Ataxia telangiectasia-mutated protein and DNA-dependent protein kinase have complementary V(D)J recombination functions

Abstract

Antigen receptor variable region exons are assembled during lymphocyte development from variable (V), diversity (D), and joining (J) gene segments. Each germ-line gene segment is flanked by recombination signal sequences (RSs). Recombination-activating gene endonuclease initiates V(D)J recombination by cleaving a pair of gene segments at their junction with flanking RSs to generate covalently sealed (hairpinned) coding ends (CEs) and blunt 5′-phosphorylated RS ends (SEs). Subsequently, nonhomologous end joining (NHEJ) opens, processes, and fuses CEs to form coding joins (CJs) and precisely joins SEs to form signal joins (SJs). DNA-dependent protein kinase catalytic subunit (DNA-PKcs) activates Artemis endonuclease to open and process hairpinned CEs before their fusion into CJs by other NHEJ factors. Although DNA-PKcs is absolutely required for CJs, SJs are formed to variable degrees and with variable fidelity in different DNA-PKcs–deficient cell types. Thus, other factors may compensate for DNA-PKcs function in SJ formation. DNA-PKcs and the ataxia telangiectasia-mutated (ATM) kinase are members of the same family, and they share common substrates in the DNA damage response. Although ATM deficiency compromises chromosomal V(D)J CJ formation, it has no reported role in SJ formation in normal cells. Here, we report that DNA-PKcs and ATM have redundant functions in SJ formation. Thus, combined DNA-PKcs and ATM deficiency during V(D)J recombination leads to accumulation of unjoined SEs and lack of SJ fidelity. Moreover, treatment of DNA-PKcs– or ATM-deficient cells, respectively, with specific kinase inhibitors for ATM or DNA-PKcs recapitulates SJ defects, indicating that the overlapping V(D)J recombination functions of ATM and DNA-PKcs are mediated through their kinase activities.

Ig and T-cell receptor (TCR) variable region exons are assembled during early lymphocyte development from variable (V), diversity (D), and joining (J) gene segments (1). This V(D)J recombination reaction is initiated by recombination-activating gene (RAG) endonuclease (1). RAG recognizes short recombination signal sequences (RSs) that flank each germ-line V, D, and J segment; in the context of an appropriate pair of V, D, or J segments, RAG introduces DNA double-strand breaks (DSBs) between V, D, or J coding sequences and flanking RSs. The RAG cleavage reaction generates two kinds of ends. Coding ends (CEs) are generated as covalently sealed hairpins, whereas the RS ends (SEs) are generated as blunt 5′-phosphorylated broken ends (1). Joining of both CEs and SEs is carried out by the ubiquitously expressed nonhomologous end-joining (NHEJ) pathway, a major DSB repair pathway in eukaryotes. NHEJ precisely joins the two SEs to form RS joins (SJs). In contrast, to form coding joins (CJs), the NHEJ pathway must open and process the hairpinned CEs before they are joined (1).

The mammalian NHEJ pathway has seven known members (2). Ku70 and Ku80 form a heterodimer (Ku) that recognizes and binds to DSBs. Among other functions, DNA-bound Ku likely protects ends from resection and also recruits other NHEJ pathway members, including DNA-dependent PK catalytic subunit (DNA-PKcs). Together, Ku70/80 and DNA-PKcs form the DNA-PK holoenzyme, which binds to and phosphorylates the Artemis endonuclease, thereby activating endonuclease function. Activation of Artemis is thought to be generally important for processing DNA ends that cannot be directly ligated (e.g., hairpinned CEs). In this regard, Artemis and to a variable extent, DNA-PKcs are not strictly required for joining ends that do not require processing (e.g., blunt 5′ phoshorylated SEs) (3, 4). X-ray repair cross complementing protein 4 (XRCC4) and Ligase 4 function as a complex to mediate the end ligation step of NHEJ (1, 5, 6). XRCC4-like factor [XLF, also called Cernunnos or nonhomologous end-joining 1 (NHEJ1)] is another recently discovered NHEJ factor (5, 6), which may have overlapping functions with ataxia telangiectasia-mutated (ATM) kinase, but its precise role is still emerging (7–9). During V(D)J recombination, Ku70, Ku80, XRCC4, and Ligase 4 are required for both CJ and SJ formation owing to their role in DSB recognition and end ligation. DNA-PKcs and Artemis are absolutely required for opening the hairpin-sealed CEs before ligation and as such, are absolutely required for CJ formation. However, Artemis is not required for SJ formation, because these ends do not require processing to be joined. SJs also can be formed in the absence of DNA-PKcs but to variable extents, indicating that DNA-PKcs has NHEJ functions beyond activation of Artemis (10).

The classical SCID mouse, which has a mutation in the DNA-PKcs kinase domain, has a block in lymphocyte development because of inability to form V(D)J CJs in developing lymphocytes, but it still forms relatively normal SJs (11–13). Genetically targeted DNA-PKcs–deficient mice showed a similar phenotype (14–16). Assays of V(D)J recombination through introduction of RAG and V(D)J recombination substrates into DNA-PKcs mutant nonlymphoid cells gave similar but not always identical results (11–17). Thus, DNA-PKcs–deficient mouse embryonic fibroblasts (MEFs) and a DNA-PKcs–deficient Chinese hamster ovary (CHO) cell line (XR-C1) had significantly decreased SJ efficiency and decreased fidelity with some SJs harboring long deletions (18–20). In contrast, DNA-PKcs–deficient ES cells and human glioma cells, reminiscent of developing lymphocytes, showed little or no defects in SJ efficiency or fidelity, although MEFs from the same animals had substantial SJ defects (16, 17).

The exact role of DNA-PKcs in SJ formation is not yet understood, although DNA-PKcs has been suggested to have synapsis functions in end joining (21). Likewise, the reason for the variable degree of RS joining in different DNA-PKcs–deficient mice and/or cell types has been a mystery. It has been proposed to reflect, at least in part, the nature of the DNA-PKcs mutations studied, which range from a point mutation in the kinase domain (in the SCID mouse previously described) to complete KO of the large DNA-PKcs protein. Another not mutually exclusive possibility is that the cell type-specific differences might reflect the variable presence of a specific factor (or factors) that compensates for DNA-PKcs function during SJ formation. Although SJs do not normally contribute to protein coding, their generation is still very important. Thus, because V(D)J recombination occurs in the context of inversion within certain Ig and TCR loci, defects in SJ formation during inversional V(D)J recombination could result in chromosomal breaks that would be highly detrimental to developing B and T cells. Likewise, unjoined SEs could be mutagenic after insertion into a chromosome (22–24) or might trigger DSB responses (25).

DSBs generated during V(D)J recombination activate the ATM-mediated DSB response (25, 26). In this regard, ATM stabilizes DNA during chromosomal V(D)J recombination to promote correct resolution of inversional V(D)J recombination reactions (27). ATM also promotes efficient CJ formation on chromosomal V(D)J recombination substrates (27). In addition, ATM and Histone H2AX, an ATM DSB response substrate, have overlapping activities with XLF in the generation of chromosomal V(D)J CJs and SJs (9). In contrast, extrachromosomal V(D)J recombination is not affected in ATM mutant cells (28). ATM and DNA-PKcs both belong to the PI3K-like kinase family, which also includes ataxia telangiectasia and Rad3 related protein (ATR) (29). Although ATR predominantly functions in S-phase repair, DNA-PKcs and ATM have significant functional overlaps in general DNA repair. Thus, mice deficient in either DNA-PKcs or ATM are live born, but DNA-PKcs and ATM double deficiency leads to early embryonic lethality (30, 31). At the molecular level, DNA-PKcs and ATM share many phosphorylation substrates during the DNA damage response, including histone H2AX (32) and potentially, Artemis (3). Finally, DNA-PKcs and ATM kinase activities were found to have overlapping functions in the context of Ig heavy-chain class switch recombination (CSR), another DSB joining process (32).

In this study, we have tested the hypothesis that ATM and DNA-PKcs have overlapping functions in SJ formation. Specifically, we generated DNA-PKcs and ATM double-deficient cells by conditional deletion of ATM from DNA-PKcs^−/− ES cells and immortalized pro–B-cell lines. Our result indicates that DNA-PKcs and ATM indeed have redundant functions in SJ formation on both extrachromosomal and chromosomal V(D)J recombination substrates. We also show that the overlapping functions of ATM and DNA-PKcs are mediated, at least in a substantial part, by their overlapping kinase activities.

Results

Generation of ATM and DNA-PKcs Double-Deficient Murine ES Cells.

To circumvent the early embryonic lethality associated with ATM and DNA-PKcs double deficiency (30, 31), we generated ES cells that were both DNA-PKcs deficient (DNA-PKcs^−/−) and homozygous for an ATM allele that could be conditionally inactivated through Cre recombinase (ATM^C/C) (33). Subsequently, ATM was inactivated in vitro using Adenovirus-encoded Cre recombinase to generate DNA-PKcs^−/−ATM^−/− ES cells as described for the generation of other double KO lines (33). We note that our initial DNA-PK^−/− targeting introduced a phosphoglycerate kinase (PGK)-neo cassette flanked by two loxP sites into exon 6 of DNA-PKcs (16) (Fig. S1A). On exposure to Cre, the PGK-neo cassette can be deleted from the DNA-PKcs locus to generate the DNA-PKcs^ΔNeo allele (Fig. S1 A–C). The resulting DNA-PKcs^ΔNeo allele contains an ∼140-bp insertion, including one LoxP site within exon 6 of DNA-PKcs, which can be identified by PCR with primers flanking exon 6 (between P4 and P5) (Fig. S1C). This insertion introduces two or more translation termination signals in all three possible reading frames (Fig. S1D). We confirmed that the DNA-PKcs^ΔNeo/ΔNeo cells did not express detectable DNA-PKcs protein by Western blotting (Fig. S1E) and that they were severely deficient for CJ formation on extrachromosomal V(D)J recombination substrates (Fig. S2). In the experiments described below, we used DNA-PKcs^ΔNeo/^ΔNeo ES cells as a control for the ATM^−/−DNA-PKcs^ΔNeo/ΔNeo ES cells. Similar deletion of the PGK-neo cassette within the DNA-PKcs locus also occurred during the generation of ATM/DNA-PK double-deficient pro-B cells by a similar Cre recombinase-mediated approach (see below).

ATM and Its Kinase Activity Play a Major Role in SJ Formation During Extrachromosomal V(D)J Recombination in DNA-PKcs–Deficient ES Cells.

To test if ATM has a role in SJ formation in the absence of DNA-PKcs, we conducted an extrachromosomal V(D)J recombination assay (8) to measure SJ formation efficiency in WT, ATM^−/−, DNA-PKcs^ΔNeo/ΔNeo, and DNA-PKcs^ΔNeo/ΔNeoATM^−/− ES cells. This assay is only semiquantitative and best used to identify a dramatic loss of V(D)J recombination efficiency. We performed multiple experiments on each line and used XRCC4-deficient ES cells, which show dramatically reduced ability to form SJs (2 ± 2% of WT levels) (Fig. 1A and Fig. S3), as a control. Consistent with prior publications, DNA-PKcs–deficient ES cells supported V(D)J recombination on SJ substrates with overall levels comparable with WT controls (i.e., TC1; 75 ± 33% of WT levels) (Fig. 1A and Fig. S3) (14, 16). Although ATM deficiency alone did not affect the efficiency of SJ formation (93 ± 15% of WT levels) (Fig. 1A and Fig. S3), ATM/DNA-PKcs double-deficiency led to a major reduction in SJ efficiency to about 10% of WT levels (9 ± 4% of WT level) (Fig. 1A and Fig. S3). Notably, treatment with an ATM kinase inhibitor (KU55933) also significantly reduced SJ formation in DNA-PKcs–deficient cells (to 10 ± 3% of WT level) but not in WT cells (153 ± 60% of WT level) (Fig. 1A and Fig. S3), indicating that the kinase activity of ATM is required for efficient SJ formation in DNA-PKcs–deficient cells.

Fig. 1.

DNA-PKcs /ATM double-deficient ES cells are defective in RS joining during extrachromosomal V(D)J recombination. (A) Summary of relative efficiency in SE joining measured by extrachromosomal V(D)J recombination assays in WT, DNA-PKcs^Δneo/Δneo, DNA-PKcs^Δneo/ΔneoATM^−/−, ATM^−/−, and XRCC4^−/− ES cells. The data presented are the average and SD of at least three independent experiments per cell type per condition. The detailed results of each experiment are presented in Fig. S2. P < 0.01 in all four pairs marked by the brackets (Student t test). (B) Fidelity analysis of the extrachromosomal signal joints recovered from DNA-PKcs^Δneo/Δneo, DNA-PKcs^Δneo/ΔneoATM^−/−, and DNA-PKcs^Δneo/Δneo cells treated with ATM inhibitor. (C) Scatter plot for the number of base pairs deleted at signal joints recovered from DNA-PKcs^Δneo/Δneo, DNA-PKcs^Δneo/ΔneoATM^−/−, and DNA-PKcs^Δneo/Δneo cells treated with ATM inhibitor. The black bar indicates the mean of each group. The exact sequence of each junction summarized in B and C is presented in Fig. S3.

Prior studies showed that DNA-PKcs mutations in lymphocytes, MEFs, and CHO cells resulted in variably reduced fidelity of SJs but that DNA-PKcs–deficient ES cells generated SJs with nearly normal fidelity (1, 11, 18–20). Precise joining of SEs within the SJ substrate (pJH200) generates a recognition site for the restriction enzyme ApaLI. Consistent with earlier findings, only 1 of 14 (7.1%) substrates recovered from DNA-PKcs–deficient ES cells was nonprecise (ApaLI-resistant); however, 4 of 12 (33.3%) and 9 of 18 (50%) substrates recovered from ATM inhibitor-treated DNA-PKcs^ΔNeo/ΔNeo ES cells or DNA-PKcs^ΔNeo/ΔNeoATM^−/− ES cells, respectively, were nonprecise (Fig. 1B and Fig. S4). To examine the nature of the imprecise junctions, we sequenced them and found significant base pair loss on one or both sides of the SJs from DNA-PKcs/ATM double-deficient cells (mean base pair loss = 8.44) or DNA-PKcs–deficient ES cells treated with ATM inhibitor (mean base pair loss = 4.33). In contrast, SJs from DNA-PKcs–deficient ES cells showed, as expected, little base pair loss (mean base pair loss = 0.07) (Fig. 1C and Fig. S4). Together, these results suggest that ATM kinase activity is required for the relatively efficient and precise extrachromosomal V(D)J SJ formation observed in DNA-PKcs–deficient cells.

ATM and Its Kinase Activity Play an Important Role in Chromosomal SJ Formation in DNA-PKcs–Deficient Cells.

To test if ATM also plays a role in SJ formation during chromosomal V(D)J recombination in DNA-PKcs–deficient cells, we generated v-abl kinase-transformed pro–B-cell lines from DNA-PKcs^−/− and DNAPKcs^−/−ATM^C/C mice. Treatment with a v-abl kinase inhibitor (STI571) arrests the v-abl–transformed pro-B cells in G1 and induces RAG expression, leading to efficient V(D)J recombination of integrated substrates in WT cells (27). We introduced a chromosomal V(D)J recombination substrate designed to test for RS joins and broken SEs (pMX-DEL^SJ) (Fig. 2A) (27). To measure SJ formation, the cells were treated with STI571 for 0 (untreated), 2, or 4 d. Formation of SJs and deletion of intervening sequences by V(D)J recombination led to the accumulation of a 4-kb band detectable by Southern blotting with a C4 probe after EcoRV digestion of the genomic DNA (Fig. 2A). In the event that the SJs could not be formed because of a repair defect, SEs accumulate as a 2.2-kb fragment visible by the same Southern analyses (e.g., as seen with XRCC4-deficient cells) (Fig. 2B).

Fig. 2.

ATM kinase activity is essential for efficient signal join formation during chromosomal V(D)J recombination of DNA-PK–deficient v-abl–transformed pro-B cells. (A) Diagram of the pMX-DEL^SJ substrate used in chromosomal V(D)J recombination assays. The diagram is adapted from Bredemeyer et al. (27). The pMX- DEL^SJ vector has a single pair of RSs. Normal V(D)J recombination between the two RSs leads to excision of the intervening sequences and formation of a chromosomal SJ. Also shown is a schematic representation of unrearranged (UR) cleavage intermediates (3′ SE) and joining products (SJ) of pMX-DEL^SJ. The long terminal repeats (LTR), IRES-hCD4 cDNA (hCD4), 5′ 12-RS and 3′ 23-RS (filled and open triangles, respectively), EcoRV (EV) site, and C4 probe are indicated. (B) Southern blot analyses of the rearrangement status of pMX-DEL^SJ V(D)J recombination substrate (A) in two pairs (DEL^SJ4 and DEL^SJ6) of DNA-PKcs^−/−ATM^C/C, DNAPKcs^−/−ATM^−/−, and XRCC4^−/− control v-Abl–transformed pro-B cells. DNA was digested with EcoRV and probed with the indicated C4 probe. The UR band is 5 kb, the SJ band is 4 kb, and the 3′SE band is 2.2 kb. (C) Southern blot analyses of the rearrangement status of pMX-DEL^SJ V(D)J recombination substrate (A) in DNA-PKcs^−/− cells with or without treatment of ATM inhibitor (15 μM). (D) Fidelity analysis of the chromosomal signal joints recovered from DNA-PKcs^−/−ATM^C/C and PKcs^−/−ATM^−/− cells with pMX-DEL^SJ substrates.

We generated two DNA-PKcs^−/−ATM^C/C pro–B-cell lines (DEL^SJ4 and DEL^SJ6), each with a single clonal integration of the pMX-DEL^SJ substrate, and then deleted ATM in those cells through Tat-Cre transduction (Fig. 2B). This approach eliminates variability associated with different sites of substrate integration. Consistent with largely normal SJ formation in vivo, v-abl transformed DNA-PKcs^−/−ATM^C/C pro-B cells rearranged the SJ substrate efficiently without accumulation of SEs (Fig. 2B). Deletion of ATM in both clones (DEL^SJ4 and DEL^SJ6), although still allowing formation of significant levels of chromosomal RS joins, led to substantial accumulation chromosomal SEs, showing that ATM plays a role in chromosomal SJ formation in DNA-PK^−/− pro–B-cell lines (Fig. 2 B and C). In addition, we found that treatment with ATM kinase inhibitor (KU55996) induced SE accumulation in DNA-PK^−/− pro–B-cell lines (Fig. 2C), suggesting that the kinase activity of ATM plays an important role in SJ formation in these DNA-PK^−/− cells. Finally, we cloned a number of SJs from either DNA-PKcs^−/−ATM^C/C:DEL^SJ4 or the isogenic DNA-PKcs^−/−ATM^−/−:DEL^SJ4 cell lines and investigated the fidelity of the SJs by ApaLI digestion and sequencing. Unlike ES cells, v-abl–transformed pro-B cells often express terminal deoxynucleotidyl transferase (Tdt); therefore, imprecise SJs could be caused by either nucleotide insertion because of Tdt activity or deletions. Again, we observed an increased frequency of nucleotide deletions in SJs from DNA-PKcs^−/−ATM^−/− cells (20%) vs. DNA-PKcs^−/−ATM^C/C cells (5.6%), without significant changes in the frequency of nucleotide insertion (29/36 = 80.5% for DNA-PKcs^−/−ATM^C/C vs. 31/40 = 77.5% for DNA-PKcs^−/−ATM^−/−) (Fig. 2D and Fig. S5). However, these findings also show that some normal chromosomal RS joins can be formed in the absence of both DNA-PKcs and ATM.

Overlapping Kinase Activities of ATM and DNA-PKcs Contribute to SJ Formation.

DNA-PKcs and ATM double-deficient ES cells and pro-B lines show more severe SJ defects than ATM-deficient cells. Studies on CJ formation have revealed potential structural functions (e.g., bridging ends) of DNA-PKcs in addition to its kinase activity (3). To test if the role of DNA-PKcs in SJ formation is caused by kinase activity or its potential structural functions, we tested if DNA-PK kinase inhibitor (NU7441 or NU7026) affected SJ formation in ATM-deficient pro-B lines. For this experiment, we generated pooled v-abl–transformed ATM^−/− pro-B cells with diverse pMX-DEL^SJ integrations and treated them with STI571 with or without DNA-PK kinase inhibitor [NU7441 (5 μM) or NU7026 (20 μM)]. As reported previously (27), ATM-deficient pro-B lines form SJs efficiently without accumulation of SEs (Fig. 3A). However, DNA-PKcs kinase inhibitor (either NU7441 or NU7026) treatment led to significant accumulation of SEs (Fig. 3A), indicating that DNA-PK kinase activity supports SJ formation in ATM-deficient cells. Sequence analyses of the SJs recovered from ATM-deficient cells with or without DNA-PK inhibitor (NU7441) revealed increased deletions in those from cells treated with inhibitor (33% vs. 0%) (Fig. 3B and Fig. S6). Finally, we generated WT cells with an integrated pMX-DEL^SJ substrate and treated them with either ATM or DNA-PKcs inhibitors alone or in combination. Notably, although neither inhibitor alone causes V(D)J SJ defects, as indicated by accumulation of SEs, combined ATM/DNA-PKcs kinase inhibitors led to accumulation of SEs after STI571 stimulation (Fig. 3C). Together, these results suggest that the overlapping kinase activities of DNA-PKcs and ATM promote precise and efficient chromosomal V(D)J SJ formation in v-Abl–transformed pro-B lines.

Fig. 3.

DNA-PKcs kinase activity promotes efficient signal join formation in ATM-deficient pro-B cells. (A) Southern blot analyses of the rearrangement status of pMX-DEL^SJ V(D)J recombination substrate in ATM^−/− v-Abl–transformed pro-B cells with or without DNA-PKcs inhibitor (either NU7026, marked as 26, or NU7441, marked as 41). (B) Fidelity analysis of the chromosomal signal joints recovered from ATM^−/− cells with pMX-DEL^SJ substrate in the presence or absence of DNA-PKcs inhibitor NU7441. (C) Southern blot analyses of the rearrangement status of pMX-DEL^SJ V(D)J recombination substrate in WT cells treated with ATM inhibitor, DNA-PKcs inhibitor, or both together. For both A and C, the DNA was digested with EcoRV and probed with the indicated C4 probe. The UR band is 5 kb, the SJ band is 4 kb, and the 3′SE band is 2.2 kb.

At endogenous antigen receptor loci, SJs formed through inversional V(D)J recombination are retained within the chromosome. To test if ATM and DNA-PKcs have a synergistic role in SJ formation during inversional V(D)J recombination, we tested whether DNA-PKcs kinase inhibitors affect SJ formation in WT or ATM-deficient v-Abl transformed pro-B cells carrying an inversional V(D)J recombination substrates (pMX-INV) (Fig. S7A). In cells that already have CJ defects (such as ATM-deficient cells or WT cells treated with ATM inhibitor), additional SJ defects during pMX-INV V(D)J recombination would lead to the accumulation of 3′SE-5′CE fragments (∼0.85 kb) that can be detected through Southern blotting with a GFP probe after EcoRV digestion of genomic DNA (Fig. S7A). Consistent with our observations with pMX-DEL^SJ substrates (Figs. 2 and 3), ATM and DNA-PKcs kinase activity also has overlapping functions in SJ formation during inversional V(D)J recombination, such that the 0.8-kb 3′SE-5′CE fragments are only observed in v-Abl–transformed pro-B cells treated with inhibitors of both the ATM and DNA-PKcs kinase activities but not cells singly treated with either ATM or DNA-PKcs inhibitors (Fig. S7 B and C). Together, our results clearly show that the kinase activities of DNA-PKcs and ATM have overlapping functions in V(D)J recombination SJ formation.

Discussion

During V(D)J recombination, both ATM and DNA-PKcs have well-documented roles in the CJ formation (2, 27). However, the role of DNA-PKcs in SJ formation has been a subject of some debate, because it varies significantly in different cells (13–16, 18–20, 34, 35), and a role for ATM in SJ formation has only been very recently observed, specifically in the context of XLF deficiency and primarily in the context of chromosomally integrated substrates (9). We now report that DNA-PKcs and ATM have overlapping functions, mediated through their kinase activities and potentially through their common downstream substrates, in promoting efficient SJ formation and preventing excess end resection on both chromosomal and extrachromosomal V(D)J recombination substrates. Based on our findings, it is possible that the variable impact of DNA-PKcs–deficiency on SJ formation in different cell types (e.g., ES cells vs. MEFs) might be related, at least in part, to the differential availabilities or activities in different cell types of the ATM kinase or downstream substrates shared with DNA-PKcs.

We observe a more substantial inhibition of extrachromosomal SE joining compared with chromosomal SE joining in the absence of ATM and DNA-PKcs (Figs. 1 and 2). This finding contrasts markedly with the impairment in chromosomal as opposed to extrachromosomal SE joining in the absence of XLF and ATM (9). Because there is still some residual chromosomal SE joining even in the absence of ATM and XLF that theoretically might represent DNA-PKcs–mediated SE joining, it is tempting to speculate that XLF might support the observed chromosomal joining in the absence of DNA-Kcs and ATM. However, there are many other possibilities.

Regarding the apparently overlapping ATM and DNA-PKcs substrates involved in SE joining, these two kinases phosphorylate a large number of shared substrates in response to DSBs, including both chromatin-associated repair factors (e.g., H2AX, SMC1, and Kap1) and members of the NHEJ pathway (e.g., KU, XRCC4, Ligase 4, Artemis, and DNA-PKcs itself) (36–38). Given the more substantial impairment of extrachromosomal SE joining in the absence of ATM and DNA-PKcs, relevant common substrates might be a core NHEJ factor or even RAG. Although previous mutagenesis analyses failed to identify a definitive role for ATM and DNA-PKcs phosphorylation sites of any single NHEJ factor during V(D)J recombination (36–38), it is possible that phosphorylation of two or more NHEJ factors might lead to overlapping functions that could mask effects of mutagenesis of a single factor. Alternatively, RAG, which generates and associates with SEs, also has potential phosphorylation sites for DNA-PKcs and ATM. Conceivably, ATM and DNA-PKs phosphorylation might regulate RAG and thereby, affect SE ligation. Finally, it still is notable the SJs that do form in the absence of DNA-PKcs and ATM have increased junctional base pair loss, suggesting a potential overlapping function in end protection, reminiscent of that recently described for H2AX in the context of unjoined CEs and SEs (9, 39). Indeed, it is quite possible that the overlapping functions of DNA-PKcs and ATM kinase activities might promote SJ formation through phosphorylation of several different substrates. Finally, the overlapping kinase activities of ATM and DNA-PKcs also are required for joining of DSBs during CSR (32), which, together with our current results, suggests that these overlapping ATM and DNA-PKcs activities might be more broadly involved in DNA repair through NHEJ.

Experimental Procedures

Mice.

DNA-PKcs–deficient mice were previously described (16). ATM-deficient and ATM conditional-deficient mice were derived from the ATM conditional targeting construct described before (33). All animal work has been conducted in a pathogen-free facility and approved by the Institute Animal Care and Use Committee of the Children's Hospital, Boston.

Derivation of ES Cells from Mice.

Three- to four-week-old female DNA-PKcs^+/−ATM^C/C mice were superovulated by i.p. injection of 5.0 IU pregnant mare serum gonadotropin (PMS; Calbiochem) followed by i.p. injection of 5.0 IU human chorionic gonadotrophin (hCG) (Sigma) 46–48 h later. The females were set up with adult DNA-PKcs^+/−ATM^C/C mice (1:1 ratio) immediately after the hCG injection. The mating was confirmed by plug checking the next morning. Four days after hCG injection, the mated females were killed, and mature blastocysts were flushed out of the uterus, treated with EmbryoMax Acidic Tyrode's Solution (Millipore) to remove the zona pellucid, and plated one per well in 96-well plates containing irradiated MEF feeders. After 9–14 d incubation, the cells were expanded to 48-well plates and then 24-well plates, at which stage one-half of the cells were frozen down and the rest of the cells were used to derive DNA for genotyping. DNA-PKcs^−/−ATM^C/C ES cells were identified by PCR and then confirmed by Southern blotting.

V(D)J Recombination Assays.

Extrachromosomal and chromosomal V(D)J recombination assays were performed as described (8, 9, 27). For ATM inhibitor treatment, KU55933 (15 μM; Tocris Bioscience) was added into the cells 12 h after the initial transfection for the extrachromosomal assay. The fidelity of the SJs was determined by ApaLI digestion and sequencing using primer 5′-GAT AAC AAT TTC ACA CAG GAA ACA GCT ATG ACC-3′ as described before (8, 27). For chromosomal V(D)J recombination analysis, the cells were exposed to v-abl kinase inhibitor (3 μM STI571; Novartis) with or without ATM inhibitor (15 μM KU55933; Tocris Bioscience) or DNA-PKcs inhibitor (5 μM NU7441 or 20 μM NU7026; Tocris Bioscience) for the indicated times. To generate DNA-PKcs^−/−ATM^−/− v-Abl–transformed pro-B lines, we transduced Tat-Cre recombinase into the DNA-PKcs^−/−ATM^C/C v-Abl–transformed cells with singly integrated pMX-DEL^SJ substrates as described previously (9).