Phosphorylation-dependent interactions of BLM and 53BP1 are required for their anti-recombinogenic roles during homologous recombination

Abstract

Mutations in bloom helicase protein (BLM) helicase cause Bloom syndrome, characterized by predisposition to almost all forms of cancer. We have demonstrated previously that endogenous BLM, signal transducer 53BP1 and RAD51 are present in a complex during replication stress. Using full-length recombinant proteins, we now provide evidence that these proteins physically interact. BLM interacts with checkpoint kinase (Chk) 1 via the kinetochore-binding domain (KBD). Wild-type (WT) Chk1 phosphorylates 53BP1 in the KBD, both in vitro and in vivo during replication stress. Chk1-mediated phosphorylation of 53BP1 enhances its binding to BLM and is required for the accumulation of 53BP1 at the site of stalled replication. 53BP1, in turn, binds to the N-terminal domain of BLM. Ataxia telangiectasia and Rad3 related (ATR)-mediated phosphorylation of BLM at Thr99 is critical for its interaction and subsequent co-localization with 53BP1. WT BLM enhances the interaction and co-localization between 53BP1 and RAD51 during replication arrest. Interactions between the three proteins have functional consequences. Non-binding or phosphorylation-deficient mutants of BLM and 53BP1 fail to demonstrate the anti-recombinogenic property of the WT counterparts. Consequently, these mutants cause elevation of endogenous RAD51 foci formation. These results provide evidence that the phosphorylation-mediated interactions between BLM, 53BP1 and RAD51 are required for their regulatory roles during homologous recombination.

Introduction

Double-strand break repair defect is a common denominator of carcinogenesis due to generation of genomic instability (1). Maintenance of genomic fidelity relies on the high precision of homologous recombination (HR). The pro-recombinogenic protein RAD51 plays a central role in the process forming recombinatory sub-nuclear structures called RAD51 foci. These foci mark a subset of cells that have entered the HR pathway and contain functional recombination complexes (2). HR needs to be finely regulated as RAD51 over-expression has been reported in several tumour-derived cells and correlates with elevated HR frequencies (3). Multiple regulatory processes are involved in the regulation of RAD51 activity. RAD51 is phosphorylated at Thr309 by the checkpoint kinase (Chk1), a process essential for the appearance of RAD51 foci (4). We have reported that the assembly of RAD51 structures are regulated by p53 (5) and by bloom helicase protein (BLM) and 53BP1 (6).

53BP1, identified as a p53-interacting protein, is involved in DNA damage-induced signal transduction (7). 53BP1 contains multiple structural elements including two breast cancer gene 1 (BRCA1) C-terminal repeats, tandem tudor domains, a glycine–arginine rich (GAR) methylation stretch, a nuclear localization signal and a 30 amino acid region termed the dimerization domain (supplementary Figure S2A is available at Carcinogenesis Online). A stretch of 381 amino acids spanning the region (1235–1616) and encompassing the dimerization domain, tudor folds and the GAR together constitute the kinetochore-binding domain (KBD) region. Among other functions, this domain is essential for the accumulation of ionizing radiation (IR)-induced 53BP1 foci (8). 53BP1 contains numerous phosphatidyl inositol-like kinase sites and is phosphorylated in vivo by ataxia telangiectasia mutated (ATM) at Ser25/Ser29 (9,10). However, the ATM-mediated phosphorylation is dispensable for its 53BP1 accumulation at the sites of double-strand breaks. We had earlier shown that 53BP1 is involved in the recruitment of BLM to the sites of hydroxyurea (HU)-induced stalled replication forks during S-phase (11). Others and we have also demonstrated that BLM is involved in the efficient localization of serine 15-phosphorylated p53 to the sites of stalled replication, whereby both the tumour suppressors (BLM and p53) together modulate HR (12,13). Finally, we have recently demonstrated that the loss of BLM and 53BP1 synergistically enhance stress-dependent HR (6).

BLM is phosphorylated by distinct kinases in different stages of the cell cycle. It is phosphorylated during mitosis by Monopolar spindle 1 and mitotic cdc2 kinase (14,15), and in S-phase by ataxia telangiectasia and Rad3 related (ATR) (16). Phosphorylation by ATR on BLM occurs on two residues, Thr99 and Thr122, and has a role in the recovery from S-phase. Incidentally, Thr99 (and probably Thr122) is also targeted by ATM after IR (17). The functional interaction between BLM and ATR is evolutionarily conserved. Mutations in Mec1 and Sgs1 (the respective homologs of ATR and BLM in budding yeast) lead to replisome instability, fork collapse and gross chromosomal rearrangements (18). BLM may regulate HR via multiple processes—(i) cooperation with Topo IIIα to resolve recombination intermediates like double Holliday junctions (19) and/or (ii) through the evolutionary conserved interaction with RAD51 (20). We have recently demonstrated that BLM and 53BP1 can independently regulate HR by modulating the assembly of RAD51 filaments (6). However, for 53BP1 to attain its full anti-recombinogenic function, the presence of BLM was a necessary factor, thereby also indicating a BLM-dependent role (6). Hence, we wanted to determine (i) the regulatory mechanisms that govern a possible BLM–53BP1 interaction and (ii) how BLM and 53BP1 can independently affect RAD51 function during HR.

Materials and methods

Plasmids and siRNA

pCMH6K53BP1 (a gift of Kuniyoshi Iwabuchi), pCMH6K53BP1 (ΔKBD), i.e. pCMH6K53BP1 (Δ1235–1616) and pCMH6K53BP1 5K (both gifted by Phillip Carpenter), glutathione S-transferase (GST)-53BP1 fragments and pCMH6K53BP1 ΔDimer, i.e. pCMH6K53BP1 (Δ1231–1270) (all gifted by Junjie Chen), pcDNA3 Flag BLM, pcDNA3 Flag BLM (1–1417, T99A, T122A) and pGEX4T-1 BLM (1–212) (all gifted by Ian Hickson). pcDNA3-HA 53BP1 was obtained by sub-cloning the amplified 53BP1 complementary DNA (cDNA) from pCMH6K53BP1 into the NheI/XhoI sites of pcDNA3.1 hygro(−) plasmid. GST BLM fragments were obtained by cloning the respective polymerase chain reaction-amplified products into EcoRI/XhoI sites for pGEX4T-1 BLM (1001–1417) or into the BamHI/XhoI sites for pGEX4T-1 BLM (191–660) and pGEX4T-1 BLM (621–1041). pcDNA3 Flag RAD51 was obtained by sub-cloning the RAD51 cDNA into the BamH1/XbaI sites of pcDNA3. GST RAD51 (1–339), GST RAD51 (1–98) and GST RAD51 (99–339) were obtained by sub-cloning appropriate polymerase chain reaction-amplified fragments into the BamH1/NotI sites of pGEX4T-1. pcDNA3 Flag BLM (1–1417, T99A), pcDNA3 Flag BLM (1–1417, T122A), silently mutated 53BP1 (53BP1*, a single nucleotide change at position 222 of 53BP1cDNA) and 53BP1* (ΔKBD) mutants were constructed by site-directed mutagenesis kit (Stratagene, CA) using corresponding BLM or 53BP1 wild-type (WT) cDNAs as the template. Based on published sequences, siRNA for Chk1 (21), ATR (22), ATM (23) and green fluorescent protein (GFP) (used as control siRNA) (22) were synthesized by Dharmacon Research, IL.

Antibodies

Anti-BLM: goat polyclonal C-18 (Santa Cruz, CA) and rabbit polyclonal NB100–161 (Novus, CO); anti-BLM Thr99 (a gift of Yves Pommier); anti-53BP1: monoclonal (BD) and rabbit polyclonal (a gift of Yasuhisa Adachi); anti-RAD51: rabbit polyclonal Ab-1 (Calbiochem); anti-actin: monoclonal C-2 (Santa Cruz); anti HA: rabbit polyclonal SG77 (Zymed Laboratories, CA) and anti-flag: monoclonal M2 (Sigma, MO). Secondary antibodies were purchased from Jackson ImmunoResearch Laboratories, PA and Southern Bio-technology Associates, AL.

Cells, culture conditions and treatments

hTERT-immortalized Bloom syndrome (BS) fibroblasts, BS and the corrected BS fibroblasts and A-15 were a gift of Jerry Shay. hTERT-immortalized normal human fibroblast (NHF) strain GM07532 (NHF), BS shRNA53BP1 (clone 19) and A-15 shRNA53BP1 (clone 1) were maintained as described (6,11,13). BS shRNA53BP1 (clone 19) was derived from BS cells in which 53BP1 was depleted. A-15 shRNA53BP1 (clone 1) was derived from A-15 cells in which 53BP1 was depleted (6). HU (Sigma, 1 mM) treatments were done as described (13). Cells were either left untreated (−HU) or treated (+HU) for 16 h. Parallel HU-treated plates were washed and incubation continued for a further 6 h [post-wash (PW)].

Western blots and immunoprecipitations

Cytoplasmic and nuclear extracts from cells were made using NE-PER Nuclear and Cytoplasmic Extraction reagent (Pierce, IL). Immunoprecipitations (IPs) were done as described previously (13) using either 2.5 μg of purified proteins or 1 mg of the nuclear extracts. The experiments were repeated at least twice and representative blots shown.

Expression and purification of proteins

Proteins were expressed according to standard protocols in Escherichia coli at 18°C. GST-tagged proteins were purified by binding to Glutathione S-Sepharose slurry (Amersham, England). HA-53BP1 (both WT and mutants) was purified with HA affinity gel (Sigma). Recombinant hexa-histidine-tagged human BLM protein was purified from yeasts transformed with the plasmid pJK1 (gifted by Ian Hickson).

GST interactions and kinase assays

pcDNA Flag BLM, pcDNA3-Flag-RAD51 and pcDNA3-HA 53BP1 were used for coupled in vitro transcription/translation of BLM, 53BP1 and RAD51, respectively. Reactions were carried out with kit (Promega, WI) using 1 μg of the respective cDNAs and 5 μCi of S³⁵ methionine for 90 min at 30°C. GST-bound target proteins (bound to Glutathione S-Sepharose beads, Amersham) were incubated with the in vitro translated interacting partner for 4 h at 4°C with constant inversion. Interaction was assayed by determining the radioactivity bound to GST beads after washing, sodium dodecyl sulphate–polyacrylamide gel electrophoresis and fluorography. The experiments were repeated three times and representative blots shown. Kinase assays were carried out with 5 ng of GST-Chk1 (Upstate, MA) and purified HA-tagged WT or mutant 53BP1 (600 ng) in the presence of 5 μCi [γ]P³² adenosine triphosphate (ATP) and 25 μM cold ATP at 30°C for 30 min. Alternately, a modified kinase assay was standardized where 100 ng of the nuclear extracts obtained from −HU, +HU or PW cells were used as the source of kinase. Modified kinase assays, done in parallel either with [γ]P³²ATP or with cold ATP, were repeated four times and representative blots shown.

Microscopy

Immunofluorescence (IF) was carried out as described previously (13). Briefly, the cells after washing were either directly fixed by 100% ice-cold ethanol or subjected to a hypotonic lysis buffer (10 mM Tris–HCl, pH 7.5, 2.5 mM MgCl₂, 1 mM phenylmethylsulphonyl fluoride, 0.5% NP-40) on ice and subsequently fixed with ethanol. After staining, cells were visualized in a Upright Axioimager M1 motorized Epifluorescence microscope equipped with a high-resolution AxioCam MRm Rev. 2 camera. The images were taken with Plan Apochromate objective 100× per 1.40 oil immersion using fluorescein isothiocyanate and/or Texas red fluorophore. At least 100 cells were analysed for each co-localization experiment. The experiments were repeated twice. Co-localization factor is defined as [(fraction of cells having co-localization) × (fraction of foci co-localized per cell)] × 100.

Plasmid transfections

All transfections were done using Lipofactamine 2000 (Invitrogen, CA). For expression of proteins, the transfections were continued for 48 h, the last 16 h in the presence of HU. For HR, assay transfection was done for either HR substrate (pBHRF) alone or pBHRF combination with different expression plasmids. Cells transfectioned with pBHRF were grown for 24 h followed by 16 h of HU treatment. Each experiment was done in triplicate and repeated three times. Cells were subsequently treated for western blotting or flow cytometry.

Flow cytometry and host-cell reactivation assay

Cells were fixed for flow cytometry according to standard protocols. Cells were analysed in BD LSR (BD Biosciences) flow cytometer for host-cell reactivation assay.

The host-cell reactivation assay to determine HR rate was carried out as described (24). Transfections were carried out with the substrate (pBHRF) encoding an intact, emission shifted, ‘blue’ variant of GFP [blue fluorescent protein (BFP)], with a 300 nucleotide stretch of homology to a non-functional copy of GFP. In the absence of HR, only BFP is present while HR can also create a functional GFP. Green and blue fluorescences were simultaneously examined by exciting the cells using a 488 nm argon laser (GFP) and a UV (350–360 nm) laser (BFP). Post-transfection incubation was continued for 36 h in full medium after which the cells were harvested and flow cytometry carried out. Each experiment was carried out at least three times. The data analysis was done in Cell Quest Pro.

Results

53BP1, BLM and RAD51 physically interact

We have recently demonstrated that BLM, 53BP1 and RAD51 are present in a complex during replication stress (6). To further characterize the 53BP1-RAD51 interaction in vivo, we carried out IF experiments in NHF left asynchronous (−HU), arrested in S-phase by HU (+HU) or allowed to proceed after washing away HU (PW) (supplementary Figure S1A is available at Carcinogenesis Online) and quantitated the degree of co-localization (supplementary Figure S1B is available at Carcinogenesis Online). The number of RAD51 foci increased 3- to 8-fold during replication stress. 53BP1 was mostly diffused and nucleoplasmic during −HU treatment and formed few foci, which co-localized with RAD51. These co-localized foci were in promyelocytic leukemia (PML) nuclear bodies (PML NBs) as a high degree of co-localization was observed for 53BP1/PML staining, both in presence or absence of prelysis involving a hypotonic lysis buffer (supplementary Figure S1C is available at Carcinogenesis Online). The apparent contradiction with an earlier report (25) is possibly due to differences in IF techniques and/or the cell types employed. In the presence of HU, the 53BP1 foci increased in a time- and dose-dependent manner and co-localized with RAD51 and proliferating cell nuclear antigen and not with PML NBs, indicating that they were present at stalled replication forks (supplementary Figure S1A, S1B and S1D is available at Carcinogenesis Online and data not shown). During PW, the few 53BP1 and RAD51 foci present again co-localized with PML NBs (data not shown). These results indicate the dynamic nature of intra-nuclear shuttling of 53BP1, BLM and RAD51 in response to replication stress.

The presence of 53BP1, BLM and RAD51 in a complex in vivo (6) and their subsequent co-localization (as demonstrated above) do not necessarily mean that these proteins physically interact with each other. Hence, we generated the respective purified proteins and carried out IP with RAD51 antibody. We found that apart from RAD51–53BP1 and RAD51–BLM complexes, a RAD51–BLM–53BP1 trimeric complex was also formed. The specificity of the interaction was monitored by carrying out an IP with the corresponding immunoglobulin G (Figure 1A). The presence of the trimeric complex was also confirmed in vivo by sequential IPs using 53BP1, BLM and RAD51 antibodies. p53 was also part of the complex when it was present, as in NHF cells (Figure 1B).

Fig. 1.

Purified BLM, 53BP1 and RAD51 form a heterotrimeric complex. (A) Coomasie gel (left) shows the purified three proteins. * indicates immunoglobulin heavy chain obtained due to immunopurification of HA-53BP1 expressed in Cos cells. BLM and GST RAD51 were purified from yeast and Escherichia coli, respectively. IP (right) was done with anti-RAD51 antibody or with or its corresponding immunoglobulin G and immunoprecipitate probed with BLM and 53BP1 antibodies. (B) Sequential immunoprecipitations show the formation of BLM–53BP1–RAD51 complex even in absence of p53. Top panel shows the levels of p53, 53BP1, BLM, RAD51 and actin in nuclear extracts of isogenic NHF and NHF E6 cells in absence and presence of HU. Bottom panel describes the IPs done on +HU-treated nuclear extracts from NHF/NHF E6 cells. IPs were done sequentially with anti-53BP1, anti-BLM, anti-RAD51 antibodies or the corresponding rabbit immunoglobulin G.

Chk1-mediated phosphorylation of the KBD in 53BP1 regulates its interaction with BLM

To determine the mechanisms governing BLM–53BP1 interaction, assays were carried out with in vitro translated S³⁵-labelled full-length BLM and 53BP1 fragments (supplementary Figure S2A is available at Carcinogenesis Online; Figure 2A, left). Only one 53BP1 fragment spanning residues 1330–1664 containing the KBD interacted with BLM in a nucleic acid-independent manner (Figure 2A, right). To prove that the KBD of 53BP1 is solely involved in the interaction in the context of the whole protein, HA-tagged full-length 53BP1 or the ΔKBD mutant proteins were over-expressed and purified from Cos cells (checked by Coomasie; Figure 2D, right). IP with recombinant BLM reveals that lack of KBD abrogated the interaction of 53BP1 with the helicase (Figure 2B). It has been recently reported that 53BP1 is methylated on multiple arginine residues in the GAR stretch, upstream of tudor motifs but within the KBD (26,27). Amino acid residues 1231–1270 have been shown to mediate dimerization of 53BP1, a process implicated for its role during DNA repair (28) (supplementary Figure S2A is available at Carcinogenesis Online). Using two mutants, 53BP1 5K (all arginines in GAR region are mutated) and 53BP1 ΔDimer (amino acids 1231–1270 are deleted), we found that BLM–53BP1 interaction occurs even in the absence of methylation or dimerization motifs. However, a slight decrease in the level of interaction was reproducibly observed (Figure 2C).

Fig. 2.

Chk1-mediated phosphorylation of 53BP1 regulates its interaction with BLM. (A) KBD of 53BP1 interacts with BLM. Expression levels of GST 53BP1 fragments (visualized by Coomasie) (left). In vitro translated S³⁵-radiolabelled BLM was incubated with equal amounts of the 53BP1 fragments or GST alone. The reactions were also done after treating GST-53BP1 (1330–1664) with DNase or after cold phosphorylating this fragment with Chk1. Input is 20% of the sample used for interaction (right). (B) BLM binds to the KBD of 53BP1. HA-tagged 53BP1 WT or ΔKBD mutant were over-expressed in Cos cells. 53BP1 (mutant and WT) were immunopurified and subsequently incubated in vitro with purified BLM obtained from yeast cells. IP was done with anti-BLM antibodies or its corresponding immunoglobulin G and interaction was checked with anti-HA antibodies to detect 53BP1. (C) Lack of methylation or dimerization motif did not alter 53BP1–BLM interaction. The proteins (WT, 5K or ΔDimer) were immunopurified from transfected extracts of Cos cells using HA beads and visualized by Coomasie (left). Using equal levels of protein interactions were done between WT or mutants of 53BP1 and in vitro translated S³⁵-radiolabelled BLM (right). (D) Chk1 phosphorylation in 53BP1 occurs almost exclusively in KBD. HA-tagged WT or 53BP1 (ΔKBD) mutant were expressed in Cos cells and immunopurified. These proteins were stained in Coomasie along with WT or kinase dead (KD) Chk1. Kinase assay in presence of [γ]P³²ATP was carried out with either WT or KD Chk1 with HA-53BP1 or HA-53BP1 (ΔKBD). * indicates a possible co-purifying protein. Phosphorylated HA-53BP1, HA-53BP1 (ΔKBD) and Chk1 are indicated. (E) GST-53BP1 (1330–1664) is phosphorylated during replication stress. Modified kinase assays were carried out with the GST-53BP1 (1330–1664) or GST alone in presence of 100 ng of nuclear extract isolated from −HU- or +HU-treated NHF (as the source of the kinase) in the presence of [γ]P³²ATP. After sodium dodecyl sulphate–polyacrylamide gel electrophoresis, the proteins were either visualized by Coomasie (left) or transferred to the nitrocellulose membrane and autoradiographed. * indicates nuclear proteins that undergo autophosphorylation. (F) Interaction between BLM and GST-53BP1 (1330–1664) depends on Chk1-mediated phosphorylation. Kinase assay of GST-53BP1 (1330–1664) was done as in (D) except cold ATP was used. Nuclear extract from Chk1 siRNA-transfected cells was also utilized in parallel. Interactions were done between in vitro translated S³⁵-radiolabelled BLM and GST-53BP1 (1330–1664) fragment, phosphorylated with nuclear extract (NE) from different conditions. (G) Chk1-mediated phosphorylation regulates the accumulation of 53BP1 at the sites of stalled replication. NHFs were transfected either with a control or with a Chk1 siRNA and subsequently treated with HU for 16 h. IF was carried out with anti-53BP1 antibody. Nucleus was stained by 4′,6-diamidino-2-phenylindole. Bars, 5 μm.

We next wanted to determine whether the KBD of 53BP1 is phosphorylated during replication stress. Since ATR-Chk1 pathway is activated by stalled replication forks, we tested whether Chk1 phosphorylates 53BP1, particularly in the KBD. WT Chk1, but not its kinase-dead counterpart, phosphorylated full-length 53BP1 (Figure 2D, left). Chk1-mediated phosphorylation of 53BP1 was almost exclusively in the KBD, as detected using [γ]P³²ATP during kinase assays using recombinant proteins (Figure 2D, right). A modified kinase assay was subsequently standardized where nuclear extracts from +HU and −HU conditions (from NHF cells) were used as the source of the kinases. We found that GST–53BP1 (1330–1664) but not GST alone was phosphorylated in +HU condition (Figure 2E). Parallel reactions with cold phosphorylated GST–53BP1 (1330–1664) and subsequent interaction with in vitro translated S³⁵-labelled BLM revealed that BLM–53BP1 complex formation was enhanced in +HU condition (Figure 2F, lane 2). However, when nuclear extracts from HU-treated and Chk1 siRNA-transfected NHF (as tested in supplementary Figure S3, available at Carcinogenesis Online) were used as the source of kinase, the interaction between BLM and 53BP1 was no longer enhanced (Figure 2F, lane 3). We next wanted to determine whether Chk1-mediated phosphorylation in the KBD affected 53BP1 foci formation. Absence of Chk1 caused complete abrogation of 53BP1 foci formation during replication stress (Figure 2G). Together these results indicate that Chk1-mediated phosphorylation of the intact KBD of 53BP1 is required for its interaction with BLM and for the accumulation of 53BP1 at the site of stalled replication foci.

ATR-mediated phosphorylation of BLM regulates its interaction with 53BP1

To determine the domains of BLM, which mediated its interaction with 53BP1, interaction studies were carried out with equivalent amounts of BLM fragments (supplementary Figure S2B is available at Carcinogenesis Online; Figure 3A, left) and in vitro translated S³⁵-labelled full-length 53BP1. The N-terminal region (1–212) of BLM was the sole interacting domain (Figure 3A, right). Presence of RNase or DNase during the interaction assay had no effect (data not shown). ATR and ATM phosphorylates BLM at Thr99/Thr122 after replication arrest and IR, respectively (16,17). Using [γ]P³²ATP, we confirmed that the N-terminal region of BLM (1–212) is indeed phosphorylated during replication stress (Figure 3B) and such a phosphorylation enhances its interaction with 53BP1 (Figure 3C, lane 2). However, the enhancement of BLM–53BP1 interaction was dependent on the presence of ATR and not ATM (Figure 3C, lanes 3 and 4; supplementary Figure S3 is available at Carcinogenesis Online). Next, we wanted to determine whether phosphorylation at Thr99 and/or Thr122 in BLM affected its interaction with 53BP1. We over-expressed the Flag-tagged BLM and phosphorylated mutants in NHF cells and then treated with HU. Interaction with endogenous 53BP1 occurred only for the full-length BLM and the Thr122 mutant (Figure 3D). Consequently, BLM phosphorylated at Thr99 co-localized extensively with 53BP1 during replication arrest. Lack of ATR decreased the formation of Thr99-phosphorylated BLM (Figure 3E). Decrease in the formation of 53BP1 foci is likely because ATR is an upstream regulator of Chk1 (29), and Chk1 regulated 53BP1 foci formation (Figure 2G).

Fig. 3.

ATR-mediated phosphorylation of BLM at Thr99 regulates its interaction and colocalization with 53BP1. (A) N-terminal region of BLM interacts with 53BP1. Expression levels of GST BLM fragments (visualized by Coomasie) (left). Four GST-tagged fragments of BLM or GST alone were incubated with in vitro translated S³⁵-radiolabelled 53BP1. Input is 10% of the sample used for interaction (right). (B) GST-BLM (1–212) is phosphorylated during replication stress. Modified kinase assays were carried out with the GST-BLM (1–212) or GST alone in presence of 100 ng of nuclear extract isolated from −HU- or +HU-treated NHF (as the source of the kinase) in the presence of [γ]P³²ATP. After sodium dodecyl sulphate–polyacrylamide gel electrophoresis, the proteins were either visualized by Coomasie (left) or transferred to the nitrocellulose membrane and autoradiographed. * indicates nuclear proteins that undergo autophosphorylation. The amount of GST in Coomasie (left) is 10% of the amount used for kinase assay (right). (C) Interaction between 53BP1 and GST-BLM (1–212) depends on ATR-mediated phosphorylation. Kinase assay of GST-BLM (1–212) was done as in (B) except cold ATP was used. Nuclear extract from ATR or ATM siRNA-transfected cells was also utilized in parallel. Interactions were done between in vitro translated S³⁵-radiolabelled 53BP1 and GST-BLM (1–212) fragment, phosphorylated with nuclear extract (NE) from different conditions. (D) Phosphorylation at Thr99 in BLM enhances interaction between helicase and 53BP1 in vivo. Flag-tagged full-length BLM and its mutants were expressed in NHF and subjected to HU treatment. IP was done with anti-53BP1 antibody or its corresponding immunoglobulin G and the interaction checked with Flag antibody to detect exogenous BLM. (E) BLM phosphorylated at Thr99 co-localizes with 53BP1. NHFs (treated either with control siRNA or with ATR siRNA) were treated with HU for 16 h. IF was carried out with anti-BLM Thr99/anti-53BP1 antibodies. Bars, 5 μm.

Chk1-mediated phosphorylation of both RAD51 and 53BP1 is required for their interaction

To demonstrate and characterize the direct interaction between RAD51 and 53BP1, in vitro interaction studies were carried out between S³⁵-labelled 53BP1 and the fragments spanning RAD51 (supplementary Figure S2C is available at Carcinogenesis Online; Figure 4A, left). The C-terminal region of RAD51 encoding two Walker motifs interacted with 53BP1 in a nucleic acid-independent manner (Figure 4A, right, and data not shown). It has been recently demonstrated that RAD51 interacts with and is phosphorylated by Chk1 at Thr309 during HU treatment (4). We confirmed the phosphorylation of full-length RAD51 during replication arrest (Figure 4B). In vitro interaction with nuclear extracts from NHF indicated that 53BP1–RAD51 interaction was enhanced in a Chk1-dependent manner (Figure 4C; supplementary Figure S3 is available at Carcinogenesis Online).

Fig. 4.

Chk1-mediated phosphorylation of both 53BP1 and RAD51 mediates their mutual interaction. (A) C-terminal region of RAD51 interacts with 53BP1. Expression levels of GST RAD51 fragments (visualized by Coomasie) (left). The interactions were done between full-length RAD51 (1–339), the two RAD51 fragments or GST and in vitro translated S³⁵-radiolabelled 53BP1. Input is 5% of the sample used for interaction (right). (B) GST RAD51 (1–339) is phosphorylated during replication stress. Modified kinase assays were carried out with the GST RAD51 (1–339) or GST alone in presence of 100 ng of nuclear extract isolated from −HU- or +HU-treated NHF (as the source of the kinase) in the presence of [γ]P³²ATP. After sodium dodecyl sulphate–polyacrylamide gel electrophoresis, the proteins were either visualized by Coomasie (left) or transferred to the nitrocellulose membrane and autoradiographed. * indicates nuclear proteins that undergo autophosphorylation. The amount of GST in Coomasie (left) is 10% of the amount used for kinase assay (right). (C) Interaction between 53BP1 and GST RAD51 (1–339) depends on Chk1-mediated phosphorylation. Kinase assay of GST RAD51 (1–339) was done as in (B) except cold ATP was used. Nuclear extract from Chk1 siRNA-transfected cells was also utilized in parallel. Interactions were done between in vitro translated S³⁵-radiolabelled 53BP1 and GST RAD51 (1–339) fragment, phosphorylated with nuclear extract (NE) from different conditions. (D) RAD51 interacts with the KBD of 53BP1 in vitro. The interactions were done between the six 53BP1 fragments or GST and in vitro translated S³⁵-radiolabelled RAD51. Input is 5% of the sample used for interaction. (E) RAD51 interacts with the KBD of 53BP1 in vivo. HA-tagged WT or ΔKBD 53BP1 were transfected into NHF and subjected to HU treatment. IP was done with anti-RAD51 antibody or its corresponding immunoglobulin G and the interaction checked with HA antibody to detect exogenous 53BP1. (F) Interaction between RAD51 and GST-53BP1 (1330–1664) depends on Chk1-mediated phosphorylation. Kinase assay of GST-53BP1 (1330–1664) was done as in (B) except cold ATP was used. Nuclear extract from Chk1 siRNA-transfected cells was also utilized in parallel. Interactions were done between in vitro translated S³⁵-radiolabelled RAD51 and GST-53BP1 (1330–1664) fragment, phosphorylated with NE from different conditions.

Using purified proteins (Figure 2A, left), we determined in vitro that the KBD of 53BP1 interacted with RAD51, though a minor interacting region in the extreme N-terminus of 53BP1 was also detected (Figure 4D). We found that WT 53BP1 but not the ΔKBD mutant was capable of binding with endogenous RAD51 in vivo (Figure 4E). This interaction was again enhanced by Chk1-dependent phosphorylation of the KBD of 53BP1 during stalling of replication forks (Figure 4F; supplementary Figure S3 is available at Carcinogenesis Online). Together these results indicated that the interactions between BLM, RAD51 and 53BP1 are all dependent on specific phosphorylation events during replication arrest.

Interaction between 53BP1 and RAD51 is enhanced by BLM

Next, we asked the question: is BLM, in any way, affecting the interaction between 53BP1 and RAD51? To test this possibility, we used nuclear extracts from A-15/BS isogenic cell lines. Reciprocal IPs with either RAD51 (Figure 5A) or 53BP1 (Figure 5B) antibody revealed that the 53BP1–RAD51 complex was enhanced in corrected A-15 cells (Figure 5A and B). IF data from the BS/A-15 cells (Figure 5C and D; supplementary Figure S4 is available at Carcinogenesis Online) also revealed higher level of 53BP1/RAD51 co-localization in HU-treated A-15 cells. However, residual 53BP1/RAD51 co-localization was detected in PW, indicative of BLM-independent basal level of interaction between RAD51 and 53BP1 (Figure 5C and D; supplementary Figure S4 is available at Carcinogenesis Online). These data indicate that BLM helps in the transport of 53BP1 to the RAD51 foci—thereby providing evidence for BLM-dependent role of 53BP1 during HR.

Fig. 5.

BLM enhances the interaction and co-localization between 53BP1 and RAD51 during replication arrest. (A and B) Presence of BLM enhanced in vivo interaction between RAD51 and 53BP1. IPs were carried out with (A) anti-53BP1 antibody and corresponding immunoglobulin G (B) anti-RAD51 antibody and corresponding immunoglobulin G antibodies. The IPs were probed with self-antibody (to check for efficiency) or with antibodies to BLM, 53BP1 and RAD51 (to check for interactions). (C) BLM enhances the co-localization between 53BP1 and RAD51 during replication stress. BS and A-15 cells were treated as in (A and B). IF was performed with anti-53BP1/anti-RAD51 antibodies. Bars, 5 μm. (D) Quantitation of C.

Phosphorylation of BLM and 53BP1 regulates their anti-recombinogenic role

To dissect how phosphorylations mediated the anti-recombinogenic functions of 53BP1 and BLM in vivo, we carried out host-cell reactivation assays in NHF after HU treatment (24). The substrate (pBHRF) encoded an intact, emission shifted, blue variant of GFP (BFP), with a 300 nucleotide stretch of homology to a non-functional copy of GFP. In the absence of HR, only BFP was present while HR can also create a functional GFP. Over-expression of RAD51 in NHF cells led to increase in HR (Figure 6A). However, over-expression of WT 53BP1, but not the ΔKBD mutant, led to a 4- to 5-fold decrease in the rate of HR, indicating an anti-recombinogenic role for 53BP1, which was dependent on the integrity of the KBD. It should be noted that the data set used in this assay for RAD51 and 53BP1 is same as in ref. (6). Absence of methylation (in 53BP1 5K mutant) does not affect the anti-recombinogenic role of 53BP1, whereas lack of dimerization (in Δ1231–1270 mutant) partially relieved the anti-recombinogenic function. Interestingly, the over-expression of BLM (T99A) mutant led enhancement in the rate of recombination compared with the substrate alone. These results indicate that the phosphorylable WT BLM and the entire KBD of 53BP1 were essential for both the proteins to attain their full anti-recombinogenic potential.

Fig. 6.

Mutants affecting phosphorylation of 53BP1 and BLM negatively regulate RAD51 function during HR. (A) Over-expression of WT BLM and 53BP1, but not mutants affecting phosphorylation, decreases HR. NHF cells were used for the experiment. pBHRF was either transfected alone or co-transfected with other expression vectors, as indicated. Post-transfection and before harvesting, the cells were treated with HU for 16 h. Cells were subsequently analysed by BD LSR flow cytometer. The fold difference of the GFP/BFP ratio is represented. The data set used for RAD51 and 53BP1 is same as in Figure 2C of ref. (6). (B) Increase of RAD51 foci due to loss of BLM and 53BP1 is reversed by WT proteins but not by phosphorylation-deficient mutants. A-15, A-15 shRNA53BP1 (Clone 1), BS, BS shRNA53BP1 (Clone 19) cells were treated with HU for 16 h. In parallel, BS shRNA53BP1 (Clone 19) cells were transfected with WT BLM, BLM (T99A), 53BP1* or 53BP1* (ΔKBD) mutants and subsequently treated with HU for 16 h. Transfection efficiency of the constructs in BS shRNA53BP1 (Clone 19) cells was normalized. IF carried out with anti-RAD51 antibodies. DNA was visualized with 4′,6-diamidino-2-phenylindole. Bars, 5 μm. (C) Quantitation of (B).

We next wanted to determine whether phosphorylation of BLM and 53BP1 affects RAD51 function. We have recently demonstrated that over-expression of WT BLM or 53BP1* (a silently mutated 53BP1 cDNA immune to RNAi effect) in BS shRNA 53BP1 cells decreased the number of cells with complete RAD51 structures [(6); supplementary Figure S5A and S5B is available at Carcinogenesis Online]. Very few cells had RAD51 structures when both 53BP1* and BLM were over-expressed. Next, we wanted to determine whether phosphorylation of BLM and 53BP1 could affect the formation of RAD51 structures. When 53BP1*ΔKBD or BLM (T99A) was over-expressed alone in HU-treated BS shRNA 53BP1 cells, the level of disruption of RAD51 structures was lower. Over-expression of 53BP1*ΔKBD with BLM Thr99 mutant completely reversed the disruption of structures and complete RAD51 polymers were re-observed (supplementary Figure S5A and S5B is available at Carcinogenesis Online).

Since the above assay was based on over-expression of proteins, we wanted to understand whether phosphorylation of BLM and 53BP1 also affects the formation of endogenous RAD51 foci in vivo. For our study, we initially used the four isogenic cell lines generated by us (6) at similar stages of the cell cycle after HU treatment (data not shown). Compared with A-15 cells (expressing both BLM and 53BP1), loss of either BLM (in BS cells) or 53BP1 (A-15 shRNA53BP1 cells) caused an increase in the number of endogenous RAD51 foci, with higher increase observed for loss of BLM alone (Figure 6B and C). Loss of both BLM and 53BP1 (in BS shRNA53BP1 cells) resulted in a further increase in the number of RAD51 foci, thereby confirming that the BLM-independent component of the anti-recombinogenic role of 53BP1 (6). The increase in the number of endogenous RAD51 foci due to loss of BLM and/or 53BP1 was reversible when WT BLM or 53BP1* but not BLM (T99A) and 53BP1* (ΔKBD) mutants were over-expressed in BS shRNA53BP1 cells (Figure 6B and C). These results indicate that in vivo phosphorylation of both BLM (by ATR) and 53BP1 (by Chk1) during stalled replication affects the formation of RAD51 foci formation and thereby negatively regulate HR.

Discussion

In this report, we have demonstrated that phosphorylation-dependent associations between BLM, 53BP1 and RAD51 were essential for their respective roles in the regulation of HR. Consequently, loss of 53BP1 and BLM resulted in a synergistic increase in the level of replication stress-induced HR (6). Our results also provide mechanistic evidence that both BLM-dependent and independent roles of 53BP1 are essential for the latter to attain full anti-recombinogenic function.

The physical interaction of 53BP1 with both BLM and RAD51 was mediated via its Chk1-phosphorylated KBD (Figures 2 and 4). Moreover, the recruitment of 53BP1 to the sites of stalled replication also depended on Chk1-mediated phosphorylation (Figure 2G). Consequently, absence of phosphorylation in KBD affected the role of 53BP1 as an anti-recombinogenic protein in HR assays (Figure 6A) and during regulation of RAD51 foci formation (Figure 6B and C). Incidentally, phosphorylation of RAD51 by Chk1 at Thr309 is required for its (RAD51's) recruitment to the site of DNA damage after HU exposure (4). Methylated lysine residues in histones (K20 in histone H4 and K79 in histone H3) modulate the accessibility of 53BP1 to the chromatin (30,31). Methylation of lysine residues on the histones may coordinate with phosphorylation events in the KBD of 53BP1, thereby regulating the access of 53BP1 to the chromatin where it can control HR by modulating RAD51 foci formation. It has been recently reported that a histone-independent pathway for the recruitment of Crb2 (53BP1 in fission yeast) exists and it involves a phosphorylation-mediated interaction with another breast cancer gene 1 C-terminal protein Cut5 (TopBP1 homolog) (32). Such phosphorylation-dependent mechanisms may also exist in human cells.

Like 53BP1, phosphorylation also affects the anti-recombinogenic role of BLM. Thr99 and Thr122 of BLM are known to be phosphorylated by both ATR during replication stress (16) and ATM during IR (17). Classically it has been thought that ATM is activated by double-strand breaks, whereas ATR is recruited to single-stranded regions of DNA. However, recent studies have demonstrated that ATM functions upstream of ATR following exposure to IR (33,34). Moreover, it has also been reported that ATR-dependent phosphorylation and activation of ATM occur in response to UV treatment or replication fork stalling (35). Such interplay between the ATR and ATM pathways raises the question about the identity phosphatidyl inositol-like kinase, which actually phosphorylates BLM under different conditions. Indeed, ATM and ATR, but not DNA-PK, appear to play a redundant role in phosphorylating BLM during camptothecin treatment (36). We have conclusively demonstrated that ATR-dependent phosphorylation on Thr99 of BLM during replication arrest led to an increase in its interaction and co-localization with 53BP1 (Figure 3). Enhancement of interaction between 53BP1 and RAD51 was dependent on the BLM (Figure 5). Finally, loss of Thr99 phosphorylation affected BLM function as an anti-recombinogenic protein in HR assays (Figure 6A) and did not allow the decrease in the number of RAD51 foci as observed by its WT counterpart (Figure 6B and C).

An earlier study has reported that the lack of phosphorylation in Th99/Thr122 did not prevent BLM from suppressing HR or co-localizing with other replication and repair factors (16). However, it should be noted that in the above study, the RAD51/BLM co-localization and Sister Chromatid Exchange experiments were done with Th99/Thr122 double mutant and that too in absence of replication stress. Extrapolating results from such double-mutant studies can be misleading as phenotypes of single mutants can be suppressed during the process. It has also been reported that BLM leaves PML NBs following HU treatment and co-localizes to the sites of stalled replication. ATR is necessary for the rapid movement of BLM to damaged replication forks (37). We also demonstrate similar dynamics of BLM relocalization during and after replication arrest (supplementary Figure S1 and data not shown, available at Carcinogenesis Online). Once recruited, BLM maybe involved in the disruption of the RAD51 filaments via a translocase activity, as in yeast DNA helicase, Srs2 (38). Alternately, the proposed translocase activity in BLM may also require phosphorylation at Thr99.

In conclusion, we have dissected the biochemical parameters required by three key members of DNA damage signal transduction and repair machinery (53BP1, BLM and RAD51) to functionally interact during replication arrest. We have provided evidence that phosphorylations of BLM, 53BP1 and RAD51 are critical requirements for their respective functions. Using non-binding and phosphorylation-deficient mutants of BLM and 53BP1, we demonstrate how RAD51 function and correspondingly HR is affected. Our results also provide mechanistic confirmation about both BLM-dependent and independent roles of 53BP1 during HR. These effects probably act via a concerted mechanism—BLM itself being brought to the site of stalled replication by 53BP1 (11), BLM in turn enhanced the interaction of 53BP1 with RAD51 (Figure 5; supplementary Figure S4 is available at Carcinogenesis Online) and once at the site of DNA damage both BLM and 53BP1 modulated RAD51 function [Figure 6 and ref. (6)]. Hence, these data provide in vivo and biochemical evidence indicating an intricate multi-regulatory network of inter-regulation between BLM and 53BP1 and demonstrate how combinatorial functions of different signal transducer proteins can happen in response to DNA damage in human cells.

Abbreviations

ATP

adenosine triphosphate

ATR

ataxia telangiectasia and Rad3 related

ATM

ataxia telangiectasia mutated

BFP

blue fluorescent protein

BLM

bloom helicase protein

BS

Bloom syndrome

cDNA

complementary DNA

Chk

checkpoint kinase

GAR

glycine–arginine rich

GFP

green fluorescent protein

GST

glutathione S-transferase

HR

homologous recombination

HU

hydroxyurea

IF

immunofluorescence

IP

immunoprecipitation

IR

ionizing radiation

KBD

kinetochore-binding domain

NB

nuclear body

NHF

normal human fibroblast

PML

promyelocytic leukemia

PW

post-wash

WT

wild-type