Ionizing radiation (IR) is an effective anticancer treatment, although failures still occur. To improve radiotherapy, tumor-targeted strategies are needed to increase radiosensitivity of tumor cells, without influencing normal tissue radiosensitivity. Base excision repair (BER) and single-strand break repair (SSBR) contribute to the determination of sensitivity to IR. A crucial protein in BER/SSBR is DNA polymerase β (polβ). Aberrant polβ expression is commonly found in human tumors and leads to inhibition of BER. Here, we show that truncated polβ variant (polβ-Δ)–expressing cells depend on homologous recombination (HR) for survival after IR, indicating that a considerable fraction of polβ-Δ–induced lesions are subject to repair by HR. Increased sensitization was found not to result from involvement in DNA-dependent protein kinase–dependent nonhomologous end joining, the other major double-strand break repair pathway. Caffeine and the ATM inhibitor Ku55933 cause polβ-Δ–dependent radiosensitization. Consistent with the observed HR dependence and the known HR-modulating activity of ATM, polβ-Δ–expressing cells showed increased radiosensitization after BRCA2 knockdown that is absent under ATM-inhibited conditions. Our data suggest that treatment with HR modulators is a promising therapeutic strategy for exploiting defects in the BER/SSBR pathway in human tumors. Cancer Res; 70(21); 8706–14. ©2010 AACR.
One of the primary treatments for many types of cancer is ionizing radiation (IR), either as a single modality or in combination with other therapies. Local tumor control, however, is strongly determined by intrinsic tumor radioresistance and normal tissue radiosensitivity, which can both limit success. To improve radiotherapy, tumor-targeted strategies are needed that exploit tumor-specific characteristics and consequently spare normal tissue. Here, we present a novel tumor-directed radiosensitization strategy, exploiting DNA repair deficiencies commonly found in tumors.
Base excision repair (BER) is the major repair system responsible for removing and repairing the majority of lesions caused by IR, namely single-strand breaks (SSB) and base damages. Repair is initiated by DNA glycosylases that excise the damaged base, followed by AP-endonuclease (APE1) generating SSB intermediates (1, 2). Direct radiation-induced SSBs require some end processing by enzymes such as PNK and APE1 (3, 4). Poly(ADP-ribose) polymerase-1 (PARP-1) is recruited to SSB sites and is thought to facilitate recruitment of repair proteins. Further repair occurs through either short-patch repair or long-patch repair (LP-BER), resulting in different numbers of incorporated nucleotides. DNA polymerase β (polβ) is one of the key players in BER and SSB repair (SSBR), although the replicative polymerases polδ/ε are also implicated in LP-BER (5). Ligase complexes complete the repair.
Polβ deficiency leads to diminished BER efficiency and causes cells to become hypersensitive to killing by alkylating agents (6). We previously showed a role for polβ in repair of radiation-induced DNA damage in both G0 and G1 cells, underlining the importance of polβ in BER after IR (7, 8). These data also implied that there are strong backup repair processes in S-phase cells that promote survival after IR in cells lacking polβ.
Polβ has been found to be altered in a considerable number of human tumors (reviewed in ref. 9) of which most alterations were derived from single mutations and truncations. Aberrant expression was frequently confined to the polymerase domain, resulting in either error-prone synthesis or abolition of polymerase activity. Expression of several proteins, such as aberrant polβ, has been shown to inhibit BER and can result in cellular transformation (10, 11). Interestingly, aberrant polβ can act in a dominant-negative manner over concomitantly expressed wild-type (WT) polβ (12, 13).
Interference in BER of radiation-induced DNA damage was shown by expression of such an aberrant polβ variant containing the DNA-binding properties but lacking the polymerase domain (14–16). In cells expressing this truncated protein (polβ-Δ), we observed increased induction of chromosome and chromatid-type aberrations after IR, indicating formation of secondary DSBs (17). Our data indicated that these arose partly during repair of IR-specific clustered damage and partly during replication. Unrepaired BER intermediates can cause the replication fork to stall and to form a DSB. These replication-associated DSBs are thought to depend mainly on repair by homologous recombination (HR; refs. 18, 19).
We therefore hypothesized that replication-associated DBS formed after radiation, resulting from interference by polβ-Δ, will be repaired by HR. Because most would be repaired successfully, these would have little influence on cell killing. However, if cells are deficient in HR, this will result in unrepaired DSBs, leading to increased cell killing. Here, we show that cells expressing polβ-Δ indeed depend on HR for survival after IR. Because a significant fraction of human tumors express such aberrant polβ proteins, this could be therapeutically exploited by targeting HR. We show here the feasibility of such a tumor-targeted strategy.
Materials and Methods
Cell lines, cell culture, and transfection
Human lung carcinoma cells (A549) were obtained from American Type Culture Collection and transduced with a truncated variant of polβ-containing LZRS-MS-EGFP expression vector (polβ-Δ) or empty vector control (LZRS; ref. 20). Cells were grown in DMEM. Chinese hamster cells V79B (WT), CLV4B (deficient in RAD51C), and CLV4B+ (complemented with full-length human RAD51 cDNA) were previously described (21, 22) and kindly provided by M.Z. Zdzienicka and B.C. Godthelp (Leiden University Medical Center, Leiden, the Netherlands) and grown in HAM-F10 medium. All cell lines were used within 20 passages before returning to the original frozen stock. Karyotype, radiosensitivity, and gene expression were regularly tested and remained constant in all cell lines. All cells were grown under standard conditions at 37°C with 5% CO2 in medium supplemented with 10% FCS and antibiotics and tested regularly for Mycoplasma. All hamster cell lines were transfected with polβ-Δ and LZRS by using FuGene6 (Roche) according to the manufacturer's protocol. Cells were harvested 48 hours after transfection and placed under puromycin selection. Cells were sorted on green fluorescent protein (GFP) expression by FACScan flow cytometry (FACScan, BD).
Gamma- and UV-C irradiation
Gamma irradiation experiments were performed using a 137Cs irradiation unit with a dose rate of 0.66 Gy/min, and UV-C irradiation was performed using a Stratagene UV-C X-linker (254 nm).
Clonogenic survival assays
Cell survival was assessed by colony formation. All hamster cells were irradiated 6 hours after plating. Human A549 cells were plated and irradiated 14 to 15 hours later. When inhibitors were used, they were added 1 hour before irradiation, and the medium was replaced by a fresh medium 23 hours later. Colonies were fixed and stained, and those consisting of 50 cells or more were counted. Survival was calculated relative to the plating efficiency of unirradiated controls. Hamster cell colonies were counted using the ColCount (Oxford Optronics) colony counter, and A549 cells were counted by eye under an inverted dissecting microscope. Each experiment was carried out at least three times. Dose enhancement factors (DEF) values were calculated as the ratio of doses to produce 37% survival (DEF37). Average DEF37 for drug-induced sensitization were calculated from second-order polynomial curve fits through the dose-response data (log survival versus radiation dose) of the individual experiments. Errors on DEF37 are SD. For the analysis of the genetically modified cells, means of the 37% survival values and SD from three or more separate experiments were calculated for each cell line and condition. To derive errors on the DEF, errors on the DEF37 values for the two groups being compared were combined using the root mean squares of the fractional errors. Sensitization factors at a given inhibitor concentration were calculated as the ratio of surviving fractions at 2 Gy versus 0 Gy.
Drugs used were the DNA-dependent protein kinase (DNA-PK) inhibitor NU7026 and the ATM-kinase inhibitor Ku55933, both purchased from Calbiochem (an affiliate of Merck KGaA). In knockdown experiments, an ATM inhibitor purchased from Tocris Biosciences was used. ATM and DNA-PK inhibitors were dissolved in DMSO. Controls without inhibitors were treated with the respective solvent concentrations. Caffeine (Sigma-Aldrich) was freshly dissolved in double-distilled water before use.
Immunohistochemistry for RAD51 foci
For RAD51 foci scoring, experiments were performed as described in ref. (23) by using a rabbit anti-Rad51 (polyclonal, H-92; Santa Cruz Biotechnology) and FITC-conjugated anti-rabbit antibody (Sigma). Cells were counterstained with 4′,6-diamidino-2-phenylindole for 10 minutes before coverslips were mounted. RAD51 foci were counted blind by using a Zeiss fluorescence microscope (Axiovert 100M) equipped with a CCD camera (MAC 200A, Photometrics). One hundred cells were counted per point and dose. Each experiment was repeated three times.
siRNA transfection and quantitative real-time-PCR
A549-LZRS and polβ-Δ cells were transfected with 25 nmol/L Hs_BRCA_7 or AllStars Negative Control siRNA from Qiagen using Hiperfect (Qiagen) transfection reagent according to the manufacturer's protocols. Forty-eight hours after transfection, cells were plated for colony assays. Cell samples of each experiment were harvested to assess knockdown efficiency by real-time-PCR (RT-PCR) after cDNA synthesis (with SuperScriptII Reverse Transcriptase, Invitrogen) on RNA prepared with the RNeasy kit (Qiagen) according to standard protocols. RT-PCR analysis was performed with SYBR Green PCR Master Mix (Applied Biosystems) and the QuantiTect primers Hs_BRCA2_1_SG and Hs_B2M_1_SG (Qiagen) using the Applied Biosystem 7500 Fast-Real-Time system equipment. Only data from experiments with a knockdown efficiency above 75% were included.
The Student's t test was used to test for significant differences between the average DEF induced by small-molecule inhibitors. A P value of <0.05 was used throughout as the level of significance.
Polβ-Δ expression leads to a transient increase in γH2AX foci and SCEs after IR
We have previously shown that expression of the dominant-negative polβ-Δ protein interferes with BER after IR and results in increased formation of residual double-strand breaks (DSB) as indicated by an increase in chromosome and chromatid aberrations (17). Here, we analyzed γH2AX foci induction at earlier time points after radiation. We treated cells with 2 Gy of irradiation and observed elevated levels of γH2AX foci at 6 and 10 hours in A549–polβ-Δ–expressing compared with vector controls (Supplementary Fig. S1). γH2AX foci numbers in the polβ-Δ–expressing cells reached control levels at later time points, indicating removal by repair.
We therefore analyzed sister chromatid exchanges (SCE), which represent “successful” repair events. A549–polβ-Δ or A549-LZRS was irradiated, and metaphases were collected 8 hours later (including 2 hours colcemid). Metaphases therefore represented cells irradiated in the S phase. Spontaneous SCE induction was not significantly different in the two cell lines. However, we observed increased amounts of SCE after 2 Gy in polβ-Δ–expressing cells (Supplementary Fig. S2), which we concluded to result from additional but repaired DNA lesions in polβ-Δ–expressing cells. Because we hypothesized that those lesions could have resulted from replication attempts at polβ-Δ–induced BER intermediates, we also expected these lesions to arise after treatment with DNA-damaging agents creating damage similar to IR (oxidative base damage, SSB) and requiring BER activity; one such agent is H2O2. A549–polβ-Δ cells showed an increase in SCE induction compared with control cells after H2O2 treatment (Supplementary Fig. S2), which was significant at a dose of 200 μmol/L H2O2. This indicates increased formation of DNA lesions after treatment with H2O2 as well as after IR.
Radiosensitization does not result from polβ-Δ interference in DNA-PK–driven processes
Polβ-Δ interference in BER has been previously documented (14, 15). Nevertheless, the observed increase in DSBs after IR could also be caused by interference of polβ-Δ in the DSB repair process itself. To test for a possible polβ-Δ involvement in nonhomologous end joining (NHEJ), one of the major DSB repair pathways, we used Chinese hamster ovary cells either WT or deficient in DNA-PK. These cells were transiently transfected with the human polβ-Δ protein or the empty vector LZRS. Western blot analyses showed that polβ-Δ expression was similar in the two lines (Supplementary Fig. S3). DNA-PK–deficient cells, due to their deficiency in DSB repair, are significantly more radiosensitive than their parental control cells. Expressing polβ-Δ in both DNA-PK–deficient as well as DNA-PK–proficient cell lines, however, resulted in equal increases in radiosensitivity with average DEFs at 37% survival (DEF37) of 1.18 (±0.21) and 1.16 (±0.11), respectively (Supplementary Fig. S4A). This indicates that polβ-Δ–induced radiosensitization was independent of DNA-PK expression. We further confirmed this by using human glioma cell lines deficient and complemented with DNA-PK (Supplementary Fig. S4B). Here, we also observed similar increases in radiosensitivity after expression of polβ-Δ and independent of NHEJ capacity, with DEF37 values of 1.05 for the DNA-PK–deficient and 1.11 for the DNA-PK–complemented cells, respectively.
Polβ-Δ–induced radiosensitization in these cell lines was small; thus, to support these findings, we also chemically inhibited NHEJ by using the DNA-PK inhibitor NU7026. This inhibitor was previously shown to effectively inhibit NHEJ and sensitize cells to IR (24, 25). As previously shown (17, 20), A549–polβ-Δ was more radiosensitive compared with A549-LZRS. Incubation with NU7026 for 24 hours did not result in significant toxicity: Exposure to 5 μmol/L NU7026 resulted in an average of 83% and 91% survival compared with untreated A549-LZRS and A549–polβ-Δ cells, respectively. The radiation dose-response curves in Fig. 1 of A549–polβ-Δ or A549-LZRS show that inhibition of DNA-PK with NU7026 increased radiosensitivity of both cell lines with a drug-induced DEF37 of 2.23 ± 0.76 and 1.89 ± 0.34 for LZRS and polβ-Δ–expressing cells, respectively. Taken together, the similar radiosensitization in DNA-PK–proficient and DNA-PK–deficient cells shows that polβ-Δ–induced radiosensitization does not result from interference with DNA-PK–dependent repair processes. In addition, due to the lack of a greater increase in radiosensitization in the DNA-PK–inhibited or DNA-PK–deficient cells, these data also indicate that NHEJ is not used as the main repair mechanism of potentially lethal lesion induced by polβ-Δ interference.
Survival of polβ-Δ–expressing cells after radiation is dependent on HR
HR is the other main pathway used for repairing DSBs after IR. In addition, it provides critical support for replication by resolving stalled or blocked replication forks. Because SCEs represent mainly HR events and are dependent on HR processes (26), increased SCE induction by the expression of polβ-Δ would indicate participation in such processes. To exclude involvement of polβ-Δ in HR, we used cells complemented or deficient in Rad51C, a crucial protein for HR. Expression of polβ-Δ in RAD51C-complemented or WT cells resulted in a modest increase in radiosensitivity (polβ-Δ–induced DEF37 of 1.05 ± 0.09 and 1.06 ± 0.15, respectively; Fig. 2A). In Rad51C-deficient cells, however, expression of polβ-Δ resulted in a significantly greater radiosensitization (DEF37 1.67 ± 0.46). This shows that, first, the polβ-Δ–induced radiosensitization is not a result of inhibition of Rad51C-driven HR processes, and second, that survival of polβ-Δ–expressing cells after irradiation strongly depends on Rad51C expression. This therefore indicates that polβ-Δ not only does not affect repair by HR, but that HR is an important backup pathway for repair of potential lethal lesions that have occurred due to expression of polβ-Δ.
Because radiosensitization by polβ-Δ was increased in HR-deficient cells, we hypothesized that inhibiting HR processes in polβ-Δ–expressing cells would lead to increased cell killing after IR. To test this hypothesis, we investigated the magnitude of radiosensitization caused by caffeine. Caffeine-induced radiosensitization has previously been shown to depend on HR status because caffeine failed to induce radiosensitization in HR-deficient cells (27, 28) and inhibited HR on SceI-induced DSBs. A549–polβ-Δ or A549-LZRS was treated with 2.5 mmol/L caffeine. Inhibition of HR by caffeine resulted in increased radiosensitivity of both cell lines but had a greater radiosensitizing effect on polβ-Δ–expressing cells (drug-induced DEF37 of 1.56 ± 0.35 and 1.81 ± 0.34, respectively; Fig. 2B). This increased radiosensitivity was not a result of increased killing by caffeine alone in the polβ-Δ–expressing cells, which had a survival of 81% compared with 40% in the controls. In contrast to IR, UV-induced DNA lesions are not repaired by BER. UV-C irradiation to 3 J/m2 resulted in a similar surviving fraction of A549 cells as after 2 Gy. Moreover, as described previously, both cell lines were equally sensitive to UV-C radiation (20). Adding caffeine increased the UV sensitivity of both cell lines similarly (Fig. 2C), thereby excluding any BER-independent changes. These studies with caffeine further support a role for HR as an important backup pathway for survival after IR in cells expressing polβ-Δ.
polβ-Δ–expressing cells show increased RAD51 foci formation after IR
Based on the results described above, we predicted an increase in HR events in polβ-Δ cells compared with their controls after IR. Nuclear foci containing RAD51 are thought to represent sites of HR and thereby can be used to indicate HR events in the cell. As anticipated, A549–polβ-Δ cells showed significantly increased levels of RAD51 foci compared with A549-LZRS 1 hour after irradiation (Fig. 3). Increased formation of RAD51 foci in cells expressing polβ-Δ was also confirmed in hamster AA8 cells expressing polβ-Δ (Supplementary Fig. S5). Together with the increased dependence on HR, we interpret this increase in RAD51 foci to represent increased HR events and use in polβ-Δ–expressing cells. This supports the hypothesis of increased HR incidence due to increased BER intermediate levels and underlines the importance of HR in polβ-Δ–expressing cells after IR.
Chemical inhibition of HR leads to polβ-Δ–specific radiosensitization
The polβ-Δ variant used in these studies and other tumor-specific polβ variants have been found to inhibit BER. Our findings could therefore have potential clinical relevance and indicate that BER-deficient tumors could specifically be targeted by inhibiting their backup repair pathways, which we hypothesized to be HR. However, no specific HR inhibitor has been reported to date. Bryant and Helleday (29) reported that ATM inhibition prevented PARP inhibitor induced HR as determined by hypoxanthine guanine phosphoribosyl transferase recombination frequencies. We therefore tested the ATM kinase inhibitor Ku55933 for radiosensitization potential in the polβ-Δ–expressing cells. Exposure to Ku55933 resulted in a small reduction of the plating efficiencies of A549-LZRS and A549–polβ-Δ by 27% and 9%, respectively. Incubation with Ku55933 for 24 hours combined with radiation resulted in decreased survival of A549-LZRS cells (DEF 1.69 ± 0.13; Fig. 4A). However, the radiosensitizing effect of this ATM inhibitor was significantly greater on polβ-Δ–expressing cells (DEF37 2.29 ± 0.38).
We next questioned if ATM inhibition resulted in the inhibition of HR-dependent processes crucial to the survival of the polβ-Δ–expressing cells. We transfected the cells with siRNA to BRCA2 and exposed the cells to 3 Gy with and without the ATM inhibitor Ku55933. Knockdown efficiencies were comparable in both cell lines, polβ-Δ and LZRS controls, with 84 ± 10% and 82 ± 6%, respectively. Consistent with the previous observation in Rad51C-deficient cells, reduction in HR resulted in increased radiosensitization in the A549 polβ-Δ cells compared with their controls (factor of 2.2 versus 1.3 in the LZRS; Fig. 4B). However, this increase in the polβ-Δ was absent under ATM-inhibited conditions, showing that ATM influences polβ-Δ–driven and BRCA2-dependent radiosensitization. The lack of BRCA2-dependent radiosensitization is consistent with the idea that HR modulation resulting from ATM inhibition causes increased radiosensitization in polβ-Δ–expressing cells.
Drug dose–dependent sensitization of polβ-Δ–expressing cells
We next investigated the optimum inhibitor concentration at a clinically relevant radiation dose of 2 Gy. Irradiated cells incubated for 24 hours with varying concentrations of caffeine or the ATM inhibitor showed an increased radiosensitization in A549–polβ-Δ compared with vector controls, starting at concentrations above 1 mmol/L for caffeine (Fig. 5A) and above 1 μmol/L for the ATM inhibitor (Fig. 5B). In particular, the ATM inhibitor greatly increased the sensitization factor for A549–polβ-Δ compared with control cells (by 1.6- to 2.0-fold at 2.5–10 μmol/L).
Surprisingly, radiosensitization of polβ-Δ–expressing cells was evident at lower drug concentrations compared with the controls. This is possibly due to the increased number of lesions, thereby increasing vulnerability to inhibition of backup repair. In terms of clinical applicability, the data show a window of drug dose that increases radiosensitivity in the polβ variant–expressing cells while leaving WT cells unaffected. As a control, we also tested the DNA-PK inhibitor NU7026 at concentrations ranging from 0.1 to 10 μmol/L and observed that inhibiting DNA-PK at any of the given concentrations does not result in polβ-Δ–dependent radiosensitization. The sensitization factor increased with increasing DNA-PK inhibitor concentration although to similar extent in both cell lines (Fig. 5C), consistent with the radiation dose-response data described above.
In summary, two compounds, caffeine and an ATM inhibitor, caused greater radiosensitization in BER-deficient cells, demonstrating the feasibility of a tumor-targeted radiosensitization strategy.
From the results of this study, we propose a new tumor-targeted strategy in which inhibitors affecting HR could be used to increase radiosensitization of tumors displaying BER deficiencies. The data presented here show that cells expressing a truncated DNA polymerase β (polβ-Δ) depend on HR for survival after IR. Moreover, we show that polβ-Δ–induced radiosensitization is not affected by DNA-PK–driven NHEJ. The data therefore strongly indicate that HR is used as the major backup pathway for repair of secondary DSBs or collapsed replication forks that were induced by expression of this truncated polβ. As judged from the radiosensitization in the DSB repair mutant cells, polβ-Δ expression does not affect DSB repair and therefore cannot be the cause of the radiosensitization observed in these cells.
To improve radiotherapy, strategies are being developed that involve the identification of molecular targets responsible for the radioresistance of cancer cells. The archetypal example to date is targeted radiosensitization through inhibition of epidermal growth factor receptor (EGFR)/RAS, phosphoinositide 3-kinase/Akt, and related signaling pathways in EGFR-overexpressing tumors (30, 31). Here, we propose a strategy that exploits differences in dealing with radiation damage in tumors compared with normal tissue, thereby realizing tumor-targeted radiosensitization. By targeting the backup pathway (HR) that is predominantly engaged in BER-deficient cells after radiation, we achieved increased radiosensitization. BER-proficient cells, such as those in normal tissues, by contrast, are capable of sufficiently repairing damage by BER, thereby reducing the formation of secondarily formed and potentially lethal lesions.
This is analogous to the use of inhibitors of PARP, a key protein in BER/SSBR, which successfully and specifically targets HR-deficient cells (BRCA1 and BRCA 2; refs. 32, 33). PARP inhibition is thus a useful therapeutic strategy for tumors displaying defects in the BRCA genes (34, 35). These studies underline the potential and impact of tumor-targeted strategies that exploit DNA repair deficiencies of tumor cells. Here, we show the reverse: Inhibitors affecting HR are more effective sensitizers on cells deficient in BER/SSBR. This is of potential clinical relevance because ∼30% of human tumors of different origins have been reported to contain mutations and/or truncations of polβ (9). Aberrant expression was found concomitant with WT polβ expression and often observed in the tumors only. Interestingly, clear cell renal carcinomas arising in kidney or cell lines did not exhibit POLB mutations (COSMIC database; ref. 36). In contrast, POLB somatic mutations in prostate cancer are hypothesized by the authors to contribute to tumorigenesis by driving mutagenesis (37). In addition to genomic mutations, altered expression can be a result of aberrant splicing, commonly observed in tumors (38). Expression of several of these polβ variants has been shown to lead to decreased BER efficiency.
This prompted us to test the proposed novel tumor-targeted radiosensitization strategy. We hypothesized that in polβ-Δ–expressing cells, HR inhibitors could be used for specific sensitization given their dependence on HR for survival. To test this, we first studied the effects of caffeine, which inhibits both ATM and ATR kinases, thereby inhibiting repair and abrogating proper DNA damage checkpoint responses after IR (39). From a lack of radiosensitization in HR-deficient cells, Wang and colleagues showed that caffeine exerts its radiosensitizing effect through HR (27, 28). Caffeine was also shown to inhibit HR, as measured by an I-SceI–induced GFP-based DSB repair assay (40). Together with changes in Rad51 foci formation, these data strongly suggest that caffeine inhibits HR-directed repair processes. Our results with caffeine therefore support the HR dependence of polβ-Δ–expressing cells.
In addition to caffeine, in an attempt to show feasibility by chemical inhibition, we also tested drugs that have been reported to indirectly affect HR. One such drug is an ATM kinase inhibitor. ATM is the main transducer of response to DSBs, and its activation triggers phosphorylation of several downstream targets that modulate cell cycle arrest and DNA repair. ATM has also been shown to regulate HR to a greater extent than NHEJ (40). Recent data from Beucher and colleagues (41) confirmed the role of ATM in HR of radiation-induced DSBs in the G2 phase of the cell cycle. Specific ATM inhibitors have been developed that increased sensitivity to radiation (42). In particular, ATM inhibition prevented PARP inhibitor–induced HR (29). This indicated that ATM is also required for triggering HR to resolve collapsed replication forks after encountering SSBs and BER intermediates caused by the inhibition of PARP, hence the use of the ATM inhibitor here. Indeed, we found that ATM inhibition increased radiosensitization of polβ-Δ–expressing cells. Despite the fact that targeting ATM will result in a broad inhibition spectrum, and although not specifically inhibiting HR, these data underline the potential of such a strategy and indicate that use of such drugs represents a feasible approach to increase sensitization of polβ-Δ–expressing tumors to IR.
A lack of polymerase activity is a common feature of tumor-specific polβ variants. In our studies, we therefore used such a truncated variant of polβ, deprived of its polymerase activity. We previously showed that the expression of this variant leads to increased radiosensitization due to interference in BER since polβ-Δ did not radiosensitize XRCC1-deficient cells (15). We and others further showed inhibition of BER by in vitro assays either after alkylating or radiation damage (12–14, 20, 43). Mutations or alterations in other proteins can also result in BER deficiency, which should lead to accumulation of similar unrepaired intermediates, hence an increased dependence to HR. The targeted radiosensitization approach as presented here may therefore be applicable to a broad spectrum of tumors with BER deficiencies.
The magnitude of the increased sensitivity to radiation with the drugs used here is less than that often seen when comparing drug-sensitive and drug-resistant cell lines. In contrast to drug-induced kill, many genetic and drug studies show that modifying radiation response will not exceed a sensitization factor of around 3. The differential radiosensitization observed here is favorable and significant compared with current radiosensitizing drugs such as AKT or PARP inhibitors. It should also be noted that based on the slopes of dose-response curves for local tumor control of clinical tumors treated with radiotherapy alone, the degree of radiosensitization seen here could translate into significant therapeutic gains. For squamous cell carcinomas of the head and neck, the reported γ50 slope values (percent gain in tumor control for each percent increase in dose) typically range between 1.5 and 2.5 (44); thus, even a small increase of cellular radiosensitization of 10% (DEF 1.1) would result in an additional 15% to 25% local control. Our data with the ATM inhibitor showed a 1.5-fold increased DEF37 in polβ-Δ–expressing versus control cells, therefore anticipating a larger effect. If applied to tumors expressing such truncated polβ variants, this could translate to significant increases in tumor control and, due to the discrimination of normal and cancerous cells, to a widening of the therapeutic window.
In conclusion, the data presented here suggest that treatment with small-molecule inhibitors, which at least partially affect HR, represent a promising new therapeutic strategy for exploiting defects in the BER/SSBR pathway such as tumors expressing polβ variants like those used in this study. This would ultimately lead to greater tumor-specific cell kill, thereby increasing the therapeutic ratio.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
We thank P. Olive (BC Cancer Agency), and B.C. Godthelp and M.Z. Zdzienicka (Leiden University Medical Center) for their generous gift of cell lines; C.M. Koch for help with SCE analysis; and L. Oomen, L. Brocks, A. Pfauth, and F. van Diepen from the NKI microscope and flow cytometry facility for their technical assistance.
Grant Support: Dutch Cancer Society (grant NKI 2002-2598).
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.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received October 27, 2009.
- Revision received June 2, 2010.
- Accepted July 20, 2010.
- ©2010 American Association for Cancer Research.