Tumor Suppressor P53 Binding Protein 1 (53bp1) Is Involved in DNA Damage–Signaling Pathways

Cells have evolved various sophisticated pathways to sense and overcome DNA damage as a mechanism to preserve the integrity of the genome. Environmental attacks like radiation or toxins, as well as spontaneous DNA lesions, trigger checkpoint activation and consequent cell cycle arrest and/or apoptosis. One key protein that coordinates DNA repair with cell cycle progression and apoptosis is the tumor suppressor protein p53. P53 is activated and posttranslationally modified in response to DNA damage (Appella and Anderson 2000). These modifications include phosphorylation by ataxia telangiectasia mutated (ATM), a protein kinase implicated in DNA damage–signaling pathways (Canman et al. 1998; Khanna et al. 1998). By transcriptionally activating genes involved in cell cycle control, DNA repair, and apoptosis, p53 participates in the maintenance of the genomic integrity after DNA damage.

P53 interacts with p53 binding protein 1 (53BP1). 53BP1 has been identified in a yeast two hybrid screen as a protein that interacts with the central DNA–binding domain of p53 (Iwabuchi et al. 1994). Similar to breast cancer susceptibility gene 1 (BRCA1; Ouchi et al. 1998; Zhang et al. 1998a; Chai et al. 1999), 53BP1 enhances p53-dependent transcription (Iwabuchi et al. 1998). Interestingly, the COOH terminus of 53BP1 contains tandem BRCA1 COOH terminus (BRCT) motifs. This motif was first identified in the COOH-terminal region of BRCA1 and has since been found in a large number of proteins involved in various aspects of cell cycle control, recombination, and DNA repair in mammals and yeast (Koonin et al. 1996; Bork et al. 1997; Callebaut and Mornon 1997). The function of the BRCT domain is not known. However, evidence suggests that BRCT domains may mediate protein–protein interactions (Zhang et al. 1998b).

The presence of BRCT domains in 53BP1 and the reported interaction with p53 prompted us to investigate whether 53BP1 is involved in DNA damage–response pathways. Here we report that 53BP1 becomes hyperphosphorylated and rapidly relocates to the sites of DNA strand breaks in response to ionizing radiation. 53BP1 foci formation is reduced in ATM-deficient cells and can be inhibited by wortmannin in ATM wild-type cells. Moreover, radiation-induced hyperphosphorylation of 53BP1 is absent in cells treated with wortmannin, as well as in ATM-deficient cells. Taken together, these results strongly suggest that 53BP1 participates in DNA damage–signaling pathways and is regulated by ATM after γ-radiation.

Cells were grown in RPMI 1640 medium supplemented with 10% fetal bovine serum at 37°C with 5% CO₂. FT169A and YZ5 cells were provided by Dr. Y. Shilon (Tel Aviv University, Ramat Aviv, Israel). Cells grown on coverslips were irradiated in a JL Shepherd ¹³⁷Cs radiation source at a rate of 1 Gy/min for doses of 1–5 Gy or 10 Gy/min for a dose of 10 Gy. UV light was delivered in a single pulse (50 J/m²) using a Stratalinker UV source (Stratagene). Before UV irradiation, the culture medium was removed and the medium was replaced immediately after irradiation. All cells were returned to the incubator for recovery and harvested at the indicated times. Genotoxic agents and other drugs were used at the indicated concentrations. After a 1-h exposure, the cells were harvested for immunostaining.

Immunoprecipitation, immunoblotting, and immunostaining were performed as described previously (Scully et al. 1997). Rabbit polyclonal anti-53BP1 serum E7 was raised against a glutathione S-transferase (GST) fusion protein encoding residues 338–671 of 53BP1. Mouse anti-53BP1 monoclonal antibodies were raised against a mix of three GST fusion proteins encoding residues 1–337, 338–671, and 1,331–1,664, respectively, of 53BP1. Antiphospho-H2AX antibody was generated as described previously (Rogakou et al. 1999).

ATM was immunoprecipitated from K562 cells using anti-ATM antibody Ab3 (Oncogene Research Products). Aliquots of the ATM–protein A Sepharose immunocomplexes were resuspended in 25 μl kinase buffer (10 mM Hepes, pH 7.4, 50 mM NaCl, 10 mM MgCl₂, 10 mM MnCl₂, 1 mM DTT, 10 nM ATP) and incubated for 20 min at 30°C with 10 μCi of [γ³²]P-ATP and 1 μg of various affinity-purified GST fusion proteins containing different fragments of 53BP1.

Several proteins, including BRCA1 and Mre11/Rad50/Nbs1, form DNA damage–regulated, subnuclear foci in the cell. To determine whether 53BP1 participates in DNA damage–signaling pathways, we examined 53BP1 localization after various types of DNA damage using several anti-53BP1 polyclonal and monoclonal antibodies generated for this study. All antibodies specifically recognize endogenous, as well as HA-tagged, full-length 53BP1 as examined by Western blotting, immunoprecipitation, and immunostaining (data not shown). As shown in Fig. 1, 53BP1 is diffusely localized in the nuclei of normal cells, but relocates to discrete subnuclear foci structures in response to ionizing radiation (e.g., 1 Gy). These 53BP1 foci can be detected as early as 5 min after irradiation (data not shown). Higher doses of radiation (e.g., 10 Gy) lead to more but smaller 53BP1 foci (Fig. 1). The number of foci reaches a peak at ∼30 min after radiation. Thereafter, the foci number slowly decreases, whereas the foci size increases (data not shown).

Foci formation is also observed in response to other DNA-damaging events. UV radiation induced the formation of numerous small foci, similar to that induced by 4NQO (a UV-mimetic agent) and hydroxyurea (Fig. 1). Treatment with the DNA topoisomerase I poison camptothecin or the topoisomerase II poison etoposide (VP16), which cause DNA single strand and double strand breaks, respectively, also resulted in the formation of 53BP1 foci. Similar results were obtained with the alkylating agent methylmethanesulfonate. However, cisplatin, a DNA cross-linking agent, induced only a few 53BP1 foci during the first hour after drug application, whereas the protein kinase inhibitor UCN-01 and the antimitotic agent paclitaxel (Taxol; Bristol-Meyers Squibb Co.) did not induce 53BP1 foci formation. Thus, different types of DNA damage trigger the recruitment of 53BP1 into discrete nuclear foci.

The time course of 53BP1 foci formation and disappearance is very similar to that recently described for phosphorylated H2AX (Rogakou et al. 1999; Paull et al. 2000). H2AX is one of the histone H2A molecules in mammalian cells and becomes rapidly phosphorylated after exposure of cells to ionizing radiation (Rogakou et al. 1999; Paull et al. 2000). Phosphorylated H2AX (γ-H2AX) appears within 1–3 min as discrete nuclear foci on sites of DNA double strand breaks (Rogakou et al. 1999). Similar to γ-H2AX (Rogakou et al. 1999), the number of 53BP1 foci showed a linear relationship with the severity of DNA damage (Fig. 1 and data not shown). As shown in Fig. 2 A, damage-induced 53BP1 foci colocalized with γ-H2AX at the various time points analyzed. The number of 53BP1 foci was identical to that of γ-H2AX throughout the course of the experiment. In addition, coimmunoprecipitation analysis revealed that 53BP1 and γ-H2AX biochemically interact after γ-radiation (Fig. 2 B). Small amounts of 53BP1 were detected in γ-H2AX immunoprecipitates prepared from irradiated HBL100 cells. In unirradiated cells, H2AX was not phosphorylated and anti–γ-H2AX antibodies did not immunoprecipitate any phosphorylated H2AX. Similarly, 53BP1 was also not present in anti–γ-H2AX immunoprecipitates prepared from unirradiated cells. These results demonstrate that 53BP1 colocalizes and interacts with γ-H2AX at the sites of DNA strand breaks after γ-radiation.

Several phosphatidylinositol 3–kinase (PI3K)-related kinases, including DNA-dependent protein kinase (DNA-PK), ATM, and ATM-related kinase (ATR), participate in DNA damage–responsive pathways (Smith and Jackson 1999; Khanna 2000). It is possible that DNA damage–induced 53BP1 foci formation may depend on one or more of these PI3K-like kinase family members.

We first examined 53BP1 foci formation in the presence or absence of DNA-PK using two derivatives of the human glioma cell line MO59 (Lees-Miller et al. 1995). No difference in the time course of 53BP1 foci appearance and disappearance was observed in these two cell lines after exposure to 1 Gy of γ-radiation (data not shown). However, comparison was hampered by the high number of 53BP1 foci in unirradiated MO59K and MO59J cells and subtle differences might be overlooked.

We then examined whether the 53BP1 response to ionizing radiation is affected in cells lacking ATM. Immortalized ATM-deficient fibroblasts (FT169A) were compared with their isogenic derivative cells, YZ5, that have been reconstituted with wild-type ATM cDNA (Ziv et al. 1997). As shown in Fig. 3 A, although irradiation with 1 Gy resulted in a rapid formation of 53BP1 foci in the ATM-reconstituted cells (ATM+), a reduced response was observed in the cells lacking wild-type ATM (ATM−). Similar results were obtained when we compared other ATM-deficient fibroblast lines (GM03189D and GM05849C) with wild-type ATM cell lines (GM02184D and GM00637H) (Fig. 3 A). The time course of the number of 53BP1 foci per cell, as calculated from three independent experiments using YZ5 versus parental FT169A cells, is illustrated in Fig. 3 B.

To further corroborate the role of ATM in 53BP1 foci formation, we pretreated HeLa cells for 30 min with wortmannin before exposure to 1 Gy of irradiation. Wortmannin is a potent inhibitor of the PI3K-related kinases, including ATM and DNA-PK (Sarkaria et al. 1998). As shown in Fig. 3 C, pretreatment with 50 μM wortmannin greatly reduced the number of 53BP1 foci evident 1 h after γ-radiation. At an even higher dose (200 μM), wortmannin completely blocked 53BP1 foci formation. These results suggest that the kinase activities of ATM or other PI3K-related kinases are required for 53BP1 foci formation.

Many proteins involved in DNA damage–response and/or DNA repair are phosphorylated upon DNA damage. To examine whether 53BP1 becomes phosphorylated in response to γ-radiation, K562 cells were irradiated (20 Gy) and harvested 1 h later. After immunoprecipitation using anti-53BP1 antisera, the samples were incubated for 1 h at 30°C in the presence or absence of λ protein phosphatase and separated on a 3–8% gradient SDS gel. Phosphatase treatment of unirradiated K562 cells revealed a faster migrating form of 53BP1 (Fig. 4 A). This indicates that 53BP1 is modified by phosphorylation in normal undamaged cells. Upon γ-radiation, 53BP1 showed an even slower mobility that was reversed by phosphatase treatment (Fig. 4 A). These results suggest that 53BP1 is phosphorylated in undamaged cells and becomes hyperphosphorylated after γ-radiation.

Since 53BP1 is hyperphosphorylated after γ-radiation, we then examined whether wortmannin would affect radiation-induced 53BP1 phosphorylation. As illustrated in Fig. 4 B, there was no detectable radiation-induced 53BP1 mobility shift in wortmannin (50 μM)-pretreated cells. In contrast, the radiation-induced 53BP1 mobility shift was readily detected in cells that had received no drug treatment before radiation. We next repeated the experiment using the ATM-deficient GM03189D and GM02184D cells expressing wild-type ATM. Again, in ATM wild-type cells, γ-radiation induced a 53BP1 mobility shift in control, but not in wortmannin-pretreated samples (Fig. 4 C). However, no radiation-induced 53BP1 mobility shift was observed in ATM-deficient cells, with or without wortmannin treatment (Fig. 4C and Fig. D). Taken together, these results strongly suggest that ATM is required for 53BP1 hyperphosphorylation after γ-radiation.

S/TQ sites have been described to be the minimal essential recognition sites for ATM (Kim et al. 1999). 53BP1 contains a total of 30 S/TQ sites, many of them clustered in the NH₂-terminal region. To examine whether 53BP1 is a substrate for ATM, and to define regions that can be phosphorylated by ATM in vitro, we designed six overlapping 53BP1 GST fragments that span the entire ORF of 53BP1 and performed a standard ATM kinase assay. As shown in Fig. 4 E, the first three NH₂-terminal 53BP1 fragments were phosphorylated by ATM in vitro, whereas no phosphorylation was observed in the last three COOH-terminal fragments, despite the fact that there are a total of 10 S/TQ sites within these 53BP1 fragments. These data suggest that 53BP1 is a substrate of ATM kinase.

Here we report that 53BP1 participates in the early DNA damage–response. Using several antibodies specifically recognizing 53BP1, we show that 53BP1 becomes hyperphosphorylated and forms nuclear foci after exposure to ionizing radiation. γ-Radiation–induced 53BP1 hyperphosphorylation and foci formation are reduced in ATM-deficient cells. Moreover, 53BP1 hyperphosphorylation, as well as foci formation, is inhibited by wortmannin, an inhibitor of the PI3K-related kinases including ATM, DNA-PK, and, to a lesser extent, ATR (Sarkaria et al. 1998). Taken together, these data suggest that ATM and other PI3K-related kinases directly phosphorylate 53BP1 and regulate its localization to the sites of DNA strand breaks.

In favor of a functional link between ATM and 53BP1, we also demonstrate that NH₂-terminal fragments of 53BP1 are effectively phosphorylated by ATM in vitro. Similarly, Xia and colleagues have recently shown that Xenopus 53BP1 and a NH₂-terminal fragment of human 53BP1 can be phosphorylated by ATM in vitro and in vivo (Xia et al. 2000), supporting our hypothesis that 53BP1 is a direct substrate of ATM. In contrast to our findings, Schultz et al. 2000 observed no difference in 53BP1 foci formation in ATM-deficient cells when compared with that in normal ATM wild-type cells (Schultz et al. 2000). We also observed that 53BP1 foci still formed, albeit with slower kinetics, in cells lacking ATM, suggesting the existence of an alternative, ATM-independent pathway for the regulation of 53BP1. However, our data presented here clearly demonstrate that ATM plays a critical role in the regulation of 53BP1 hyperphosphorylation and foci formation after γ-radiation.

53BP1 rapidly colocalizes with γ-H2AX in response to ionizing radiation. H2AX is a histone H2A variant that becomes phosphorylated and forms foci at sites of DNA strand breaks after DNA damage (Rogakou et al. 1999; Paull et al. 2000). The number, as well as appearance and disappearance of 53BP1 foci, matched almost completely with that of γ-H2AX. Moreover, 53BP1 and γ-H2AX physically interact after ionizing radiation, suggesting that 53BP1 relocates to the sites of DNA double strand breaks in response to γ-radiation. Similar to 53BP1, γ-H2AX foci formation is inhibited by wortmannin treatment (Rogakou et al. 1999; Paull et al. 2000) and is reduced in ATM-deficient cells (Rappold, I., and J. Chen, unpublished observation). It is possible that phosphorylation of H2AX may mediate the relocalization of 53BP1 to DNA strand breaks. If this is the case, ATM-dependent hyperphosphorylation of 53BP1 may be a secondary event that is not required for 53BP1 foci formation. This possibility will be examined in future studies using phosphorylation-deficient mutants of 53BP1.

Upon relocalizing to the sites of DNA damage, 53BP1 could participate in chromosome remodeling that makes DNA lesions accessible to DNA repair proteins. Alternatively, 53BP1 could be involved in the recruitment of repair proteins like BRCA1 and Rad51 to these DNA lesions. Both of these proteins colocalize with 53BP1 several hours after exposure to ionizing radiation (Rappold, I., and J. Chen, unpublished observations). In addition, BRCA1 biochemically interacts with 53BP1 after γ-radiation (Rappold, I., and J. Chen, unpublished observations).

53BP1 contains two BRCT motifs at its COOH terminus. 53BP1 BRCT motifs are closely related with those of BRCA1 and Saccharomyces cerevisiae Rad9 (scRad9) protein. Insight into the potential role of scRad9 comes from studies of its association with scRad53. ScRad53 is the homologue of mammalian Chk2 or Schizosaccharomyces pombe Cds1. After DNA damage, scRad9 is phosphorylated and this phosphorylated scRad9 associates with the forkhead homology–associated (FHA) domain of scRad53 (Sun et al. 1998; Vialard et al. 1998). Mutations in either the scRad53 FHA domain (Sun et al. 1998) or scRad9 BRCT motifs (Soulier and Lowndes 1999) prevent scRad53 activation after DNA damage. Although the mammalian homologue of scRad9 has not been identified, a scRad9 homologue likely exists in mammals. Because of the close homology of their BRCT motifs, two candidate scRad9 homologues are BRCA1 and 53BP1. Based on yeast studies, one would predict that the activation of Chk2, the homologue of scRad53, should depend on this scRad9 homologue in mammalian cells. However, DNA damage–induced phosphorylation of Chk2 was observed in BRCA1-deficient cells (Matsuoka et al. 1998), suggesting that BRCA1 may not be the mammalian homologue of scRad9. Experiments using 53BP1-deficient cells will be performed to examine whether 53BP1 is the scRad9 homologue in mammals.

In conclusion, our data demonstrate that 53BP1 participates early in DNA damage–signaling pathways and is regulated by ATM after γ-radiation. The exact role of 53BP1 in these pathways remains to be resolved. Given the importance of these DNA damage–signaling pathways in cancer prevention, it will be interesting to examine whether 53BP1 is mutated in tumors.

We thank Drs. Scott Kaufmann, Larry Karnitz, and Jann Sarkaria for stimulating conversations. We also thank the Mayo Clinic Flow Cytometry and Protein Core facilities for their assistance. Initial studies of this work were performed in the laboratory of Dr. David M. Livingston. We specially thank Dr. Livingston for help and encouragement during the course of this study.

This work was supported by the Mayo Foundation, Mayo Cancer Center, Division of Oncology Research, and an Eagles grant to J. Chen.