL ethidium bromide. The variables of the “comets” were quantified with the use of the software Comet Assay 2 (Perceptive Instrument). Duplicate slides were processed for each experimental point. The tail moment is defined as the product of the percentage of DNA in the tail and the displacement between the head and the tail of the comet (28).

Histology and immunodetection. At various times, animals were sacrificed and tumors were excised and divided in two parts: one half was stored in frozen nitrogen to be used for immunodetection and the other half was placed in neutral buffered formalin before to be embedded in paraffin. Sections (7 μm) were cut and stained with hematoxylin, eosin, and safran. The extent of necrosis (indicated by increased cell size, indistinct cell border, eosinophilic cytoplasm, loss or condensation of the nucleus, or associated inflammation) is expressed as the proportion (%) of the surface area of the tissue section analyzed that was necrotic. The number of mitotic cells and apoptotic cells were estimated from representative nonnecrotic fields of ∼1,000 cells analyzed at high power. Sections of frozen sample were fixed with 4% paraformaldehyde for 20 min before immunodetection by overnight incubation with mouse monoclonal anti-phospho-histone H2AX (Ser¹³⁹; Upstate) diluted 1:200 in 1× PBS, 2% bovine serum albumin. For immunostaining of cells grown on coverslips, we used rabbit polyclonal anti-53BP1 (Cell Signaling Technology), rabbit polyclonal anti-NBS1 (Novus Biologicals), and mouse monoclonal anti-phospho-histone H2AX (Ser¹³⁹; Upstate) diluted 1:200, 1:100, and 1:500, respectively. All secondary antibodies conjugated with Alexa 555, 633, or 488 (Molecular Probes) were used at a dilution of 1:200 and DNA was stained with 4′,6-diamidino-2-phenylindole.

Dbait and irradiation treatments in mice. Hep2, SK28, and LU1205 xenograft tumors were obtained by injecting 10⁶ tumor cells into the flank of adult female nude mice (Charles River). The animals were housed in the laboratory at least 1 week before commencing experiments. There were 6 animals per cage under controlled conditions of light and dark cycles (12 h:12 h), relative humidity (55%), and temperature (21°C). Food and tap water were available ad libitum. After ∼8 days for Hep2 and LU1205 and 12 days for SK28, when the subcutaneous tumors measured 150 to 200 mm³, the mice were separated into homogeneous groups to receive different treatments. Irradiation was done in a ¹³⁷Cs unit (0.5 Gy/min) with a shield designed to protect about two-thirds of the animal's body. Doses were controlled by thermoluminescence dosimetry. A total dose of 30 Gy was delivered in 15 sessions at intervals of 3 sessions of 2 Gy/wk during 5 weeks (Hep2) or in 6 sessions at intervals of 3 sessions of 5 Gy/wk during 2 weeks (LU1205, SK28). Dbait molecules with in vivo-jet polyethylenimine (PEI) reagent (Polyplus Transfection) at the N/P ratio 6 were diluted in 100 μL of 5% glucose. The Dbait-PEI mixture was incubated for 15 min at room temperature before injection. Intratumoral injections of the indicated amount of Dbait were done 5 h before each radiotherapy session. Mock treated animals were injected with 100 μL of 5% glucose according to the protocol of the associated assays. Tumor size was assessed by caliper measurements every 3 days and size was calculated by the formula: 0.5 × length × width². Mice were weighed and pictures of tumors were taken every week. For ethical reasons, the animals were sacrificed when tumors reached 2,000 mm³. This endpoint used in survival analysis was death day. The Local Committee on Ethics of Animal Experimentation approved all experiments.

Statistical analysis. Descriptive analyses of the tumor response were done for each treatment and each tumor type for 150 days after the first treatment session (day 1). Median lifetime was estimated according to the Kaplan-Meier method. Tumor growth delay (TGD) was calculated by subtracting the mean tumor volume quadrupling time of the control group from tumor volume quadrupling times of individual mice in each treated group. The mean TGD was calculated for each treated group using the individual measurements. Overall survival curves were assessed by Kaplan-Meier estimates and compared using the nonparametric log-rank test because the data do not follow a normal distribution. The analysis was done using S-Plus 6.2 version software (MathSoft) and statEL (ad Science). A global log-rank test was first done for each group with the same tumor type, and treatments with Dbait were compared with the mock-treated control. The number of animals (n), the relative risk (RR), and the P value are reported in Table 1.

Design and in vitro screening of the Dbait molecules. We designed small DNA molecules (called Dbait) that mimic DSBs. To favor double-stranded DNA structure in cells, the Dbait molecules were hairpin with one blunt end and the two strands of the other end joined by a hexaethyleneglycol linker. The blunt end was protected from exonuclease attack by substituting the three 3′ and 5′ terminal nucleotide residues with phosphorothioate nucleotides (29). Molecules were screened for DNA-PK activation in cell extracts. As observed previously with unmodified DNA (30, 31), we found that both Dbait32Hc and Dbait32H molecules, which are 32 bp in length and differ only in their sequences, efficiently activate DNA-PKcs kinase activity of the purified enzyme and in cell extract (Supplementary Fig. S1). Short (8 bp) molecules (Dbait8H) as well as single-stranded 32-nucleotide oligonucleotides (Dbait32ss) that did not activate kinase activity were used as negative controls.

Induction of H2AX phosphorylation by Dbait molecules. One of the earliest steps in the cellular response to DSBs is the phosphorylation of Ser¹³⁹ of histone H2AX by the phosphatidylinositol 3-kinases (24, 32, 33). The appearance of the phosphorylated form of γ-H2AX in nuclei is often used as an indicator of the presence of DNA DSBs produced by ionizing radiation (34). We therefore assayed for the presence of γ-H2AX in cells treated with Dbait32Hc, designed to mimic DSBs. High levels of γ-H2AX with a pan-nuclear distribution was observed in 40% to 60% of cells transfected with Dbait32Hc (Fig. 1A) but not in cells transfected with Dbait8H inactive control, indicating that Dbait32Hc molecules are recognized as DSBs and induce kinase activation in transfected cells. The γ-H2AX mean level was 6- to 10-fold higher in Dbait-transfected cells than after 10 Gy irradiation and persisted for at least 48 h (Fig. 1B and C). Irradiation of Dbait32Hc-transfected cells increased the mean level of γ-H2AX. Similar results were obtained in all cell lines (HeLa, Hep2, MRC5, MO59K, LU1205, and SK28) and primary fibroblasts examined (Fig. 1B and C; data not shown).

ATM and DNA-PK have been shown to function redundantly to phosphorylate H2AX after irradiation (24). To test if both kinases were also involved in the phosphorylation of H2AX after Dbait32Hc transfection, we measured γ-H2AX in ATM (AT5B1) and DNA-PK (MO59J) mutants (Fig. 1C). Induction of γ-H2AX by Dbait32Hc transfection was observed in AT5B1 but not in MO59J, indicating that, for Dbait32Hc-induced signaling, ATM is not able to compensate for the DNA-PK defect in MO59J mutant as it does for irradiation-induced signaling.

Inhibition of repair foci formation by Dbait32Hc. Formation of γ-H2AX foci is important for recruiting and/or stabilizing numerous DSB recognition and repair factors at the break site, including DNA damage checkpoint proteins, chromatin remodelers, and cohesions. Several groups have shown that γ-H2AX foci colocalize with DNA repair proteins including NBS1, 53BP1, MDC1, RAD51, and BRCA1 repair proteins involved in homologous recombination repair pathway (20, 21, 32, 35, 36) in addition to DNA-PKcs and XRCC4. To determine how the Dbait32Hc-induced pan-nuclear distribution of γ-H2AX on chromosomes would affect repair foci formation at DSB sites, we monitored foci formed by 53BP1 (37, 38) or NBS1, a component of the MRE11-RAD50-NBS1 complex (32) after irradiation. In irradiated cells, the number of foci formed by 53BP1 and NBS1 was significantly lower in cells with high γ-H2AX content (in Dbait-transfected cell nuclei) compared with cells with normal level of γ-H2AX in the same population (illustrated in Fig. 2A). To confirm that γ-H2AX immunolabeling did not affect foci detection, we compared the number of cells without 53BP1 or NBS1 foci after irradiation in Dbait32Hc populations treated with γ-H2AX antibodies or not. Results were similar in both cases with at least 20% and 15% of irradiated cells that did not form 53BP1 or NBS1 foci, respectively, in Dbait32Hc-transfected population compared with <1% without foci in populations that were not transfected or transfected with Dbait8H.

Inhibition of damage repair by Dbait32Hc. Because defects in “repair foci” formation have been shown to be associated with DNA repair defects, we tested DNA repair in Dbait32Hc-transfected cells using a single-cell alkaline comet assay that measures DSBs as well as single-strand breaks and base modifications. Dbait32Hc transfection without irradiation did not induce apoptosis (Supplementary Fig. S2). However, after irradiation, repair of DNA damage was delayed in the Dbait32Hc-transfected population and some cells persisted without significant repair for several hours, indicating that Dbait32Hc prevented damage repair in these cells (Fig. 2B; Supplementary Fig. S3).

We analyzed the ability of cells transfected with various amounts of Dbait32H to form clones. Transfection treatment induced a small dose-independent lethality in Dbait32Hc-transfected cells as well as in Dbait8H (Fig. 2C) or Dbait32ss (data not shown) transfected cells. Survival after irradiation decreased in transfected cells according to the amount of Dbait32Hc but was not affected by Dbait8H at any concentration (Fig. 2C). The dose resulting in 50% survival in the absence of Dbait (LD₅₀ = 3 Gy; data not shown) or in the presence of the inactive Dbait8H (LD₅₀ = 2.95 Gy) was 3-fold reduced in the presence of the active Dbait32H (LD₅₀ = 1 Gy; Fig. 2D). Only Dbait32H and Dbait32Hc molecules that activate DNA-PK were able to sensitize cells to irradiation (Supplementary Fig. S4). This Dbait radiosensitizing effect was probably underestimated because, due to low transfection efficiency under cell culture conditions, only half of the transfected population received enough Dbait to induce H2AX phosphorylation. In vivo assays were designed to analyze the effects of Dbait on tumor radiosensitivity with the aim of increasing the observed radiosensitivity using repeated Dbait treatments.

Dbait treatment increases Hep2, LU1205, and SK28 xenografted tumor growth control. We tested the effect of Dbait32Hc on tumor growth using two different protocols of treatment according to the type of tumors treated: head and neck tumors (Hep2) received fractionated treatment in 15 sessions (2 Gy/session) over 5 weeks, whereas the melanoma tumors (SK28 and LU1205) were treated with a hypofractionated protocol of 6 sessions (5 Gy/session) over 2 weeks. Dbait32Hc was combined with PEI (39–42) to facilitate cellular delivery. The Dbait32Hc-PEI combination persisted in the tumor for at least 5 days (Supplementary Figs. S5 and S6). Formulation with PEI was required to observe the Dbait radiosensitizing effect, indicating that Dbait molecules have to enter the cells to be active (Supplementary Fig. S6). In the three tumor models, single treatment with irradiation or with Dbait32Hc (3 nmol/session) inhibited tumor growth but failed to permanently prevent it, whereas combination of both treatments led to efficient growth control (Fig. 3). The median survival of animals that received combined treatment was significantly increased compared with survival of animals treated only with radiotherapy (RR = 0.3; P < 3.9 × 10⁻⁴ in Hep2 and RR = 0.26; P < 6.4 × 10⁻⁵ in SK28) and (RR = 0.47; P < 1.2 × 10⁻² in LU1205; see also Table 1). Interestingly, up to 12% of tumors treated with Dbait combined with irradiation showed complete necrosis followed by a rapid healing and no recurrence of tumor growth over >1 year (Table 1; illustrated in Supplementary Fig. S7). Such efficient regression has not been observed when animals were treated with only radiotherapy or Dbait32Hc treatment.

Histologic analysis revealed large areas of necrosis that cover 30% and 50% of the tumors treated with radiotherapy or Dbait treatment alone, respectively, and up to 75% of the tumors treated with the combination of Dbait32Hc and radiotherapy. These observations were confirmed by magnetic resonance imaging analysis (Supplementary Fig. S8A). The tissue outside the necrotic area showed a significant increase of apoptotic cells and a drastic reduction of the number of mitotic cells (Supplementary Fig. S8B).

Dbait induces H2AX phosphorylation in tumors. Tumors were treated with Dbait32Hc fluorescent molecules during 1 week and analyzed for Dbait and γ-H2AX distribution (Fig. 4A). In parallel, cells from nonnecrotic tumoral tissues were dissociated and analyzed by Western blot (data not shown) or flow cytometry (Fig. 4B and C). The three independent methods of analysis indicated a significant increase of the level of phospho-H2AX after Dbait32Hc treatment. Irradiation alone and 8H treatment did not induce a significant increase of γ-H2AX in tumors (Fig. 4C). The H2AX phosphorylation induced in tumors treated with Dbait32Hc confirms that the Dbait molecules retain the ability to mimic DSBs and activate damage signaling kinase activity in vivo.

Radiosensitizing effect of Dbait depends on its length of the molecule but not on its sequence. Dbait molecules that do not activate DNA-PK in vitro such as single-stranded Dbait32ss, and short Dbait8H molecules displayed no radiosensitization activity in vivo (Table 1). To confirm that the sequence is not involved in the radiosensitizing effect of Dbait, we used another molecule, Dbait32H, differing only in the sequence from Dbait32Hc. Both molecules enhanced the efficacy of radiotherapy in a similar manner (Table 1), confirming that the sequence is not a significant part of the therapeutic activity of Dbait. These results are in agreement with our in vitro observations that Dbait32H and Dbait32Hc similarly activate DNA-PKcs (Supplementary Fig. S1) and inhibit DNA repair (Supplementary Fig. S3).

Radiosensitization is Dbait dose-dependent.In vitro, the radiosensitivity of the cells is proportional to the amount of Dbait-transfected cells (Fig. 2C). To determine if a similar dose dependence is observed in vivo, we compared the efficacy of various amounts of Dbait32H for treating Hep2 tumors (Fig. 5A). Administration of Dbait32H alone had no effect at a dose of 1 nmol (RR = 0.62; P < 0.24). Tumor growth was reduced at a dose of 3 nmol (RR = 0.41; P < 0.016) and the best efficiency was observed at a dose of 6 nmol (RR = 0.14; P < 10⁻⁵; Fig. 5A, left; Table 1). Similarly, in association with radiotherapy, increasing amounts of Dbait improved tumor growth control, resulting in a median survival that increased from 61 days with radiotherapy alone to 86 days (RR = 0.43; P < 1.9 × 10⁻²), 129 days (RR = 0.25; P < 2.2 × 10⁻⁴) and 150 days (RR = 0.13; P < 10⁻⁵) with doses of Dbait32H of 1, 3, and 6 nmol, respectively (Fig. 5A, right; Table 1). Interestingly, although 1 nmol Dbait had no significant effect when administered alone, it significantly enhanced radiotherapy efficiency when it was administered before irradiation, indicating a synergistic effect of the combined treatment at low doses. Radiosensitization by Dbait is not dependent on fractionation of the radiotherapy and was observed either in association with 10 sessions of 3 Gy (10 × 3 Gy), 6 sessions of 5 Gy (6 × 5 Gy), or 2 sessions of 15 Gy (2 × 15 Gy; Supplementary Fig. S9). With all three radiotherapy protocols, administration of Dbait increased the efficacy of the radiotherapy with a concurrent increase in median survival, 36 days with 10 × 3 Gy (RR = 0.4; P < 0.13), 79 days with 6 × 5 Gy (RR = 0.28; P < 10⁻⁵), and 20 days with the 2 × 15 Gy radiotherapy protocol (RR = 0.33; P < 0.02).

Lack of toxic effect or immune response in healthy tissues. Large DNA molecules and oligonucleotides (43, 44) carrying CpG sequences have been reported to induce immune responses. To determine if Dbait induced immune responses, we measured the levels of interleukin (IL)-2, IL-4, IL-5, IL-6, IL-10, IL-12P70, IFN-γ, and tumor necrosis factor-α in blood samples at various times after repeated intravenous or subcutaneous injections of Dbait32Hc or Dbait32H in BALB/c mice. The molecules differ only in their sequence and were equally efficient in controlling tumor growth (Table 1). Animals received seven injections of 6 nmol Dbait within 24 days without developing any toxicity or skin inflammation. We found that injection of the CpG-rich Dbait32H induced a rapid response of IL-6 and a more delayed response of IL-12p70, whereas Dbait32Hc, which is devoid of CpG, did not induce any cytokine production irrespective of whether the delivery was intravenous or subcutaneous (Supplementary Fig. S10). Moreover, the combination of Dbait32Hc and radiotherapy in fractionated as well as hypofractionated treatments had no detectable effect on the treated healthy skin. Cytology analysis of the treated tissue showed only a slight inflammation similar to the irradiated area treated without or with Dbait injections.

Systemic administration of Dbait. Various authors have shown that DNA formulated with PEI and administered by intraperitoneal injection accumulates in subcutaneous tumors (42, 45). We analyzed the efficacy of intraperitoneal injections of Dbait32Hc formulated with PEI in controlling tumor growth. Animals with subcutaneous SK28 tumors received 6 nmol Dbait32Hc before irradiation (6 sessions). All treated tumors rapidly underwent large central necrosis. Cytologic analysis of the tumors sampled 1 week after the end of treatment confirmed areas of large necrosis in tumors treated with intraperitoneal injection of Dbait (Fig. 5C). Although a lower efficiency than intratumoral administration was observed, systemic delivery of the Dbait-PEI complex increased the median survival of the treated animals (RR = 0.4; P < 0.009) compared with animals that received only radiotherapy (Fig. 5B).

We have developed new molecules (Dbait) that mimic DSBs and disorganize DNA damage signaling. The molecular characteristics for Dbait activities were similar for activation of the protein kinase, inhibition of DNA repair, and enhanced sensitivity of tumors to γ-irradiation. We found that the optimal Dbait molecules were 32 bp double-stranded DNA with one free end. The sequence had no influence on the activities tested, indicating that the effects of Dbait were due to the molecular structure as a mimetic substrate rather than to the targeting of a specific sequence. The phosphorothioate and hexaethyleneglycol added to the DNA molecule did not account for the biological effect because Dbait8H, which contains all, did not show any significant activity. The high level of γ-H2AX in cells treated with Dbait32Hc indicates that the Dbait molecules activate a kinase involved in DNA damage signaling. The increase of cell radiosensitivity on transfection with Dbait32Hc may be explained by Dbait binding and sequestering DNA-PK or by a more complex mechanism in which the “false” damage signaling induced by Dbait disturbs the organization or functioning of the DSB repair complexes.

In this work, we show that Dbait radiosensitizes different types of radioresistant tumors. The fact that Dbait molecules trigger histone H2AX phosphorylation in tumors (Fig. 4) suggests that they could act in vivo as observed in vitro (Fig. 2). As Dbait targets the general mechanism of DNA repair, its application is likely to be applied to irradiation as well as other treatments that induce genotoxic damage. How does Dbait act in the absence of genotoxic treatment? Dbait could activate the DNA damage response in tumors that in turn would stop tumor proliferation through cell cycle arrest or death mechanisms (senescence or apoptosis). Alternatively, the standalone effect of Dbait could reflect that tumor cells also require repair for proliferation. Actually, it is well known that tumor cells encounter repetitive chromosome breaks during the cell cycle and require efficient DNA repair to survive. Conversion of this damage and replication stress into fatal lesions (unrepaired DSBs) by Dbait-induced perturbation of damage signaling and repair is expected to trigger cell death.

Although inhibition of DNA repair kinases has been widely discussed as a strategy for sensitizing tumors to antineoplastic treatments (7, 13, 15, 16, 46), to date, a DNA-PK inhibitor NU7441 and an ATM inhibitor KU55933 are the only DSB repair inhibitors in preclinical development (47). Our Dbait molecules offer an alternative avenue for the enhancement of radiotherapy by interfering with DSB repair activity. Dbait differs fundamentally from kinase inhibitors as it acts by disorganizing the spatiotemporal response to damage rather than targeting a key protein of the repair pathway. Moreover, the fact that it targets a whole complex and not a unique enzyme reduces the possibility of developing resistance during treatment.

C. Agrario and A. Herbette: employees of DNA Therapeutics. M. Quanz: recipient of Ph.D. fellowship cofinanced by DNA Therapeutics. M. Dutreix, J-S. Sun, L. Larue, and J-M. Cosset: cofounders of DNA Therapeutics.

The magnetic resonance imaging analyses were done by Andreas Volk and Carole Thomas (Institut Curie). Confocal microscopy was done on the Institut Curie imagery platform.

We thank Elisabetta Marangoni and Nathalie Neboud for establishing the treatment protocols, Nathalie Dogna for animal care, Alexia Savignoni (Department of Statistics, Institut Curie Hospital) for supervising the statistical analysis, and Daniel Louvard and Pierre Bey for support during this work.