A Double-Strand Break Repair Defect in ATM-Deficient Cells Contributes to Radiosensitivity

INTRODUCTION

DNA double-strand breaks (DSBs) represent a major threat for the maintenance of genomic integrity. These lesions can be produced by exogenous agents such as ionizing radiation and some chemicals but also arise endogenously during replication, V(D)J recombination, and meiosis (1 , 2) . If left unrepaired, they can result in permanent cell cycle arrest, induction of apoptosis or mitotic cell death (3) , and if repaired incorrectly, can lead to carcinogenesis through directly induced or delayed chromosomal rearrangements (4) . Nonhomologous DNA end-joining (NHEJ) is the major mechanism for the repair of ionizing radiation-induced DSBs in mammalian cells and also effects rearrangements during V(D)J recombination (5) . NHEJ involves the DNA end-binding heterodimer Ku70/Ku80, the catalytic subunit of the DNA-PK, the XRCC4 gene product, and DNA ligase IV (reviewed in Refs. 6 , 7 ). Recently, an additional protein, Artemis, was also shown to be involved in NHEJ during V(D)J recombination (8 , 9) . Cell lines with mutations in any of these genes are radiation sensitive and show, except for Artemis-defective cells, marked deficiencies in DSB repair after ionizing irradiation (10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20) .

Recently, hypomorphic mutations in DNA ligase IV were identified in some human patients displaying immunodeficiency and developmental delay (21) , identifying a disorder called LIG4 syndrome. Cells from one of the patients, 180BR cells, were extensively studied and found to be homozygous for a missense mutation in the LIG4 gene (22) . This mutation was shown to impair the adenylation and ligation activities of DNA ligase IV in vitro (23) . The patient did not display any abnormal clinical characteristics, including a lack of immunodeficiency, but developed leukemia at age 14 years and dramatically overresponded to radiotherapy. Because DNA ligase IV knockout mice are embryonic lethal (24) , it was suggested that the mutated protein in 180BR cells has some residual function in vivo that allows proficient V(D)J recombination and apparently normal development but is insufficient to repair the large number of DSBs introduced during radiotherapy (22 , 23) .

Ataxia telangiectasia mutated (ATM), the protein defective in the hereditary disorder ataxia telangiectasia (AT) (25) , is a serine-threonine kinase that is central to the signal transduction process initiated by DSBs (26 , 27) . AT is a neurodegenerative disease with progressive cerebellar ataxia and telangiectasia, and AT patients display clinical radiosensitivity, immunodeficiency, and show an elevated risk for cancer induction (28) . ATM is activated after ionizing irradiation (29) and phosphorylates proteins involved in cell cycle checkpoint control and DNA repair. Consequently, AT cell lines show pronounced radiosensitivity and defective activation of cell cycle checkpoints after radiation exposure (30) . This defect involves an inability to arrest at the G₁-S, S, and G₂-M checkpoints. Failure to effect the S-phase checkpoint results in radiation-resistant DNA synthesis, which represents a hallmark of AT cells (31 , 32) . Additionally, AT cells fail to arrest before mitosis if they are in the G₂ phase at the time of irradiation (33) . The severe checkpoint deficiencies of AT cells contrast with a rather small defect in the repair of DSBs. Whereas NHEJ-deficient cells show proficient checkpoint activation but fail to repair a large fraction (40–60%) of the initially induced DSBs (5) , AT cells repair most of the induced breaks at a normal rate and have, at best, only a small elevation in the number of unrepaired DSBs (e.g., Refs. 34 , 35 ). Moreover, AT cells are usually more radiosensitive than NHEJ-deficient cells, arguing that the radiosensitivity and most of the defects in AT cells must result from checkpoint failures or other deficiencies in signal transduction and cannot be ascribed to a minor defect in the repair of DSBs.

Notwithstanding these compelling arguments, cytogenetic experiments with nondividing confluent AT cells provide indirect evidence for a more prominent DSB repair defect in AT cells. Confluent primary human AT fibroblasts held in the G₁-phase during the entire repair period were analyzed by premature chromosome condensation and exhibited a level of unrepaired chromosome breaks sufficient to account for their radiosensitivity (Refs. 36 , 37 ; see Refs. 38 and 39 for further discussion). Additionally, delayed plating survival experiments, in which confluent cells are irradiated and incubated for repair before plating is performed, usually result in a large increase in radioresistance in DSB repair-proficient cells, but this is not observed in AT cells (40, 41, 42) . On the basis of this evidence, it is surprising that AT cells represent one of the most radiosensitive cell lines available and yet show a defect in DSB repair that is many fold smaller than that of the cell lines deficient in NHEJ factors. However, it is important to consider that all DSB repair measurements described thus far were performed at radiation doses well above those used for cytogenic studies and radiosensitivity measurements.

Recent studies in our laboratories have demonstrated that enumerating γ-H2AX foci (a phosphorylated histone, see Refs. 43 , 44 ) can be used to measure the repair of radiation-induced DSBs in primary human fibroblasts (45) and Chinese hamster ovary cells (46) at physiologically relevant doses. In the present study, we used γ-H2AX foci analysis in parallel to pulsed-field gel electrophoresis (PFGE) to investigate the capacity of G₁-arrested NHEJ- and ATM-deficient cells for repairing DSBs induced by ionizing radiation doses between 0.02 and 80 Gy. We show that AT cells fail to repair a subset of DSBs (∼10%), irrespective of the initial number of lesions induced, whereas DNA ligase IV-deficient cells exhibit an increased repair capacity with decreasing radiation dose, are able to repair DSBs for many days after irradiation, and can reach a level of unrepaired breaks that is lower than that of AT cells. The repair measurements are supported by cell survival studies where NHEJ-deficient but not ATM-defective cells show pronounced recovery in delayed plating experiments. These data reconcile the high radiosensitivity of AT cells with their seemingly small defect in DSB repair and provide an in vivo demonstration of the residual DNA ligase IV activities in cells with hypomorphic LIG4 mutations.

MATERIALS AND METHODS

Cell Culture and Survival Experiments.

Primary human fibroblasts from the lung (MRC-5, wild-type, European Collection of Animal Cell Cultures no. 84101801) and the skin (AT1BR, deficient in ATM, European Collection of Animal Cell Cultures no. BM0020; 180BR, deficient in DNA ligase IV; Ref. 22 ) were grown in a humidified 5% CO₂ atmosphere at 37°C in MEM supplemented with 10% (MRC-5) or 20% (180BR) FCS and 2 mm glutamine and antibiotics or in Nut Mix HAM’s F10 medium (AT1BR) supplemented with 15% FCS and antibiotics. For survival experiments, cells were irradiated in the exponential growth phase (Fig. 3A) ⇓ or were grown to confluence before irradiation (Fig. 3, B and C) ⇓ , plated for colony formation either immediately (Fig. 3, A and B) ⇓ or 4 days after irradiation (delayed plating experiments, Fig. 3C ⇓ ), and stained with crystal violet 3–4 weeks after plating (colonies of >30 cells were scored as survivors). The primary skin fibroblasts 48BR (wild type), 1BR (wild type), AT7BI (deficient in ATM), 411BR (deficient in DNA ligase IV), and 2BN (Fig. 5) ⇓ were cultured and analyzed for radiosensitivity as described previously (21 , 47) . All DSB repair and immunofluorescence experiments with primary human fibroblasts were performed with nondividing confluent cultures (at least 98% of the cells in G₁ as determined by flow cytometry).

The human pre-B cells N114P2 (null for LIG4; Ref. 48 ) and the parental cell line, Nalm-6, were cultured in RPMI 1640 with 10% FCS, glutamine, and antibiotics. Immunofluorescence experiments were performed with plateau-phase cultures at a density of ∼2 × 10⁶ cells/ml. At this density, the cell cycle distribution as determined by flow cytometry is as follows: ∼60% G₁-; ∼20% S-; and ∼20% G₂-phase cells. After radiation exposure and repair incubation, the G₁-fraction slightly increases (up to ∼70% after 8 h) because cells in G₂ at the time of irradiation progress into G₁ during repair incubation and because cells in G₁ at the time of irradiation delay entry into S (measured by two-parameter flow cytometry with bromodeoxyuridine and propidium iodide). For γ-H2AX immunofluorescence analysis, cells with a G₁ DNA content were scored (they were discriminated under the microscope from G₂-phase cells by the size of the nuclei). However, we cannot exclude the possibility that some cells that were scored as G₁-phase cells after repair incubation were in the G₂ phase at the time of irradiation. Because NHEJ-deficient cells show improved DSB repair in G₂ compared with G₁ (46) , this effect could lead to a slight overestimation of the repair capacity of G₁-phase N114P2 cells.

Irradiation.

For PFGE measurements and survival experiments, X-irradiation was performed at 95 kV and 25 mA with a 1.3-mm aluminum filter at a dose rate of ∼5 Gy/min (Figs. 1A ⇓ , 3A ⇓ , and 4B ⇓ , left panel), and acute γ-irradiation was performed with a ⁶⁰Co γ-ray source at a dose rate of 8 Gy/min (Fig. 3, B and C) ⇓ or as described previously (49) . Cells in flasks were irradiated in ice-cold PBS [137 mm NaCl, 2.7 mm KCl, 8 mm Na₂HPO₄, and 1.5 mm KH₂PO₄ (pH 7.45); for PFGE measurements] or in cell culture medium at room temperature (for survival experiments). For immunofluorescence experiments, X-irradiation was performed at 90 kV, either with a 1-mm aluminum filter (for 0.2 and 2 Gy doses, dose rate 2 Gy/min) or a 1-mm copper and a 1-mm aluminum filter (for 0.02 Gy doses, dose rate 0.06 Gy/min), and γ-irradiation (Fig. 5) ⇓ was performed as described previously (49) . Cells were irradiated on coverslips (or in vials for pre-B cells) at room temperature in cell culture medium. For low dose rate experiments (Fig. 4, C and E) ⇓ , cells in flasks were irradiated for 14 days with ⁶⁰Co γ-rays inside an incubator under repair conditions (dose rate of 5.7 Gy/day). Control experiments had shown that confluent primary human fibroblasts irradiated with 80 Gy remain metabolically active without any sign of DNA degradation for many weeks if the medium is changed regularly every 7–10 days (determined by flow cytometry analysis, trypan blue exclusion, and 3-(4,5-dimethylthiazol-2-yl)-2.5-diphenyl tetrazolium bromide analysis (50) . Therefore, experiments with repair times of 14 days (Fig. 4) ⇓ were briefly interrupted after day 7 to change the medium. The dose rates of the different radiation sources were determined by chemical dosimetry and with an ionization chamber.

Immunofluorescence Analysis.

MRC-5, 180BR, and AT1BR fibroblasts grown on coverslips were fixed in 2% paraformaldehyde for 15 min, washed in PBS for 3 × 10 min, and permeabilized for 5 min on ice in 0.2% Triton X-100. Pre-B cells in suspension were spotted onto glass slides (Superfrost Plus), dried briefly, and fixed with 100% methanol for 20 min at −20°C. Samples were blocked in PBS with 1% BSA or 1% FCS for 3 × 10 min at room temperature. Samples were incubated with anti-γ-H2AX antibody (R&D Systems, Minneapolis, MN) for 1 h, washed in PBS, 1% BSA (or 1% FCS) for 3 × 10 min, and incubated with Alexa Fluor 488-conjugated goat antirabbit secondary antibody (Molecular Probes, Eugene, OR) for 1 h at room temperature. Cells were washed in PBS for 4 × 10 min and mounted using Vectashield mounting medium with 4,6 diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA). Fluorescence images were captured using a Zeiss Axioplan 2 imaging epifluorescent microscope equipped with a charge-coupled device camera and ISIS software (MetaSystems, Altlussheim, Germany). Optical sections through the nuclei were captured at 0.2-μm intervals, and images were obtained by projection of the individual sections. For quantitative analysis, foci were counted by eye during the imaging process using a ×63 objective. In a single experiment, cell counting was performed until at least 40 cells and 40 foci were registered/sample. For data points that were derived from a single experiment, the error bars represent the SE from the analysis of the number of cells analyzed. For data points that were derived from more than one experiment, the error bars represent either the SE from the number of cells analyzed in the single experiments or the SE between the different experiments (whichever is highest). Immunofluorescence analysis with 48BR, 411BR, AT7BI, and 2BN fibroblasts (Fig. 5) ⇓ was performed as described earlier (51) .

DSB Repair Measurements.

After irradiation and repair incubation, cells were harvested, embedded in agarose plugs, and lysed. After washing the plugs in TE buffer [10 mm Tris-HCl, 1 mm Na₂EDTA (pH 8)], PFGE was carried out with a CHEF DRIII system (Bio-Rad, Hercules, CA) in 0.8% agarose gels. The gels were run at 14°C with linearly increasing pulse times from 50 to 5000 s for 66 h at a field strength of 1.5 V/cm. Gels were stained with ethidium bromide and photographed with a charge-coupled device camera system under UV transillumination. Quantitative analysis was performed with AIDA software (raytest, Straubenhardt, Germany). The fraction of DNA entering the gel was quantified. Samples irradiated with various doses and not incubated for repair served as a calibration to determine the percentage of remaining DSBs in the repair samples from the fraction of DNA entering the gel (for details, see Ref. 52 ). A specialized PFGE assay was performed as described earlier (52 , 53) , and a 3.2-Mbp NotI restriction fragment on chromosome 21 was analyzed by DNA probe D21S1 for the determination of correct DSB rejoining events with the criterion of restriction fragment reconstitution.

RESULTS

Deficient DSB Repair in LIG4 and ATM Mutant Cells after Physiological X-Ray Doses.

In the present study, we investigated the repair of DSBs in confluent wild-type (MRC-5), DNA ligase IV-deficient (180BR), and ATM-deficient (AT1BR) primary human fibroblasts by PFGE and γ-H2AX analysis. Analysis of the time course for DSB repair by PFGE (Fig. 1A) ⇓ shows that 180BR cells have a pronounced repair defect after ionizing radiation doses between 10 and 80 Gy, consistent with previous determinations (45 , 53) . AT1BR cells, in contrast, show initially normal repair kinetics, followed by a slight but significant defect for repair times up to 24 h. The magnitude of the repair deficiency (∼10% unrepaired DSBs) is consistent with previous measurements (Ref. 54 , see Ref. 38 for a review). γ-H2AX focus formation was investigated at several time points after 2 Gy of irradiation (Fig. 2A) ⇓ . Although ionizing radiation-induced foci are visible in MRC-5 and 180BR cells 3 min after irradiation, AT1BR cells do not form foci at this time point. After 15 min incubation, the number of foci observed in AT1BR cells is similar to that of MRC-5 cells (Fig. 1B) ⇓ , suggesting that foci formation is slightly delayed but quantitatively similar between wild-type and ATM-deficient cells. This finding is in agreement with our recent work showing that H2AX phosphorylation after ionizing radiation exposure of nonreplicating cells can be carried out by ATM and DNA-PK in a redundant, overlapping manner. 3 For repair times of 4 h and longer, AT1BR cells show significantly more residual foci than MRC-5 cells, in agreement with the slight DSB repair defect observed by PFGE analysis. 180BR cells, in contrast, exhibit substantially more foci than MRC-5 and AT1BR cells for all time points up to 24 h. Significantly, for all three cell lines, the kinetics of foci disappearance after 2 Gy closely resemble the kinetics of DSB repair determined by PFGE after doses between 10 and 80 Gy. This similarity provides additional evidence that γ-H2AX foci analysis can be applied as a quantitative measure of DSB repair in NHEJ- and ATM-deficient cells.

Because 180BR and AT1BR cells are radiosensitive in asynchronous populations (55 , 56) , we wished to compare asynchronous cells to confluent G₁-phase cells. Cell survival experiments confirm that both cell lines exhibit substantial (∼3–4 fold) radiosensitivity in asynchronous populations (Fig. 3A) ⇓ , and show that radioresistance is only marginally restored in confluent G₁-phase cells plated immediately after irradiation (Fig. 3B) ⇓ . When cell survival is assessed in G₁-phase cells that are plated 4 days after irradiation (delayed plating experiments), 180BR but not AT1BR cells show significant recovery compared with immediate plating conditions (Fig. 3C) ⇓ . The lack of significant recovery in delayed plating experiments with AT1BR cells is in agreement with literature data (40 , 41) and may indicate that DSBs remain unrepaired and contribute to the radiosensitivity of AT cells. The significant difference in survival of 180BR cells between delayed and immediate plating conditions, conversely, suggests that substantial DSB repair may occur during the extra time available in delayed plating experiments.

ATM Mutant Cells Fail to Repair a Subset of DSBs.

To investigate the capacity of 180BR and AT1BR cells to repair DSBs many days after irradiation, we have analyzed the time course for DSB repair by PFGE up to 14 days after 80 Gy X-irradiation (Fig. 4A ⇓ , Lanes 8–13). AT1BR cells show no significant DSB repair between days 1 and 14 and exhibit the same value of ∼10% unrepaired breaks during the entire incubation period. 180BR cells, in contrast, show substantial repair between days 1 and 14 and reach a level of unrepaired DSBs after day 14, which is similar to that of AT1BR cells (Fig. 4B) ⇓ . Low dose rate experiments, where the cells are continuously exposed to γ-radiation inside an incubator for 14 days at a dose rate of 5.7 Gy/day (for a total dose of 80 Gy; Fig. 4A ⇓ , Lanes 14 and 15), show proficient repair in 180BR cells but a significant fraction of unrepaired DSBs (∼5%) in AT1BR cells (Fig. 4C) ⇓ . Analysis of the time course for DSB repair between days 1 and 14 after 2 Gy of X-rays by γ-H2AX focus formation shows substantial repair up to 14 days in 180BR cells but little repair and a significant fraction of unrepaired DSBs (∼4 foci/cell, i.e., ∼6% unrepaired DSBs) after day 14 in AT1BR cells (Fig. 4D) ⇓ . Taken together, these experiments demonstrate that 180BR cells have the capacity to repair DSBs many days after irradiation, whereas AT1BR cells fail to show significant repair and exhibit a substantial fraction of unrepaired DSBs many days after irradiation. Significantly, after low dose (Fig. 4D) ⇓ or low dose rate conditions (Fig. 4C) ⇓ , the level of unrepaired DSBs in 180BR falls below that of AT1BR cells, consistent with the increased survival of 180BR cells under delayed plating conditions (Fig. 3C) ⇓ .

Using a specialized PFGE assay in which the reconstitution of a large genomic restriction fragment is quantified by Southern hybridization (57) , it is possible to identify correctly rejoined DSB ends produced by ionizing radiation. This approach in combination with conventional PFGE, which measures correctly as well as incorrectly rejoined DSB ends, allows the quantification of incorrect rejoining events, termed misrejoining (57) . We previously showed that low dose rate conditions as described above (Fig. 4C) ⇓ lead to no detectable misrejoining in wild-type MRC-5 cells but reveal a substantial misrejoining frequency in 180BR cells (53) . This finding is confirmed in the present study. Additionally, we observed that AT1BR cells show much less misrejoining than 180BR cells (Fig. 4E) ⇓ . Because elevated misrejoining frequencies are thought to result from decreased DSB repair capacity, the elevated misrejoining level in 180BR cells likely reflects the impaired but continuous DSB repair in these cells. AT1BR cells, in contrast, which show initially normal DSB repair but exhibit a significant fraction of unrepaired breaks, fail to show substantial misrejoining. Thus, we conclude that AT1BR cells have a DSB repair defect that is quantitatively and qualitatively different from the defect in 180BR cells. The latter cell line repairs the majority of DSBs with strongly compromised kinetics, which leads to an increase in the error rate, exemplified in the present study by an increase in the DSB misrejoining frequency. AT cells repair the majority of breaks with normal kinetics, leading to a low error rate, but fail to repair a subset of the induced DSBs.

We also investigated another AT line, AT7BI, and an additional DNA ligase IV-deficient cell line, 411BR, for their capacity to repair DSBs several days after 2 Gy of γ-irradiation. γ-H2AX analysis shows that AT7BI cells exhibit, compared with 411BR cells, fewer unrepaired DSBs after short repair times but more residual breaks after 2 and 3 days (Fig. 5A) ⇓ , demonstrating that the observations with 180BR and AT1BR cells represent a general feature of ATM- and DNA ligase IV-deficient cell lines. In the same experiments, we also analyzed a cell line, 2BN, with features that closely resemble those of NHEJ-deficient cells. These cells are highly radiosensitive, strongly impaired in DSB repair measured by PFGE, and show reduced V(D)J recombination. Because no deficiencies were observed in the known NHEJ factors, it was suggested that the defect of 2BN cells lies in a yet unidentified component of the NHEJ pathway (47) . Significantly, these cells show a substantial DSB repair defect by γ-H2AX analysis but, as with 180BR cells, continue to repair for several days and reach a level of unrepaired DSBs lower than that of AT7BI cells (Fig. 5A) ⇓ . This finding supports our conclusion that AT cells are defective in repairing a subset of DSBs that can be repaired by cells with defects in NHEJ. The biological importance of the ongoing repair in 411BR cells is demonstrated by the increase in cell survival if plating is performed 1 or 4 days after irradiation (Fig. 5B) ⇓ , similar to that observed with 180BR cells (Fig. 3C) ⇓ .

DSB Repair Deficiency in LIG4 and ATM Mutant Cells after Very Low X-Ray Doses.

We next investigated the capacity of 180BR and AT1BR cells for DSB repair after radiation doses that introduce only a few DSBs (Fig. 2B) ⇓ . The number of γ-H2AX foci initially induced by a 0.2-Gy X-ray dose is ∼7/cell, in agreement with 70 initial foci/cell after a dose of 2 Gy. After 0.02 Gy, an average of 0.7 foci/cell is introduced, i.e., the majority of the cells have one or no focus (Fig. 6A) ⇓ . The time course of repair determined by γ-H2AX analysis after 0.2 or 0.02 Gy is similar, yet not identical, to that after 2 Gy. AT1BR cells show initially normal repair but exhibit more unrepaired breaks than wild-type cells 8 and 24 h after irradiation, whereas 180BR cells show substantially more foci than wild-type cells at all time points up to 24 h (Fig. 6A) ⇓ . However, a detailed comparison of the repair kinetics of 180BR cells measured by PFGE and γ-H2AX analysis between 0.02 and 80 Gy indicates an improved repair capacity with decreasing radiation dose (compare Fig. 1 ⇓ with Fig. 6A ⇓ ). Whereas >40% of the initially induced breaks remain unrepaired 24 h after 80 Gy, this value decreases to ∼25% for the dose range between 2 and 20 Gy and even further to ∼10% for doses of 0.02 and 0.2 Gy. AT1BR cells, in contrast, show a nearly dose-independent value of ∼10% unrepaired DSBs after 24 h. Despite the improved repair capacity of 180BR cells at lower radiation doses, these cells do show a significant repair defect after 0.02 and 0.2 Gy. This observation together with the finding that 180BR cells repair DSBs for many days after irradiation (Fig. 4) ⇓ is consistent with the idea that mutated DNA ligase IV molecules in 180BR cells can perform the ligation step in NHEJ repair but do this at a reduced rate even for individually occurring breaks. At higher doses, the mutation in 180BR cells leads to a more severe repair defect, indicating that the mutated DNA ligase IV protein can handle a small but not an excessive number of DSBs.

To further test the idea that the hypomorphic mutation in 180BR cells confers residual DNA ligase IV activity that leads to a DSB repair capacity, which depends on the initial number of lesions, we have investigated a human pre-B cell line, N114P2, with targeted disruptions in both LIG4 alleles (48) . Because 180BR cells show improved DSB repair at doses of ≤0.2 Gy, γ-H2AX foci formation was analyzed in N114P2 and their corresponding wild-type cells, Nalm-6, after a dose of 0.2 Gy (Fig. 6B) ⇓ . The initial number of induced DSBs as well as the time course for DSB repair is very similar between wild-type primary fibroblasts and wild-type pre-B cells, indicating that a comparison between the two different cell types may be justified. The time course of the LIG4 null pre-B cells, however, is significantly impaired in comparison with the time course of the DNA ligase IV-deficient primary fibroblasts (Fig. 6B) ⇓ . The difference is most pronounced for repair times of 8 and 24 h when N114P2 cells show, compared with 180BR cells, more than twice as many unrepaired DSBs. Because of methodological difficulties related to synchronizing suspension cells in G₁ (see “Materials and Methods”), this result by itself cannot serve to demonstrate the residual DNA ligase IV activity in 180BR cells. However, in combination with the described dose dependence and the continuous repair activity of 180BR cells over many days, this experiment supports our conclusion that mutated DNA ligase IV molecules in 180BR cells have a physiological function that provides a reasonable repair capacity for a limited number of DSBs.

DISCUSSION

By using γ-H2AX foci analysis as an approach for DSB repair measurements, we conducted the first study of DSB repair in AT cells using physiologically relevant radiation doses. Our results show that AT cells held in the G₁ phase of the cell cycle have a significant repair defect that is detected both by PFGE and γ-H2AX foci analysis. Although AT cells repair the majority of DSBs with kinetics similar to those of wild-type cells, a significant fraction of the induced breaks remains unrepaired. Several days after irradiation, the residual breaks in AT cells outnumber the unrepaired DSBs in NHEJ-deficient cells, which continue to repair DSBs for many days and thus finally reach a level of unrepaired DSBs similar to that of wild-type cells (Figs. 4 ⇓ and 5) ⇓ . These findings agree qualitatively with previous PFGE studies that show that AT cells exhibit a small but reproducible repair defect (34 , 35 , 58) and that NHEJ-deficient cells can repair all of the DSBs after prolonged incubation (59 , 60) . Additional evidence for prolonged DSB repair in NHEJ-deficient cells is the substantial recovery of radiosensitivity if plating is performed 4 days after irradiation (Fig. 3) ⇓ . AT cells, in contrast, show little recovery in delayed plating experiments, demonstrating that repair ceases in these cells at earlier times after irradiation. The correlation between unrepaired DSBs and survival in these experiments suggests that the striking radiosensitivity of AT cells can be attributed to a defect in repairing a subset of DSBs, as opposed to impaired cell cycle checkpoint activation.

We also detected a difference in the quality of DSB rejoining between ATM- and NHEJ-deficient cells by examining the ability of cells to reconstitute a genomic restriction fragment to its original size. In this assay, NHEJ- but not ATM-deficient cells show a high level of DSB misrejoining (Fig. 4E) ⇓ . We propose that the misrejoining occurs as a consequence of the slow rejoining. In contrast, AT cells, by not carrying out any slow rejoining, do not show elevated misrejoining. It is noteworthy that our experiments involve plateau-phase cells and therefore cannot be attributed to any impact of cell cycle checkpoints. Overall, our findings are consistent with cytogenetic studies involving premature chromosome condensation of G₁-arrested primary human AT fibroblasts (36 , 37) or fluorescence in situ hybridization of mitotic cells (61 , 62) . In these studies, AT cells exhibit a substantially increased level of unrepaired chromosome breaks after irradiation compared with control cells and a smaller increase of exchange-type chromosome aberrations. The latter can be considered equivalent to the misrejoining events measured in this study. In summary, our data demonstrate that AT cells display a substantial repair defect, and we provide evidence for its link to their radiosensitivity. The slow residual rejoining in DNA ligase IV-deficient cells appears to be associated with misrejoining but elevated survival, providing evidence that misrejoining may not necessarily be a lethal event.

Radiation quantities of 0.02 Gy are frequently encountered during the use of X-rays in the clinic for diagnostic purposes. Such doses introduce approximately one DSB/cell. It is, therefore, important to investigate the factors required for the repair of an individual DSB in a cell. A repair defect can be clearly demonstrated in AT1BR and 180BR cells after 0.02 Gy (Fig. 6A) ⇓ . Thus, γ-H2AX foci analysis is a suitable approach to screen for DSB repair deficiencies using clinically relevant doses. For AT1BR cells, the fraction of unrepaired DSBs (∼10%) appears to be dose independent. For example, after the induction of 70 DSBs by a dose of 2 Gy, the average number of unrepaired DSBs/cell is 7. After 0.02 Gy, which induces on average 0.7 DSBs/cell, ∼1 of 10 cells with a DSB fails to repair the break (i.e., 10%). This suggests that the defect in AT cells cannot be attributed to a failure to detect a low number of breaks. Instead, the data suggest that ATM is required for the repair of a subset of radiation-induced breaks which arise in a dose-independent manner.

180BR cells carry a homozygous mutation (R278H) in the LIG4 gene that allowed normal development of the patient until the onset of leukemia at age 14 years but conferred marked clinical radiosensitivity (22) . On the basis of cellular and biochemical analyses of the mutant protein, it was suggested that the residual activity present in 180BR cells may be sufficient to repair the low number of DSBs introduced during V(D)J recombination but not the excessive number of DSBs after radiotherapy (23) . Here, we show that the repair capacity of 180BR cells, in contrast to AT cells, is dose dependent. After doses of 0.02 and 0.2 Gy, which introduce 0.7 and 7 DSBs/cell, respectively, the kinetics of DSB repair are faster compared with those at higher doses. As a consequence, the defect relative to AT1BR cells is dose and time dependent. For example, 180BR cells have more unrepaired DSBs than AT cells 24 h after doses of 2 Gy and higher (Fig. 1) ⇓ but exhibit a similar level of unrepaired breaks 24 h after doses of 0.02 and 0.2 Gy (Fig. 6A) ⇓ . However, notwithstanding the improved repair capacity of 180BR cells after low doses, there is still a substantial defect observed even for individually occurring breaks. This finding provides an explanation for the apparent paradox that the patient showed pronounced clinical radiosensitivity but did not display any marked immunodeficiency and additionally highlights the necessity to measure DSB repair for both mechanistic and clinical purposes at physiological radiation doses. We also examined a human pre-B LIG4^−/− cell line and found a repair defect after a dose of 0.2 Gy that is greater than that of 180BR cells (Fig. 6B) ⇓ , providing additional evidence that there is residual DNA ligase IV-dependent repair capacity in 180BR cells. However, analysis of these cells at prolonged repair times is hindered by difficulties to hold them in a noncycling state.

Taken together, our experiments using γ-H2AX foci analysis in combination with PFGE demonstrate a considerable DSB repair defect in confluence-arrested G₁-phase primary AT fibroblasts that lies in the inability of these cells to repair a subset of radiation-induced breaks. Our ability to use low doses for DSB repair experiments has allowed us to examine the contribution of these unrepaired breaks to survival using delayed plating experiments. We argue that the DSB repair defect underlies a substantial component of the radiosensitivity of AT cells.