Ataxia telangiectasia and Rad3-related (ATR) is a phosphoinositol-3-kinase like kinase (PIKK) that initiates a signal transduction response to replication fork stalling. Defects in ATR signalling have been reported in several disorders characterized by microcephaly and growth delay. Here, we gain insight into factors influencing the ATR signalling pathway and consider how they can be exploited for diagnostic purposes. Activation of ATR at stalled replication forks leads to intra-S and G2/M phase checkpoint arrest. ATR also phosphorylates γ-H2AX at single-stranded (ss) DNA regions generated during nucleotide excision repair (NER) in non-replicating cells, but the critical analysis of any functional consequence has not been reported. Here, we show that UV irradiation of G2 phase cells causes ATR-dependent but replication-independent G2/M checkpoint arrest. This process requires the Nbs1 N-terminus encompassing the FHA and BRCT domains but not the Nbs1 C-terminus in contrast to ATM-dependent activation of G2/M arrest in response to ionizing radiation. Thus, Nbs1 has a function in ATR signalling in a manner distinct to any role at stalled replication forks. Replication-independent ATR signalling also requires the mediator proteins, 53BP1 and MDC1, providing direct evidence for their role in ATR signalling, but not H2AX. Finally, the process is activated in Cockayne's syndrome but not Xeroderma pigmentosum group A cells providing evidence that ssDNA regions generated during NER are the ATR-pathway-specific activating lesion. Replication-independent G2/M checkpoint arrest represents a suitable assay to specifically identify patients with defective ATR signalling, including Seckel syndrome, Nijmegen breakage syndrome and MCPH-1-dependent primary microcephaly.
The signalling response to DNA damage is regulated by one of two phosphoinositol-3-kinase like kinases (PIKKs); ataxia telangiectasia mutated (ATM)-dependent signalling is activated in response to DNA double-strand breaks (DSBs), whereas ataxia telengiectasia and Rad3-related (ATR) signalling responds to single-stranded (ss) regions of DNA generated at stalled replication forks (1,2). ATM signalling, which has been better studied than ATR signalling, requires the Mre11/Rad50/Nbs1 (MRN) complex as a sensor protein to activate ATM (3). An early step in the signalling pathway is phosphorylation of H2AX, generating γ-H2AX, which facilitates the retention of damage response mediator proteins including MDC1, 53BP1 and Brca1 at the sites of damage, which collectively serve to amplify the ATM signal (4). ATM phosphorylates a range of substrates which regulate ATM-dependent cell cycle checkpoint arrest (1).
The available data suggest that ATR is activated by replication protein A (RPA) binding to ssDNA at stalled replication forks (2) and promotes replication fork restart (5). Although it has been suggested that ATR is only activated in S phase following fork stalling, ATR-dependent γ-H2AX can form following UV irradiation of non-replicating cells at ssDNA generated during nucleotide excision repair (NER) (6–9). Phosphorylation of checkpoint proteins has also been reported following NER although the activating PIKK was not established (10). Since ssDNA tracts generated during NER are rapidly filled in and since ATR-dependent γ-H2AX in G1 cells occurs as a diffuse signal compared with the large, defined foci induced by ionizing radiation (IR), it is unclear whether replication-independent ATR activation leads to checkpoint arrest (11,12). In short, the critical analysis of whether UV can activate ATR-dependent replication- independent checkpoint arrest has not been examined and, indeed, is questioned (6).
Nbs1, the protein defective in Nijmegen breakage syndrome (NBS), is a component of the MRN complex, which functions as a DNA DSB sensor protein during ATM signalling (3). Nbs1 is also required for ATR signalling following replication fork stress (12–14). Since Nbs1 prevents DSB formation at collapsed replication forks, it is possible that its role in ATR signalling is indirect reflecting a function in processing stalled forks rather than a direct role in ATR activation or signal amplification (15,16). Interestingly, NBS patients display growth delay and microcephaly, features which are characteristic of patients defective in ATR signalling rather than the typical features observed in ataxia telangiectasia (A-T) patients (17). Given these surprising clinical features, it is important to define the roles of Nbs1 in the ATR versus ATM signalling response.
Following exposure to IR, ATM activates G2/M checkpoint arrest within 1 h due to DSBs generated in G2 phase cells (18). At later times post-IR, G2/M arrest is ATR dependent due to ATR activation in irradiated S phase cells (18,19). ATR-dependent G2/M arrest is normally examined 5–24 h post-treatment with UV or agents that stall replication forks (e.g. hydroxyurea) to allow S phase cells to progress to the G2/M checkpoint (see 7,12). We reasoned that if ATR is significantly activated by ssDNA generated during NER in G2, G2/M arrest should occur within 1–2 h post-treatment with low UV doses, which do not directly induce DSBs. Here, we examine G2/M arrest at 90 min post-UV (5 Jm−2), a time which precludes any contribution from S phase cells. Exclusion of S phase cells was further supported by addition of aphidicolin (APH), to prevent their progression into G2. While ATR clearly activates checkpoint arrest following replication, our findings here provide the first demonstration that ATR can regulate a replication-independent checkpoint in G2. We show that Nbs1 is required for replication-independent ATR checkpoint activation, demonstrating a direct role for Nbs1 in ATR signalling that cannot be attributed to any role in replication fork processing, and further we delineate regions of Nbs1 required for ATR versus ATM signalling. Additionally, we show that 53BP1 and MDC1 are required for replication-independent ATR checkpoint activation suggesting that these two mediator proteins amplify ATR signalling in a similar manner to their role in ATM signalling (20). Finally, we show that whereas cells which fail to generate ssDNA at the site of UV-induced lesions due to an inability to carry out the incision step of NER (XP-A cells) are defective in activating G2/M checkpoint arrest after UV irradiation, Cockayne's syndrome cells, which are defective solely in the repair of UV-induced dimers at transcriptionally active sites (transcription coupled repair, TCR), are able to activate the arrest. Since XP-A cells show high predisposition to sun-induced cancer in contrast to Cockayne's syndrome, the ability to effect G2/M checkpoint arrest in response to UV irradiation provides a provocative distinction between these defects which correlates with their tumour predisposition. ATR signalling defects have been observed in several disorders, which share common characteristics of microcephaly and developmental delay (7,12,17,21–24). Seckel syndrome (SS) is particularly important in this context and, indeed, a subset of patients (designated ATR-SS) have mutations in ATR itself (7). Moreover, although SS is genetically and clinically heterogeneous, the majority of cell lines derived from SS patients display defective ATR signalling (21,25). The replication- independent, ATR-dependent G2/M checkpoint assay represents an assay that has utility for diagnostic purposes to identify a subset of SS patients.
ATR-dependent, replication-independent checkpoint activation
To examine whether replication-independent ATR activation leads to checkpoint arrest, we examined whether G2/M arrest is activated 90 min post-UV. We utilized lymphoblastoid cell lines (LBLs) derived from a normal individual (wild type, WT), an individual with A-T and an ATR-SS patient with a hypomorphic mutation in ATR (ATR-SS). Asynchronously growing LBLs were UV irradiated (5 Jm−2) in the presence of APH, an inhibitor of the replicative polymerase, to prevent S phase cells progressing into G2 during analysis and the mitotic index (MI) evaluated. While S phase cells show strong activation of γ-H2AX after APH treatment (data not shown), APH-treated G2 cells, identified using a G2-specific staining protocol (CENP-F staining), showed a minimal increase in γ-H2AX (Supplementary Material, Fig. S1A). A comparable response for phosphorylation of Chk1 was also observed (Supplementary Material, Fig. S1B). Additionally, UV treatment of G2 phase cells caused γ-H2AX formation to a similar extent in control or APH-treated cells (Supplementary Material, Fig. S1A). Thus, APH alone does not activate ATR signalling in G2 phase and does not impair the response of G2 phase cells to UV irradiation. The MI markedly decreased in UV-treated control and A-T LBLs while UV treatment did not perturb the MI in ATR-LBLs, demonstrating ATR-dependent checkpoint arrest in irradiated G2 cells (Fig. 1A). Similar results were obtained following UV irradiation in the absence of APH (Supplementary Material, Fig. S2). To verify that cells entering mitosis had not originated from S phase cells, we added BrdU to cells treated with APH to mark S phase cells. Ninety minutes later, we examined control and ATR-SS cells using α-BrdU and α-CENP-F antibodies. We observed cells strongly staining for either BrdU or CENP-F, but no cells jointly staining for both markers in the presence of APH (Supplementary Material, Fig. S3). Even in the absence of APH, only ∼20% of CENP-F-positive cells were BrdU positive at 2 h. Thus, few (<20%) S phase cells progress to G2 phase cells within the 90 min allowed for activation of G2 arrest. Additionally, the concentration of APH utilized (1.5 µm) was expected to completely abolish replication. To verify this, we examined entry into mitosis 24 h post-APH treatment. We failed to observe any mitotic cells in either control LCLs or ATR-SS (Supplementary Material, Fig. S4). Thus, we conclude that the APH treatment utilized prevents the replication even of ATR-SS cells.
Nbs1, 53BP1 and MDC1 are required for replication-independent G2/M arrest
Previous studies have shown that Nbs1 is required for ATR-dependent G2/M checkpoint arrest at later times (5–24 h) post-UV. Since this is sufficient time for S phase cells to progress to G2, this most likely monitors the role of Nbs1 in replication-dependent ATR signalling (12,13). To examine the requirement for Nbs1 in replication-independent G2/M arrest, we examined an SV40 transformed NBS fibroblast (nbs) derived from a patient expressing the common 675del5 mutation and a derivative corrected with full-length Nbs1 (nbs+p95). Ninety minutes post-UV treatment (3 Jm−2) in the presence of APH, nbs+p95 cells showed a marked decrease in MI in contrast to NBS cells (Fig. 1B). The absence of APH did not affect the result (Supplementary Material, Fig. S5A). Thus, we conclude that Nbs1 is required for ATR signalling in G2 phase cells, a role that cannot be attributed to any function of Nbs1 in replication fork stability.
Since Nbs1 has been described as a mediator protein for ATM signalling as well as having an upstream role in ATM activation, we examined whether MDC1 and 53BP1 might also be required for ATR-dependent replication-independent G2/M arrest. WT, 53BP1−/− and MDC1−/− knockout mouse embryo fibroblasts (MEFs) were UV irradiated in the presence of APH and the MI estimated 90 min later. While WT MEFs showed a marked decrease in MI following exposure to 2.5, 5 or 10 Jm−2, the MI of MDC1−/− or 53BP1−/− MEFs exposed to 2.5 or 5 Jm−2 was only mildly decreased compared with untreated cells (Fig. 1C). Following exposure to 10 Jm−2, the MI was significantly decreased although to a lesser extent than observed in WT cells.
We also examined UV-induced G2/M checkpoint arrest in H2AX−/− MEFs and, surprisingly, found a normal response (Fig. 1C). They displayed the anticipated defect in ATM-dependent G2/M arrest after IR, namely a defect specifically at low doses of IR (Supplementary Material, Fig. S6A); but also showed a normal response after UV in replication-dependant G2/M arrest (Supplementary Material, Fig. S6A). Since the retention of 53BP1 at DSBs requires H2AX, we examined 53BP1 retention at ssDNA regions generated by NER (i.e. after UV irradiation). Since signalling after UV irradiation is more pan nuclear compared with the defined foci observed after IR, it was difficult to distinguish specific retention at damage sites from the 53BP1 signal observed in untreated cells. We, therefore, exploited filters to achieve localized spots of UV irradiation. Using filters with a small pore size, we observed small regions of enhanced 53BP1 accumulation which co-localized with γ-H2AX formation in control cells and with larger pore size filters observed larger regions of 53BP1 accumulation (Supplementary Material, Fig. S6B). The background staining of 53BP1 outside of these regions appeared to decrease compared with untreated cells, most likely due to the relocalization of 53BP1 to the UV-irradiated sites. Importantly, 53BP1 relocalization but not γ-H2AX formation occurred in H2AX−/− cells. As expected, 53BP1 foci formation after IR was defective in H2AX−/− cells. These data strongly suggest that 53BP1 retention at the site of DSBs has different factor requirements to that occurring at ssDNA regions, providing an explanation for the lack of requirement for H2AX for replication-independent UV-induced G2/M checkpoint arrest.
The N- but not the C-terminus of Nbs1 is required for ATR-dependent G2/M arrest
Previously, we showed that the C-terminus of Nbs1 (Nbs652) was dispensable for ATR-dependent phosphorylation events following replication fork stalling providing evidence that the function of Nbs1 in ATR activation at stalled replication forks is distinct to its role in ATM signalling (26). Here, we extend the domain analysis of Nbs1 using a range of Nbs1 constructs expressed in transformed NBS fibroblasts to examine its role in replication-independent G2/M arrest. We utilized Nbs cells expressing full-length Nbs1, Nbs1 mutated at two identified PIKK phosphorylation sites (nbs+PP), Nbs1 lacking the C- or N-terminus (nbs+652, nbs+FR5, respectively) and an Nbs1 construct encompassing the Mre11 interaction motif but lacking the extreme C-terminus ATM interaction motif (nbs+dATM) (Fig. 2A and B). We examined G2/M arrest at 90 min post-UV (3 Jm−2) in the presence of APH and observed efficient arrest in cells expressing Nbs1 constructs lacking the Nbs1 C-terminus (nbs+652 and nbs+dATM) or nbs+PP while cells expressing nbs+FR5 failed to activate G2/M arrest (Fig. 2C). Results with Nbs deficient cells (nbs) and cells complemented with full-length Nbs1 (nbs+p95) which were examined in the same experiments are shown in Fig 1B. Thus, the N- but not the C-terminus of Nbs1 is required for replication-independent UV-induced arrest.
The ability to examine ATR-dependent G2/M arrest in cells in G2 at the time of UV irradiation allowed a direct comparison with ATM-dependent G2/M arrest following γ-irradiation. Following exposure to varying doses of IR, the MI was monitored at 90 min in NBS cells and its derivatives (Fig. 2D). For direct comparison, the experiments were carried out in the presence of APH but similar findings were obtained in the absence of APH (Supplementary Material, Figs S5 and S7). The results were strikingly distinct to those following UV irradiation (Figs 1B and 2C). NBS cells show impaired G2/M checkpoint arrest following exposure to low doses of IR but a normal response following irradiation with higher doses (3 Gy) (27, and references therein). Consistent with previous findings, NBS cells expressing nbs+652 or nbs+dATM remain impaired for IR-induced G2/M arrest demonstrating the role for the Nbs1 C-terminus in ATM signalling (28). In contrast, cells expressing the C-terminal Nbs1 fragment were largely proficient for ATM signalling albeit slightly impaired compared with control cells. This important finding verifies that the FR5 fragment of Nbs1 is correctly folded and functionally active. It is noteworthy, that NBS cells (nbs) show nearly normal G2/M checkpoint arrest following exposure to 3 Gy IR, which is consistent with previous findings that NBS cells display a ‘leaky’ phenotype (27).
Our finding that FR5 is functionally active encouraged us to additionally examine the domain requirements for Nbs1 in replication-dependent ATR signalling. We examined G2/M checkpoint arrest 5 h following UV treatment, likely reflecting G2/M checkpoint arrest following replication fork stalling (Fig. 3A). We also examined Chk1 phosphorylation following HU or UV irradiation (Fig. 3B and C) and utilized an immunofluorescence (IF)-based procedure to monitor the ability of stalled replication forks to restart DNA synthesis, another event that has been shown to be ATR-dependent (Fig. 3D) (12). We observed a similar pattern in all assays, namely efficient ATR-dependent responses in cells expressing constructs that lack the Nbs1 C-terminus (nbs+652; nbs+dATM) but an impaired response in cells lacking the Nbs1 N-terminus (nbs+FR5).
XP-A but not CS-A or B cells are defective in replication-independent G2/M arrest
The activation of ATR in G0/G1 has previously been shown to require XPA, consistent with the generation of ssDNA during NER (7–10). We observed that replication-independent G2/M arrest after UV is abolished in a transformed XP-A fibroblast cell line (Fig. 4) suggesting that the process requires XP-A-dependent incision. XP-A LBLs also failed to activate G2/M arrest after UV (data not shown). This is consistent with our previous demonstration that plateau phase XP-A cells fail to activate γ-H2AX phosphorylation following UV treatment (7). In contrast, transformed CS-A and CS-B fibroblasts, which are proficient in global genome repair but specifically defective in TCR, were proficient in UV-induced G2/M arrest in the presence (Fig. 4) or absence of APH (data not shown).
Although ATR can be activated at ssDNA generated during NER, the critical examination of whether this leads to checkpoint arrest and hence is of functional significance has not been undertaken (7–9). By assessing G2/M checkpoint arrest 90 min post-UV treatment with a low dose (5 Jm−2) that does not directly induce DSBs, coupled with the addition of APH to prevent S phase cells progressing to the G2/M checkpoint, we demonstrate that ATR activation in G2 cells following exposure to modest doses of UV can lead to checkpoint arrest, demonstrating a functional impact of ATR activation in non-S phase cells. ATM is similarly activated following exposure to IR. Indeed, while IR-induced arrest at early times (1–5 h) is ATM-dependent, at later times it is ATR dependent due to the progression of irradiated S phase cells into G2 (18,19). Recently, we presented evidence that ATM-dependent G2/M checkpoint arrest significantly contributes to the maintenance of genomic stability by preventing chromosome breakage (29). Our findings here raise the possibility that UV-induced ATR-dependent G2/M arrest may be similarly important. Although ssDNA gaps or UV photoproducts may not directly induce chromosome breakage in mitosis, genomic instability may subsequently arise in the following S phase. Thus, while NER may be important to enhance survival post-UV, checkpoint arrest may serve an important function in limiting genomic instability following UV exposure. Here, we show that XP-A but not CS transformed cells show impaired G2/M arrest, which is consistent with a previous study showing UV-induced phosphorylation of checkpoint proteins in G1/G2, although checkpoint arrest and the responsible PIKK was not examined (10). Additionally, the use of a high dose in this study (20 Jm−2) allowed the possible indirect induction of DSBs due to overlapping repair tracts. Provocatively, XP-A but not CS, is associated with cancer predisposition (30). Although ATR has also been reported to phosphorylate H2AX in G0/G1 cells, there is controversy as to whether Chk1 is also activated and G1/S arrest has not been examined (7–10). ATR-dependent G1/S arrest might be as, or more important, than G2/M arrest in the maintenance of genomic stability. Conversely, if G1/S arrest is not activated, a larger burden is placed on the G2/M checkpoint to prevent progression of cells with ssDNA regions.
Our discovery of ATR-dependent G2/M arrest at 90 min post-UV treatment allowed a direct comparison of factors required for ATR versus ATM signalling. While ATM signalling requires the ATM- and Mre11-interaction sites in the Nbs1 C-terminus, both are dispensable for ATR signalling (28). In contrast, we show here that the N-terminal region of Nbs1 which contains the FHA and BRCT domains of Nbs1, which is largely dispensable for ATM signalling, is required for ATR signalling (31). This is a striking result that distinguishes domain function requirements of Nbs1 in its two roles. The fact that NBS cells expressing the FHA/BRCT domains show substantial ATM-dependent G2/M arrest provides evidence that the fragment is folded correctly and capable of significant function, enhancing the significance of the defect in ATR signalling. These findings strongly suggest that the FHA/BRCT domains themselves may be required for ATR signalling but further domain analysis would be required to verify this. Unfortunately, we have observed that point mutational changes in either domain exert dominant negative effects precluding further analysis. We also show that the two mediator proteins, 53BP1 and MDC1 are required for replication-independent ATR signalling, representing the first evidence that 53BP1 functions in ATR signalling. The mediator proteins have been reported to function in ATM signal amplification being required for checkpoint arrest after low but not higher IR doses (20,32). It is difficult to assess the dose dependency of ATR signalling because ATR-SS cells show some checkpoint arrest after high UV doses, potentially attributable to their hypomorphic mutation or to DSB induction after high doses. Nonetheless, the phenotypic similarity between cells deficient for Nbs1, 53BP1 or MDC1 raises the possibility that Nbs1 functions as a mediator protein in ATR signalling and the presence of a BRCT domain, a feature common to mediator proteins, adds weight to this suggestion.
In contrast to our finding with 53BP1−/− and MDC1−/− cells, H2AX appeared to be dispensable for replication-independent ATR signalling. This was surprising given the fact that after IR, H2AX is required for the stable retention of 53BP1 at the DSB site. Following UV treatment, we showed that 53BP1 is recruited to the ss regions of DNA generated independently of H2AX. Interestingly, 53BP1 has previously been reported to bind ssDNA regions (33). Collectively, these findings suggest that 53BP1 functions as a mediator protein in this response, similarly to its role following IR but its mode of recruitment appears to differ with HAX being dispensable for its retention at ssDNA regions.
Finally, defects in downstream steps of ATR signalling have been observed in a number of disorders including SS, MCPH-defective primary microcephaly, NBS, Miller-Dieker lissencephaly syndrome and Williams–Beuren syndrome (7,12,21,23,34–36). Previously, checkpoint arrest has been examined 5–24 h post-UV treatment. Entry into mitosis at this prolonged time post-UV irradiation likely reflects a failed ability to arrest at both the S phase and G2 phase checkpoints. However, this assay can be difficult to interpret in cell lines with slow rates of cell cycle progression. The assay characterized here to monitor replication-independent G2/M arrest is a more reliable assay that monitors solely a G2/M checkpoint arrest. Interestingly, we have shown that it is defective in most of the SS cell lines we have examined from clinically diagnosed SS patients (data not shown) making it a convenient assay to identify this class of SS patients.
MATERIALS AND METHODS
GM02188, GM3189D, DK0064 and LB708 were control, A-T, ATR-SS and XPA LBLs, respectively. NBS-ILB1 is an SV40-transformed NBS fibroblast line from Dr M. Zdzienicka. Nbs+p95, nbs+FR5, nbs+dATM, nbs+PP and nbs-ILB1+Nbs652 are derivatives retrovirally complemented with full-length or modified human Nbs1 (Fig. 2A).
Wild type, MDC1−/− and 53BP1−/− MEFs were kindly supplied by Dr J. Chen. 1BRneo (control), XP12RO-SV40 (XP-A), CS1AN-G2 (CS-A) and CS3BE-G2 (CS-B) are transformed human fibroblasts.
Treatment with DNA damaging agents and ATM inhibitor
Irradiation was carried out using a UVC source (0.6 Jm−2/s). Hydroxyurea was purchased from Sigma-Aldrich (Poole, UK). Localized UV irradiation was delivered through Isopore TMTP membrane filters (Millipore, MA, USA) of 3 or 8 µm pore size.
IF and immunoblotting
α-Chk1Ser317antibodies were purchased from Cell Signaling Technology (Beverley, MA, USA). α-H2AXSer139 and α-Histone H3Ser10 antibodies were from Upstate Technology (Buckingham, UK), α-CENP-F (H-260) antibodies were from Santa Cruz Biotechnology (CA, USA). α-rabbit and α-mouse secondary antibodies were from Dako (Glostrup, Denmark). Procedures were as previously described (12).
G2/m checkpoint arrest
Cells were exposed to the indicated dose of UV or IR and incubated for 1.5 or 5 h in the presence of 1 µm APH or 1.5 µm Nocodazole when indicated, followed by processing for IF. Mitotic cells were detected by α-Histone H3Ser10 antibodies. APH was purchased from Sigma-Aldrich. All results represent the mean and SD of at least three experiments.
Replication fork stability assay
The procedure followed was as previously described with modifications (12,37,38). Cells were labelled with CldU (50 µm) for 20 min, pelleted, washed with PBS and resuspended in complete medium with 10 µm APH and incubated for 2 h. Cells were then pre-incubated with IdU (50 µm) for 20 min in the presence of 10 µm APH. Cells were then pelleted, swollen in 75 mm KCl for 10 min to allow fork re-initiation and IdU incorporation. IF was performed as described except that after permeabilization cells were incubated with 2 M HCl for 30 min to denature the DNA. Cells were blocked for 1 h with 10% FCS in PBS, and then incubated with both primary antibodies overnight at 4°C. The first label was detected with red secondary antibodies, the second with green secondary antibodies.
We acknowledge support from the Medical Research Council, the Department of Health, the Association for International Cancer Research, the Leukaemia Research Fund and the EU [Integrated Projects no. LSHG-CT-2005-512113 (DNA repair) and FI6R-CT-2003-508842 (RiscRad)]. M.O'D. is a Cancer Research UK Senior Fellow, whose lab is supported by the CRUK and Medical Research Council.
Conflict of Interest statement. None declared.