BRCA1 and CtIP Are Both Required to Recruit Dna2 at Double-Strand Breaks in Homologous Recombination

Homologous recombination plays a key role in the repair of double-strand breaks (DSBs), and thereby significantly contributes to cellular tolerance to radiotherapy and some chemotherapy. DSB repair by homologous recombination is initiated by 5’ to 3’ strand resection (DSB resection), with nucleases generating the 3’ single-strand DNA (3’ssDNA) at DSB sites. Genetic studies of Saccharomyces cerevisiae demonstrate a two-step DSB resection, wherein CtIP and Mre11 nucleases carry out short-range DSB resection followed by long-range DSB resection done by Dna2 and Exo1 nucleases. Recent studies indicate that CtIP contributes to DSB resection through its non-catalytic role but not as a nuclease. However, it remains elusive how CtIP contributes to DSB resection. To explore the non-catalytic role, we examined the dynamics of Dna2 by developing an immuno-cytochemical method to detect ionizing-radiation (IR)-induced Dna2-subnuclear-focus formation at DSB sites in chicken DT40 and human cell lines. Ionizing-radiation induced Dna2 foci only in wild-type cells, but not in Dna2 depleted cells, with the number of foci reaching its maximum at 30 minutes and being hardly detectable at 120 minutes after IR. Induced foci were detectable in cells in the G2 phase but not in the G1 phase. These observations suggest that Dna2 foci represent the recruitment of Dna2 to DSB sites for DSB resection. Importantly, the depletion of CtIP inhibited the recruitment of Dna2 to DSB sites in both human cells and chicken DT40 cells. Likewise, a defect in breast cancer 1 (BRCA1), which physically interacts with CtIP and contributes to DSB resection, also inhibited the recruitment of Dna2. Moreover, CtIP physically associates with Dna2, and the association is enhanced by IR. We conclude that BRCA1 and CtIP contribute to DSB resection by recruiting Dna2 to damage sites, thus ensuring the robust DSB resection necessary for efficient homologous recombination.

Introduction streptomycin (Nacalai Tesque). Doxycycline was added at a final concentration of 100 ng/ml to inactivate the expression of the tetracyclin repressible promoter.

Plasmid Constructs and Transfection
All primers used for plasmid construction are shown in Table 1. To generate a DNA2 knockout construct, genomic DNA was amplified with the following primers: F1 and R1 (2.9 kb) for the left arm and F2 and R2 (2.7 kb) for the right arm. Each product cloned was combined on pBLUESCRIPT-SK (pBS-SK) plasmid, after which the histidinol dehydrogenase (hisD) gene was introduced with BamHI. To generate the D245A (aspartic acid replaced by alanine at position 245) knockin construct, the mutation was introduced by polymerase chain reaction (PCR) using two sets of primers: F1, F3, R3, and R1. The knockout and knockin constructs were digested with XhoI for linearization. Linearized plasmid (30 μg) was transfected into 10 7 cells. The genomic DNA was digested with StuI and BamHI for Southern blotting. 0.6 kb DNA fragment amplified with F4 and R4 from genomic DNA was used as probes for Southern blotting. The expression of DNA2 -/D245A transcript was confirmed by RT-PCR using F5 and R5. Each amplicon, 430 bp for D245A fragment, was digested using BstUI. To construct a plasmid for the inducible expression of DNA2, cDNA2 was cloned downstream from the tetracycline promoter with SalI and BamHI. Transfection for chicken DT40 cells was performed as previously described [12].
To generate a CtIP knockout construct, genomic DNA was amplified with F6 and R6 for the left arm, which was cloned in pBS-SK with NotI and BamHI. To generate the N183A,R187A knockin construct, the mutation was introduced by PCR using two sets of primers: F7, F8, R7, and R8 and the product was cloned into pBS-SK with BamHI and XhoI. The construct was digested with XhoI for linearization. The expression of CtIP N183A,R187A/-/transcript was amplified by RT-PCR using F9 and R9 and digested with BfaI.

Cell Counting, Cell Cycle Analysis, and Cell Synchronization
Cells were counted and cell-cycle analysis was carried out using a flow cytometer (Becton Dickinson, US). Immediately before being counted, cells were mixed with reference beads (Polyscience Ins., US) and propidium iodide. Cells were treated with BrdU (final concentration 20 μM) for 10 minutes and then fixed with 70% EtOH for the cell-cycle analysis. Cells were then incubated with FITC-conjugated anti-BrdU (1/1000, Pharmingen, US) and propidium iodide. Fluorescence data were displayed as dot plots using Cell Quest and Diva software. HeLa cells were synchronized using a double thymidine block as described previously [23]. In brief, cells were incubated with thymidine (final concentration 2 mM) for 18 hours, followed by washing with PBS and then releasing cells into fresh medium without thymidine. After incubating for 9 hours to release, cells were again incubated with 2 mM thymidine to arrest at early S-phase for 17 hours, followed by releasing into fresh medium at the time zero. Cells were harvested after 8 hour incubation to obtain G2 fraction.
siRNA Knockdown Assay siRNA oligos were purchased from QIAGEN (CtIP) and Dharmacom (set of 4, ON-TARGETplus, DNA2). Transfection was done using Lipofectamine 2000 (Life Technologies) according to the attached protocol. After transfection, cells were cultured in fresh media for 48 hours prior to being harvested for Western blotting and immunostaining assays.

Chromosome Aberration Analysis
Cells were treated with 0.1 μg/ml colcemid (Invitrogen, Carlsbad, CA) for 3 hours. Cells were suspended in 75mM potassium chloride for 15 minutes, washed with Carnoy's solution (a 3:1 mixture of methanol and acetic acid), dropped on slides, and stained with a 5% Giemsa solution for 10 minutes.

Protein Purification and In Vitro Association of Human CtIP and Dna2
Human CtIP purification from insect cells: Cells were lysed in buffer A (20 mM Tris pH 8.0, 100 mM NaCl, 10% glycerol, and 1 mM DTT) supplemented with 0.5% Triton, 2.5 mM Na 4 P 2 O 7 , 1 mM Glycerol-ß-phosphate, 5 mM PMSF. Lysate was sonicated on ice until homogenous and insoluble material removed by centrifugation (100,000xg) for one hour. The supernatant was bound to 2 ml of anti-Flag M2 agarose resin slurry (Sigma) at 4°C for one hour followed by washing on a column with 20 ml buffer A, 10 ml 0.5 M LiCl, and 20 ml buffer A. The protein was eluted with 10 ml buffer A containing Flag peptide (Sigma) at a concentration of 100 μg/ml.
Eluted protein was loaded onto a 1 ml Hitrap SP column (GE), washed with 20 ml buffer A and eluted with buffer A containing 0.6 M NaCl. The peak fractions were dialyzed twice against fresh 300 ml buffer A for one hour and flash frozen in liquid nitrogen. Dna2 protein was expressed in Sf21 insect cells using the baculovirus expression system. Dna2 expressing cells were lysed by homogenization and sonicated three times for 20 seconds in A buffer containing 0.5% tween-20 and 1mM PMSF (phenylmethylsulfonyl fluoride). The lysate was centrifuged for 1 hour at 35,000 rpm at 4°C. The supernatant was incubated with~1ml M2 anti-Flag antibodyconjugated agarose resin (Sigma) with rotation at 4°C for 1 hour. After incubation the lysate with resin was centrifuged for 3 min at 1000g. After removing the supernatant, the remaining resin was washed with 20 ml of A buffer twice and was eluted with 5 ml of A buffer containing 0.8mg/ml 3X Flag peptide (Sigma). The peptide was incubated with the resin for 20 min before elution. The Flag eluent was then loaded onto 1 ml HiTrap SP column (G.E.) and washed with buffer A then eluted with buffer A containing 500 mM NaCl. The eluted protein fractions were dialyzed in A buffer and the dialyzed fractions were aliquoted, frozen in liquid nitrogen, and stored at -80°C. Purified recombinant CtIP and Dna2 were incubated on ice for 1 hour in a 100 μl reaction containing buffer A. After incubation, CtIP was immunoprecipitated using 5 μg of anti-CtIP antibody (Bethyl) and 10 μl of Protein A resin (Sigma). The beads was washed 3 times with buffer A and separated on a 6% SDS-PAGE. The gel was transferred onto a PVDF membrane, which was blotted against anti-Dna2 (Pierce: PA5-23691), and further visualized with horseradish peroxidase-conjugated anti-rabbit secondary antibody by chemiluminescence (Pierce).

Dna2 Nuclease Activity is Essential for HR
The chicken DNA2 gene is located on chromosome VII (18 exons;~14.5 kb) and encodes a 992 amino-acid protein.
The D245A substitution of chicken Dna2 corresponds to the D294A substitution of human Dna2, which mutation completely abolishes its nuclease activity [24]. We confirmed the expression of DNA2 D245A mRNA but not of wild-type DNA2 mRNA at 48 hours after doxycycline treatment (S1E Fig To investigate the cause of cell mortality, we conducted a chromosomal analysis of mitotic cells. We classified chromosomal aberrations into three categories: exchange, isochromatidtype, and chromatid-type breaks (S2B Fig). The DNA2 -/and DNA2 D245A/mutants displayed very similar increases in the number of chromosomal aberrations in mitotic chromosome spreads at 72 hours after doxycycline treatment (Fig 1C, 1D and 1E). This phenotype is reminiscent of that seen in cells deficient in HR-related genes such as MRE11, CtIP, and RAD51, suggesting that Dna2 plays an essential role in HR, as do Mre11, CtIP, and Rad51 [12,13,25].
To examine the role of Dna2 in HR-dependent DSB repair, we exposed the DNA2 mutants to 2 Gy IR at 24 hours after addition of doxycycline, when the expression of the tet-DNA2 transgene was undetectable ( Fig 1A, lower panels) and Dna2-depleted cells proliferated normally. We then enriched the mitotic cells by adding colcemid, harvesting the cells 3 hours later. This protocol allows for the selective evaluation of the HR efficiency of IR-induced DSBs during the G 2 phase, where HR predominates in chicken DT40 cells [13,26]. As with the HRdeficient RAD54 -/cells, the number of chromosomal breaks in the two DNA2 mutants (null and nuclease-dead) was a few times greater than in the wild-type cells (Fig 2A), indicating that Dna2 plays an important role in the HR-dependent repair of IR-induced DSBs. Similarly, mitotic chromosomal breaks induced by olaparib, a PARP inhibitor, were four to five times higher in the two DNA2 mutants than in the DNA2 +/cells (Fig 2B). We conclude that the nuclease activity of Dna2 is essential for HR-dependent DSB repair in chicken DT40 cells.
To evaluate Dna2-dependent DSB resection, we monitored IR-induced Rad51-focus formation over time. In wild-type and DNA2 +/cells, the number of Rad51 foci reached its maximum (8.4 per cell) at one hour after IR (Fig 2C and S3A Fig). The null and nuclease-dead DNA2 mutants showed a very severe defect in Rad51-focus formation (Fig 2C). Thus we conclude that the nuclease activity of Dna2 plays an essential role in DSB resection in chicken DT40 cells.

Epistatic Relationship Between CtIP and Dna2 in DSB Resection
We next performed siRNA-mediated depletion of Dna2 in human U2OS cells (Fig 3A). The depletion of Dna2 resulted in a moderate accumulation of cells in the late S and G 2 phases (S3B Fig) and caused a significant decrease in the number of IR-induced Rad51 foci (p value < 0.001; Fig  3B), as did the inactivation of Dna2 in chicken DT40 cells (Fig 2C). We obtained very similar results when we examined Rad51-focus formation in cyclin-A-positive S and G 2 -phase cells ( Fig  3C and 3D). Likewise, depletion of CtIP in human U2OS cells (Fig 3A) attenuated Rad51-focus formation (Fig 3C and 3D), as did the inactivation of CtIP in chicken DT40 cells [11,12,14]. The strong defect in IR-induced Rad51-focus formation in both the CtIP and Dna2 mutants suggests that collaboration between the two resecting enzymes is required for robust DSB resection.
To investigate the functional relationship between CtIP and Dna2, we co-depleted them and subsequently measured Rad51-focus formation. Co-depletion reduced the two proteins to the same degree as did the single depletion of either Dna2 or CtIP (Fig 3A). Remarkably, depletion of both Dna2 and CtIP resulted in a phenotype equivalent to that caused by the depletion of either single gene, in terms of both percentage of Rad51-foci-positive cells and the number of Rad51 foci (Fig 3C and 3D). This implies that the roles of CtIP and Dna2 in DSB resection may be interdependent. To further investigate this interdependent relationship, we examined subnuclear RPA foci as well as the phosphorylation of RPA32 (pS4/8) following IR treatment. Note that the phosphorylation reflects DSB resection [27]. Co-depletion of both CtIP and Dna2 reduced the number of RPA foci to the same extent as did the depletion of either CtIP or Dna2 (Fig 3E and S3D Fig). Likewise, the level of phosphorylated RPA32 was significantly reduced by the depletion of either CtIP or Dna2, whereas co-depletion of both CtIP and Dna2 did not reduce RPA phosphorylation any more than did the single depletion of either CtIP or Dna2 (Fig 3F, compare lanes 3 and 4 with lane 5). We therefore conclude that there may be a mutual dependency between CtIP and Dna2 in DSB resection.

BRCA1 and CtIP are Required for Recruitment of Dna2 to DNA Damage Sites
A simple interpretation of this interdependency is that CtIP and Dna2 function together spatiotemporally at DSB sites. To examine the physical interaction between CtIP and Dna2, we immuno-precipitated the Dna2 protein of whole-cell extract prepared from the human U2OS and Mre11-deficient ATLD2 cell lines [28] (Fig 4A, left panels). In U2OS cells, CtIP protein co-immunoprecipitated with Dna2 even in the presence of ethidium bromide (Fig 4A, right panels, and Fig 4B), suggesting that CtIP does interact with Dna2. Next, we explored the involvement of Mre11 in this interaction, since CtIP physically and functionally interacts with the MRN complex [11]. The extract prepared from Mre11-deficient ATLD2 cell line shows the immuno-precipitation comparable to that from the U2OS cell line (Fig 4A, right panels), suggesting that CtIP forms a robust complex with Dna2 even in the absence of Mre11. This idea is supported by the data that purified recombinant CtIP can physically associate with Dna2 (S2D Fig, and Fig 4C). Importantly, the interaction between CtIP and Dna2 is significantly enhanced by IR (Fig 4A, right panels). We conclude that this interdependency may be based on the physical interaction between CtIP and Dna2. This led us to hypothesize that CtIP is required for the recruitment of Dna2 to DSB sites. To test this hypothesis, we examined the recruitment of Dna2 onto IR-induced DSB sites. We checked whether or not any commercially available anti-human-Dna2 antibody could be used to immuno-cytochemically stain Dna2 at DSB sites in γ-ray-irradiated chicken DT40 as well as human cells. We successfully detected Dna2 subnuclear foci in the DT40 cells only after IR with Dna2-focus formation reaching its maximum at 30 minutes and being hardly detectable at 120 minutes after irradiation (Fig 5A and S3C Fig). Dna2 foci were detectable also in γ-irradiated human U2OS and HeLa cells reaching maximal numbers at one hour following irradiation ( Fig 5B, and S3E Fig). The absence of discrete subnuclear Dna2 foci in a Dna2-deficient background in both human and chicken cells confirmed the specificity of the immunostaining (Fig 5A, 5B and S3C Fig). IR-induced Dna2 foci were observed only in human cells synchronized in the G 2 phase, but not in the G 1 phase (Fig 5C, 5D and 5E). Taken together, we conclude that this novel immunostaining allows for the reliable detection of Dna2 localizing at DSB sites in both chicken and human cells.
To test whether CtIP is required to recruit Dna2 to DSB sites, we examined Dna2 foci in CtIP-depleted CtIP -/-/-/tet-CtIP cells when they were still capable of proliferating with normal kinetics. Remarkably, Dna2-focus formation was significantly impaired in the chicken CtIP-depleted cells (Fig 6A, 6B and S3C Fig). Consistently, CtIP depletion by siRNA in human U2OS cells also showed a significant reduction of Dna2 foci (Fig 6C). We therefore conclude that CtIP is required to load Dna2 onto DSB sites. Next, since CtIP physically associates with BRCA1 [29,30], we tested whether or not BRCA1 is also required to load Dna2 onto DSB sites. Dna2-focus formation was compromised in BRCA1 -/-DT40 cells (Fig 6A, 6B and S3C Fig) as well as in human cells deficient in BRCA1 (Fig 6D). We analyzed BRCA1 -/-/53BP1 -/-DT40 cells, since defective Rad51-focus formation in BRCA1-deficient mice is significantly reversed by the additional inactivation of 53BP1 [2,19]. The BRCA1 -/-/53BP1 -/cells showed a higher number of Dna2 foci than did the BRCA1 -/cells (Fig 6A, 6B and S3C Fig), indicating that BRCA1 and 53BP1 control DSB resection by regulating the recruitment of Dna2 to DSB sites. In summary, BRCA1 and CtIP are required for the efficient loading of Dna2 onto DSB sites.
Recent studies indicate that CtIP contributes to DSB resection through its non-catalytic role but not as a nuclease [15,16]. To exclude catalytic role in the recruitment of Dna2 to DSB sites, we generated nuclease-dead CtIP mutant cells by inserting N183A and R187A mutations into one of the three allelic CtIP gene in DT40 cells [12] (S4A-S4C Fig). The resulting CtIP N183A, R187A/-/cells were more sensitive to topoisomerase poisons, camptothecin and etoposide, compared with CtIP +/-/cells (S4D and S4E Fig), as indicated previously [15]. As expected, We therefore conclude that CtIP contributes to the recruitment of Dna2 to DNA damage sites through its non-catalytic role, most likely physical interaction with Dna2, but not as a nuclease.

Discussion
Our current study shows that the immuno-cytochemical analysis of Dna2 allows for the detection of Dna2 localizing at DNA-damage sites (Fig 5). Dna2-focus formation reached its maximum quickly, at 30 minutes post-IR, and hardly detectable at 120 minutes ( Fig 5A). The following evidences suggest that Dna2 foci form at the DSBs that are subjected to Dna2mediated resection. First, Dna2 foci were detectable only Dna2 proficient cells but not deficient ones (Fig 5A, 5B and S3 Fig). Second, Dna2 foci were undetectable in the G 1 phase, where no DSB resection occurs (Fig 5C and 5E). Third, the number of Dna2 foci was similar to that of Rad51 foci, and a few times smaller than that of γH2AX foci (Fig 5C, 6C and S3C Fig). Moreover, Dna2 focus formation preceded Rad51 focus formation, which reaches its maximum at 60 minutes post-IR. Forth, prominent Dna2 focus formation was observed only in the cells that are capable of resecting DSBs, including 53BP1 -/and BRCA1 -/-/53BP1 -/cells but not BRCA1 -/or CtIP -/cells (Fig 6). We therefore propose that the Dna2 focus formation is an excellent biomarker for ongoing DSB resection. We observed that only few fractions of the IR-induced Dna2 foci colocalize with γH2AX and Rad51 foci (Fig 5C, 6C and S3C Fig). Such differential localization is also often seen between IR-induced Rad51 and γH2AX foci [31]. The differential localization between Dna2 and γH2AX foci may reflect spatial separation of resected DNA from γH2AX in the vicinity of individual DSB sites.
The novel assay of detecting Dna2 foci suggests that CtIP is essential for the recruitment of Dna2 to DNA-damage sites. We also show that CtIP and Dna2 function interdependently in DSB resection (Fig 3C, 3D and 3E). We therefore conclude that CtIP-dependent recruitment of Dna2 is likely required for the DSB resection necessary for efficient HR. By contrast, the functionality of Exo1 appears to be independent of CtIP, since co-depletion of CtIP and Exo1 inhibits DSB resection to a higher extent than that depletion of CtIP alone [32]. Collectively, CtIP is essential for the generation of 3' overhangs that are long enough to efficiently initiate HR. We thus propose that CtIP and Dna2 function interdependently of one another for the generation of 3' overhangs that are long enough to efficiently initiate HR.
How does CtIP recruit Dna2 to DNA-damage sites? There are two possible mechanisms underlying the collaboration of the two nucleases. First, CtIP, together with Mre11, might perform short-range resection, with the resulting 3' single-strand tails efficiently recruiting Dna2, just as the yeast MRX complex stimulates Dna2 resection by creating DNA substrates that efficiently associate with Dna2 [33]. Second, CtIP might facilitate the recruitment of Dna2 to DSB sites via a physical interaction between the two proteins. The latter is more likely since two recent manuscripts [15,16] demonstrated the catalytic and non-catalytic roles of CtIP endonuclease, with only the non-catalytic role contributing to DSB resection. Our manuscript now defines the non-catalytic role as the recruitment of Dna2 nuclease, the major resecting nuclease to the site of DSBs (S4 Fig). Both CtIP and Dna2 seem to be subject to stringent control by kinases involved in cell-cycle regulation and the damage-checkpoint response [34][35][36]. The phosphorylation of CtIP and Dna2 might control the interaction between CtIP and Dna2 as well as their enzymatic activity. Moreover, the impaired Dna2-focus formation in BRCA1 -/cells but not in BRCA1 -/-/53BP1 -/cells (Fig 6B) suggests that 53BP1 and BRCA1 may control CtIP and thereby affect DSB resection by Dna2.
Accumulating evidence points to three distinct functions for BRCA1. First, our current result suggests that BRCA1 contributes to DSB resection by promoting the recruitment of Dna2 to DNA-damage sites (Fig 6A, 6B and 6D). Second, the epistatic relationship between BRCA1 and BRCA2 suggests that the two BRCA proteins collaboratively facilitate polymerization of Rad51 at the resected 3' overhangs [2]. Third, BRCA1 promotes the elimination of chemical adducts from DSB ends. This third function is suggested by the fact that a mutation at Ser327 in CtIP (equivalent to Ser322 in chicken CtIP) abolishes the tight association of BRCA1 with CtIP and also sensitizes cells to topoisomerase poisons without impairing DSB resection [12,37]. This third function is also suggested by the finding that RAP80, one of the BRCA1 interacting proteins, contributes to cellular tolerance to the topoisomerase II poison, which generates DSBs covalently associating with polypeptides at their 5' ends [38]. In summary, BRCA1 contributes to cellular tolerance to various chemotherapeutic agents through three distinct functions. (C) RT-PCR using a couple of primers flanking the mutation region was done 96 hours after addition of doxycycline (Dox). The RT-PCR product amplified from the nuclease-dead mutant (CtIP N183AR187A /-/-) mRNA, but not wild-type mRNA, was digested with BfaI. The silent mutation that generates BfaI site was introduced nearby N183AR187A mutation. (D, E, F) The DT40 nuclease-dead CtIP cells are sensitive to epotoside (D) and campothecin (E), but not to IR (F). The survival curves were plotted using colony formation assays. (G) Representative images of Rad51 foci (green) (upper) and the average number of Rad51 foci per cell in indicated genotypes at one hour after 2 Gy γ-rays (bottom). (H) Representative images of Dna2 foci (Red) (upper) and the percentage of cells with at least four Dna2 foci in chicken DT40 clones carrying the indicated genotypes (bottom). Positive cells were counted at 30 minutes following 2 Gy γ-ray irradiation. (EPS)