ATM Localization and Heterochromatin Repair Depend on Direct Interaction of the 53BP1-BRCT2 Domain with γH2AX

Abstract

53BP1 plays multiple roles in mammalian DNA damage repair, mediating pathway choice and facilitating DNA double-strand break repair in heterochromatin. Although it possesses a C-terminal BRCT2 domain, commonly involved in phospho-peptide binding in other proteins, initial recruitment of 53BP1 to sites of DNA damage depends on interaction with histone post-translational modifications--H4K20me2 and H2AK13/K15ub--downstream of the early γH2AX phosphorylation mark of DNA damage. We now show that, contrary to current models, the 53BP1-BRCT2 domain binds γH2AX directly, providing a third post-translational mark regulating 53BP1 function. We find that the interaction of 53BP1 with γH2AX is required for sustaining the 53BP1-dependent focal concentration of activated ATM that facilitates repair of DNA double-strand breaks in heterochromatin in G1.

Graphical abstract

Figure 1

53BP1 BRCT₂ Domain Binds γH2AX (A) Fluorescence polarization assay (closed circle) showing binding of His₆-SUMO-53BP1-BRCT₂ to a fluorescein-tagged phosphopeptide (Flu-SGGKKATQApSQEY) corresponding to the last 13 residues of γH2AX. Data are means of four replicates with error bars showing 1 SD, and the K_D (1.4 μM ± 0.15) was calculated by least-squares fitting of a one-site binding model. Removal of the phosphate group on Ser139 by addition of λ-phosphatase (open circle) abrogates the interaction. The dashed line indicates a signal level consistent with non-specific interaction. (B) As (A) but for His₆-SUMO-MDC1-BRCT₂ binding to the fluorescein-tagged γH2AX phosphopeptide. (C) As (A) but with a fluorescein-tagged phosphopeptide derived from the major ATM phosphorylation site on the heterochromatin protein KAP-1 (Noon et al., 2010). No binding was observed. (D) As (A) but for γH2AX peptides in which the C-terminal α-carboxyl group of Tyr142 is modified by either amidation (closed circle) or addition of a further glycine residue (open circle). Both modifications substantially reduce the affinity, and no K_D could be determined. (E) As (A) but for H2AX peptides in which the side chain of Tyr142 is phosphorylated, either alone (closed circle) or in the presence of the γH2AX Ser139 phosphorylation (open circle). In both cases, phosphorylation of Tyr142 substantially diminished binding to His₆-SUMO-53BP1-BRCT₂. (F) Overview of crystal structure of a phosphopeptide corresponding to the last four residues of γH2AX bound to the BRCT₂ domain of 53BP1. The core DNA-binding domain of P53, which is required for crystallogenesis, makes no interaction with the γH2AX peptide and is omitted for clarity. (G) Detail of interactions made by the tail of γH2AX with the 53BP1-BRCT₂ domain. Carbon atoms in γH2AX are white; those of 53BP1 are rainbow colored to reflect their relative position within the protein sequence. The basic side chains of 53BP1 residues Lys1814 and Arg1811 provide neutralizing hydrogen bonding interactions with the acidic phosphate of γH2AX-pSer139 and the α-carboxyl of Tyr142, which are both required for binding of γH2AX to 53BP1-BRCT₂; see (A) and (C). (H) As (A) but showing binding of the fluorescent γH2AX peptide to His₆-SUMO-53BP1-BRCT₂ with either an R1811E mutation (closed circle) or a K1814E mutation (open circle).

Figure 2

53BP1 BRCT₂ Domain Localizes to DNA Damage through Phospho-Specific Interactions (A) HeLa cells were reverse transfected with 53BP1 siRNA. 48 hr later, cells were transfected with eYFP-SV40_NLS or eYFP-SV40_NLS-53BP1-BRCT₂ constructs containing either wild-type or mutant BRCT domains, or eYFP-SV40_NLS-MDC1-BRCT₂. 16 hr after this cells were damaged using laser micro-irradiation along a line indicated by the white arrowheads, and images were recorded over 3 min. Scale bar, 10 μm. (B) eYFP fluorescence was tracked over 3 min to observe localization. Profiles of fluorescence intensity along the laser track in (A) were generated using Slidebook6 software (n = 30 from three experimental repeats). Error bars, 1 SD. (C) Cells were treated with either KU55933 (ATMi), NU7441 (DNA-PKi), or both for 1 hr prior to irradiation with 3 Gy of IR. Cells were incubated for a further 30 min before fixing and staining for γH2AX. Scale bar, 10 μm. (D) Cells were processed as before with the addition of either DNA-PKi, ATMi, or both for 1 hr prior to laser micro-irradiation. eYFP-fluorescence was tracked over 3 min to observe localization. Fluorescence intensity profiles were generated using Slidebook6 software (n = 30 from three experimental repeats). Error bars, 1 SD. (E) H2AX^+/+ and H2AX^−/− mouse embryonic fibroblasts were reverse transfected with 53BP1 siRNA. 24 hr later, cells were transfected with either the eYFP-SV40_NLS-53BP1-BRCT₂ construct or the eYFP-SV40_NLS control. 16 hr after this, cells were damaged using laser micro-irradiation. eYFP fluorescence was tracked over 3 min to observe localization. Scale bar, 10 μm. (F) Fluorescence intensity profiles from (E) were generated using Slidebook6 software (n = 30 from three experimental repeats). Error bars, 1 SD. (G) eYFP-SV40_NLS-53BP1-BRCT₂ was immunoprecipitated using GFP-trap beads (ChromoTek), from benzonase-treated, γ-irradiated HeLa cell lysates, and eYFP-SV40_NLS-53BP1-BRCT₂ or γH2AX detected by western blot. Inputs represent 1/1,000 of the total sample used in the immunoprecipitation.

Figure 3

Co-localization of pATM Is Disrupted in 53BP1-BRCT₂ Phospho-Binding Mutants (A) HeLa cells were reverse transfected with 53BP1 siRNA. 48 hr later, cells were transfected with an siRNA-resistant construct containing wild-type HA-53BP1, one of the BRCT mutants, or a Tudor domain mutant (Y1502L). 16 hr after this, cells were exposed to either 8 Gy for mock and wild-type (WT) or 3 Gy of ionizing radiation for the mutants and allowed to recover for 24 hr before staining for HA and NBS1. Scale bar, 5 μm. (B) As (A) but stained for pATM. (C) Cells processed in (A) and (B) were analyzed by measuring foci intensity of HA, pATM, and NBS1 (n = 30). Foci intensities were normalized to the average wild-type foci intensity for their respective channel. Normalized HA foci intensities from cells transfected with either WT or BRCT mutants indicates comparable expression of the plasmid constructs. Graph shows average of three experiments with 1 SD.

Figure 4

Slow Repair of Damage in Heterochromatin Is Defective in 53BP1-BRCT₂ Phospho-Binding Mutants (A) HeLa cells were reverse transfected with 53BP1 siRNA. 48 hr later, cells were transfected with an siRNA-resistant construct containing wild-type HA-tagged 53BP1 or one of the BRCT mutants. 16 hr later, cells were exposed to 3 Gy of ionizing radiation and allowed to recover for the indicated time before staining for HA-53BP1 and γH2AX. Higher numbers of γH2AX foci were evident 24 hr after irradiation in siRNA-treated cells and siRNA-treated cells transfected with 53BP1 mutant constructs. (B) Quantitation of γH2AX focus persistence. Graphs show the mean of three experiments; error bars, 1 SD. 53BP1 constructs lacking the BRCT domain or with point mutations that abrogate γH2AX binding in vitro show significantly higher levels of persistent γH2AX foci than wild-type (one-way ANOVA). (C) As (A) but with 53BP1 and/or KAP-1 siRNA. Persistence of significantly higher levels of γH2AX foci in the presence of 53BP1 mutants is suppressed when KAP-1 is also knocked down, indicating that persistence is due to a defect in KAP-1 phosphorylation by ATM as a result of the failure of the 53BP1 mutants to localize pATM to heterochromatinized sites of damage.