Distinct domains in Nbs1 regulate irradiation-induced checkpoints and apoptosis

To determine the role of various Nbs1 domains in affecting ATM activity, we used BAC recombineering to mutate a BAC clone (255A7) containing the human Nbs1 gene, which was previously demonstrated to rescue the embryonic lethality and completely reconstitute Nbs1 knockout mice (7). Five BAC transgenes were crossed into an endogenous Nbs1-deficient background (Fig. 1 A): (a) Nbs1^WT, which expresses the human WT BAC (7); (b) Nbs1^657Δ5, which is a 5-bp deletion that leads to the expression of a hypomorphic allele that is described in 95% of NBS patients (7, 34, 35); (c) Nbs1^H45A, which is a point mutation in the FHA domain (7, 37); (d) Nbs1^tr735, which is a truncated derivative lacking the C-terminal 20 aa residues; and (e) Nbs1^S278A/S343A, which carries two mutations in which the serine S278 and S343 residues are modified to alanine. All of these constructs rescued the embryonic lethality of the Nbs1-null mutation, as mice were born at predicted Mendelian frequencies in intercrosses between transgene⁺Nbs1^+/− mice.

We confirmed the presence of the mutations in the BAC by sequence analysis (not depicted) and verified protein expression and size by Western blot analysis (Fig. 1 and Fig. 2 A). Nbs1^tr735 and Nbs1^S278A/S343A proteins were expressed at similar levels (within 1.5-fold) as the full-length Nbs1^WT transgene (Fig. 1, B and C). Nbs1^tr735 mice produced a truncated protein, as indicated by the fact that antibodies against the Nbs1 C terminus (aa 735–754) (50) failed to detect the protein, whereas the protein was detected using an antibody that recognizes the N terminus (aa 395–743; Fig. 1 B). Phoshospecific antibodies generated against S278 and S343 detected phosphorylated forms of Nbs1 in Nbs1^WT mice but not in Nbs1^S278A/S343A mice (Fig. 1 C). Moreover, we detected a substantial reduction of IR-induced mobility shift in Nbs1 when thymocytes from Nbs1^S278A/S343A were irradiated (Fig. 1 C). Nevertheless, it is possible that other sites within Nbs1 besides S278 and S343 could be phosphorylated in Nbs1^S278A/S343A mice in response to IR (22).

To determine whether Nbs1 mutants can still form a functional complex with Mre11/Rad50, we performed immunoprecipitation and immunofluorescence analyses on mouse embryo fibroblasts (MEFs; Fig. S1 and Fig. S2). Although the interaction with Nbs1 and nuclear localization of Mre11 were not affected in Nbs1^H45A and Nbs1^tr735 mutant cells, foci formation was dependent on the intact FHA domain but not on the C terminus, as expected (7, 25, 36, 37).

We have previously reported that Nbs1^H45A mice share many overlapping phenotypes with Nbs1^657Δ5 mice, including defective thymocyte and oocyte development, genomic instability, and a reduction in phosphorylation of the ATM downstream target Chk2 at low IR doses (7). These phenotypes are similarly observed, but more severe, in ATM knockout mice. In contrast, Nbs1^S278A/S343A and Nbs1^tr735 mice were fertile and did not exhibit any detectable defect in B cell or T cell development (not depicted). Moreover, Nbs1^tr735 MEFs did not show more γ IR–induced chromosomal aberrations relative to Nbs1^WT (Fig. S3). To examine the cellular DNA damage signaling response, we measured ATM autophosphorylation and substrate phosphorylation in primary B cells in response to γ IR (Fig. 2 A). In Nbs1^S278A/S343A mice, there was no detectable defect in phosphorylation of ATM, Chk1, or Chk2. In contrast, a small reduction in the magnitude of some ATM substrate phosphorylations, particularly Chk1-S317p and Chk2p was apparent in Nbs1^657Δ5, Nbs1^H45A, and Nbs1^tr735 B cells treated with a low dose (2 Gy) of IR. Consistent with inefficient substrate phosphorylation at low but not high doses of IR, the kinetics of Chk2 phosphorylation were perturbed in primary Nbs1^tr735 B cells relative to WT controls after 2 Gy (Fig. 2 B), but normal after 8 Gy, of IR (not depicted).

The finding that DSB signaling was not profoundly affected in Nbs1^tr735 mice was particularly surprising given that transfection of Nbs1ΔC into NBS cell lines failed to complement ATM and substrate phosphorylations (25, 26). To determine whether γ IR–induced signaling was indeed ATM dependent, we pretreated B cells with a small molecule inhibitor of ATM (KU55933) (51) before IR. Treatment with the ATM inhibitor blocked the phosphorylation of ATM, SMC1, Chk1, and Chk2 in Nbs1^WT, Nbs1^657Δ5, Nbs1^H45A, and Nbs1^tr735 mice (Fig. S4), indicating that ATM mediates substrate phosphorylations in these mice.

To determine if the Nbs1 C-terminal domain is required for interaction with ATM, we performed immunoprecipitation analysis (Fig. 3, A and B). We were able to coimmunoprecipitate ATM and Nbs1 in WT and Nbs1^tr735 mutant cells, indicating that the deletion of the Nbs1 C terminus clearly did not abrogate the Nbs1–ATM interaction (Fig. 3, A and B). To determine whether ATM is still capable of being recruited to DNA damage in Nbs1^tr735 cells, we induced DSBs in primary and mutant fibroblasts using a laser microbeam. The assembly of ATM coincident with γ-H2AX–marked sites of DSBs was tracked by fluorescence microscopy. We found rapid recruitment of ATM to DSBs independent of Nbs1 status (Fig. 3 C). The fact that the Nbs1 C-terminal domain is dispensable for interaction with ATM and ATM recruitment to DSBs may explain why DSB signaling is not ablated in Nbs1^tr735 mice.

NBS cells, which express a hypomorphic allele identical to our BAC-encoded Nbs1^657Δ5 transgene, show cell-cycle checkpoint defects and sensitivity to IR. To determine which of the Nbs1 domains contribute to DNA damage-induced cell-cycle checkpoints, we treated primary B cells with γ IR (Fig. 4, A and B). Nbs1^WT, Nbs1^S278A/S343A, Nbs1^tr735, Nbs1^H45A, Nbs1^657Δ5, Nbs1^Δ/− (B cell–specific conditional null; reference 14), and Atm^−/− B cells were assayed over a dose range of 0.5–10 Gy 1 h after IR. Although cells from Nbs1^Δ/− mice showed a reduction in the mitotic index similar in magnitude to Atm^−/− cells, the G2/M checkpoint arrest was normal in Nbs1^WT, Nbs1^tr735, and Nbs1^S278/343A mice. Intermediate defects were observed in cells from Nbs1^657Δ5 and Nbs1^H45A mice (Fig. 4 A). Similar results were obtained when we assayed for thymidine incorporation after IR-induced DNA damage (Fig. 4 B). Again, only Nbs1^657Δ5, Nbs1^H45A, and Nbs1^Δ/− cells showed a reduction in the rate of DNA replication. Collectively, these results indicate that cell-cycle checkpoints are regulated by the DSB-induced activation of ATM through the MRN complex and that the Nbs1–FHA domain participates in eliciting a full cell-cycle checkpoint arrest. In contrast, neither the Nbs1 C terminus nor ATM-dependent phosphorylations at S278 and S343 are necessary for efficient checkpoint responses.

One hallmark feature of Atm^−/− mice is the resistance of their immature thymocytes to IR-induced apoptosis (32). To examine the potential role of different Nbs1 domains in the apoptotic response, we subjected mice to whole-body IR (Fig. 4, C–E). Flow cytometric analysis of thymocytes from WT mice 24 h after treatment with 4 Gy confirmed that they were sensitive to IR. We observed a 93% reduction in the total number of thymocytes (Fig. 4 E), and an ablation of the CD4/CD8 double-positive (DP) thymocyte population from 80–85% in control mice to 15–30% (average 18%) remaining after IR (Fig. 4, C and D). As expected, ATM^−/− thymocytes were more resistant to apoptosis, showing a threefold increase in viability relative to the WT (Fig. 4 E) and a persistence of DP thymocytes accounting for 76% of the population (Fig. 4 D). Nbs1^tr735 mice were about threefold more resistant to apoptosis than WT mice (Fig. 4 E), and on average 51% of the irradiated thymocytes were DP (Fig. 4 D). Consistent with this, the number of apoptotic cells was reduced in Nbs1^tr735 thymi relative to the WT, as determined by apoptosis assays and the activation of caspase 3 in situ (Fig. S5), and there was a reduction in phosphorylation at residue S15 of p53 (Fig. S6). In contrast, Nbs1^H45A and Nbs1^657Δ5 thymocytes were exquisitely sensitive to IR, exhibiting an almost complete depletion in the CD4/CD8 DP thymocyte population and reduced viability of total thymocytes compared with WT littermates (Fig. 4, C–E). Thus, the N and C terminus domains of Nbs1 have opposite effects on the induction of IR-induced thymocyte apoptosis.

By using a human BAC harboring endogenous regulatory elements of the Nbs1 gene, we have been able to assess the contribution of various domains of Nbs1 in stimulating divergent ATM-mediated DSB response pathways in vivo in the complete absence of other aberrant forms of the Nbs1 protein. As such, we arrive at surprisingly different results from what has as yet been reported (Fig. 5) (25, 26). In particular, the C terminus of Nbs1 was previously shown to bind to ATM and was required for recruitment of ATM to sites of DNA damage. Thus, one would predict that deletion of the Nbs1 C terminus in vivo would result in an AT-like phenotype. On the contrary, we find that Nbs1^tr735 mice harboring a premature truncation of Nbs1, resulting in a protein lacking this reported ATM binding domain, do not show any defects in lymphocyte or germ cell development in vivo and neither checkpoint defects nor genomic instability in vitro. However, we did observe attenuated Chk2 and Chk1 phosphorylations in response to low-dose IR. This indicates either that the activity of ATM is modestly compromised or that the C terminus regulates the optimal binding between ATM and specific substrates. Nevertheless, the association between Nbs1ΔC and ATM is sufficient for recruitment of ATM to sites of DNA damage and for nearly normal ATM activation.

One possibility for the observed discrepancy could be that in earlier studies (25, 26), Nbs1ΔC protein was expressed in hypomorphic NBS cells in contrast to a Nbs1^−/− background. It would be possible that simultaneous expression of two hypomorphic alleles could exacerbate the phenotypes. To address this issue, we have generated mice doubly transgenic for the tr735 and the 657Δ5 or H45A allele, respectively (Fig. S7). Despite the persistence of a defect in Chk2 phosphorylation at low doses of IR in cells from Nbs1^tr735/H45A and Nbs1^tr735/657Δ5 double-mutant mice, the cell-cycle checkpoint defects, ovarian failure, and thymocyte developmental defects observed in Nbs1^H45A and Nbs1^657Δ5 mice were completely rescued in Nbs1^tr735/H45A and Nbs1^tr735/657Δ5 mice (Fig. S7 and not depicted). Thus, expressing the tr735 mutation in the context of the Nbs1^657Δ5 mutation is not fundamentally distinct from reconstitution of the Nbs1-null cells/mice.

Because conditional deletion of Nbs1 nearly abolishes ATM activity (7), there are likely to be other regions within Nbs1 or within the MRN complex that can compensate for the lack of the extreme C terminus. Consistent with this idea, gel filtration analysis suggests that ATM makes multiple independent direct contacts with Mre11/Rad50 and Nbs1 (6). Because the MRN complex is intact (Fig. S1), remains nuclear, and the interaction between ATM and Nbs1 is not abrogated in cells lacking the Nbs1 C terminus (Fig. 3, A and B), ATM should still be capable of binding Mre11 and Rad50 in Nbs1^tr735 mice. Another possibility is that mutant Nbs1 may be compensated for by the activity of other accessory proteins. For example, suppression of 53BP1 was found to decrease ATM activation in NBS but not WT cell lines (52). Consistent with this idea, recent studies by Lee et al. have demonstrated that although MRN(ΔC) is by itself incapable of stimulating ATM kinase activity in vitro, addition of 53BP1 and Brca1 allows for efficient ATM phosphorylation of substrates (Paull, T., personal communication). Thus, one possible explanation for the observed discrepancy between our results and those based on transformed NBS cells is that these lines have acquired secondary mutations that limit the ability of adaptor proteins to affect ATM activity.

Besides functioning upstream of ATM, Nbs1 is itself a target of ATM that becomes phosphorylated at sites of DSBs (53). Surprisingly, mice expressing a nonphosphorylatable form of Nbs1 at S278 and S343 fail to demonstrate any detectable AT-like phenotype in vivo, nor do they exhibit a defect in the intra–S phase checkpoint. In contrast, NBS cell lines reconstituted with S278A and S343A failed to complement the S phase checkpoint defect (20–23). Interestingly, in NBS lymphoblastoid cells, Nbs1^p70 is still phosphorylated in response to IR (54). Similarly, there is no impairment in Nbs1 phosphorylation in AT-like disorder lymphoblasts that carry a mutation in Mre11 but nevertheless express full-length Nbs1 (54). Despite normal Nbs1 phosphorylation, both NBS and AT-like disorder cells are impaired in the induction of the S phase checkpoint (54). Collectively, these results indicate that Nbs1 phosphorylation at S278 and S343 is neither sufficient nor necessary for inducing the S phase checkpoint. Although we cannot exclude that other phosphorylation sites in Nbs1 contribute to the intra–S phase checkpoint, we speculate that the major phenotypic overlap between AT and NBS patients is caused by a role of Nbs1 upstream rather than downstream of ATM.

Although both the Nbs1^H45A and the Nbs1^tr735 mutants show a defect in Chk2 phosphorylation, only the FHA domain is required for eliciting complete DSB induced cell- cycle checkpoints. In contrast, the C terminus is required for efficient IR-induced apoptosis. This suggests that Nbs1 may act as a molecular switch that determines alternative (cell cycle arrest vs. cell death) cellular fates in response to DSBs. This dual role may require multiple interactions between MRN and ATM and/or other mediators of the DSB response.

Bacteria containing the BAC RPCI21-255A7 (Research Genetics) were targeted using the two-step recombination method, and mice were generated as previously described (55). The targeting was verified by sequencing of the BAC. Mice were screened as previously described (7). For crosses of Nbs1^tr735 mice with either Nbs1^H45A or Nbs1^657Δ5 mice, tail screening was performed using the following primers: NBS–open reading frame (ORF; 5′-CAACATGCAAAAGAAGAGTCTC-3′), NBS-F (5′-CAACATGCATGAGAATAGTGAC-3′), and NBS-R (5′-AATCGTAATTCTACTCAGAGCTG-3′). NBS-ORF and NBS-R amplify a PCR product in Nbs1^H45A or Nbs1^657Δ5 transgenes, whereas NBS-F and NBS-R amplify a PCR product in the Nbs1^tr735 transgene.

For Western blotting, the following primary antibodies were used at the indicated dilutions: mouse anti–mouse Atm S1987p (1:500), rabbit anti–human ATM S1981p (1:400; Rockland Immunochemicals), mouse anti–human unphosphorylated ATM S1981 (1:2,500; Rockland Immunochemicals), mouse anti-ATM 5C2 (1:500; Novus), rabbit anti-Chk1 317p (1:500; Bethyl Laboratories), mouse anti-Chk2 (1:1,500; Upstate Biotechnology), mouse antitubulin (1:10,000; Sigma-Aldrich), and anti-p53 and anti-p53 S15p as previously described (5). Anti–mouse Atm 1987p monoclonal antibodies were raised by immunization with the synthetic peptide SPTFEEGSpQGTTISS (Becton Dickinson). Phosphospecific antibodies against S343-pNbs1 and S278-pNbs were a gift from E. Lee (University of Texas Health Sciences Center, San Antonio, TX). For immunofluorescence, MEFs were irradiated with a 364-nm UV laser as previously described (56), fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X-100 for 5 min, and stained with rabbit anti–γ-H2AX (1:1,000; a gift from W. Bonner, National Cancer Institute, Bethesda, MD) and mouse anti-ATM 2C1 (1:200; Santa Cruz Biotechnology, Inc.) antibodies. For foci formation, MEFs were fixed in methanol 24 h after IR and stained with mouse anti–γ-H2AX (1:1,000; Upstate Biotechnology) and rabbit anti-Mre11 (raised against glutathione S-transferase [GST]–Mre11 fusion protein; Spring Valley Laboratories).

30 min after 10 Gy of IR, B cells were collected, washed once in PBS, and then lysed in TNG buffer low (50 mM Tris-HCl [pH 7.5], 200 mM NaCl, 50 mM β-glyceroposphate, 0.5% Tween 20, 0.1% NP-40) containing protease mix and PMSF. Cell lysis was done on ice for 30 min, followed by centrifugation at 13,000 rpm for 10 min. Protein concentrations of the supernatant were determined by a BCA assay kit (Pierce Chemical Co.). 1 mg of protein was precleared with protein A/G beads (Ultralink; Pierce Chemical Co.) at 4°C for 1 h and incubated with either 3 μg S1981 (Rockland Immunochemicals) or Nbs1 (Novus) antibody overnight at 4°C. 20 of Ultralink protein A/G beads were added for 1 h at 4°C to recover antibody-bound complexes. The beads were pelleted, as per the manufacturer's instructions, and washed twice with TNG buffer low and twice with PBS. The washed pellet was resuspended with 1× Lamelli buffer (10% Glycerol, 3% SDS, and 0.5× Upper Tris [0.5 M Tris-HCl, 0.4% SDS] containing no dye or β-mercaptoethanol) and boiled for 10 min with agitation. The supernatant was recovered, and half was mixed with NuPAGE sample buffer (Invitrogen) and dithiothreitol and run on a NuPAG 4–12% gradient gel, transferred, and blotted for ATM (5C2; Novus). The remaining fraction was run on a 6% Tris-glycine SDS gel, transferred, and blotted with anti–human Nbs1 (Novus) or anti–mouse Nbs1, as previously described (50). To determine the MRN complex assembly, immunoprecipitations were performed as described above, but protein lysis was performed in RIPA buffer using 10 μl anti-Mre11 serum generated by immunizing rabbits with a GST-Mre11 fusion protein (Spring Valley Laboratories).

B cells were purified by CD43 depletion of splenocytes and cultured for 2 d in the presence of LPS and IL-4. For intra–S phase and G2/M checkpoints, B cells were challenged with different doses of IR and measured the thymidine uptake and mitotic index, as previously described (7). Primary MEFs were harvested 24 h after IR, and metaphase spreads were prepared and analyzed by DAPI staining.

Postnatal day 5 littermates were transcardially perfused with 4% paraformaldehyde in PBS 4 h after IR with 4 Gy from a cesium irradiator (at a rate of 391 rad/min). Tissue was removed and cryoprotected in 25% PBS-buffered sucrose. 10-μm-thick sections were cut on a cryostat (HM500M; Microm) and used for histology and immunohistochemistry. Hematoxylin and eosin staining was performed according to standard procedures. For active caspase 3 (1:500; BD Biosciences) immunohistochemistry, sections were treated with citrate buffer for antigen retrieval after incubation with cleaved caspase 3 antibody overnight after quenching endogenous peroxidase using 0.6% hydrogen peroxide, and immunoreactivity was visualized with a substrate kit (VIP; Vector Laboratories), according to the manufacturer's direction. Sections were counterstained with 0.1% methyl green, dehydrated, and mounted in DPX (Fluka). Apoptosis was assessed by single-stranded DNA (ssDNA) immunoreactivity and TdT-mediated dUTP-biotin nick-end labeling (TUNEL). Cryosections were incubated with ssDNA antibody (1:300; IBL Co., Ltd.) overnight and visualized with indocarbocyanine (Cy3; Jackson ImmunoResearch Laboratories) mounted with Vectashield containing DAPI (Vector Laboratories). TUNEL was performed using ApopTag (Chemicon) according to the manufacturer's protocol. Images were captured with a digital camera (SPOT; Diagnostic Instruments) and processed using Photoshop CS software (Adobe). Ovaries were fixed in 10% formalin, and paraffin sections were stained with hematoxylin and eosin.

In Fig. S1, we examined the MRN interaction in cells from WT and mutant Nbs1 mice by immunoprecipitation, and Fig. S2 shows the Mre11 and γ-H2AX foci formation in Nbs1^WT, Nbs1^657Δ5, Nbs1^H45A, Nbs1^tr735, and Nbs1^S278/343A MEFs 24 h after 0 or 10 Gy of IR. Fig. S3 shows chromosomal aberrations in MEFs from Nbs1^WT, Nbs1^H45A, Nbs1^657Δ5, and Nbs1^tr735 lines before and after DNA damage induction. Fig. S4 depicts ATM substrate phosphorylation in WT and Nbs1 mutants in the presence or absence of an ATM inhibitor. Fig. S5 shows regional resistance in the cortex of the Nbs1^tr735 mutant thymus to IR-induced apoptosis, as measured by apoptosis assays. Fig. S5 shows Western blots of p53 stabilization and phosphorylation in thymocytes from Nbs1^WT, Nbs1^tr735, and Atm^−/− mice after whole-body IR and their quantitative analysis. Fig. S7 shows that the Nbs1^tr735 allele rescues cell-cycle checkpoint, oocyte, and thymocyte developmental defects but not inefficient Chk2 phosphorylation of Nbs1^657Δ5 mice.

Thanks to Eva Lee for providing antibodies; David Winkler, Hua-Tang Chen, Steve Jay, and Nada Radoja for technical assistance; and Tanya Paull for sharing unpublished results.

This work was supported by the Intramural Research Program of the National Institutes of Health, the National Cancer Institute, the Center for Cancer Research, the National Institute of Aging, and a grant from the A-T Children's Project.

The authors have no conflicting financial interests.