STAT-1 facilitates the ATM activated checkpoint pathway following DNA damage

To ensure that cells pass on accurate copies of their genomes on to the next generation, a series of surveillance pathways – the so called cell-cycle-checkpoint protein kinases – are activated following DNA damage to allow appropriate time for DNA repair to take place. The ATM kinase in conjunction with adaptor proteins (transducers/mediators) plays a pivotal role in initiating the checkpoint-cascade pathway following DNA damage by phosphorylating and activating a subset of ATM substrates including Chk2^-T68, p53^-S15, NBS1^-S343 and BRCA1^-S1387 (Melochionna et al., 2000; Chehab et al., 2000; Lim et al., 2000; Abraham, 2001; Shiloh, 2003). In mammalian cells, proteins 53BP1 and MDC1 (mediator of DNA damage checkpoint 1) that contain a BRCA1 C-terminal (BRCT), have been termed adaptors/mediators, because they play a central role in regulating ATM activation and ATM-mediated pathways (Shultz et al., 2000; DiTullio et al., 2002; Fernandez-Capetillo et al., 2002; Wang et al., 2002; Abraham, 2002; Goldberg et al., 2003; Lou et al., 2003; Stewart et al., 2003; Xu and Stern, 2003; Peng and Chen, 2003).

Following DNA damage, Chk2 kinase is also effective in phosphorylating its own subset of substrates on alternative sites to those phosphorylated by ATM, including p53^-S20, BRCA1^-S988, Cdc25A^-S123, Cdc25C^-S216, PML and E2F-1 (Chehab et al., 2000; Hirao et al., 2000; Lim et al., 2000; Bartek and Lukas, 2003; Yang et al., 2002; Stevens et al., 2003). Activation of Chk2-Cdc25A has been implicated in both G1-S phase and G2-M transition checkpoint-control (Abraham, 2001; Falck et al., 2001; Falck et al., 2002; Bartek and Lukas, 2003; Shiloh, 2003), whereas Chk2-p53, Chk2-PML and E2F-1 activated pathways have been reported to play a role in the apoptotic pathway (Hirao et al., 2000; Yang et al., 2002; Takai et al., 2003; Stevens et al., 2003).

In addition to BRCA1, ATM also phosphorylates NBS1 on serine 343 (Lim et al., 2000). NBS1 is a component of the multifunctional MRE11-RAD50-NBS1 (MRN) complex that is involved in the repair of DNA double-strand breaks (DSBs) and is also required for Chk2 phosphorylation (D'Amours and Jackson, 2001; Buscemi et al., 2001). Following DNA damage, the histone variant H2AX is one of the earliest proteins to be phosphorylated and forms positive nuclear foci at sites of DSBs (Fernandez et al., 2002; Coleste et al., 2003). This is followed by recruitment of 53BP1 and MDC1 that colocalise with phosphorylated γH2AX at DNA DSBs (Fernandez et al., 2002; Wang et al., 2002; Stewart et al., 2003). Finally, the recruitment of the MRN complex facilitates the binding of DNA repair factors (D'Armours and Jackson, 2001). Recent studies have reported that the response to DNA damage leads to distinct, branched pathways that are activated via the phosphorylation of specific ATM downstream targets and shows a regulatory hierarchy that converges to control-processes such as DNA repair, cell cycle or apoptosis (Wang et al., 2002; Foray et al., 2003).

The signal transducer and activator of transcription 1 (STAT-1) protein is essential for signalling of interferons (IFNs) (Broomberg and Darnell, 2000; Ihle, 2001), which, in addition to their role in innate immunity, serve as potent inhibitors of cell growth and promoters of apoptosis. Although STAT-1-deficient mice develop no spontaneous tumours, they are highly susceptible to chemical, carcinogen-induced tumourigenesis (Durbin et al., 1996; Kaplan et al., 1998). Crossing the STAT-1 knockout into a p53-deficient background yields animals that develop tumours more rapidly, and with a broader spectrum of tumour types than is seen with p53 single-mutants (Kaplan et al., 1998), suggesting that STAT-1, like p53, may have tumour suppressor properties.

p53 plays an important role in mediating the apoptotic programme (Vousden and Lu, 2002). Recently, STAT-1, like p53, has been directly implicated in modulating apoptosis. For example, cells lacking STAT-1 are less susceptible to tumour necrosis factor α-induced cell death than cells containing STAT-1 (Kumar et al., 1997). STAT-1-deficient cells are also resistant to hypoxia-induced cell death (Janjua et al., 2002) and STAT-1 promotes apoptosis in cardiac myocytes exposed to ischaemia/reperfusion injury (Stephanou et al., 2000; Stephanou et al., 2001; Stephanou et al., 2002).

Our recent work has shown that STAT-1 can interact with p53, modulate its activity by enhancing p53-responsive genes and can induce apoptosis (Townsend et al., 2004). Moreover, levels of p53 are reduced in cells lacking STAT-1, and STAT-1 is a negative regulator of the p53 inhibitor Mdm2 (Townsend et al., 2004). However, the mechanism of how STAT-1 can inhibit cell growth is unclear. In this study, we investigated the role of STAT-1 in the DNA-damage response pathway in both murine and human cells lacking STAT-1, and found that its absence is associated with defects in the cell-cycle checkpoint and also with a reduction in a subset of ATM-dependent, phosphorylated downstream substrates after DNA damage.

Wild-type STAT-1^+/+ and STAT-1^–/– mouse embryonic fibroblasts (Durbin et al., 1996) were kindly provided by David E. Levy (National Institutes of Health, Bethesda, MD) and maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (GIBCO). The human fibrosarcoma cell lines 2fTGH and U3A, and U3A-derived cells stably expressing STAT-1 were kindly provided by Ian Kerr (Cancer Research UK, London) (McKendry et al., 1991) and cultured in DMEM supplemented with 10% FBS. HCT-15 cells, stably expressing HA-Chk2-wt or HA-Chk2-kd, were maintained as described previously (Lou et al., 2003).

The STAT-1 RNA interference (RNAi) vector was constructed using the protocol described previously (Paddison et al., 2002). Briefly, the forward and reverse primers were, 5′-ccagaacgaa tgagggtcctc-3′ and 5′-gagggaccctcattcgttctgg-3′, respectively. The MDC1 RNAi and green fluorescent protein (GFP)-MDC1 vectors were constructed as described previously (Peng et al., 2003). The MDC1 promoter (2.0 kb) was PCR-amplified from human DNA using the following forward and reverse primers 5′-gtaccttgggtgcgctgggc-3′ and 5′-gatctgggaaggatacacatt-3′, respectively, and cloned into the pGL3-basic luciferase reporter construct (Promega, UK).

Rates of DNA synthesis after γ-irradiation (IR) of 2 or 10 Gy were measured with the two-isotope radioresistant DNA synthesis (RDS) assay. Twenty hours after plating, cells were pulsed with 5 nCi/ml [¹⁴C]-thymidine (Amersham Biosciences, UK) and incubated a further 24 hours. The medium was then removed and the cell were washed and exposed to γ-IR of 2 or 10 Gy. After a 30-minute incubation, the medium was replaced with fresh medium containing 20 μCi/ml [³H]-thymidine (Amersham Biosciences, UK) and cells were incubated a further 30 minutes. Cells were then harvested and the ³H:¹⁴C ratios were measured to assess the rates of DNA synthesis.

To assess in more detail changes in S phase, we performed the standard BrdU assay. Briefly, for BrdU dynamic cell-cycle analysis, cells were first pulsed for 30 minutes at 37°C with 10 mM BRDU (Sigma) followed by extensive washing with PBS containing 1% BSA, 10 mM Azide (PBA). Cells were then irradiated or not, followed by a second incubation for 0-48 hours. After this, cells were trypsinised, fixed in ice-cold 70% ethanol and washed with PBA. 2M HCl was added for 20 minutes, washed off and excess acid neutralised with 0.1 M Na₂B₄O₇. Cells were washed one further time in PBA before adding anti-BrdU monoclonal antibody (mAb) (BD Pharmingen) diluted in PBS with 0.5% BSA and 0.5% Tween-20 (PBT) for 45 minutes at room temperature. Cells were washed again with PBA and incubated with 10 μg/ml propidium iodide (PI) for 30 minutes to stain DNA before assessment by flow cytometry. Analysis of samples was performed with a FACScan flow cytometer (Becton Dickinson) using a 488 nm argon laser for excitation, and a 560 nm dichroic mirror and 600 nm band-pass filter (bandwidth 35 nm) for detection. Red fluorescence data was expressed on a linear scale, and green FL1 on a log scale.

To measure changes in mitosis specifically, we used the standard two-parameter flow cytometry assay to measure DNA and phosphorylated histone H3. Briefly, cells were plated, incubated, fixed and harvested in the same way as detailed above, incubated with a specific rabbit polyclonal antibody (Upstate) in PBT for 45-120 minutes, washed and incubated with a goat anti-rabbit FITC-conjugated Ab (Sigma). Cells were then washed, stained with PI and assessed by flow cytometry as before.

For survival assays, cells (2.0 ×10⁶) were exposed to 10 Gy γ-IR and incubated for 72 hours. Cells were then washed with 1 × PBS before staining with Crystal Violet (0.2% Crystal Violet 2% EtOH). Viable cells were calculated as a percentage to control cells that were not exposed to γ-IR.

Cells were exposed to γ-IR (2 or 10 Gy) or left untreated and cell extracts were prepared in lysis buffer (150 mM NaCl, 50 mM Tris base, 0.5% SDS, 1% NP-40). Samples were then boiled in SDS sample buffer for 5 minutes and separated on a 10% SDS PAGE gel. Samples were transferred to nitro-cellulose filters and subjected to western blotting. Antibodies against Chk2, and the phosphorylated forms Chk2^-T68 and p53^-S20 were purchased from Cell Signaling. Antibody against phosphorylated NBS1^-S343 was purchased from Oncogene. Antibodies against phosphorylated γ-H2AX, phosphorylated BRCA1^-1497, ATM and phosphorylated ATM-1981 were purchased from Upstate. Anti-STAT-1 and anti-Cdc25A antibodies were purchased from Santa Cruz.

For immunostaining, cells were grown on gelatin-coated coverslips and left either untreated or were treated with 50 ng/ml IFN-γ or 10 μM cisplatin for 4 hours. After fixation in –20°C methanol, coverslips were incubated 60 minutes in PBS with 3% BSA at room temperature, followed by incubation in PBS with 1% BSA containing mouse anti-Chk2^-T68 Ab(1:200) and rabbit anti-STAT-1 Ab (1:200) (Santa Cruz) for 60 minutes. After three washes in PBS, Alexa488-goat anti-mouse Ab (1:2000) (Molecular probes) and Alexa568-goat anti-rabbit Ab (1:1000) (Molecular probes) were added together in PBS with 1% BSA, with Hoechst 33258 (Sigma) for 30 minutes. After three washes in PBS, coverslips were mounted with DAKO fluorescent mounting medium. Images were collected using a Leica TCS SP2 confocal microscope, and absence of antibody cross-reaction and bleed-through of fluorophore was verified on control slides.

We previously reported that STAT-1-deficient cells showed reduced p53-mediated responses (Townsend et al., 2004). Since cells that lack p53 exhibit various defects in the cell cycle following DNA damage, we examined whether cells lacking STAT-1 also displayed similar defects. Initially, we assessed whether cells that lack STAT-1 showed any defects in S-phase checkpoint response to DNA damage. Radioresistant DNA synthesis (RDS) normally occurs in ATM- and Chk2-deficient cells exposed to DNA-damaging agents owing to a defective S-phase checkpoint. 2fTGH parental cells exposed to 5 Gy γ-IR resulted in approximately 80% inhibition of DNA synthesis (Fig. 1A). By contrast, U3A cells lacking STAT-1 showed only a 40-50% inhibition of DNA synthesis, consistent with an RDS phenotype. U3A-ST1 cells that had STAT-1 stably reintroduced, had similar γ-IR-induced inhibition of DNA synthesis to that seen for 2fTGH parental control cells. Moreover, the effects on DNA inhibition following γ-IR were dose-dependent (Fig. 1B). STAT-1 expression in 2fTGH, U3A or U3A-ST1 cells and also their responsiveness to γ-interferon is shown in the upper panel of Fig. 1A. The level of STAT-1 in 2fTGH and U3A-ST1 cells was very similar and also the induction of STAT-1 phoshorylation was very comparable in 2fTGH and U3A-ST1 cells. Also notice that, no expression or induction of STAT-1 is observed in the U3A cell line, demonstrating that the 2fTGH and U3A-ST1 cells respond to STAT-1 activation to a similar degree. These data suggest that STAT-1 has a role in the RDS checkpoint-response following DNA damage.

We next examined whether STAT-1 has a role in the G2-M checkpoint responses to DNA damage. Once again we compared the STAT-1-expressing 2fTGH and U3A-ST1 cells with the STAT-1-deficient U3A cells and measured the level of mitosis following exposure to γ-IR. As shown in Fig. 1C (and Fig. S1 in supplementary material for FACScan data), the levels of mitotic cells were much higher in U3A cells lacking STAT-1 when compared with 2fTGH and U3A-ST1 cells, which were similar following DNA damage.

To confirm whether a delayed G2-arrest was apparent in U3A cells, dynamic cell-cycle analysis was performed in a pulse-chase BrdU assay. Following a short pulse with BrdU, cells were irradiated and then incubated again for a further 24 hours. Using BrdU-specific antibodies, S-phase cells, which incorporated the BrdU during the short pulse period, were identified and tracked over time after γ-IR. It was clear that 24 hours after γ-IR, approximately twice as many BrdU-labelled U3A cells were present in the G2-M compartment, compared with the 2fTGH or U3A-ST1 cells (∼62% and ∼34%, respectively). By contrast, approximately twice as many BrdU-labelled STAT-1-expressing cells had reached G1 compared with the U3A cells (46% compared with 21%), indicating that they had bypassed G2-arrest (supplementary material, Fig. S1; Fig. 1D). These data demonstrate that STAT-1 expression facilitates a bypass of G2-arrest after γ-IR. Overall, these results demonstrate that cells lacking STAT-1 have both an enhanced RDS phenotype indicative of a defective intra-S-phase checkpoint, and an enhanced G2-M checkpoint.

Because many studies reported γ-IR hypersensitivity in cells with defects in the their ATM pathway, we also tested whether cells lacking STAT-1 are radiosensitive following exposure to γ-IR. As shown in Fig. 1E, cells lacking STAT-1 were more resistant to cell death than cells expressing STAT-1, suggesting in our case that STAT-1-deficient cells are not radiosensitive.

The ATM-Chk2-Cdc25A pathway plays an active role in both S-phase and G2-M-phase checkpoint responses to DNA damage (Abraham, 2001; Shiloh, 2003). To evaluate the molecular nature of the cell-cycle defects observed in STAT-1-deficient cells, we examined whether STAT-1 has a role in modulating the ATM-Chk2 pathway. Therefore, we examined Chk2^-T68 phosphorylation after DNA damage in cells that express or lack STAT-1. Western blot analysis demonstrated enhanced Chk2^-T68 phosphorylation in both the 2fTGH and U3A-ST1 cells after γ-IR (Fig. 2A). By contrast, Chk2^-T68 phosphorylation was completely absent in U3A cells lacking STAT-1. Levels of unphosphorylated Chk2 were similar whether cells expressed or lacked STAT-1, suggesting that, in STAT-1-deficient cells, the defect in Chk2^-T68 phosphorylation is not owing to STAT-1 regulating endogenous Chk2 levels. Similarly, reduced nuclear staining of phosphorylated Chk2^-T68 was also observed in situ by immunofluorescence analysis after γ-IR of U3A cells and 2fTGH cells (Fig. 2B).

Since activated phosphorylated Chk2^-T68 is involved in phosphorylation and degradation of Cdc25A (Falck et al., 2001, Falck et al., 2002; Bartek and Lukas, 2003), we also examined the levels of Cdc25A in 2fTGH and U3A cells exposed to γ-IR. As shown in Fig. 2A, the defect in U3A cells was associated with sustained Cdc25A levels – compared with 2fGTH and U3A-ST1 cells – after γ-IR (Fig. 2A). These studies suggest that STAT-1 may be important in regulating the ATM-Chk2-Cdc25A pathway in response to DNA damage.

An important early event in response to γ-IR-induced DNA DSBs is activation of ATM, involving autophosphorylation on serine 1981 and the conversion of inactive ATM dimers into active monomers (Bakkenist and Kastan, 2003). The Mre11-Rad50-NBS1 (MRN) complex is also recruited to DNA DSBs very early, and recent studies have shown that the MRN complex is required for activation of ATM (Carson et al., 2003; Uziel et al., 2003; Lee and Paull, 2004). Recently, NBS1 and BRCA1 have been shown to function via two independent, branched pathways that require H2AX to initiate both NBS1 and BRCA1 phosphorylation events (Fernadez-Capetillo et al., 2002). By contrast, 53BP1 is required for phosphorylation of BRCA1 but not NBS1 (Wang et al., 2002).

We therefore examined whether ATM activation and other components of the ATM pathway is modulated by STAT-1. After γ-IR, ATM phosphorylation on serine 1981 is dramatically reduced in U3A cells (without STAT-1) compared with U3A-ST1 cells (Fig. 2C). The levels of total ATM remained unchanged in U3A-ST1 and U3A cells, suggesting that the defect in ATM phosphorylation is not attributable to differences in ATM expression (Fig. 2C). Interestingly, the reduced activation of ATM after γ-IR-exposure was also associated with decreased phosphorylation of p53^-S15 and NBS1^-S343 but not H2AX or BRCA1^-1497 in STAT-1-deficient U3A cells versus U3A-ST1 cells (Fig. 2C). Similarly, immunostaining demonstrated reduced nuclear levels of the phosphorylated forms of NBS1^-S343 and ATM^-S1981 but not γ-H2AX (Fig. 2D). These results show that STAT-1 can modulate distinct ATM regulatory pathways. The DNA-damage checkpoint pathway has been suggested to be branched and shows regulatory hierarchical pathways. The complexity of this hierarchical checkpoint pathway could be because of other ATM-like members (ATR or DNA-PK) that compensate when one pathway is blocked and/or the extent of DNA damage.

The so-called DNA-damage adaptors/mediators 53BP1 and MDC1 have been reported to play a role in the initial activation of ATM as well as in phosphorylation of downstream ATM mediated pathways following DNA DSBs (Lou et al., 2003; Xu and Stern, 2003; Peng and Chen, 2003; Mochan et al., 2003). To investigate therefore, the mechanism of how STAT-1 is able to modulate the ATM–Chk2 and/or ATM–NBS1 pathways we examined whether the lack of ATM activation and ATM-mediated pathways is associated with changes in 53BP1 and MDC1 after DNA damage in STAT-1-deficient cells. Western blot analysis shows that the levels of both 53BP1 and MDC1 are reduced in cells lacking STAT-1 but were restored in U3A-ST1 cells (Fig. 3A). Similar results, showing reduced expression of 53BP1 and MDC1, were also obtained at the mRNA level in STAT-1-deficient U3A cells (Fig. 3B). Likewise, immunfluorescent staining of 53BP1 and MDC1 was also reduced in U3A cells compared with 2fTGH cells. Furthermore, 53BP1 and MDC1 expression levels were restored in U3A-ST1 cells (Fig. 3C). Because Chk2^-T68, p53 and NBS1 phosphorylation is abolished in cells lacking 53BP1 or MDC1 (Lou et al., 2003; Peng and Chen, 2003), our data suggest that STAT-1 regulates the expression of the crucial upstream mediators/adaptors that are required after DNA damage for ATM activation and for separate ATM, branched, downstream pathways.

Examination of the MDC1 promoter region using the Transfact programme (version 4.0) showed the presence of several potential DNA binding sites of STAT-1. To determine whether STAT-1 directly regulates the transcription of the MDC1 gene, we cloned a 2-kb fragment of the MDC1 promoter upstream of the transcriptional start site into the pGL3-basic luciferase reporter construct. As shown in Fig. 3D, the basal activity of the MDC1-reporter construct was much higher in STAT-1^+/+ than in STAT-1^–/– MEF cells. Co-transfection of a full-length STAT-1 construct enhanced MDC1-reporter activity in both STAT-1^+/+ and STAT-1^–/– cells. However, a STAT-1 construct without the C-terminal transactivation domain (STAT-1β), only partially enhanced MDC1-reporter activity. Similar data were also observed in STAT-1-expressing U3A-ST1 and the STAT-1-deficient U3A cells (Fig. 3E). These data agree with previous reports by us and others that the C-terminal domain of STAT-1 is required for its effects on transcription (Levy, and Darnell, 2002; Stephanou et al., 2002; Stephanou et al., 2003); but most importantly, these findings indicate that the MDC1 gene is a direct target of STAT-1.

Because STAT-1 binds to p53 and modulates its activity (Townsend et al., 2004) and because 53BP1 has been shown to interact with p53 and Chk2, we investigated whether STAT-1 also associates with Chk2, 53BP1, MDC1 or BRCA1. Under normal, non-stressed conditions STAT-1 did not interact with any of these factors in 2fTGH-cell lysates. However, following γ-IR, STAT-1 interacted with Chk2, MDC1 and 53BP1 but not with BRCA1 (Fig. 4B). An association between STAT-1 and Chk2 was also detected after γ-IR in HCT15 cells that stably expressed HA-Chk2 (Fig. S2a, supplementary material). Furthermore, a Chk2 kinase-dead (Chk2-KD) construct, containing a mutated kinase domain, was still able to interact with STAT-1 (supplementary material Fig. S2b), suggesting that the kinase activity of Chk2 is not essential for the interaction with STAT-1 following DNA damage. In addition, we found that STAT-1 signalling in HCT15 cells that stably express wild-type Chk2, is similar in HCT15 cells that stably express the Chk2 kinase-dead construct, which also suggests that the kinase activity of Chk2 does not affect the functional activity of STAT-1 (data not shown). We also performed similar STAT-1 immunoprecipitation experiments in U3A STAT-1-deficient cells and found no co-precipitation with any of the above factors (data not shown), which confirms that the binding of STAT-1 is specific to the above interacting factors in STAT-1-expressing cells. These data thus demonstrate that, STAT-1 can interact with Chk2, MDC1 and 53BP1 and this association is probably important for the regulation and function of Chk2.

To determine whether MDC1 is required for mediating ATM phosphorylation of downstream substrates in cells lacking STAT-1, we examined the effects of overexpressing MDC1 in U3A cells and measured the phosphorylation levels of p53, Chk2 and NBS1 mediated by ATM after inducing DNA damage. Enforced synthesis of MDC1 increased levels of phosphorylated ATM, and the phosphorylation of Chk2^-T68, p53^-S15 and NBS1^-S343. However, enforced synthesis of a mutated MDC1 that lacks the forkhead-homology-associated (FHA) domain, stimulated phosphorylated ATM, and Chk2^-T68 p53^-S15 and NBS1^-S343 phosphorylation only partially (Fig. 5A), indicating that the FHA domain is important for mediating distinct phosphorylation of downstream substrates by ATM. Overexpressing wild-type MDC1 had a dose-dependent effect on enhancing ATM^-S1981, Chk2^-T68 and p53^-S15 phosphorylation (Fig. 5B). By contrast, MDC1 lacking the FHA domain had a much weaker dose-dependent effect on enhancing the phosphorylation of ATM¹, Chk2 and p53 (Fig. 5C,D). To demonstrate whether the defects in ATM phosphorylation of downstream substrates in the U3A cells can also be observed in other STAT-1-deficient cells, we also performed similar MDC1 overexpression experiments in STAT1^–/– MEF cells after γ-IR. Once again, enforced expression of wild-type MDC1 but not of the mutant MDC1 gene lacking the FHA domain enhanced p53^-S15 phosphorylation (Fig. 5E).

To further investigate the effects of MDC1 in enhancing the activity of ATM, we inhibited the expression of MDC1 with RNAi. As shown in Fig. 5F, cells overexpressing the MDC1-RNAi construct showed a significant reduction in the expression of MDC1. In 2fTHG cells that expressed STAT-1, transfection of the MDC1-RNAi construct reduced the phosphorylation of Chk2^-T68, p53^-S15 and NBS1^-S343 following DNA damage (Fig. 5G). These results further support the notion that STAT-1 and its association with MDC1 plays a role in mediating DNA damage checkpoint responses.

Interestingly, constitutive activation of phosphorylated Chk2^-T68 and 53BP1 has been reported in p53-deficient or mutant cell lines, whereas inhibition of 53BP1 by RNAi reduced phosphorylated Chk2^-T68 in these cells (DiTullio et al., 2002). Moreover, staining of constitutively phosphorylated Chk2^-T68 was also observed in both lung and other cancer tissues that were p53 mutant, thus implying that 53BP1 is important for mediating ATM-dependent checkpoint pathways (DiTullio et al., 2002). Since our studies show that STAT-1 is also required for mediating a subset of ATM-dependent checkpoint pathways, we examined whether the STAT-1 status was altered in p53-deficient or mutant cell lines. As shown in Fig. 6A, expression of STAT-1 was significantly enhanced in Soas2 and HCT15 cells lacking functional p53, which correlated with increased expression of MDC1 and 53BP1 levels compared to IMR90 and SKNSH cells which both have functional p53 (Fig. 6A). Furthermore, in both HCT15 and Soas2 cells the increased MDC1 and 53BP1 levels were also associated with constitutively phosphorylated ATM^-S1981, Chk2^-T68 and NBS1. Additionally, expression of STAT-1 was also significantly enhanced in p53^–/– MEF cells compared with p53^+/+ MEF cells of the same genotype (Fig. 6B) These studies show that STAT-1 levels are enhanced in cells that lack or have an inactive p53, which correlates with our finding that STAT-1 expression is necessary for the activation of ATM-dependent pathways by regulating the expression of MDC1 and 53BP1.

To determine whether the expression-status of STAT-1 is indeed associated with the constitutively phosphorylated ATM^-S1981, we inhibited the expression of STAT-1 in Soas2 and HCT15 cells that had been transfected with a STAT-1 RNAi construct (Fig. 6C,D). In these cells, phosphorylated ATM^-S1981 was significantly reduced compared with cells that had been transfected with a control RNAi construct (Fig. 6D), which also correlated with the reduced expression of MDC1 and 53BP1. Overall, these studies demonstrate that, in cells defective in p53, overexpression of STAT-1 is associated with enhanced ATM activity. Furthermore, we also found that, in IMR90 cells, suppression of STAT-1 with STAT-1 RNAi also resulted in an RDS-like phenotype following DNA damage (data not shown). We therefore exclude the possiblility that the results of the RDS assays in U3A cells are artifacts arising from the use of an immortalised cell line.

To show whether the STAT-1 status is involved in modulating ATM activation in cells with are wild-type p53, we overexpressed STAT-1 in the IMR90 cell line. As shown in Fig. 6E, overexpression of STAT-1 resulted in enhanced expression of MDC1 and 53BP1, which also associated with constitutive phosphorylation of ATM. Thus, STAT-1 is able to modulate ATM activity presumably via the increased protein expression of the ATM mediators MDC1 and 53BP1.

In response to DNA DSBs, distinct ATM-mediated regulatory pathways are activated and appear to play an important role in transducing DNA-damage signals to downstream effectors to control processes such as DNA repair, checkpoint arrest or apoptosis (Abraham, 2001; Shiloh, 2003). Distinct checkpoint pathways involved in DNA-damage-dependent S-phase responses are known to cooperate following DNA damage by inhibiting DNA replication. These include the ATM-Chk2-Cdc25A pathway (Falck et al., 2001; Falck et al., 2002; Bartek and Lukas, 2003) and the ATM-NBS1 pathway which jointly contribute to the inhibition of DNA synthesis after γ-IR. The mechanistic role of Cdc25A in the inhibition of DNA synthesis is well known; phosphorylated NBS1 seems to mediate the phosphorylation of the downstream structural maintenance of chromosome-1 (SMC1) protein following DNA damage (Yazdi et al., 2002).

Our data demonstrate that STAT-1 is able to modulate the phosphorylation of ATM and its downstream substrates Chk2^-T68 and NBS1^-S343, suggesting that the RDS phenotype, observed in cells lacking STAT-1, may be attributed to a defect in both the ATM-Chk2-Cdc25A and the ATM-NBS1-SMC1 pathways. Previous studies have shown that cells that lack functional NBS1 still have an intact ATM-Chk2-Cdc25A pathway in response to DNA damage (Yazdi et al., 2002), implicating the existence of a separate and distinct ATM-NBS1-SMC1 pathway involved in S-phase checkpoint-control. Furthermore, the ATM-NBS1-SMC1 pathway seems to be BRCA1-independent, because in cells that lack functional BRCA1, phosphorylation of SMC1 is not affected (Yazdi et al., 2002). Our studies show that activation of either the ATM-Chk2-Cdc25A or the ATM-NBS1-SMC1 pathway partly depends on STAT-1. How STAT-1 can mediate these effects is not clear, but it might involve activating the expression of both MDC1 and 53BP1, factors known to be required for mediating downstream activation of ATM-dependent pathways.

A role of MDC1 in mediating Chk2^-T68 phosphorylation was previously reported in cells following studies using RNAi to silence the MDC1 gene (Lou et al., 2003; Peng and Chen, 2003). By contrast, others have shown that RNAi silencing of MDC1 had no effect on activation of the Chk2^-T68-Cdc25A pathway (Goldberg et al., 2003). This discrepancy may be owing to cell-type-specific effects or the RNAi protocol from different studies, which showed a variable effect on MDC1 suppression. However, MDC1 silencing has also been shown to reduce the phosphorylation of Chk2 on serines 33 and 35 (Mochan et al., 2003). Interestingly, MDC1 and 53BP1 function in parallel pathways, and suppression of both these factors has a greater effect on abolishing Chk2^-T68 phosphorylation than those seen by inhibition of either 53BP1 or MDC1 (Peng and Chen, 2003). Moreover, MDC1 physically associates with ATM and the MRN complex, and studies have suggested that MDC1 is an ATM/ATR-dependent organizer that recruits DNA-checkpoint-signalling- and repair-proteins to the sites of DNA damage (Goldberg et al., 2003; Xu and Stern, 2003). This is consistent with our data here, which shows that cells lacking STAT-1 show reduced expression of both 53BP1 and MDC1, and this is associated with reduced ATM-dependent activated pathways following DNA damage. Our studies also confirm that STAT-1 is a direct activator of the MDC1 promoter and that STAT-1 probably is an important regulator of the MDC1 gene.

The MRN complex, together with ATM activation, is the earliest event that occurs at DNA DSBs (D'Amours and Jackson, 2001). The order in which ATM and MRN act in the early phase of DSB response is unclear. However, recent studies have shown that functional MRN is required for ATM activation and for ATM-mediated pathways because, after DNA damage cells lacking active MRE11 or NBS1 show a weaker response in activated–autophosphorylated–ATM and its downstream (ATM-dependent) pathways activated by phosphorylation (Carson et al., 2003; Uzeil et al., 2003; Lee and Paull, 2004). Furthermore, MDC1 is required for recruitment of NBS1 to sites of DNA DSBs, and the MRN complex is required for ATM activation (Xu and Stern, 2003). Both MDC1 and the MRN complex, together with ATM all form a large complex at sites of DNA DSBs (Xu and Stern, 2003) which mediates autophosphorylation of ATM at serine 1981 and dissociation of inactive ATM dimers into active monomers (Bakkenist and Kastan, 2003).

More recent studies have placed 53BP1 and MDC1 upstream of ATM by showing that both factors are independently recruited to sites of DNA DSBs and that these events are independent of ATM (Mochan et al., 2003). In cells with wild-type NBS1, suppression of 53BP1 expression had no effect on phosphorylation of ATM^-S1981 but was associated with increased recruitment of MDC1 and NBS1 to sites of DNA DSBs, demonstrating that a reduction of 53BP1 is associated with a compensatory increase in MDC1-NBS1 activity (Mochan et al., 2003). By contrast, suppression of MDC1 resulted in a decrease of ATM^-S1981 phosphorylation in cells expressing NBS1 following DNA damage (Mochan et al., 2003). Thus, 53BP1 and MDC1-NBS1 function in parallel pathways, which are able to cross-talk in order to activate the ATM-response to DNA damage. Additionally, these data demonstrate that the components of the MRN complex have a function upstream of ATM. Activation of ATM can then phosphorylate the MRN-complex-component NBS1, which mediates events downstream of ATM (ATM-NBS1-SMC1 pathway). Thus, depending on the phosphorylation status of NBS1, it can function upstream or downstream of ATM.

Our data also demonstrate that STAT-1 interacts with MDC1, 53BP1 and Chk2 following DNA damage. This is in contrast to previous data, showing that MDC1 or 53BP1 associate with Chk2 under normal conditions and that this association is abolished in response to γ-IR (Lou et al., 2003; Wang et al., 2002). Thus, STAT-1 might be recruited to sites of DNA DSBs together with MDC1, thereby facilitating the recruitment and phosphorylation of Chk2 through activated ATM.

Recently, we have shown that STAT-1 can interact with p53 and modulate p53-mediated transcriptional effects as well as modulate apoptosis (Townsend et al., 2004). Our previous work and the data presented here demonstrate that, STAT-1, like p53, is involved in processes that mediate cell-cycle arrest or apoptosis. An important finding from this study is the mechanism of how STAT-1 may inhibit cell growth after genotoxic stress: inhibition might be mediated by STAT-1 regulating the ATM-Chk2-Cdc25A and ATM-NBS1-SMC1 pathways, which jointly contribute to the rapid inhibition of DNA synthesis after DNA damage. Interestingly, defects in the ATM pathways have been shown to be associated with radiosensitivity (Falck et al., 2001). We show here, that STAT-1-deficient cells are more resistant to cell death following γ-IR than STAT-1 expressing cells (Fig. 1D), indicating that cells that lack STAT-1 are not radiosensitive.

Both MDC1 and 53BP1 have been suggested to play a role as an adaptor protein in a similar fashion to the yeast proteins Rad9 and Rad53 (Rad53 is a homologue of human Chk2), which play a central role in transducing and amplifying DNA-damage-signals by activating the kinase Rad53 (Toh and Lowndes, 2003). Our data show that STAT-1 might also play a role as an adaptor protein for Chk2 by modulating its kinase activity after DNA damage. This is consistent with our data: cells lacking STAT-1 show reduced phosphorylation of p53^-S20, a downstream substrate of Chk2 kinase that is activated following γ-IR (data not shown). However, further studies are required to confirm whether STAT-1-Chk2 interaction can enhance the functional activity of Chk2 or whether Chk2/STAT-1 association alters STAT-1 functional activity.

A further key finding from this study is that the reduced ATM-dependent checkpoint-pathway in cells lacking STAT-1, is associated with reduced expression of MDC1 and 53BP1. This implies that STAT-1 modulates the normal expression levels of 53BP1 and MDC1. Furthermore, we show for the first time that STAT-1 can transactivate the MDC1 promoter. Thus, STAT-1 is probably an important regulator of the MDC1 gene, which plays a crucial role in transducing the DNA damage checkpoint response.

The ATM-Chk2-p53 pathway contributes to apoptosis following genotoxic stress. The state of phosphorylation of p53 is known to be crucial in the apoptotic programme following genotoxic stress (Vousden and Lu, 2002). Our present data shows reduced levels of phosphorylated p53^-S15 in STAT-1-deficient cells after γ-IR, which correlates with resistance to apoptosis following DNA damage. Other established substrates for Chk2, include PML and E2F-1 (Yang et al., 2002; Stevens et al., 2003). Both factors have been reported to play a role in promoting apoptosis in cells exposed to DNA-damaging agents. Furthermore, the functional activity of Chk2 to promote apoptosis, as well as its effects on RDS, requires the presence and association with MDC1 via its forkhead-homolgy-associated domain (Lou et al., 2003; Goldberg et al., 2003). Recent studies have suggested that the FHA domain is a phosphoamino acid binding-domain and interacts with other factors involved in DNA damage response pathways (Huang and Elledge, 2000). This is also consistent with our data in that, STAT-1 plays a role in regulating the expression of MDC1 as well as interacting with MDC1, which may therefore be crucial to sensitise cells towards undergoing apoptosis.

Taken together, these findings identify STAT-1 as a major player in modulating the cell-cycle-checkpoint responses following DNA damage. Our data also show an enhanced STAT-1 expression in cells that are defective in p53 and also have constitutively phosphorylated Chk2-^T68. This combination of events might be common in human cancer, especially, because STAT-1 and STAT-3 are were shown to be overexpressed in different cancers (Bromberg, and Darnell, 2000; Turkson and Jove, 2000). The relative levels of STAT-1 or STAT-3 might also be important to determine cell-fate in response to a number of stressful stimuli. This implies that STAT-1 is a novel and important molecular target for the development of cancer therapies, warranting further investigation in how STAT-1 deregulation and cancer are linked in checkpoint processes.

We are grateful to David Levy for the STAT-1^–/– MEF cells, Ian Kerr for the 2fTGH, U3A and U3A-ST1 cell lines, James Darnell for the STAT-1 constructs, Junjie Chen for the HCT15 cell lines and the anti-MDC antibody and Thanos Halazonetis for the anti-53BP1 antibody. This work was supported by the British Heart Foundation and the Wellcome Trust. S.B. and J.M. are supported by British Heart Foundation Research Studentships.