CtIP-mediated resection is essential for viability and can operate independently of BRCA1

The analysis of CtIP functions in vivo has been hampered by the lethality of CtIP knockout mice, with mutant embryos dying at embryonic day E3.5 (Chen et al., 2005). To circumvent this, we generated a B cell–specific deletion of CtIP by intercrossing CD19^cre mice (Rickert et al., 1997) with mice carrying conditional and null alleles (CtIP^co/−; Chen et al., 2008; Bothmer et al., 2013). The resulting CtIP-null (CtIP^Φ/−) B cells expressed little or no CtIP protein as determined by Western blot analysis (Fig. 1 A). Upon ex vivo stimulation with LPS and IL-4 cells, splenic B cells enter the cell cycle and undergo class switch recombination (CSR). Despite the fact that CtIP-null mice invariably experience early embryonic death (Chen et al., 2005), CtIP^Φ/− B cells were able to proliferate, exhibited a normal cell cycle distribution (Fig. 1 B), and underwent a similar number of cellular divisions as WT cells, as determined by CFSE dye dilution (Fig. 1 C). Moreover, CtIP-deficient B cells have similar level of class switching compared with WT (Bothmer et al., 2013).

Although CtIP deletion in B cells did not lead to a marked defect of cellular proliferation in live cells, we detected heightened phosphorylation of KAP-1 and p53 (Fig. 1 A), indicative of constitutive DNA damage signaling. To determine whether there was a selection against CtIP-deficient cells in culture, we compared the efficiency of CD19-CRE–mediated excision in CtIP^co/− versus CtIP^co/+ B cells by crossing these mice with the reporter line ROSA26-STOP-EYFP, in which YFP expression is induced upon CRE mediated excision. We found that there was a gradual depletion of ROSA26^YFP/+CD19^cre/+ CtIP^co/− but not ROSA26^YFP/+CD19^cre/+CtIP^co/+ B cells after 2 d in culture, indicating that the CtIP deletion confers a growth disadvantage (Fig. 1 D). We therefore restricted our analyses to day 2 stimulated B cells in subsequent experiments to maximize the number of CtIP-deficient and actively cycling cells.

The observed constitutive phosphorylation of KAP-1 and p53 (Fig. 1 A) suggested that loss of CtIP might lead to spontaneous chromosomal damage. To test this, we analyzed metaphase spreads generated from stimulated WT or CtIP^Φ/− B cells. We observed that CtIP-deficient B cells showed spontaneous accumulation of chromosome aberrations (Fig. 1, E and F). Indeed, ∼40% of CtIP^Φ/− B cells exhibited chromosomal aberrations (Fig. 1 F), and on average, there were 1.5 aberrant chromosomes per mutant metaphase, whereas no instability was observed in controls (Fig. 1 E). To exclude the possibility that the instability we observed is dependent on AID expression induced by activating B cells with IL-4 and LPS, we stimulated CtIP-deficient and WT B cells with RP105. Upon RP105 activation, B cells do not express AID and do not undergo CSR. We found similar levels of spontaneous genomic instability in CtIP^Φ/− B cells independent of stimulation conditions (unpublished data). Because CtIP has been implicated in HR, we monitored IR induced Rad51 focus formation in CtIP–deficient B cells. We found that the percentage of RAD51 focus-positive cells was decreased on average to 50% of the WT level (Fig. 1 G), which is consistent with decreased resection activity in CtIP-deficient cells (Sartori et al., 2007). Thus, CtIP loss leads to considerable DNA damage accumulation associated with defective HR.

Cells with impaired HR are hypersensitive to compounds that induce replication-associated breaks. For example, BRCA-deficient cells are hypersensitive to PARPi (Bryant et al., 2005; Farmer et al., 2005; Jackson and Bartek, 2009). Consistent with this, treating cells with the PARPi (olaparib) resulted in a nearly twofold increase in the total amount of chromosomal aberrations in CtIP-null B cells (Fig. 1 H), whereas such PARPi treatment lead to little or no induction of aberrations in control cells. Although the total amount of DNA damage induced by PARPi was equivalent in both CtIP-deficient and BRCA1-deficient cells, radial structures accumulated at a greater frequency in the absence of BRCA1 (Fig. 1, H and I). We conclude that BRCA1- and CtIP-deficient B cells are both hypersensitive to PARPi, but CtIP-deficient cells accumulate considerably more spontaneous damage.

Because loss of 53BP1 rescues the viability of BRCA1-deficient mice (Cao et al., 2009), we tested whether deletion of 53BP1 could similarly rescue the embryonic lethality of CtIP^−/− mice. However 53BP1^−/−CtIP^+/− intercrosses did not generate any of the expected 15 CtIP/53BP1 double knockout mice out of 60 live births (Table 1). In Schizosacchoromyces pombe, the hypersensitivity of CtIP (Ctp1)-deficient cells to camptothecin (which yields DSBs during S-phase that rely on resection and HR for their repair) is rescued by loss of KU (Langerak et al., 2011). To see whether a similar relationship exists in mammals, we intercrossed CtIP^+/−KU80^+/− mice to determine if deletion of KU80 could rescue CtIP^−/− embryonic lethality. Again, no CtIP^−/− mice were generated among 129 offspring (Table 1). Collectively, these findings revealed that neither 53BP1-deficiency nor KU-deficiency can rescue embryonic lethality in CtIP^−/− mice.

Loss of 53BP1 normalizes the HR defects in BRCA1-deficient cells, at least in part by increasing DSB resection (Bunting et al., 2010). To determine what impact 53BP1 deletion might have on CtIP^−/−-associated genomic instability, we generated CtIP/53BP1 double knockout B cells by crossing CD19^cre/+CtIP^co/− with 53BP1^−/− mice (Fig. 2 A). We found that CtIP/53BP1 double-mutant B cells harbored spontaneous chromosomal aberrations and constitutive DNA damage signaling similar to the levels observed in CtIP single-mutant cells (Fig. 2, A and B). Thus, in contrast to its ability to rescue BRCA1-associated phenotypes, loss of 53BP1 does not restore genomic integrity in the absence of CtIP.

Loss of 53BP1 increases resection in BRCA1-deficient cells (Bunting et al., 2010, 2012). To determine whether and how 53BP1 impacts on resection in CtIP-deficient cells, we used a previously described nondenaturing immunofluorescence protocol, in which MEFs are pulsed with BrdU and ssDNA accumulation after irradiation (IR) is monitored by flow cytometry with an anti-BrdU antibody (Bunting et al., 2012). WT, 53BP1^−/−, BRCA1^Δ11/Δ1153BP1^−/−, and BRCA1^Δ11/Δ11 MEFs were infected with a short hairpin RNA (shRNA) specifically targeting the CtIP mRNA (shCtIP; Bunting et al., 2010; Fig. 2 C). The extent of knockdown differed in each cell line, making it difficult to compare the effects of CtIP deletion across the nonisogenic lines (Fig. 2 C). However, in all individual cell lines loss of CtIP expression decreased the extent of IR-induced ssDNA (Fig. 2 D). Thus, DNA damage stimulates an increase in end-resection activity in BRCA1- and 53BP1-deficient cell lines that is CtIP dependent.

To determine the impact of CtIP deficiency on RAD51 foci formation, we crossed CD19^cre/+CtIP^co/−, BRCA1Δ11^f/f and 53BP1^−/− mice to generate triple CtIP–BRCA1–53BP1 mutant B cells. Consistent with the decrease of ssDNA in CtIP depleted BRCA1–53BP1 MEFs (Fig. 2 D), we found that CtIP deficiency in B cells led to a decrease in IR induced RAD51 foci formation relative to WT (Fig. 2 E). Moreover, PARPi treatment of CtIP–BRCA1–53BP1 triply deficient cells led to a 13-fold increase in the level of DNA damage relative to BRCA1Δ11^Φ/Φ 53BP1^−/− (Fig. 2 F). Together with the finding that the IR-induced increase in resection in BRCA1–53BP1 double-knockout cells was CtIP dependent (Fig. 2 D), these results suggest that CtIP-mediated resection is critical for the rescue of HR in BRCA1–53BP1 mutant cells.

In the two-step model for resection, CtIP carries out initial end processing, which is followed by EXO1- or DNA2/BLM-mediated extension of the resected tracks (Symington and Gautier 2011). Because EXO1 recruitment and exonuclease activity are reported to be CtIP dependent (Eid et al., 2010), we tested whether the rescue of HR in BRCA–53BP1-deficient cells was EXO1 dependent. To do this, EXO1^+/− mice were intercrossed with CD19^cre/+Brca1Δ11^f/f53BP1^−/− mice, and the resulting CD19^cre/+Brca1Δ11^f/f53BP1^−/−EXO1^+/− mice were intercrossed to generate triple BRCA1–53BP1–EXO1 mutant B cells. We found that BRCA1Δ11^Φ/Φ 53BP1^−/−EXO1^−/− cells were as insensitive to PARPi as BRCA1Δ11^Φ/Φ 53BP1^−/− cells and much less sensitive than BRCA1Δ11^Φ/Φ or triple mutant CtIP–BRCA1–53BP1 cells (Fig. 2, F and G). These data thus establish that CtIP but not EXO1 mediates the rescue of genomic stability of BRCA1–53BP1 deficient cells.

The aforementioned findings suggested that CtIP could act independently of BRCA1 to promote resection. Consistent with this, we found that BRCA1 was not required for recruitment of CtIP to DNA damage sites (Fig. 3 A). To explore the physiological role of the CtIP–BRCA1 interaction mediated by CDK-dependent phosphorylation of CtIP at S327 (Yu and Chen, 2004), we generated BAC transgenic mouse models. By introducing the human gene coding for CtIP (hereafter, referred to as CtIP^WT) as a transgenic copy carried by a bacterial artificial chromosome (BAC RP11-104H10), we were able to rescue the lethality caused by homozygous deletion of mouse CtIP (Table 2). CtIP^WT was expressed at a similar level compared with endogenous CtIP (Fig. 3 B). Moreover, B cells in CtIP^WTCtIP^−/− mice were equivalent to CtIP^+/+ with respect to RAD51 foci formation and genome stability (Fig. 3, C and D).

We next generated BAC transgenic mice expressing a CtIP mutant in which Ser-327 was substituted for Ala (CtIP^S327A) to abrogate phosphorylation by CDK at this site. Like CtIP^WT, expression of mutant CtIP^S327A rescued the lethality of CtIP knockout mice (Table 2). Indeed, CtIP^S327ACtIP^−/− mice developed normally, without notable defects in growth, weight, lymphocyte development, or fertility (unpublished data). Moreover, in stark contrast to CtIP deficiency, CtIP^S327ACtIP^−/− B cells were not markedly hypersensitive to PARPi, did not harbor constitutive DNA damage signaling, and formed normal levels of IR–induced RAD51 foci (Fig. 3, B–D). These results are consistent with a recent characterization of CtIP-S326A knock-in mice, which concluded that CDK-dependent phosphorylation of CtIP–BRCA1 interaction domain is dispensable for HR (Reczek et al., 2013).

In addition to S327, the other CDK-dependent phosphorylation site of CtIP implicated in DSB resection is Thr-847 (Huertas and Jackson, 2009). To assess whether this residue affects CtIP function, we used BAC recombineering to mutate threonine into either alanine (CtIP^T847A) or a potentially phospho-mimicking glutamic acid (CtIP^T847E), and then generated transgenic mice. We found that the CtIP^T847A transgene did not rescue the lethality of CtIP knockout mice, whereas CtIP^T847E did (Table 2). By isolating embryos from timed pregnancies, we determined that CtIP^T847ACtIP^−/− embryos died earlier than E9.5 (Table 3), consistent with the early embryonic lethality of CtIP^−/− mice (Chen et al., 2005). Thus, phosphorylation of CtIP at T847, in contrast to S327, is essential for embryonic viability.

To examine the consequence of CtIP^T847A protein expression, we generated CtIP^T847ACD19^cre/+CtIP^co/− B cells. Western blot analysis confirmed that CtIP^T847A was expressed in these cells at similar levels as CtIP^WT but lower than in CtIP^T847E (Fig. 3 E). CtIP^T847A mutant B cells exhibited high levels of spontaneous genomic instability, persistent DNA damage signaling, and defective IR-induced RAD51 focus formation similar to those seen with CtIP deficiency (Fig. 3, E–G). In contrast, CtIP^T847E mutant B cells were normal in all these respects (Fig. 3, E–G). The finding that CtIP^T847A phenocopies CtIP deficiency thereby implies that CDK-dependent phosphorylation of T847 is essential for CtIP functions in mice.

Given that phosphorylation of CtIP^T847 is essential for efficient resection (Huertas and Jackson, 2009), we wondered whether CDK-dependent phosphorylation favors the recruitment of CtIP to damaged chromatin. To address this, we generated FLAG-tagged CtIP^WT, CtIP^T847A, and CtIP^T847E retroviral constructs, introduced them into WT MEFs, and then induced targeted DNA damage by laser microirradiation. Using an antibody that recognizes exogenously expressed FLAG-tagged proteins (Fig. 3 H) we could readily detect CtIP^WT, CtIP^T847A, and CtIP^T847E at DNA damage sites. Given that CtIP^T847A can be recruited to DSB (Fig. 3 H), we conclude that the phosphorylation of CtIP-T847, although critical for resection, is not required for CtIP recruitment to damage regions. Instead, chromatin binding of CtIP appears to require DNA damage inducible phosphorylation at CtIP-T859 by ATR kinase (Peterson et al., 2013).

Because the CtIP-T847E mutant has been reported to be proficient in allowing DSB resection (Huertas and Jackson, 2009), we asked whether this phospho-mimic version of the protein could rescue the sensitivity of BRCA1-deficient cells to PARPi. To do so, we crossed CD19^cre/+BRCA1Δ11^f/f mice with either CtIP^WT or CtIP^T847E transgenic mice to generate BRCA1-deficient B cells expressing either CtIP^WT or CTIP^T847E protein (Fig. 3 I). We found that the genomic instability induced by PARPi treatment was significantly reduced in BRCA1Δ11 mutant B cells expressing CtIP^T847E relative to that found in Brca1Δ11^Φ/Φ cells (Fig. 3 I). CtIP^WT, which expressed slightly lower levels of CtIP than CtIP^T847E (Fig. 3 I), also reduced the level of chromosomal aberrations in BRCA1-deficient mice, but this was not significant (Fig. 3 I). These data therefore indicate that overexpression of CtIP can partially overcome 53BP1’s block to resection in BRCA1-deficient cells.

Previous studies have suggested that BRCA1 plays a role in resection and that this is mediated by the CtIP-BRCA1 interaction. These conclusions were based on the findings that BRCA1-deficient cells exhibit a mild decrease in IR-induced RPA focus formation (Chen et al., 2008; Escribano-Díaz et al., 2013), that BRCA1 and CtIP inhibit RIF1 end-blocking activity in S/G2 (Escribano-Díaz et al., 2013; Zimmermann et al., 2013), and that cells expressing CtIP-327A are defective in HR (Yun and Hiom, 2009). However, subsequent studies in chicken (Nakamura et al., 2010), frog (Peterson et al., 2011), and mouse cells (Reczek et al., 2013; analogous to our CtIP-null cells reconstituted with CtIP-S327A), and direct measurements of gene conversion (Chandramouly et al., 2013), argue that the CtIP-BRCA1 interaction is dispensable for HR. Moreover, our finding that restoration of HR in BRCA1–53BP1-deficient cells and increased resection that occurs in the absence of 53BP1 are CtIP dependent, demonstrates that CtIP-mediated resection is BRCA1 independent. These results likely explain why loss of 53BP1 cannot reverse the embryonic lethality or genomic instability of CtIP-deficient mice. Localization of CtIP to damage sites must take place by mechanisms other than association with BRCA1, such as direct association of CtIP with DNA (You et al., 2009) or through phosphorylation by the ATR kinase (Peterson et al., 2013).

If CDK-mediated phosphorylation of CtIP is essential for its function (Table 1 and Fig. 3, E–G), whereas BRCA1 is dispensable for resection, how does BRCA1 promote HR? We have previously suggested that BRCA1’s principal role in DSB repair is to antagonize 53BP1-dependent end protection (Bunting et al., 2010). In this model, in the absence of BRCA1, CtIP activity is limited by 53BP1 and downstream cofactors RIF1 and PTIP, which results in a modest defect in RPA focus formation (Escribano-Diaz et al., 2013). Nevertheless CtIP-mediated resection can partly overcome 53BP1 end blocking activity, evident by our finding that mimicking constitutive CtIP phosphorylation (CtIP^T847E) partially alleviated genomic instability of BRCA1-deficient/53BP1^+/+ cells. Furthermore, although 53BP1 antagonizes CtIP function, complete absence of 53BP1 allows full access and activity of CtIP, which significantly increases ssDNA formation. Thus, we propose that BRCA1 facilitates CtIP mediated resection in the presence of 53BP1 but is no longer needed in its absence. In addition, BRCA1 plays functions in meiosis and cross-link repair that are 53BP1 independent (Bunting et al., 2012).

The pro-NHEJ (non-homologous end joining) functions that promote class switch recombination, and the antirecombination functions that lead to loss of genome stability in BRCA1-deficient cells are mediated by distinct phospho-dependent interactions with 53BP1 (Callen et al., 2013). PTIP associates with phosphorylation sites at the extreme N terminus of 53BP1 to suppress HR in S phase, whereas RIF1 binds to independent residues to promote NHEJ in G1 (Callen et al., 2013). We have proposed that PTIP and RIF1 could interfere with a distinct set of nucleases (Callen et al., 2013). Consistent with this idea, we have found that the rescue of genome stability in BRCA1–53BP1-deficient cells is dependent on ATM (Bunting et al., 2010) and CtIP (Fig. 2 F), but occurs independently of EXO1 (Fig. 2 G). In contrast, extensive resection during class switching in 53BP1^−/− cells is ATM-independent (Yamane et al., 2013), and is mediated by both CtIP and EXO1 (Bothmer et al., 2013). Similarly, unrepaired G1-phase breaks during V(D)J recombination are processed by CtIP (Helmink et al., 2011). These results suggest that, whereas CtIP promotes nucleolytic processing of DSBs in all cell cycle phases, other factors that stimulate or antagonize the extension of resected tracks may be distinct and regulated in a cell cycle–dependent manner.

CtIP^co/− (Bothmer et al., 2013), CtIP^+/− (Chen et al., 2005), 53BP1^−/− (Ward et al., 2004), BRCA1^{f(Δ11)/f(Δ11)} (here reported BRCA1Δ11^f/f; NCI mouse repository), Ku80^−/− (Nussenzweig et al., 1996), CD19^Cre (Rickert et al., 1997), EXO1^−/− (Wei et al., 2003), and ROSA26-STOP-EYFP (Srinivas et al., 2001) mice have been described. BAC RP11-104H10 containing human CtIP genomic sequence was used to generate transgenic mice. S327A, T847A, and T847E mutations were targeted using BAC recombineering as previously described (Difilippantonio et al., 2005). Transgenic CtIP was detected by PCR screening of tail DNA using the following primer pairs: hEx1F 5′-AGCACAACCACACTGAATGC-3′ and hEx1R 5′-CACCACAGGTATTCTCACACGG-3′, which amplify a product of 305 bp in the intronic sequence specific for the human gene. All experiments with mice were conducted in agreement with protocols approved by the National Institutes of Health Institutional Animal Care and Use Committee.

B cells were isolated from WT or mutant spleen with anti-CD43 Microbeads (anti-Ly48; Miltenyi Biotec) and were cultured with LPS (25 µg/ml; Sigma-Aldrich), IL-4 (5 ng/ml; Sigma-Aldrich), and RP105 (0.5 µg/ml; BD). Cell proliferation was analyzed by CFSE (5 µM; Molecular Probes) labeling at 37°C for 10 min. For cell cycle analysis, B cells were fixed in methanol and stained with propidium iodide (PI). BrdU incorporation and detection was performed as described (Bunting et al., 2012). Samples were acquired on a FACSCalibur (BD). For genomic instability analysis, B cells were harvested after 2 d in culture; metaphase spreads were generated and processed for FISH analysis as previously described (Callen et al., 2013). PARP inhibitor (KU58948; Astra Zeneca) was added to cells stimulated ex vivo for one day.

Coding sequences of the mouse CtIP was cloned into PMX-IRES-GFP plasmid (CtIP^WT). A FLAG-tag was cloned by PCR at the N terminus of CtIP coding sequence (FLAG-CtIP^WT). Mutant FLAG-CtIP^T847E and FLAG-CtIP^T847A were generated by targeting Flag-CtIP^WT using QuikChange II XL Site-Directed Mutagenesis kit (Agilent Technologies), and mutations were confirmed by sequencing. Infected MEFs were grown and selected in medium containing 2 µg/ml puromycin for 1 wk, and then maintained in 1 µg/ml puromycin.

Western blotting was performed with the following primary antibodies: mouse anti-tubulin (Sigma-Aldrich), mouse anti–FLAG-M2 (Sigma-Aldrich), mouse anti-CtIP (mAb 14–1; gift from R. Baer, Columbia University, New York, NY), rabbit anti-CtIP (developed in collaboration with Epitomics), rabbit anti-KAP-1 pS824 (Bethyl Laboratories), rabbit anti–p53 pS15 (Cell Signaling Technology), and rabbit anti-53BP1 (Novus). For RAD51 immunofluorescence, cells were irradiated at 10 Gy allowed to recover for 4 h, and then fixed and processed as previously described (Celeste et al., 2003). For microirradiation, cells were presensitized in DMEM media containing 0.1 µg/ml of Hoechst 33342 for 60 min before replacing with phenol red free media containing 5 mM Hepes, and then irradiated with the 364-nm laser line on a LSM510 confocal microscope (Carl Zeiss, Inc.) equipped with a heated stage. Cells were allowed to recover for the indicated time before processing for immunofluorescence (Celeste et al., 2003). Primary antibodies for immunofluorescence were rabbit anti-CtIP and anti-53BP1, mouse anti-FLAG-M2, mouse anti–γ-H2AX (Upstate Biotechnology), and rabbit anti-RAD51 (Santa Cruz Biotechnology, Inc.).

We thank Davide Robbiani for the CtIP shRNA construct, Junjie Chen for 53BP1^−/− mice, Winfried Edelmann for EXO1^−/− mice, and the members of Nussenzweig laboratory for helpful discussions.

The work was supported by the Intramural Research Program of the National Institutes of Health, the National Cancer Institute, and the Center for Cancer Research and by a Department of Defense grant to A. Nussenzweig (BC102335).

The authors declare no competing financial interests.