Genetic analysis of the DNA-dependent protein kinase reveals an inhibitory role of Ku in late S-G2 phase DNA double-strand break repair.

Two major complementary double-strand break (DSB) repair pathways exist in vertebrates, homologous recombination (HR), which involves Rad54, and non-homologous end-joining, which requires the DNA-dependent protein kinase (DNA-PK). DNA-PK comprises a catalytic subunit (DNA-PKcs) and a DNA-binding Ku70 and Ku80 heterodimer. To define the activities of individual DNA-PK components in DSB repair, we targeted the DNA-PKcs gene in chicken DT40 cells. DNA-PKcs deficiency caused a DSB repair defect that was, unexpectedly, suppressed by KU70 disruption. We have shown previously that genetic ablation of Ku70 confers RAD54-dependent radioresistance on S-G(2) phase cells, when sister chromatids are available for HR repair. To test whether direct interference by Ku70 with HR might explain the Ku70(-/-)/DNA-PKcs(-/-/-) radioresistance, we monitored HR activities directly in Ku- and DNA-PKcs-deficient cells. The frequency of intrachromosomal HR induced by the I-SceI restriction enzyme was increased in the absence of Ku but not of DNA-PKcs. Significantly, abrogation of HR activity by targeting RAD54 in Ku70(-/-) or DNA-PKcs(-/-/-) cells caused extreme radiosensitivity, suggesting that the relative radioresistance seen with loss of Ku70 was because of HR-dependent repair pathways. Our findings suggest that Ku can interfere with HR-mediated DSB repair, perhaps competing with HR for DSB recognition.


Two major complementary double-strand break (DSB)
repair pathways exist in vertebrates, homologous recombination (HR), which involves Rad54, and non-homologous end-joining, which requires the DNA-dependent protein kinase (DNA-PK). DNA-PK comprises a catalytic subunit (DNA-PKcs) and a DNA-binding Ku70 and Ku80 heterodimer. To define the activities of individual DNA-PK components in DSB repair, we targeted the DNA-PKcs gene in chicken DT40 cells. DNA-PKcs deficiency caused a DSB repair defect that was, unexpectedly, suppressed by KU70 disruption. We have shown previously that genetic ablation of Ku70 confers RAD54-dependent radioresistance on S-G 2 phase cells, when sister chromatids are available for HR repair. To test whether direct interference by Ku70 with HR might explain the Ku70 ؊/؊ /DNA-PKcs ؊/؊/؊ radioresistance, we monitored HR activities directly in Ku-and DNA-PKcsdeficient cells. The frequency of intrachromosomal HR induced by the I-SceI restriction enzyme was increased in the absence of Ku but not of DNA-PKcs. Significantly, abrogation of HR activity by targeting RAD54 in Ku70 ؊/؊ or DNA-PKcs ؊/؊/؊ cells caused extreme radiosensitivity, suggesting that the relative radioresistance seen with loss of Ku70 was because of HR-dependent repair pathways. Our findings suggest that Ku can interfere with HR-mediated DSB repair, perhaps competing with HR for DSB recognition.
DNA double-strand breaks (DSBs) 1 can occur during normal cell division or be induced by ionizing radiation (IR). Verte-brates possess two major, complementary DSB repair pathways, homologous DNA recombination (HR) and nonhomologous DNA end-joining (NHEJ). DSB repair by HR uses homologous sequence provided by either a homologous chromosome or sister chromatid, whereas NHEJ joins DNA ends through a process that is largely independent of terminal homologies and that can produce junctions of varying sequence. The HR pathway requires genes of the RAD52 epistasis group, whose products include Rad51 and Rad54. NHEJ is necessary for V(D)J recombination, which generates antibody diversity and is essential for the development of the immune system. The NHEJ pathway requires the DNA-dependent protein kinase, DNA-PK (1), XRCC4 (2), and ligase IV (3)(4)(5) proteins. The DNA-PK holoenzyme is a serine-threonine protein kinase comprising a large catalytic subunit (DNA-PKcs) and a heterodimeric component, Ku, which consists of the 70-kDa Ku70 and 80-kDa Ku80 proteins (6 -8). DNA-PKcs is a member of a family of large proteins characterized by a carboxyl-terminal phosphatidylinositol 3-kinase-like domain (1,9). The nature of its physiological target(s) in the NHEJ pathway is still uncertain. The Ku70/Ku80 heterodimer binds to DSBs, recruits DNA-PKcs, and eventually stimulates its kinase activity (1,10).
Gene disruption experiments in mice have probed the functional relationships within the NHEJ proteins. Although mice deficient in DNA-PK proteins consistently exhibit elevated radiosensitivity and impaired V(D)J recombination, a defect in Ku appears to disrupt end-joining more profoundly than one in DNA-PKcs, because neither V(D)J coding nor signal joints are observed in Ku-deficient mice (11,12), whereas signal joints are found in DNA-PKcs-deficient mice (13,14). Furthermore, the dwarf phenotype and apparent replicative senescence described for Ku70-and Ku80-deficient mice (11,(15)(16)(17), but not DNA-PKcs-deficient mice, raise the possibility that there exist other activities of Ku, such as controlling cell growth or telomeric function. Remarkably, mice deficient in either XRCC4 or ligase IV show even more severe phenotypes than DNA-PKdeficient mice, exhibiting early embryonic lethality and extensive chromosomal aberrations (18,19). It is not clear whether these phenotypes can be explained solely by impaired end-joining. NHEJ and HR play complementary roles in repairing DNA damaged by IR; NHEJ dominates during G 1 to early S phase of the cell cycle, and HR is used in late S to G 2 phases (20,21). However, the overlap between these two pathways raises the possibility of competition between them. A recent model for vertebrate DSB repair proposed from a biochemical study that competition between NHEJ and HR proteins for the initial binding of a DNA lesion determines the eventual outcome of repair, i.e. whether the lesion is repaired by HR or by NHEJ (22). Ku, as the DNA-binding component of NHEJ, is a good candidate for the "switch" controlling entry to this pathway. To test this model and to gain further insight into the interplay between the DNA-PK subunits and HR, we engineered chicken DT40 cells with defective Ku70, DNA-PKcs, or both and examined their IR-induced DSB repair capacities at various cellcycle stages. Furthermore, because a major prediction of the "competitive" model was that HR frequencies should be elevated in the absence of Ku, we tested the effects of Ku on cellular HR capability directly using expression of the restriction endonuclease I-SceI to generate DSBs.

EXPERIMENTAL PROCEDURES
Construction of Targeting and Expression Vectors-Chicken genomic DNA from the PRKDC locus was amplified using primer pairs designed to amplify between exons 61-63 and 65-69 (23) and then used to generate targeting vectors carrying neomycin, hygromycin, or histidinol resistance cassettes. Targeting with these constructs was expected to replace amino acids 2888 to 3012 of DNA-PKcs, which are upstream of the phosphatidylinositol 3-kinase homology domain. The previously described chicken KU70 targeting construct pKu70-puro (20) was modified to carry a blasticidin resistance cassette (pKu70-bsr) and a RAD54 vector carrying blasticidin resistance, pRad54-bsr, was cloned from pRad54-neo (24). To make the Ku70 expression constructs, the fulllength chicken Ku70 cDNA was ligated into pAneo (25), yielding pAneo-GdKu70. The conditions of cell culture were described previously (26). mg/ml histidinol (Sigma). Genomic DNA was extracted from each clone by standard procedures, and clones that had undergone targeted recombination were identified by PCR or Southern blot analysis. The number of targeting events per analyzed clones following transfection of targeting vectors containing the neomycin, hygromycin, and histidinol resistance genes is 3/4, 4/24, and 3/21, respectively. To generate KU70 Ϫ/Ϫ /DNA-PKcs Ϫ/Ϫ/Ϫ double mutants, a DNA-PKcs Ϫ/Ϫ/Ϫ clone was sequentially transfected with pKu70-puro and pKu70-bsr. The  (20). Numbers indicate the percentage of cells in the gated regions. 4 (B) or nine (C) h after the removal of nocodazole, most cells are synchronized in G 1 -early S phase or late S-G 2 phase, respectively. At these two time points, the radiosensitivity of synchronized populations was examined, as described for Fig. 2. Error bars show the mean Ϯ S.D. for at least three separate experiments. Two independently targeted clones of each genotype show the same sensitivity to ␥-rays (data not shown).

Inhibition of S-G 2 Phase HR Repair by Ku 44415
number of targeting events per analyzed puro-and bsr-resistant clones is 16/37 and 3/81, respectively. To make DNA-PKcs Ϫ/Ϫ/Ϫ /RAD54 Ϫ/Ϫ double mutants, the DNA-PKcs Ϫ/Ϫ/Ϫ clone was sequentially transfected with pRad54-puro and pRad54-bsr. The number of targeting events per analyzed puro-and bsr-resistant clones is 3/40 and 2/85, respectively. Colony Formation Assays and Cell-cycle Analysis-Serially diluted cells were plated in triplicate onto 6-well clusters with 5 ml/well of 1.5% (w/v) methylcellulose (Aldrich) plates containing Dulbecco's modified Eagle's medium/F-12 (Life Technologies, Inc.), 15% fetal calf serum, 1.5% chicken serum, and 10 Ϫ5 M ␤-mercaptoethanol. Subsequently, ␥-radiation of the cells was performed using 137 Cs (0.02 Gy/s; Gammacell 40; Atomic Energy of Canada Limited Industrial Products, Ontario, Canada). Colonies were counted at 7 days after irradiation treatment. Percentage survival was determined relative to numbers of colonies from untreated cells. For cell-cycle analyses, cells were labeled for 10 min with 20 M BrdUrd (Amersham Pharmacia Biotech). They were then harvested and fixed at 4°C overnight with 70% ethanol and successively incubated as follows: (i) in 4 N HCl, 0.5% Triton X-100 for 30 min at room temperature; (ii) in fluorescein isothiocyanate-conjugated anti-BrdUrd antibody (Pharmingen, San Diego, CA) for 1 h at room temperature; (iii) in 5 g/ml phosphatidylinositol in phosphatebuffered saline. Between each incubation, cells were washed with phosphate-buffered saline containing 2% fetal calf serum and 0.1% sodium azide. Subsequent flow cytometric analysis was performed on a FACScan (Becton Dickinson, Mountain View, CA). Fluorescence data were displayed as dot plots using Cell Quest software (Becton Dickinson).
Western Blot Analysis-Western blot analysis of DNA-PKcs and Ku70 was performed as described previously (26). Briefly, 10 6 cells were lysed in 20 l of SDS lysis buffer. Following sonication and boiling, the lysates (10 l/lane) were separated by 5% SDS-polyacrylamide gel electrophoresis gel. After transfer to a nylon membrane, proteins were detected by anti-DNA-PKcs antiserum (Neomarkers, Fremont, CA) and horseradish peroxidase-conjugated anti-mouse IgG antibody and Super Signal chemiluminescent substrate (Pierce, Rockford, IL). The same lysates (10 l/lane) were separated by 10% SDS-polyacrylamide gel electrophoresis gel. After transfer to a nylon membrane, proteins were detected by anti-GdKu70 antiserum (20,26) and horseradish peroxidase-conjugated anti-rabbit IgG antibody and Super Signal chemiluminescent substrate. (27) was inserted into the previously described OVALBUMIN gene construct and then targeted into the OVALBUMIN locus in wild-type, KU70 Ϫ/Ϫ , and DNA-PKcs Ϫ/Ϫ/Ϫ DT40 cells. In transient transfections, 5 ϫ 10 6 cells suspended in 0.5 ml of phosphate-buffered saline were mixed with each of the following plasmid DNA (30 g) without linearization: pBluescript SK, I-SceI expression vector (pCBASce), and chicken Ku70 expression vector and electroporated at 250 V, 960 microfarads. 24 h after electroporation, cells were transferred to 96-well clusters containing 2.0 mg/ml G418. Cells were grown for 5-7 days, and surviving G418 resistant colonies were counted.

Ku Deficiency Confers Radiotolerance upon Wild-type and DNA-PKcs Ϫ/Ϫ/Ϫ Cells in Late S to G 2 Phase-
To monitor the ability of the cells to carry out DSB repair, we performed clonogenic survival assays following DSB induction by ␥-irradiation. Asynchronous DNA-PKcs Ϫ/Ϫ/Ϫ cells were more radiosensitive than either wild-type or KU70 Ϫ/Ϫ cells (Fig. 2). Asynchronous KU70 Ϫ/Ϫ cells showed a biphasic pattern of radiosensitivity as IR dose increased (Fig. 2), reflecting the presence of two distinct fractions, as we have described previously (20). In nocodazole-synchronized populations, KU70 Ϫ/Ϫ cells in G 1 to early S phase of the cell cycle, where end-joining is used exclusively (20), are extremely sensitive to IR (Fig. 3, A  and B). KU70 Ϫ/Ϫ cells in late S-G 2 phase, where HR is used preferentially over end-joining for DSB repair, are more resistant to IR than wild-type cells (Fig. 3C). KU70 Ϫ/Ϫ and DNA-PKcs Ϫ/Ϫ/Ϫ cells showed equal radiosensitivity in G 1 to early S phase (Fig. 3B), demonstrating that the NHEJ pathway is impaired to the same extent in either mutant line. However, in late S-G 2 phase, DNA-PKcs Ϫ/Ϫ/Ϫ cell populations were slightly more radiosensitive than wild-type cells, which contrasts markedly with the enhanced radioresistance of KU70 Ϫ/Ϫ cells (Fig. 3C) (20).
To test whether the radioresistance of Ku-deficient cells is caused by the absence of Ku or by the presence of DNA-PKcs, we analyzed the radiosensitivity of KU70 Ϫ/Ϫ /DNA-PKcs Ϫ/Ϫ/Ϫ cells. Strikingly, cells that lack both DNA-PKcs and Ku70 showed the same pattern of radiosensitivity as KU70 Ϫ/Ϫ cells in both asynchronous and synchronized populations (see Fig. 2 and Fig. 3, B and C). Importantly, the expression of KU70 cDNA in KU70 Ϫ/Ϫ /DNA-PKcs Ϫ/Ϫ/Ϫ cells increased radiosensitivity to the level of DNA-PKcs Ϫ/Ϫ/Ϫ cells (Fig. 2). These findings reveal that the absence of Ku70, rather than the presence of Ku-deficient DNA-PKcs, increases the cellular DSB repair capacity in late S to G 2 phases. Because NHEJ is disrupted to the same extent in KU70 Ϫ/Ϫ and DNA-PKcs Ϫ/Ϫ/Ϫ cells (Fig. 3B), these observations suggest that the other major DSB repair pathway, HR, acts more efficiently in the absence of Ku70 in late S-G 2 .
DNA ligase IV-deficient DT40 cells show significantly higher IR sensitivity than DNA-PKcs Ϫ/Ϫ/Ϫ or wild-type cells. Deletion of Ku70 in the DNA ligase IV-deficient cells makes them relatively more radioresistant with a survival profile that matches that of KU70 Ϫ/Ϫ cells (30). This result supports our data that Ku proteins suppress homologous recombination in late S-G 2 phase.
A possible complication was potentially aberrant cell-cycle checkpoint response in the mutants described here. Because DT40 cells are p53-deficient, they lack a G 1 /S arrest. In yeast, Ku70 proteins are involved in G 2 /M arrest after DNA damage (31). However, after treatment with ionizing radiation, all cell lines described in this paper showed similar G 2 /M arrest and inhibition of DNA synthesis (data not shown), indicating that our results are not because of any difference in cell-cycle checkpoint.
HR Deficiency in KU70 Ϫ/Ϫ and DNA-PKcs Ϫ/Ϫ/Ϫ Cells Causes Severe Radiosensitivity-If the model that the different IR sensitivities of KU70 Ϫ/Ϫ and DNA-PKcs Ϫ/Ϫ/Ϫ cells are because of different HR efficiencies is correct, disruption of HRmediated repair in the two mutants should result in identical IR sensitivity in them. To test this hypothesis, we disrupted HR-mediated DSB repair by targeting the Rad54 gene in KU70 Ϫ/Ϫ and DNA-PKcs Ϫ/Ϫ/Ϫ cells, thus generating KU70-Ϫ/Ϫ /RAD54 Ϫ/Ϫ/Ϫ and DNA-PKcs Ϫ/Ϫ/Ϫ /RAD54 Ϫ/Ϫ clones. Consistent with our model, both clones exhibited the same high level of radiosensitivity (Fig. 4). Therefore, the increased IR tolerance of KU70 Ϫ/Ϫ cells in late S to G 2 phase requires functional HR-mediated repair. This conclusion supports the idea that Ku can interfere with HR-mediated DSB repair.
Ku70, but Not DNA-PKcs, Deficiency Elevates HR Frequencies in an Artificial Substrate-To investigate directly the Ku protein's suppression of HR-mediated repair, we measured DSBinduced HR using the SCneo substrate (27), which we integrated into the OVALBUMIN locus of wild-type, KU70 Ϫ/Ϫ , and DNA-PKcs Ϫ/Ϫ/Ϫ cell lines by gene targeting. In each clone containing the modified SCneo, a single DSB was introduced at the I-SceI site by transient expression of the I-SceI endonuclease. Because HR generates a functional neomycin resistance (neo ϩ ) gene from SCneo, the number of HR events can be determined by counting the number of neo-resistant colonies (27,32,33). A KU70 Ϫ/Ϫ clone transfected with an I-SceI expression vector showed significantly higher HR frequencies than a wild-type clone (paired t test, p ϭ 0.0047 Ͻ 0.01; see Table I) and a DNA-PKcs Ϫ/Ϫ cell line (p ϭ 0.0409 Ͻ 0.05). Notably, the expression of Ku70 reduced the recombination frequency by 25-40% in KU70 Ϫ/Ϫ cells but not in wild-type DT40s (Table II). The higher level of HR in KU70 Ϫ/Ϫ cells correlates with the increased late S to G 2 phase radiotolerance seen in Ku-deficient cells relative to wild-type cells (Fig. 3C). Therefore, Ku70 expression decreases the frequency of DSB-induced HR, providing a mechanistic explanation for the radiotolerization induced by Ku deficiency.
To define how the association of Ku with DSBs interferes with HR, we measured the induction of IR-induced subnuclear Rad51 foci and gene targeting efficiency in the presence and absence of Ku. We found no delay in kinetics of Rad51 focus formation in response to IR in KU70 Ϫ/Ϫ cells when compared with wild-type and DNA-PKcs Ϫ/Ϫ/Ϫ cells (data not shown). Thus, the Ku proteins do not necessarily affect the assembly of Rad51 in nucleoprotein filaments at DSBs. There was no significant difference in gene targeting efficiencies in wild-type and KU70 Ϫ/Ϫ cells, whereas a slight reduction of random integration frequencies was observed in the absence of Ku70 (data not shown). Thus, the Ku proteins do not suppress all HR reactions but specifically interfere with HR-mediated DSB repair. Presumably, the association of Ku with DSBs may disturb D-loop formation, i.e. interaction of the nucleoprotein filaments with homologous sequences and/or inhibit DNA synthesis following D-loop formation.
HR and NHEJ Activities in DT40 and Mammalian Cell Lines-Although genetic analyses of DNA repair in DT40 cells have recapitulated consistently the results gleaned from mammalian systems, there seems to be a quantitative difference in the relative usage of the two DSB repair pathways in late S to G 2 phase, with HR apparently playing a more important role in DSB repair in DT40 cells than in mammalian cell lines (20,34). In contrast, deletion of the Ku genes in mammalian cell lines including murine embryonic stem (ES) cells consistently increases their IR sensitivity (13,35). However, asynchronous Ku70-deficient ES cells show a biphasic survival curve with increasing radiation dose, being highly radiosensitive up to 2 Gy and less sensitive at doses of 2-4 Gy. This biphasic curve likely reflects two distinct fractions with different IR sensitivities, as observed in Ku70-deficient DT40 cells, although no data exist for synchronized ES cells. DNA-PKcs-null mutant ES cells exhibit no radiosensitivity, suggesting that DNA-PKcs is not required for end-joining in ES cells (13). However, because Gao et al. (13) also showed that primary fibroblasts derived from DNA-PKcs-null mice were significantly more radiosensitive than those from wild-type mice, as has been observed in other DNA-PKcs-deficient mammalian cell lines, the requirement for DNA-PKcs in NHEJ may vary between tissues TABLE I DSB-induced recombination frequencies of NHEJ-deficient cells PsP OVA ϩ SCneo-puro was stably integrated into the OVALBUMIN locus of wild-type, KU70 Ϫ/Ϫ and DNA-PKcs Ϫ/Ϫ/Ϫ cell lines by gene targeting. 5 ϫ 10 6 cells of each genotype were transfected with 30 g of either the I-SceI expression vector pCBASce (38) or the control plasmid, pBluescript SK, and subsequently selected in G418. The ratio of recombination frequency was calculated as the number of G418 R clones in each cell line relative to that in the wild-type cell line. One clone of each genotype was examined. Wild-type and KU70 Ϫ/Ϫ cells containing psP OVA ϩ SCneo-puro were transfected with 30 g of pCBASce or 30 g of pCBASce ϩ 30 g of pAneo-GdKu70 as in Table I. In each cell line, the ratio of neoϩ colonies was calculated as the number of G418 R clones obtained following transfection with pCBASce ϩ pAneo-GdKu70 relative to the number obtained after transfection with pCBASce alone. One clone of each genotype was examined. and cell lines. Therefore, the occasional discrepancies of between the results of IR sensitivity we have obtained using DT40 and those obtained in mammalian cells may actually prove useful in choosing appropriate experimental models for investigating NHEJ. Finally, more prominent enhancement of HR caused by defective Ku70 than by DNA-PKcs deficiency has been also observed in mammalian cells. 2 Accumulating evidence points to the use of differential HR versus NHEJ depending on the cell-cycle phase, although cell type may further influence the competitive balance between these two complementary DSB repair pathways. Defective Rad51 focus formation following IR in G 1 phase implies the presence of regulatory mechanisms that can specifically suppress HR at the step of formation of nucleoprotein filaments (36). A defect in this suppression might lead to ectopic recombination and heteroallelic recombination, possibly causing the loss of heterozygosity and eventually tumorigenesis. Conversely, a defect in suppressive regulation of the end-joining pathway could interfere with other repair pathways, as well as the HR-mediated repair pathway (37). As an example of this, we found that Ku70-deficient DT40 cells are rather more tolerant than wild-type cells of the chemotherapeutic DNA damage-inducing agent cisplatin, which does not directly induce DSBs (20). Thus, the molecular regulation of the DSB repair pathway to be used may determine the sensitivity of various tumor cells to chemotherapy and radiotherapy. Further understanding of this balance will grant us insight into the chromosomal events preceding malignant transformation.