BRCA1 Mutation-Specific Responses to 53BP1 Loss-Induced Homologous Recombination and PARP Inhibitor Resistance

SUMMARY

BRCA1 functions in homologous recombination (HR) both up- and downstream of DNA end resection. However, in cells with 53BP1 gene knockout (KO), BRCA1 is dispensable for the initiation of resection, but whether BRCA1 activity is entirely redundant after end resection is unclear. Here, we found that 53bp1 KO rescued the embryonic viability of a Brea1^ΔC/ΔC mouse model that harbors a stop codon in the coiled-coil domain. However, Brca1^ΔC/ΔC;53bp1^−/− mice were susceptible to tumor formation, lacked Rad51 foci, and were sensitive to PARP inhibitor (PARPi) treatment, indicative of suboptimal HR. Furthermore, BRCA1 mutant cancer cell lines were dependent on truncated BRCA1 proteins that retained the ability to interact with PALB2 for 53BP1 KO induced RAD51 foci and PARPi resistance. Our data suggest that the overall efficiency of 53BP1 loss of function induced HR may be BRCA1 mutation dependent. In the setting of 53BP1 KO, hypomorphic BRCA1 proteins are active downstream of end resection, promoting RAD51 loading and PARPi resistance.

Graphical Abstract

In Brief

Using a Brca1^ΔC mouse model and a panel of BRCA1 mutant cancer cell lines, Nacson et al. show that 53BP1 loss of function induced homologous recombination and PARP inhibitor resistance is suboptimal in the absence of hypomorphic BRCA1 proteins that retain the coiled-coil domain and ability to interact with PALB2.

INTRODUCTION

BRCA1 germline mutations predispose carriers to an increased lifetime risk of developing breast and ovarian cancer. Malignancies often demonstrate loss of heterozygosity, where the wild-type allele is lost but the mutant allele remains present. The ability of BRCA1 mutant alleles to generate protein products that contribute to biological processes normally undertaken by wild-type BRCA1 are beginning to emerge, and although mutant proteins lack tum or-suppressor activity, several truncated BRCA1 protein isoforms have been shown to promote residual DNA repair and chemotherapy resistance (Drost et al., 2011, 2016; Johnson et al., 2013; Shakya et al., 2011; Wang et al., 2016a, 2016b).

BRCA1 is an 1863-amino-acid protein with several conserved domains that function in homologous recombination (HR) DNA repair. The RING domain is located toward the N-terminal region of the protein and is required for BRCA1-BARD1 heterodimer formation. Exon 11 is the largest exon within the BRCA1 gene and encodes the central portion of the protein. The coiled-coil (CC) region directly interacts with PALB2, forming a larger BRCA1-PALB2-BRCA2-RAD51 protein complex (Sy et al., 2009; Xia et al., 2006, 2007; Zhang et al., 2009a, 2009b). BRCA1 harbors two BRCT repeats toward the C-terminal end of the protein that mediate interactions with phosphorylated proteins. The BRCT domain accounts for the BRCA1-CtIP interaction, potentially enhancing CtIP localization to double-strand DNA breaks (DSBs) and activating MRE11-RAD50-NBS1 (MRN) nuclease activity (Huen et al., 2010; Roy et al., 2011).

BRCA1 has been shown to contribute to HR at distinct steps, first promoting DNA end resection (Bouwman et al., 2010; Bunting et al., 2010) and later recruiting PALB2-BRCA2-RAD51 complexes to resected single-stranded DNA (ssDNA) (Sy et al., 2009; Zhang et al., 2009a, 2009b). When a DSB initially arises, 53BP1 is rapidly localized to surrounding chromatin and blocks the initiation of DNA end resection. BRCA1 promotes the displacement of 53BP1 from chromatin surrounding DSBs and consequently activates DNA end resection (Chapman et al., 2012; Densham et al., 2016). After DNA end resection, BRCA1 directly interacts with and recruits PALB2 to DSB sites, where BRCA2-RAD51 form s a com plex with PALB2, and mutations that block the BRCA1-PALB2 interaction have been shown to reduce the efficiency of RAD51 foci formation and HR (Anantha et al., 2017; Foo et al., 2017; Sy et al., 2009).

Cells that carry deleterious BRCA1 mutations have a reduced capacity to perform HR; consequently, BRCA1 mutant cancers are highly sensitive to DNA-damaging chemotherapies and PARP inhibitors (PARPis) (Bryant et al., 2005; Farmer et al., 2005; Moynahan et al., 1999, 2001). Several PARPis have been incorporated into the standard of care for the treatment of ovarian and breast cancer patients. BRCA1 and BRCA2 mutation carriers treated with PARPi therapy have demonstrated improved outcomes compared to patients without these mutations. However, PARPi remains ineffective in a portion of BRCA1-mutation-carrying patients, and most tumors that are responsive eventually acquire resistance (Ledermann et al., 2016; Robson et al., 2017).

Mutations or downregulation of proteins that regulate DNA repair pathway choice have been shown to promote HR and PARPi resistance in the absence of BRCA1 activity (Bouwman et al., 2010; Bunting et al., 2010; Tkáč et al., 2016; Xu et al., 2015). Specifically, when 53BP1 is depleted or knocked out, DNA end resection and HR ensue in the absence of BRCA1 activity and result in PARPi resistance (Bouwman et al., 2010; Bunting et al., 2010). 53bp1 knockout (KO) was shown to be capable of fully rescuing embryonic development in mice harboring homozygous Brca1 mutant alleles (Bunting et al., 2010, 2012; Cao et al., 2009; Li et al., 2016). Moreover, derived cells demonstrated Rad51 foci formation and PARPi resistance, potentially indicating that Brca1 activity may be dispensable for Rad51 loading in the setting of 53bp1 deficiency.

We and others have shown that HR and PARPi resistance can occur when BRCA1-mutation-carrying alleles produce truncated proteins that are capable of promoting RAD51 foci formation (Drost et al., 2016; Johnson et al., 2013; Wang et al., 2016a, 2016b). Furthermore, previous studies demonstrating 53bp1-KO-induced rescue of embryonic viability utilized Brca1^Δ11 and Brca1^Δ2 alleles that have been shown to produce exon-11-deficient and RING-domain-deficient hypomorphic Brca1 proteins (Bunting et al., 2010, 2012; Cao et al., 2009; Li et al., 2016). To explore the relationship between 53BP1 KO and BRCA1 protein expression, we generated a novel Brca1^ΔC allele that harbors a stop codon in the CC domain and fails to generate a detectable Brca1 protein. Herein, we assessed the relationship between BRCA1 mutational status and 53BP1 loss of function in promoting HR and PARPi resistance.

RESULTS

53BP1 Depletion Augments the Activity of BRCA1 CC-Containing Proteins

BRCA1 mutant alleles are often capable of generating truncated BRCA1 proteins that lack functional domains usually present in the full-length protein (Drost etal., 2016; Hill et al., 2014; Johnson et al., 2013; Wang et al., 2016a, 2016b; Zhou et al., 2003). To examine the DSB recruitment dynamics of BRCA1 proteins that are deficient for various functional domains, we examined γ-irradiation-induced foci (IRIF) formation kinetics. In addition, we measured the impact of BRCA1 protein status on 53BP1 foci formation. MDA-MB-436 cells have a BRCA1 5396+1G > A mutation that results in loss of BRCA1 protein expression (Elstrodt et al., 2006), undetectable RAD51 foci, and exquisite PARPi sensitivity (Johnson et al., 2013; Wang et al., 2016a). We therefore used this cell line to add back BRCA1 proteins that lacked the following peptide regions: exon 11 coding (BRCA1-Δ11q), BRCT domain (BRCA1-ΔBRCT), and both the CC and BRCT domains (BRCA1-ΔCC+ΔBRCT) (Figure 1A). We first confirmed the cytoplasmic and nuclear expression of BRCA1 proteins (Figure S1A). As expected, full-length BRCA1 was entirely nuclear, while BRCA1-Δ11q that lacks an exon 11 located nuclear import signal was expressed similarly in both cytoplasmic and nuclear fractions (Chen et al., 1996; Korlimarla et al., 2013). The BRCA1-ΔBRCT and BRCA1-ΔCC+ΔBRCT proteins were detected in cytoplasm ic fractions but were predom inantly nuclear (Figure S1A). Of note, all truncated proteins demonstrated similar or higher nuclear expression than ectopic full-length BRCA1, indicating that observations related to loss of activity would not result from reduced nuclear expression. As expected, mCherry-expressing negative control cells lacked BRCA1 foci. Interestingly, full-length BRCA1 and BRCA1-Δ11q proteins that retain BRCT and CC domains formed IRIF with similar kinetics (Figure S1B). In contrast, slower kinetics were observed, and fewer cells expressing the BRCA1-DBRCT protein demonstrated foci positivity. These dynamics were further decreased in cells expressing the BRCA1-ΔCC+ΔBRCT protein (Figure S1B). Although BRCA1 has been shown to exclude 53BP1 to the periphery of the focus (Chapman et al., 2012; Densham et al., 2016), we did not observe differences between cell lines in the overall 53BP1 IRIF kinetics in our MDA-MB-436 isogenic model (Figure S1B).

Figure 1. Effects of 53BP1 Depletion on BRCA1 Protein Activity.

(A) Cartoon showing BRCA1 constructs expressed in MDA-MB-436 cells: BRCA1 full-length, BRCA1-Δ11q, BRCA1-ΔBRCT, and BRCA1-ΔCC+BRCT. Retained peptide domains and protein interactions are indicated. Red indicates the RING domain location, orange indicates exon-11-encoded amino acids, black indicates coiled-coil domain amino acids, green shows BRCT domain locations, and blue shows non-assigned-domain amino acids.

(B) Western blot analyses of MDA-MB-436 cells engineered to express mCherry, BRCA1 full-length (FL) protein, BRCA1-Δ11q (Δ11q), BRCA1-ΔBRCT (ΔBRCT), and BRCA1-ΔCC+BRCT (ΔCC+BRCT) truncated proteins as well as non-target (NT) or 53BP1 shRNA (53). Arrows indicate truncated BRCA1 proteins, and asterisks indicate nonspecific bands.

(C) Cells from (B)were assessed for RPA32 IRIF formation using immunofluorescence assays. Forcom parison, FL-BRCA1-expressing cellsare included assolid black bars in graphs showing data for mCherry-expressing cell lines. At the indicated time points post-γ-irradiation (IR) (10Gy), cells were fixed and those with at least 10 foci per nucleus counted positive for IRIF . For each protein ,a minimum of 100 nuclei per time point and cell line were counted per experiment. Mean and SEM percentages of cells containing RPA32 foci are shown from three independent experiments. Solid lines represent NT shRNA, and dashed lines represent 53BP1 shRNA. Data were analyzed using two-way ANOVA. Significant differences between NT and 53BP1 shRNA at each time point are denoted (***p < 0.001; **p < 0.01; *p < 0.05). See Figure S1C for representative images.

(D) Cells were assessed as in (C) for mean and SEM percentages of cells containing at least 5 BRCA1 foci per nucleus from three independent experiments. See Figure S2 for representative images.

(E) Cells were assessed as in (C) for mean and SEM percentages of cells containing at least 5 RAD51 foci per nucleus from three independent experiments. See Figure S2 for representative images.

(F) Cells were incubated with rucaparib and assessed for colony formation. Mean and SEM colony formation was calculated as a percentage of DMSO-vehicle-treated cells from three independent experiments. Full-length BRCA1 colony formation data (black line) is shown as a comparator in graphs with mCherry expressing cells. NT and 53BP1 shRNA-treated cells are shown with solid and dashed lines, respectively.

We next measured the effects of 53BP1 depletion on truncated BRCA1 protein activity. MDA-MB-436 cells expressing the above proteins were further engineered to express nontarget (NT) or 53BP1-targeting short hairpin RNA (shRNA) (Figure 1B), and foci formation dynamics were compared to MDA-MB-436 cells that expressed full-length BRCA1. BRCA1 and 53BP1 both regulate the resection of DSBs, and we measured RPA32 IRIF as an indicator of DNA end-resection proficiency. Here, mCherry-, BRCA1-Δ11q-, BRCA1-ΔBRCT-, and BRCA1-ΔCC+ΔBRCT-expressing cell lines all had reduced RPA32 IRIF compared with full-length BRCA1-expressing cells. However, 53BP1 shRNA significantly elevated RPA32 IRIF, regardless of BRCA1 protein status, to similar kinetics as that observed in full-length-BRCA1-expressing cells (Figures 1C and S1C). 53BP1 shRNA did not affect BRCA1 IRIF in mCherry-negative control cells. However, 53BP1 shRNA markedly increased the number of BRCA1-Δ11q and BRCA1-ΔBRCT IRIF-positive cells compared with NT shRNA-expressing cells. The CC domain appeared crucial for 53BP1 shRNA-promoted IRIF, as the number of BRCA1-foci-positive cells expressing the BRCA1-ΔCC+ΔBRCT protein was similar between NT and 53BP1 shRNA treatments (Figures 1D and S2). Furthermore, 53BP1 shRNA had no impact on RAD51 IRIF in mCherry-expressing cells but dramatically increased RAD51 IRIF in BRCA1-Δ11q- and BRCA1-ΔBRCT-expressing cells. Similar to BRCA1 IRIF, the ability of 53BP1 shRNA to promote RAD51 IRIF appeared dependent on the BRCA1 CC domain, as 53BP1 shRNA failed to elevate RAD51 IRIF in BRCA1-ΔCC+ ΔBRCT-expressing cells (Figures 1E and S2).

Next, the effect of 53BP1 depletion in combination with truncated BRCA1 protein expression on PARPi sensitivity was measured and compared to full-length BRCA1-expressing cells in colony-formation assays. Here, 53BP1 shRNA failed to significantly impact PARPi resistance in mCherry-expressing cells. However, BRCA1-Δ11q- and BRCA1-ΔBRCT-expressing cells had a 4-fold (p = 0.013, unpaired t test) and 21-fold (p < 0.001, unpaired t test) increase, respectively, in rucaparib LC50 (lethal concentration reducing colony formation by 50%) values compared with NT shRNA-expressing cells. In contrast, 53BP1 shRNA did not significantly impact rucaparib sensitivity in cells expressing BRCA1-ΔCC+ΔBRCT (Figure 1F). All in, these data show that BRCA1 protein status does not affect the ability of 53BP1 depletion to stimulate DNA end resection and RPA32 foci formation. However, in the MDA-MB-436 cell line isogenic system, BRCA1 truncated proteins that retain the CC domain were critical for 53BP1-depletion-induced RAD51 foci formation and robust PARPi resistance.

Brca1^ΔC/ΔC Embryonic Viability Is Rescued by 53bp1 KO

Previous studies showed that 53bp1 KO rescued HR and the embryonic viability of mice harboring hypomorphic Brca1^Δ11/Δ11 and Brca1^Δ2/Δ2 mutant alleles. Furthermore, mice had normal lifespans and did not demonstrate increased tum or susceptibility (Bunting et al., 2010, 2012; Cao et al., 2009). Because 53BP1 depletion failed to induce RAD51 foci and PARPi resistance in BRCA1-ΔCC+ΔBRCT-expressing MDA-MB-436 cells, we explored the possibility that similar Brca1 mutations in mice would result in an impotent 53bp1-KO-induced rescue of HR and embryonic development. We aimed to introduce mutations in the Brca1 CC-encoding region using clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 genetic editing. Here, we generated mice harboring a Brca1 c.4080del4 frameshift mutation. This mutation results in a p.1361fs28X stop codon that blocks translation in the CC domain, predicted to cause loss of the C-terminal region of the protein, that includes both the CC and BRCT domains; we therefore refer to this allele as Brca1^ΔC (Figure 2A). Moreover, this mutation is downstream of exon 11 and unlikely to be removed by alternative splicing (Ludwig et al., 1997; Xu et al., 1999). As expected, homozygosity for the Brca1^ΔC allele resulted in embryonic lethality, and homozygous embryos were unable to be detected at embryonic day 9.5 (E9.5), indicating that Brca1^ΔC/ΔC induced early embryonic lethality (Table 1).

Figure 2. Generation and Assessment of Brca1^ΔC/ΔC;53bp1^−/− Mice.

(A) Brca1 coiled-coil genomic region targeted using CRISPR/Cas9. Red arrow indicates single guide RNA (sgRNA)-targeting location and the resulting c.4080–4083 4-bp deletion mutation. The Brca1 coiled-coil wild-type peptide region is shown with the critical L1363 residue (equivalent to human L1407) highlighted green. Amino acid changes resulting from the 1361fs28X mutation are red. Bottom : cartoon of predicted truncated protein product.

(B) Representative photograph of 6-week-old littermates with indicated genotypes. See also Table 1.

(C) Weights of individual mice with the indicated genotypes at 6 weeks of age. **p < 0.01, ***p < 0.001, NS, not significant (one-way ANOVA with multiple comparisons).

(D) Kaplan-Meier survival curves of mice with the indicated genotypes.

(E) Representative photograph of Brca1^ΔC/ΔC;53bp1^−/− mouse with thymic lymphoma. H&E of thymus, liver, bone marrow, and kidney showing tumor infiltrates (*). Scale bar, 20 μm.

(F) Primary MEFs (before passage 4) with the indicated genotypes were assessed for Brca1 protein expression by western blotting. Arrows indicate the expected gel migrations of Brca1 w ild-type (WT) (top arrow), a potential Brca1^ΔC protein (m iddle arrow), and the Brca1^Δ11 protein (lower arrow). Asterisk indicates a nonspecific band.

Table 1.

Predicted and Observed Brcal and 53bp1 Genotype Frequencies

Brca1+/ΔC × Brca1+/ΔC
	Brca1+/+	Brca1+/ΔC	Brca1ΔC/ΔC
E9.5 embryos expected	6.75	13.5	6.75
(27 screened) observed	13	14	0
Pups expected	27	54	27
(108 screened) observed	36	72	0
Brca1+/ΔC 53bp1+/− × Brca1+/ΔC 53bp1−/−
	Brca1ΔC/ΔC 53bp1+/+ or Brca1ΔC/ΔC 53bp1+/−	Brca1+/+ 53bp1−/− or Brca1+/ΔC 53bp1−/−	Brca1ΔC/ΔC 53bp1−/−
Pups expected	24.75	24.75	8.25
(132 screened) observed:	0	26	3
Brca1+/ΔC 53bp1−/− × Brca1+/ΔC 53bp1−/−
	Brca1+/+ 53bp1−/−	Brca1+/ΔC 53bp1−/−	Brca1ΔC/ΔC 53bp1−/−
Pups expected	27.25	54.5	27.25
(109 screened) observed	32	52	25

Contrary to our supposition that loss of the CC domain would abrogate 53bp1-KO-induced embryonic development, interbreeding Brca1^+/AC;53bp1^−/+ or Brca1^+/ΔC;53bp1^−/− mice resulted in Brca1^ΔC/ΔC;53bp1^−/− progeny at close to the expected Mendelian ratios (Table 1). Brca1^ΔC/ΔC;53bp1^−/− mice develop normally and are similar to Brca1^+/+;53bp1^−/− mice in terms of size (Figure 2B) and weight (Figure 2C). However, in contrast to previous reports describing Brca1^Δ11/Δ11;53bp1^−/− homozygous mice (Bunting et al., 2010; Cao et al., 2009), the lifespan of Brca1^ΔC/ΔC;53bp1^−/− mice was significantly shorter (Figure 2D), with 4 out of 4 mice that were subjected to pathological analyses demonstrating thymic lymphomas at 3–6 months of age (Figure 2E), similar to those reported in older Brca1^+/+;53bp1^−/− mice (Ward et al., 2003).

Brca1^Δ11/Δ11 and Brca1^Δ2/Δ2 mutant mouse embryonic fibroblasts (MEFs) were previously shown to express Brca1 hypo-morphic proteins that lack either the exon 11 coding region or the RING domain, respectively, with both proteins retaining the CC domain (Li et al., 2016; Xu et al., 1999). Due to severe early lethality, we were unable to generate Brca1^ΔC/ΔC;53bp1^+/+ MEFs;however, Brca1^ΔC/ΔC;53bp1^−/− MEFs were readily generated. We detected full-length Brca1 in wild-type MEFs as well as the hypomorphic Brca1^Δ11 protein in Brca1^Δ11/Δ11MEFs. However, we were routinely unable to detect Brca1 protein in Brca1^ΔC/ΔC;53bp1^−/− MEFs (Figure 2F). Brca1 mRNA was similarly expressed in Brca1^ΔC/ΔC;53bp1^−/− and wild-type MEFs (Figure S3A), suggesting that nonsense-mediated mRNA decay (NMD) did not account for loss of protein expression. We predict that the Brca1^ΔC mutation induces protein folding defects that result in proteasomal degradation. However, relatively few hours of MG132 treatment killed Brca1^ΔC/ΔC;53bp1^−/− MEFs, with concentrations and time points that enabled live cells to be collected insufficient to detect truncated Brca1 protein by western blotting (Figure S3B). In summation, 53bp1 KO rescued the embryonic development of Brca1^ΔC/ΔC mice, but mice are susceptible to tum or development. Moreover, in contrast to hypomorphic Brca1^Δ11 and Brca1^Δ2 alleles that generate CC-domain-containing proteins, the Brca1^ΔC allele does not generate a detectable protein product.

Brca1^ΔC/ΔC;53bp1^−/− Mice Have Impaired HR

We next examined HR DNA repair in Brca1 and 53bp1 mutant mice. MEFs that are Brca1 mutated or DNA repair defective often senesce at early passages due to oxidative DNA damage and activation of p53-p21 (Cao et al., 2009; Parrinello et al., 2003). A previous study reported that senescence occurred in early passages of Brca1^Δ11/Δ11;53bp1^+/+ MEFs but was rescued in Brca1^Δ11/Δ11;53bp1^−/− cells (Cao et al., 2009), presumably because 53bp1 KO restored HR DNA repair. In our experiments, over 80% of Brca1^ΔC/ΔC;53bp1^−/− cells were positive for senescence-associated β-galactosidase (SA-β-gal) staining by passage 6 (Figure S3C). Senescence was delayed when cells were maintained in 2% oxygen (Figure S3C), suggesting that Brca1^ΔC/ΔC;53bp1^−/− MEFs are sensitive to oxidative DNA damage. Therefore, in contrast to Brca1^Δ11/Δ11 MEFs, 53bp1 KO did not rescue the ability of Brca1^ΔC/ΔC MEFs to proliferate beyond six passages, pointing toward the possibility that Brca1^ΔC/ΔC;53bp1^−/− MEFs remain DNA-repair defective. To reduce the onset of senescence, all subsequent experiments were set up when cells were within three passages of their derivation.

In foci formation experiments, Brca1^Δ11/Δ11 MEFs had fewer Rpa32 and Rad51 IRIF-positive cells than wild-type cells (Figure 3A). Similar to previous reports, 53bp1 KO fully rescued both Rpa32 and Rad51 foci formation in cells harboring the Brca1^Δ11/Δ11 allele. In contrast, although Brca1^ΔC/ΔC;53bp1^−/− MEFs had robust Rpa32 IRIF, there were 5-fold (p < 0.001, unpaired t test) fewer Rad51-foci-positive cells compared to wild-type MEFs (Figure 3A). Of note, Brca1^ΔC/ΔC;53bp1^−/− cells had marginally higher Rad51 foci positivity compared to Brca1^Δ11/Δ11 MEFs (Figure 3A). We also directly assessed HR DNA repair using a previously described CRISPR/Cas9-based reporter assay targeting the Rosa26 locus (Wyatt et al., 2016) (Figures S3D - S3G). Here, Brca1^ΔC/ΔC;53bp1^−/− MEFs had 2-fold (p < 0.001, unpaired t test) fewer gene conversion events compared with Brca1^Δ11/Δ11;53bp1^−/− MEFs (Figure 3B). Differences in Rad51 IRIF or HR proficiency were unlikely to be a result of disparate cell-cycle distributions, as MEFs had similar S/G2 portions (Figure S3H).

Figure 3. Assessment of HR and PARPi Sensitivity.

(A) MEFs were subject to 10-Gy IR and assessed for Rpa32 and Rad51 foci formation as described in Figure 1. The number of foci-positive cells (mean and SEM) from three independent experiments is shown (*p < 0.05, ***p < 0.001, compared to WT, unpaired t test). Representative images of MEFs with the indicated genotypes are shown. Scale bar, 10 μm.

(B) Cell lines were electroporated with plasmids expressing Rosa26-targeting sgRNA and Cas9 cDNA, plasmid enhanced green fluorescent protein (pEGFP), and a homology donor sequence. The number of gene conversion events from three independent repeats is shown. ***p < 0.001 (unpaired t test). See Figures S3D–S3G for additional controls and information.

(C) Metaphase spreads were prepared from the indicated genotypes of MEFs treated with 500 nM rucaparib for 24 hr. The mean number of breaks and radial chromosomes per 10 metaphases are shown. The number of metaphases assessed were 29 fo r Brca1^+/+;53bp1^+/+, 46 for Brca1^Δ11/Δ11,53bp1^+/+, 29 for Brca1^Δ11/Δ11 ;53bp1^−/−, and 56 for Brca1^ΔC/ΔC;53bp1^−/− MEFs. Scale bars, 10 μm.

(D) MEFs were treated with rucaparib for 7 days and stained with Sulforhodamine B (SRB). Solubilized stains were quantified, and cell growth (mean and SEM) is expressed as percentage of vehicle-treated cells from three independent experiment.

(E) Mice were treated with vehicle or 200 mg/kg rucaparib twice daily for 5 continuous days, and survival was measured by Kaplan-Meier analyses.

(F) Peripheral blood from mice described in (E) was assessed at the end of vehicle (V) or rucaparib (R) treatment for neutrophil and lym phocyte cell numbers.

**p < 0.01 compared with vehicle (unpaired t test).

Inappropriate recombination often results in deleterious chromosomal rearrangement events commonly manifested in the formation of multi-radial chromosome structures. In analyses of metaphase chromosomes from cells treated with 500 nM rucaparib, Brca1^Δ11/Δ11;53bp1^+/+ and Brca1^ΔC/ΔC;53bp1^−/− MEFs demonstrated 17- and 3.3-fold greater numbers of radial chromosomes and 6.4- and 3.9-fold greater numbers of chromosome breaks, respectively, compared with Brca1^Δ11/Δ11;53bp1^−/− MEFs (Figure 3C). Furthermore, in cell growth experiments, the mean GI50 (growth inhibition by 50%) value for rucaparib was 11,131-fold (p < 0.001, unpaired t test) lower in Brca1^Δ11/Δ11;53bp1^+/+ MEFs than wild-type MEFs (0.4 versus 4,440 nM). In contrast, no significant differences in GI50 values were observed between Brca1^Δ11/Δ11;53bp1^−/− and wild-type MEFs. The mean GI50 value for rucaparib was also 1,776-fold (p < 0.001, unpaired t test) lower in Brca1^ΔC/ΔC;53bp1^−/− MEFs than wild-type MEFs (2.5 versus 4,440 nM). However, Brca1^ΔC/ΔC;53bp1^−/− MEFs were also 6.3-fold (p < 0.001, unpaired t test) less sensitive to rucaparib than Brca1^Δ11/Δ11;53bp1^+/+ MEFs (2.5 versus 0.4nM) (Figure 3D). These data show that in contrast to the Brca1^Δ11/Δ11;53bp1^−/− genotype, Brca1^ΔC/ΔC;53bp1^−/− cells are PARPi sensitive. Nevertheless, the latter cells were also mildly less PARPi responsive than Brca1^Δ11/Δ11;53bp1^+/+ MEFs, indicating that 53bp KO likely provides some residual HR in the setting of Brca1^ΔC/ΔC.

To determine if PARPi sensitivity extended beyond MEFs, mice were treated with 200 mg/kg rucaparib twice daily for 5 days; this dose is routinely well tolerated by NSG mice in xenograft experiments. While rucaparib had no impact on Brca1^+/+;53bp1^+/+, Brca1^+/+;53bp1^−/−, or Brca1^Δ11/Δ11;53bp1^−/− mice, 5 out of 5 Brca1^ΔC/ΔC;53bp1^−/− mice treated with rucaparib demonstrated sickness that required euthanasia according to institutional guidelines within 9 days of treatment initiation (Figure 3E). Additionally, rucaparib-treated Brca1^ΔC/ΔC;53bp1^−/− mice had significantly decreased peripheral blood neutrophil and lymphocyte counts (Figure 3F). Taken together, these data show that the Brca1^ΔC/ΔC;53bp1^−/− genotype confers defective HR and PARPi sensitivity.

Because Brca1^ΔC/ΔC;53bp1^−/− mice showed markers of HR deficiency, we explored potential HR-independent mechanisms that might account for 53bp1-KO-associated rescue of embryonic development. p53 KO is an established mechanism that rescues Brca1 mutant embryo development (Hakem et al., 1997; Ludwig et al., 1997; Xu et al., 2001), and 53bp1 KO was previously shown to abrogate p53 transcriptional activity in human cancer cell lines (Cuella-Martin et al., 2016). However, we did not find evidence of reduced p53 transcriptional activity in 53bp1 KO MEFs, with γ-irradiation and nutlin treatments providing robust increases in p21 mRNA expression in all MEF genotypes (Figure S4A). Furthermore, in line with previous reports (Ray Chaudhuri et al., 2016), 53bp1 KO did not rescue Brca1-mutation-associated replication fork protection defects (Figure S4B). We therefore considered the possibility that 53bp1 KO restored the viability of mice through residual HR. Despite low levels, we were nevertheless routinely able to detect Rad51 foci in a minority of Brca1^ΔC/ΔC;53bp1^−/− cells (Figure 3A). We hypothesized that 53bp1 KO-induced hyperactivation of DNA end resection might be sufficient to stimulate residual Brca1-independent Rad51 loading that in turn provides a level of HR activity capable of supporting organismal development. Indeed, RNAi-mediated depletion of Palb2 or Brca2 completely abolished Rad51 IRIF in Brca1^ΔC/ΔC;53bp1^−/− cells (Figures S4C-S4E). These findings indicate that Brca1^ΔC/ΔC;53bp1^−/− cells have decreased HR proficiency and are more sensitive to PARPi treatment than Brca1 wild-type or Brca1^Δ11/Δ11;53bp1^−/− cells. However, 53bp1 KO likely promotes some residual HR through Brca1-independent but Palb2-Brca2-dependent and inefficient Rad51 loading.

53BP1 Depletion Promotes PARPi Resistance in Subsets of Cancer Cell Lines

We aimed to gain broader insight into the ability of 53BP1 loss of function to promote HR and PARPi resistance in human BRCA1 mutant cancers. In addition to full-length BRCA1, human cell lines often express the BRCA1-Δ11q splice isoform that is homologous to the mouse Brca1-Δ11 protein and lacks the majority of exon-11-coding amino acids (Figure 4A) (Hill et al., 2014; Wang et al., 2016a; Xu et al., 1999). We confirmed that BRCA1 exon-11-mutation-containing SUM149PT, UWB1.289, and L56BRC1 cell lines do not express full-length BRCA1 but do express the BRCA1-Δ11q splice variant (Figures 4A and 4B). In contrast, MDA-MB-436, HCC1395, and SUM1315MO2 cells have BRCA1 mutations located outside of exon 11 that result in low or undetectable levels of truncated BRCA1 proteins due to folding issues and proteasomal degradation (Figures 4A and 4B). Cell lines expressed similar levels of 53BP1, RPA32, PALB2, BRCA2, and RAD51 (Figure 4B).

Figure 4. 53BP1 Depletion in BRCA1 Mutant Cancer Cell Lines.

(A) Cartoon showing the FL BRCA1 protein and the relative location of mutations (*) in BRCA1 mutant cancer cell lines. The predicted effects of mutations are indicated. L56BRC1, SUM149PT, and UWB1.289 harbor exon 11 mutations and are capable of expressing the BRCA1-Δ11q splice isoform (shown below).

(B) Western blot analyses of the indicated proteins in MDA-MB-231 (231), SUM149PT (149), UWB1.289 (UWB), L56BRC1 (L56), MDA-MB-436 (436), HCC1395 (1395), and SUM1315MO2 (1315) cells. Full-length and BRCA1-Δ11q proteins are indicated by arrows.

(C) Western blot analyses of 53BP1 protein expression after the indicated cell lines were subject to NT or 53BP1 shRNA sequence 1 or 2.

(D) Cell lines were treated with increasing concentrations of rucaparib and colony formation assessed. BRCA1 WT MDA-MB-231 cells were also assessed for rucaparib sensitivity as a comparator. Values represent mean and SEM. LC50 values for 3 independent experiments are shown. ***p < 0.001; **p < 0.01; *p < 0.05 compared with NT shRNA for each cell line (unpaired t test). See Figure S5A for fold change values.

(E) Cells expressing NT or 53BP1 shRNA#2 were assessed for RPA32 IRIF by immunofluorescence as described in Figure 1. The number of foci-positive cells (mean and SEM) for 3 independent experiments is shown, **p < 0.01; *p < 0.05 compared to NT (unpaired t test). See Figure S5C for representative images.

(F) Cells were assessed for RAD51 IRIF as described in Figure 1. MDA-MB-231 BRCA1 WT cells were included as a positive control for RAD51 foci. To account for potential cell cycle or growth differences between cell lines, RAD51-foci-positive cells were normalized to geminin-positive cells. The number of foci-positive cells (mean and SEM) that were also geminin staining positive for 3 independent experiments is shown. **p < 0.01; *p < 0.05 compared to NT (unpaired t test). See Figure S5D for representative images.

BRCA1 mutant cell lines were engineered to express either NT or two unique 53BP1 shRNAs (Figure 4C), and PARPi sensitivity was com pared to BRCA1 wild-type control MDAMB-231 cells that are known to be HR proficient and PARPi resistant. Two individual 53BP1-targeting shRNAs significantly increased the LC50 values of rucaparib-treated SUM149PT,UWB1.289, and L56BRC1 cells compared to NT shRNA, with cells demonstrating similar or higher levels of resistance than BRCA1 wild-type MDA-MB-231 cells (Figure 4D). In contrast, although MDA-MB-436, HCC1395, and SUM1315MO2 cells gained relatively small increases in LC50 values with 53BP1 shRNA, cells remained between 106- and 311-fold more sensitive to rucaparib treatment than MDA-MB-231 cells (Figures 4D and S5A). Of note, observed phenotypes were not due to differential effects of shRNA on cell-cycle distributions (Figure S5B).

We investigated the impact of 53BP1 depletion on the initial steps of HR by measuring markers of DNA end resection and RAD51 loading. As expected, 53BP1 shRNA significantly increased RPA32 IRIF, a marker of DNA end resection, in all cell lines (Figures 4E and S5C). In contrast, RAD51 IRIF significantly increased in SUM149PT, UWB1.289, and L56BRC1 cells but remained low in MDA-MB-436, HCC1395, and SUM1315MO2 cells treated with 53BP1 shRNA (Figures 4F and S5D).

We also examined the importance of BRCA1 on the ability of 53BP1 depletion to rescue RAD51 IRIF in a panel of non-transformed and cancer cell lines. In non-transformed cell lines, BRCA1 siRNA reduced RAD51 IRIF, but combined BRCA1 and 53BP1 siRNA did not significantly increase RAD51 foci (Figure S6A). In BRCA1 wild-type cancer cell lines, although mild RAD51 IRIF increases could be detected with combined 53BP1 and BRCA1 siRNA relative to BRCA1 depletion alone in OVCAR5 and MCF7 cell lines, RAD51 IRIF remained undetectable when only BRCA1-foci-negative cells were counted (Figure S6B), indicating that 53BP1 siRNA-induced RAD51 IRIF was associated with residual BRCA1 foci. In BRCA1 mutant cancer cell lines, 53BP1 siRNA did not rescue RAD51 IRIF in BRCA1 siRNA-treated UWB1.289 cells. In contrast, we were able to detect a portion of SUM149PT cells that lacked BRCA1 foci but did contain RAD51 foci in combined BRCA1 and 53BP1 siRNA-treated cells. Furthermore, HCC1937 cells completely lacked BRCA1 foci regardless of siRNA treatment, and cells demonstrated robust RAD51 IRIF (Figure S6C). Altogether, these data point toward cell line to cell line variation in the absolute dependence on BRCA1 for RAD51 loading, with the majority of cell lines tested requiring BRCA1 for efficient RAD51 IRIF, including cells with 53BP1 depletion.

BRCA1 Hypomorphs Promote 53BP1-Gene-KO-Induced PARPi Resistance

It is possible that incomplete RNAi-mediated depletion of 53BP1 accounted for milder rescue phenotypes observed in a subset of cell lines. To compare the effects of RNAi to a more comprehensive loss of 53BP1 activity in human cancer cell lines, we generated CRISPR/Cas9-mediated 53BP1 gene KOs in BRCA1 mutant SUM149PT and MDA-MB-436 cells. For each cell line, three clones with loss of detectable 53BP1 protein expression were selected for further characterization (Figure 5A). Here, SUM149PT 53BP1 KO clones 3, 12, and 18 displayed 55-, 54-, and 67-fold (p < 0.05, unpaired t test) increased rucaparib LC50 values compared to sg_GFP control cells. In contrast, MDA-MB-436 53BP1 KO clones did not demonstrate any significant level of PARPi resistance (Figure 5B). Furthermore, SUM149PT 53BP1 KO clones exhibited increased RAD51 IRIF, whereas RAD51 IRIF remained undetectable in MDA-MB-436 53BP1 KO clones (Figure 5C).

Figure 5. 53BP1 Gene KO Using CRISPR/Cas9.

(A)Cells were treated with sg_G FP orsg_53BP1. Forsg_53BP1-treated cells,we established multiple clones per cell lineand assessed 53BP1 protein expression by western blotting. Selected clones with undetectable 53BP1 protein expression are shown for protein expression relative to the sg_GFP control cells.

(B) Cells from (A) were treated with increasing concentrations of rucaparib and colony formation assessed. Mean and SEM colony formation is shown from 3 independent experiments.

(C) Cells from (A) were assessed for RAD51 IRIF as described in Figure 1. The number of foci-positive cells (mean and SEM) from 3 independent experiments and representative images are shown. *p < 0.05, **p < 0.01 compared to GFP (unpaired t test). MDA-MB-436+BRCA1 FL proteins were also assessed as a positive control. Scale bar, 10 μm.

(D) SUM149PT sg_53BP1 KO clone#12 cells were subjected to scram bled (Sc) or each of 3 different BRCA1-targeting siRNAs. Top: western blot showing BRCA1-Δ11q depletion. Bottom : mean and SEM colony formation of cells treated with 1 μM rucaparib normalized to vehicle-treated cells from 3 independent experiments. *p < 0.05, **p < 0.01 compared to Sc (unpaired t test).

(E) Cells from (D) were assessed for RPA32 and RAD51 IRIF as described in Figure 1. The number of foci-positive cells (mean and SEM) from 3 independent experiments and representative images are shown. ***p < 0.001 compared to Sc (unpaired t test). Scale bar, 10 μm.

SUM149PT cells harbor a BRCA1 exon that blocks the synthesis of full-length BRCA1, but cells are capable of expressing the BRCA1-Δ11q splice variant. To determine if 53BP1 KO-induced PARPi resistance required BRCA1 hypomorphic protein activity, SUM149PT+53BP1KO#12 cells were treated with scrambled or 3 individual BRCA1 siRNAs capable of depleting the BRCA1-Δ11q protein, as well as 1 μM rucaparib. BRCA1 siRNA resulted in 4-, 6-, and 46-fold (p < 0.05, unpaired t test) decrease in the number of colonies forming in the presence of rucaparib compared to scrambled-siRNA-treated cells (Figure 5D). Furthermore, BRCA1 siRNA had little impact on RPA32 IRIF, but RAD51 IRIF was reduced 4-, 4-, and 11-fold (p < 0.001, unpaired t test) compared to scrambled-siRNA- treated cells (Figure 5E). Therefore, 53BP1 gene KO produced phenotypes similar to RNAi-mediated depletion, whereby BRCA1 hypomorphic protein expression was required for robust RAD51 foci and PARPi resistance.

BRCA1 and 53BP1 Mutational Status Impacts Tumor PARPi Sensitivity

The BRCA1 CC domain facilitates the direct interaction between BRCA1 and PALB2 and enables the formation of a larger BRCA1-PALB2-BRCA2-RAD51 complex. We investigated the importance of the BRCA1-PALB2 interaction for HR and PARPi resistance in the setting of BRCT-less BRCA1 protein expression and 53BP1 depletion. The BRCA1 L1407P missense mutation, previously shown to abrogate the BRCA1-PALB2 interaction (Anantha et al., 2017; Sy et al., 2009), was introduced into the BRCA1-ΔBRCT constructs and expressed in MDA-MB-436 cells. We confirmed that full-length BRCA1+L1407P was unable to interact with PALB2 and consequently BRCA2 and RAD51 in co-immunoprecipitation experiments. However, RING-and BRCT-domain-mediated interactions with BARD1 and CtIP, respectively, remained intact (Figure 6A). Additionally, we confirmed protein interactions retained by the BRCA1-ΔBRCT protein, where loss of the BRCT domain results in failure to interact with associated proteins such as CtIP but BARD1, PALB2, BRCA2, and RAD51 interactions were intact. In contrast, BRCA1-ΔBRCT+L1407P was unable to com plex with PALB2, BRCA2, RAD51, and CtIP and only retained the RING domain interaction with BARD1 (Figure 6A).

Figure 6. Effect of BRCA1 and 53BP1 Mutational Status on Tumor PARPi Sensitivity.

(A) Western blot analyses of MDA-MB-436 cells engineered to express mCherry (mCh), FL BRCA1 (FL), and BRCA1-ΔBRCT (ΔBRCT)that were WT for the coiled-coil region (WT) or harbored the L1407P(L-P) missense mutation and subjected to immunoprecipitation using hemagglutinin (HA) antibody and immunoblotted for the indicated proteins.

(B) Western blot analyses of MDA-MB-436 cells expressing either BRCA1 FL or BRCA1-DBRCT (DBRCT) proteins, which are eithercoiled-coil WT or harbor the L1407P (L-P) mutation. Additionally, cells expressed NT or 53BP1 (53) shRNA.

(C) Cells from (B) were assessed for RPA32 and RAD51 IRIF as described in Figure 1. Left: representative images and number of foci-positive cells (mean and SEM) are shown from 3 independent experiments. ***p < 0.001 compared with relative NT shRNA cells (unpaired t test). Scale bar, 10 μm.

(D) Cells from (B) were assessed for colony formation in the presence of 100 nM rucaparib. Mean and SEM from 3 independent experiments normalized to colony formation of BRCA1 FL-expressing cells are shown. **p < 0.01 compared with relative NT shRNA cells (unpaired t test).

(E) MDA-MB-436 tumor xenografts expressing or BRCA1-ΔBRCT proteins that are either coiled-coil WT or harbor the L1407P (L-P) mutation and treated with NT or 53BP1 shRNA targeting. BRCA1 and 53BP1 expression status in tumor xenografts was confirmed by IHC staining. MDA-MB-436 cells with mCherry and BRCA1 FL add-back tumors were included as controls for staining. Scale bar, 10 μm.

Next, we examined the ability of 53BP1 shRNA to restore RPA32 and RAD51 loading in cells that expressed either BRCA1-ΔBRCT wild-type or BRCA1-ΔBRCT+L1407P (Figure 6B). Here, while 53BP1 shRNA increased RPA32 IRIF regardless of BRCA1 status, RAD51 IRIF only increased in BRCA1-ΔBRCT-expressing cells, with BRCA1-ΔBRCT+L1407P demonstrating a complete absence of RAD51 IRIF (Figure 6C). Furthermore, 53BP1 shRNA increased the number of PARPi-resistant BRCA1-ΔBRCT wild-type-expressing cells to a similar level as that observed in cells expressing full-length-BRCA1. However, 53BP1 shRNA had no impact on PARPi sensitivity in cells expressing BRCA1-ΔBRCT+L1407P (Figure 6D). Similar results were obtained with BRCA1-Δ11q+L304P-expressing cells (Figures S7A and S7B).

The latter observations prompted us to explore the role of BRCA1 and 53BP1 mutational and expression status on PARPi responsiveness in BRCA1 mutant tumors. We subcutaneously injected MDA-MB-436 cells that expressed mCherry or BRCA1-full-length controls as well as BRCA1-ΔBRCT or BRCA1-ΔBRCT+L1407P with NT or 53BP1 shRNA into the flanks of NSG mice. BRCA1 and 53BP1 expression status was confirmed in tumor xenografts by immunohistochemical (IHC) staining (Figure 6E). Mice were treated with 200 mg/kg rucaparib twice daily for 10 days. Rucaparib strongly inhibited the growth of mCherry-expressing tumors but had little impact on full-length BRCA1-expressing control tumors (Figure 6F). Rucaparib treatment mildly but significantly delayed tumor growth in BRCA1-ΔBRCT+NT shRNA-expressing tumors but had no impact on the growth of BRCA1-ΔBRCT+53BP1 shRNA-expressing tumors. In contrast, rucaparib resulted in a marked delay of tumor growth in BRCA1-ΔBRCT+L1407P-expressing tumors, regardless of 53BP1 status (Figure 6F). Altogether, these data indicate that 53BP1 knockdown requires an intact BRCA1-PALB2 complex for maximal tumor PARPi resistance.

DISCUSSION

BRCA1 plays a key role in preserving genome integrity through its involvement in HR DNA repair, with several well-characterized functions, including the initiation of DNA end resection, RAD51 loading, and homologous DNA pairing (Bouwman et al., 2010; Bunting et al., 2010; Sy et al., 2009; Zhang et al., 2009a, 2009b; Zhao et al., 2017). However, potential ambiguity regarding whether BRCA1 activity is obligatory for the completion of each of these steps arises from variation in cell lines, organisms, and genetic backgrounds used to study BRCA1. Previous reports demonstrated that 53bp1 gene KO was capable of rescuing embryonic viability, Rad51 loading, HR repair, and PARPi resistance in Brca1-deficient mice (Bunting et al., 2010, 2012; Cao et al., 2009). These studies point toward the initiation of DNA end resection as a key function of BRCA1 and suggest that BRCA1 may be dispensable after end resection when RAD51 loading occurs. In the current study, we report that in spite of 53BP1-KO-induced DNA end resection, RAD51 loading and HR remain suboptimal in the absence of BRCA1 activity and that protein products generated from mutant BRCA1 alleles influence the ability of 53BP1 KO to induce maximal HR and PARPi resistance.

In line with previous reports demonstrating rescue of HR in Brca1^Δ11 mice that express the Brca1-Δ11 protein (Cao et al., 2009), we discovered that 53BP1 KO efficiently rescued HR in BRCA1 exon 11 mutant cancer cell lines that express the BRCA1-Δ11q splice isoform. 53bp1 KO was also previously shown to rescue HR and embryonic viability in Brca1^Δ2/Δ2 mice, initially thought to be Brca1 null due to a stop codon early in the reading frame (Bunting et al., 2012). However, a subsequent report showed that cells harboring the Brca1^Δ2 allele expressed a RING-less Brca1 protein that retains the CC domain and formed robust Rad51 IRIF (Li et al., 2016). In another study, 53BP1-KO-induced PARPi resistance was shown in cells harboring the Brca1^SCo allele, which deletes exons 5 and 6 (Bouwman et al., 2010). We speculate it may be possible for these cells to also produce a Brca1 RING-less protein from the Met294 in-frame translation start site in exon 11, equivalent to human Met297 (Drost et al., 2016; Wang etal., 2016b). Additional studies have assessed PARPi sensitivity in mammary tumor cells derived from mice harboring a more deleterious Brca1^Δ5–13 allele where any protein product would lack the Brca1 CC and be unable to interact with Palb2. Here, similar to our results, 53bp1 shRNA-treated Brca1^Δ5–13 cells had markedly lower levels of Rad51 IRIF compared to Brca1 wild-type cells (Jaspers et al., 2013).

We were unable to detect any truncated Brca1 protein product in Brca1^ΔC/ΔC;53bp1^−/− MEFs, indicating that the Brca1^ΔC mutation is (or is close to) a null allele. Furthermore, Brca1^ΔC/ΔC;53bp1^−/− MEFs demonstrated low levels of Rad51 foci, increased numbers of chromosomal rearrangements, and fewer gene conversion events and were sensitive to PARPi treatment, all indicative of inefficient HR. Despite the latter observations, intriguingly, 53bp1 KO rescued the viability of homozygous Brca1^ΔC mice, raising the possibility that HR-independen 53bp1 activities contribute to embryonic lethality. However, we were unable to detect a reduction in p53 transcriptional activity in Brca1^+/+;53bp1^−/− or Brca1^ΔC/ΔC;53bp1^−/− cells, and Brca1^ΔC/ΔC;53bp1^−/− MEFs senesced in normoxic conditions, suggesting p53 activity was intact. In line with previous reports, 53bp1 KO also did not rescue replication fork degradation associated with Brca1-deficient cells (Ray Chaudhuri et al., 2016). 53bp1 KO might also direct repair toward the resection-dependent and RAD51-independent single-strand annealing (SSA) or alternative end-joining (Alt-EJ) pathways (Bhargava et al., 2016), although these are highly mutagenic and unlikely to support organismal viability. In contrast, some level of residual HR likely essential for organismal and cellular viability. Recently, BRCA2-induced HR was shown to be critical for the viability of human mammary epithelial cells (Feng and Jasin, 2017). We hypothesize that Brca1^ΔC/ΔC;53bp1^−/− mice are viable, because 53bp1 KO directs DSBs toward HR, and that low levels o endogenous DNA breaks, which arise during embryo development, are tolerated due to residual Brca1-independent Rad51 loading and HR. Nevertheless, cells with inefficient Rad51 loading are unable to withstand greater numbers of DNA breaks encountered in the presence of oxidative damage or PARPi. Indeed, although Rad51 IRIF was impaired, we were able to detect some Rad51-positive Brca1^ΔC/ΔC;53bp1^−/− MEFs, and residual foci formation was dependent on Palb2 and Brca2 activity (Figure S7C).

Despite surviving partial or complete embryonic development, both Brca1^Δ11/Δ11;53bp1^+/+ and Brca1^ΔC/ΔC;53bp1^−/− cells are sensitive to PARPi. Residual HR is afforded by either the Brca1^Δ11 protein in the case of Brca1^Δ11/Δ11;53bp1^+/+ cells or loss of 53bp1 protein activity in the case of Brca1^ΔC/ΔC;53bp1^−/− cells. However, HR is further enhanced to a level that supports embryonic development and PARPi resistance in Brca1^Δ11/Δ11;53bp1^−/− mice when HR-promoting alleles are combined resulting in both Brca1^Δ11 expression and 53bp1 deficiency. Of note, despite exhibiting PARPi resistance, meiotic defects were previously reported in Brca1^Δ11/Δ11;53bp1^−/− and Brca1^Δ2/Δ2;53bp1^−/− mice (Cao et al., 2009; Li et al., 2016), suggesting that HR remains suboptimal or is not regulated correctly compared with Brca1 and 53bp1 wild-type animals. Parallel to observations in mice, low levels of HR are sufficient to support human viability, as evidenced by Fanconi anemia patients, who are also highly sensitive to DNA-damaging chemotherapy treatments.

53BP1 depletion had a mild effect on PARPi resistance in human BRCA1 mutant HCC1395, SUM135MO2, and MDA-MB-436 breast cancer cell lines that have low levels of truncated BRCA1 protein expression. Furthermore, while CRISPR/Cas9-mediated 53BP1 KO resulted in substantial PARPi resistance in SUM149PT cells, 53BP1 KO did not increase PARPi resistance in MDA-MB-436 cells. In contrast, a previous report using transcription activator-like effector nucleases (TALENs) to generate 53BP1 KO MDA-MB-436 cells found that a 53BP1 KO clone showed increased PARPi resistance (Yang et al., 2017). These differences may be a result of variations in the expression levels of truncated BRCA1 between MDA-MB-436 cells or clones. However, similar to our findings, a recent report showed that CRISPR/Cas9-mediated PARP1 KO provided PARPi resistance in SUM149PT, but not MDA-MB-436, cells. The authors also demonstrated that PARP1 KO-induced PARPi resistance in SUM149PT cells was dependent on residual activity provided by the BRCA1-Δ11q protein (Pettitt et al., 2018).

We found that loss of 53BP1 activity alone did not support tumoral PARPi resistance in in vivo experiments using the MDA-MB-436 breast cancer cell line xenograft system. Here, an intact BRCA1-PALB2 interaction and loss of 53BP1 expression were both required for robust tumor growth in the presence of rucaparib. Supporting this observation, we previously discovered the co-occurrence of increased BRCA1 hypomorphic protein expression and 53BP1 loss of function in therapy-resistant BRCA1 mutant cancer cells and patient tumors (Johnson et al., 2013). However, it is important to note that although MDA-MB-436 cells strictly required BRCA1 for RAD51 foci, we demonstrate cell line to cell line variation in the absolute dependency for BRCA1 at the RAD51 loading step of HR. For example, in contrast to MDA-MB-436 cells, Brca1^ΔC/ΔC;53bp1^−/− MEFs had low but consistently detectable levels of Rad15 IRIF, despite being null for Brca1 protein expression. Whereas HCC1397 cells demonstrated robust RAD51 IRIF in the absence of a detectable BRCA1 protein. Such cell line variations may stem from the activity of additional factors that contribute to PALB2 recruitment to DNA breaks such as RNF168, ATR or RPA (Luijsterburg et al., 2017; Murphy et al., 2014; Yazinski et al., 2017). Moreover, KO of DNA end resection inhibitory proteins involved in the shieldin complex were recently shown to promote PARPi resistance in a range of human and mouse BRCA1 mutant cell lines (Barazas et al., 2018; Dev et al., 2018; Mirman et al., 2018; Noorder-meer et al., 2018; Tomida et al., 2018). We anticipate that similar to 53BP1 KO, the presence of BRCA1 hypomorphs in combination with shieldin KO may facilitate a greater level of PARPi resistance compared with shieldin KO and the absence of mutant BRCA1 protein expression.

In conclusion, we report that suboptimal HR is capable of supporting organismal viability but is not sufficient for tumor suppression or robust PARPi resistance. Furthermore, our results indicate that specific BRCA1 mutant alleles and their protein products may impact the route of PARPi resistance acquired by cancer cells. 53BP1 KO has previously been shown to have no effect on the PARPi sensitivity of cells containing BRCA2 or PALB2 mutations (Bowman-Colin et al., 2013). BRCA1 CC domain disruptions may represent another mutational subgroup that is less likely to develop 53BP1 loss-of-function-associated PARPi resistance. Future work will be necessary to assess potential relationships between specific BRCA1 mutations and particular PARPi resistance pathways in the clinical setting.

STAR ★METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse Anti-53BP1 Monoclonal antibody	Millipore	Cat# MAB3802; RRID: AB_2206767
Rabbit Anti-BARD1 Polyclonal antibody	Bethyl	Cat# A300–263A; RRID: AB_2061250
Rabbit Anti-BRCA1 Polyclonal antibody (C-terminal)	Millipore	Cat# 07–434; RRID: AB_2275035
Mouse Anti-BRCA1 Monoclonal antibody (N-terminal)	Millipore	Cat# OP92T-10UG; RRID: AB_564282
Anti-Brca1 antibody (recognizing mouse Brca1)	Gift from A. Nussenzweig
Rabbit Anti-BRCA2 Polyclonal antibody	Bethyl	Cat# A303–434A; RRID: AB_10952240
Rabbit Anti-CtIP Polyclonal antibody	Bethyl	Cat# A300–488A; RRID: AB_2175262
Mouse Anti-Geminin Monoclonal antibody	Abnova	H00051053-M01; RRID: AB_606308
Rabbit Anti-Geminin Polyclonal antibody	Protein Tech	Cat# 10802–1-AP; RRID: AB_2110945
Mouse Anti-HA Monoclonal antibody	Cell Signaling	Cat#2367S; RRID: AB_10691311
Rabbit Anti-PALB2 Polyclonal antibody	Bethyl	Cat# A301–247A; RRID: AB_890608
Rabbit Anti-Rad51 Polyclonal antibody	GeneTex	Cat# GTX100469; RRID: ABJ951602
Rabbit Anti-RPA32 Polyclonal antibody	Santa Cruz	Cat# sc-28709; RRID: AB_2238546
Rat Anti-RPA32/RPA2 Monoclonal antibody	Cell Signaling	Cat# 2208; RRID: AB_2238543
Mouse Anti-RPA2 (Ab-2) Monoclonal antibody	CalBiochem	Cat# NA18–100UG; RRID: AB_213121
Rabbit Anti-Tubulin Polyclonal antibody	Cell Signaling	Cat# 2148; RRID: AB_2288042
Rabbit Anti-Histone3 Polyclonal antibody	Cell Signaling	Cat# 2148; RRID: AB_331563
Alexa Fluor 594-AffiniPure Donkey Anti-Mouse IgG	Jackson ImmunoResearch Labs	Cat# 715-585-150; RRID: AB_2340854
Alexa Fluor 594- AffiniPure Donkey anti-rabbit antibody	Jackson ImmunoResearch Labs	Cat# 711-585-152; RRID: AB_2340621
Fluorescein (FITC)-AffiniPure Donkey Anti Mouse	Jackson ImmunoResearch Labs	Cat# 715-095-150; RRID: AB_2340792
Fluorescein (FITC)-AffiniPure Donkey Anti-Rabbit	Jackson ImmunoResearch Labs	Cat# 711-095-152; RRID: AB_2315776
Donkey Anti-Rabbit IgG ECL Antibody, HRP Conjugated	GE Healthcare	Cat# NA9340-1ml; RRID: AB_772191
Sheep Anti-Mouse IgG ECL Antibody, HRP Conjugated	GE Healthcare	Cat# NA9310-1ml; RRID: AB_772193
Rabbit anti-Mouse IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 594	Thermo Fisher Scientific	Cat# A-11062; RRID: AB_2534109
Chicken anti-Rat IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 488	Thermo Fisher Scientific	Cat# A-21470; RRID: AB_2535873
Mouse anti-BrdU antibody	BD Biosciences	Cat# 347580; RRID: AB_400326
Rat anti-BrdU antibody	Bio-Rad	Cat# OBT0030; RRID: AB_609568
Dako EnVision+ System- HRP Labeled Polymer Anti-Rabbit	DAKO	Cat# K4002
Dako EnVision+ System- HRP Labeled Polymer Anti-Mouse	DAKO	Cat#K4000
Chemicals, Peptides, and Recombinant Proteins
Rucaparib	Clovis Oncology	N/A
Hydroxyurea	Sigma	Cat#H8627
Nutlin-3a	Selleckchem	Cat#S8059
Bpu10I	New England Biolabs	Cat#R0649S
BssSI	New England Biolabs	Cat#R0587
AscI	New England Biolabs	Cat#R0558
Vectashield mounting medium with DAPI	Vector	Cat#H-1200
5-lodo-2-deoxyuridine	Sigma	Cat#l7125
5-Chloro-2-deoxyuridine	Sigma	Cat#C6891
Tween-20	Sigma	Cat#1379
Triton X-100	Sigma	Cat#X100
Paraf ormaldehyde	Sigma	Cat#158127
OneTaq® Hot Start 2X Master Mix	New England Biolabs	Cat#M048S
Lipofectamine RNAimax	Invitrogen	Cat#13778075
Transit-LT1	Mirus	Cat#MIR2304
Sulforhodamine B	Sigma	Cat#230162
Sodium hydroxide	Sigma	Cat#221465
Sodium deoxycholate	Sigma	Cat#D6750
DAKO Target Retrieval Solution	DAKO	Cat#S1699
Dako Liquid DAB+ Substrate Chromogen System	DAKO	Cat#K3467
DA VINCI green diluent	BIOCARE	Cat#PD900L
Background Sniper	BIOCARE	Cat#BS966H
Meyer’s Hematoxylin	Sigma	Cat#MHS32
Polybrene	Boston Bioproducts	Cat#BM-862W
Alt-R Cas9	IDT	Cat# 1074181
MG-132	MedChem Express	Cat#HY-13259
Critical Commercial Assays
FxCycle PI/RNase Staining Solution	Thermo Fisher	Cat#F10797
QuikChange Lightning Site-Directed Mutagenesis Kit	Agilent	Cat#210518
Gateway LR Clonase II Enzyme Mix	Invitrogen	Cat#11791
GuidelT sgRNA/n Vitro transcription Kit	Takara	Cat#632636
Pierce Classic IP Kit	Thermo Fisher	Cat#26146
NE-PER Nuclear and Cytoplasmic Extraction Kit	Thermo Fisher	Cat#78833
MEGAclear Transcription Clean-Up	Invitrogen	Cat#AM1908
SA-β-Gal Staining	Sigma	Cat#CS0030
RNeasy Plus Mini Kit	QIAGEN	Cat#74134
Hprt expression Taqman assay	Life Technologies	Mm00446968_m1
Experimental Models: Cell Lines
Human: MDA-MB-231	ATCC	Cat#CRM-HTB-26; RRID: CVCL_0062
Human: SUM149PT	ASTERAND	Cat#CS-07; RRID: CVCL_3422
Human: UWB1.289	ATCC	Cat#CRL-2945; RRID: CVCL_B079
Human: L56Br-C1	Fox Chase Cancer Center	RRID: CVCL_DF84
Human: MDA-MB-436	ATCC	Cat#HTB-130; RRID: CVCL_0623
Human: HCC1395	ATCC	Cat#CRL-2324; RRID: CVCL_1249
Human: SUM1315MO2	ASTERAND	Cat#CS-014; RRID: CVCL_5589
Human: MCF7	ATCC	Cat#HTB-22; RRID: CVCL_0031
Human: OVCAR3	ATCC	Cat#HTB-161; RRID: CVCL_0465
Human: OVCAR5	Fox Chase Cancer Center	RRID: CVCL_1628
Human: HCC1937	ATCC	Cat#CRL-2336; RRID: CVCL_0290
Human: h-TERT RPE1	ATCC	Cat#CRL-4000; RRID: CVCL_4388
Human: IMR-90	ATCC	Cat#CCL-186; RRID: CVCL_0347
Human: MCF10F	Fox Chase Cancer Center	RRID: CVCL_0598
Human: HEK293T	Fox Chase Cancer Center	RRID: CVCL_0063
Experimental Models: Organisms/Strains
B6;129-Trp53bp1tm1Jc/J(Trp53BP1−/−)	The Jackson Laboratory	Cat#006495
B6C3F1/J	The Jackson Laboratory	Cat# 100010
B6C3F1/J Brca1ΔC	This paper	N/A
STOCK Brca1tm2.1Cxd/Nci (Δ11)	NCI	Cat#01XC9
NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ	FCCC Laboratory Animal Facility	Cat#005557
Oligonucleotides
Flexitube BRCA1 siRNA 9	QIAGEN	Cat#SI 00299495
Flexitube BRCA1 siRNA 13	QIAGEN	Cat#SI 02654575
Flexitube BRCA1 siRNA 15	QIAGEN	Cat#SI 02664368
SMARTpool: ON-TARGETplusTP53BP1 siRNA	Dharmaeon	Cat#L003548
SMARTpool: ON-TARGETplus Palb2 siRNA	Dharmaeon	Cat#L042993
SMARTpool: ON-TARGETplus Brca2 siRNA	Dharmaeon	Cat#L167423
Primers for site directed mutagenesis, mouse genotyping, qRT-PCR and oligosfor 53BP1 shRNA and HR assay, see Table S1.	This study
Recombinant DNA
Gateway pENTR 1A vector	Invitrogen
pDest-IRES-GFP Destination vector	Life Technologies
pCW-Cas9	Fox Chase Cancer Center	Gift Christoph Seeger
pLX-SG1	Fox Chase Cancer Center	Gift Christoph Seeger
VSV-G/pMD2.G	Addgene	Cat# 12259
psPAX2	Addgene	Cat# 12260
Software and Algorithms
Prism Software v6.0	Graph Pad Software
Adobe Photoshop CS v8.0	Adobe
NIS Elements	Nikon
FlowJo v10.1	FlowJo

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Contact for Reagent and Resource Sharing

Further information and requests for reagents should be directed to and will be fulfilled by the Lead Contact, Neil Johnson (neil. johnson@fccc.edu)

Cell Lines

Cell lines were obtained from the ATCC or Asterand and tested negative for mycoplasma. Cell line identities were confirmed by IDEXX analysis. All cell lines were grown in media with 10% FBS and penicillin/streptomycin except otherwise mentioned. MDA-MB-231, MCF7, MDA-MB-436, OVCAR5 cells were grown in RPMI1640, for the OVCAR3 cell line glutam axand insulin were added. HCC1395 and L56BRC1 were grown in RPMI media supplemented with HEPES and Sodium Pyruvate. SUM1315MO2 and SUM149PT were grown in Hams F-12 supplemented with insulin, EGF, HEPES and hydrocortisone. HEK293T, IMR-90 and MEFs were grown in DMEM. HEK293T were grown without penicillin/streptomycin and for the IMR-90 and the MEFs, media was supplemented with 15% FBS, non-essential amino acids, sodium pyruvate and L-glutamine. MCF10F was grown in high calcium media. H-TERT RPE1 was grown in DMEM:F-12 supplemented with L-Glutamine, HEPES, sodium pyruvate and sodium bicarbonate. With the exception of the MEFs which were grown in a 2% oxygen incubator all cell lines were incubated at 37°C with 5% CO2 in a humidified incubator.

Mice

All experiments involving animals were approved by the Fox Chase Cancer Center (FCCC) Institutional Animal Care and Use Committee. The mouse Brca1^ΔC allele was generated by the FCCC transgenic mouse facility (detailed below). Before initiating experiments, germline transmission of the allele was confirmed. Embryos harboring the Brca1^Δ11 allele (Xu et al., 2001) were obtained from the NCI and mice with the Trp53bp1^−/+ (Ward et al., 2003) allele were obtained from the Jackson Laboratory. For xenograft experiments, 6–8 weeks old NOD-scid IL2Rg^null females were acquired from the FCCC Laboratory Animal Facility.

METHOD DETAILS

BRCA1 mutant cDNA generation

HA-BRCA1 cDNA was cloned into the pENTRIA Gateway Entry vector (Invitrogen) and shuttled into a pDest-IRES-GFP Destination vector (Life Technologies) using Gateway LR Clonase II Enzyme Mix (Invitrogen). To generate cDNA that would translate to C-terminal truncated proteins an AscI site was introduced at the end of BRCA1 exon 24 by site-directed mutagenesis (Agilent). Restriction digestion was made using the AscI site (BRCA1 5773 nt) and a BssSI site present at BRCA1 4771 nt. This was followed by site- directed mutagenesis with the insertion of a stop codon at the end of the exon 11 coding region (c.4099G > T) generating BRCA1-ΔCC+ΔBRCT or at the end of the exon 15 coding region (c.4675G > T) generating BRCA1-ΔBRCT. The BRCA1-Δ11q construct was previously described (Wang et al., 2016a). The L1407P missense mutation was generated by site directed mutagenesis (Agilent). See Table S1 for primer sequences.

Lentivirus Production

HEK293T were transfected with either pDest-IRES-GFP, pLKO.1, pCW-Cas9 or pLX-SG1 with psPAX2 packaging and VSV-G/pMD2.G envelope plasmids usingTransit-LT1 (Mirus). Culture media was collected 60 hr later and filtered through a 0.45 μm filter.

Lentivirus cell infection

Cell lines were infected with lentivirus using polybrene (Boston Bioproducts) and infected cells were selected based on either GFP expression or antibiotic resistance. Clones were generated from the pCW-Cas9 infected cells to select for cells with high Cas9 expression. Cas9 expression was conditionally induced by the addition of doxycycline (4 μg/ml) to culture media. sgRNA targeting 53BP1 or GFP was cloned into the pLX-SG1 vector using BsmBI sites. After blasticidin selection clones were generated and 53BP1 protein expression was evaluated by immunoblot.

siRNA transfection

For siRNA experiments cells were transfected using 10 to 40 nM of siRNA with Lipofectamine RNAimax (Invitrogen). All the experiments were carried out at least 48 hr after siRNA transfection. See Key Resources Table for details of siRNA sequences.

RNA extraction and qRT-PCR

RNA was extracted using the RNeasy Plus Mini Kit (QIAGEN) following manufacture instructions. For quantitative RT-PCR, RNA was tested for quality on a Bio-analyzer (Agilent). RNA concentrations were determined with a spectrophotometer (NanoDrop; Thermo Fisher Scientific). RNA was reverse transcribed using Moloney murine leukemia virus reverse transcriptase (Ambion, Thermo Fisher Scientific) and a mixture of anchored oligo-dT and random decamers. Two reverse transcription reactions were performed for each sample using either 100 or 25 ng input RNA. Assays were used in combination with Taqman Universal Master Mix or Power SYBR Green Master Mix and run on a 7900 HT Sequence Detection System (all from Applied Biosystems, Thermo Fisher Scientific). Cycling conditions were 95° C for 15 min, followed by 40 (2-step) cycles (95°C, 15 s; 60°C, 60s). Ct values were converted to quantities (in AU) using a standard curve (4 points, 4-fold dilutions) established with a calibrator sample. For H prt expression the Taqman assay (Life Technologies: Mm00446968_m1) was used. Primers used for p53, p21, Palb2, Brca2 and Brca1 are listed in Table S1.

PARPi resistance assays

MDA-MB-436 cells expressing different BRCA1 proteins were seeded in 6 well plates in the following seeding densities: 2×10⁴, 1×10⁴, 0.5×10⁴, 0.25×10⁴, 1.25×10³ and 0.625×10³ cells/well. The cells were treated with 100 nM of rucaparib for 3 weeks. Media containing rucaparib was changed every 48–72 hr. After 3 weeks resistant colonies were fixed with 3:1 methanol: acid acetic solution and stained with crystal violet. The mean colony formation from 3 different experiments were expressed as the mean percentage of resistant colonies ± SEM relative to BRCA1 full-length expressing PARPi treated cells.

Colony formation assays

Cells were seeded in 6-well plates at 500 and 1000 cells per well in the presence of increasing concentrations of PARPi. Colony formation was assessed after 2 weeks with crystal violet staining. Lethal concentration (LC₅₀) values (concentration required to reduce colony formation by 50%) were calculated using GraphPad Prism software. Experiments were repeated at least 3 times.

Sulforhodamine B (SRB) assays

Depending on growing rate MEFs were seeded at densities ranging from to 2000 to 5000 cells per well in 24 well plates. The next day increasing concentrations of PARPi was added to culture media. Media was changed every 3 days. After 10 days cells were fixed with 50% Trichloroacetic acid followed by 30 min incubation at room temperature with 0.4% SRB in a 1% acetic acid solution. Plates were washed and SRB solubilized with 10 mM Tris base pH 10.5. Staining intensity was read on a plate reader at 510 nm wavelength. Cell growth was calculated as a percentage of the DMSO vehicle treated control wells for each group.

DNA replication fork protection

To measure replication fork stability MEFs were labeled with 50 μM CldU for 30 min, washed and exposed to 250 μM IdU for 30 min. This was followed by exposure of 2 μM of hydroxyurea for 3 hr (Sigma). Cells were collected, and resuspended in PBS. Cells were lysed (200 mM Tris-HCl, 50 mM EDTA, 0.5% SDS, pH 7.4), and DNA fibers stretched onto glass slides (Superfrost Microscope Slides precleaned, Fisher Scientific) by tilting slides 30 degrees, followed by fixing in a methanol: acetic acid solution (3:1) for 10 min. The fibers were denatured with 2.5 M HCl for 2.5 hr, washed with PBS and blocked with 2% BSA in PBST for 40 min. Labeled DNA with CldU and IdU were stained with anti-BrdU antibodies recognizing CldU and IdU (Novus 1:500 and BD Bioscience 1:100, respectively). Anti-mouse Alexa 594 (1:300) and anti-rat Alexa 488 (1:300) were used for secondary antibodies. Images were captured with Nikon NIU Upright Fluorescence Microscope. Fiber length was quantified using ImageJ software.

SA-β-Gal Staining

MEFs were seeded in a 12 well plate and SA-β-Gal Staining was performed using a commercial staining kit (Sigma). Cells were incubated overnight at 37°C in a low CO₂ incubator.

Cell cycle analyses

Cells were harvested and fixed with 50% ethanol. Cells were then washed with PBS and re-suspended in 0.5 mL FxCycle PI/RNase Staining Solution (Life Technologies, Thermo Fisher Scientific) and incubated at room temperature for 30 min. Data were acquired using a BD LSR II Flow Cytometer and analyzed using FlowJo software.

Immunofluorescence and microscopy

Cells were fixed with 4% paraformaldehyde (PFA) and permeabilized with 1% Triton-Xin PBS for 10 min. Primary antibodies (BRCA1 MS110 Millipore, 1:1000, RAD51 GeneTex 1:1000, RPA32 CalBiochem 1:1000) were incubated overnight at 4 degrees in 5% goat serum. Alexa Fluor or FITC conjugated antibodies (Jackson ImmunoResearch Labs 1:1000) were incubated for 1 hr at room tem perature. For RPA32 foci experiments, we pre-extracted cells prior to fixation as previously described (Di Micco R, d ‘Adda di Fagagna F. Indirect immunofluorescence with preextraction (in situ cell fractionation). Protocol Exchange. 2006). Briefly, we treated cells with ice-cold cytoskeleton buffer (10 mM Pipes pH 6.8,100 mM NaCl, 300 mM sucrose, 3mM MgCl₂,1m M EGTA, 0.5% Triton X-100) for 5 min on ice followed by 5 min incubation with ice-cold cytoskeleton stripping buffer (10 mM TrisHCl pH 7.4,10 mM NaCl, 3mM MgCl₂, 1% Tween 40(v/v), 0.5% sodium deoxycholate) then fixed in 4% PFA and continued to be processed as described above. To detect mouse RPA32 we incubated with Rat Anti-RPA32 (4E4, Cell Signaling) and Chicken anti-Rat IgG (Thermo Fisher Scientific) as secondary antibody. Slides were mounted with mounting media containing DAPI. Images were captured using a Nikon NIU Upright Fluorescence Microscope and generated images using Nikon NIS Elements software. For IR experiments, we routinely fixed cells 8 hr after treatment with 10 Gy. For analyses, we counted a minimum of 100 cells per condition per cell line. Foci positive cells were nuclei that contained 5 foci for BRCA1 and RAD51 or 10 foci for RPA32 and 53BP1. Each experiment was performed at least 3 times with biological replicates.

Western Blotting and Immunoprecipitation

Nuclear extracts were derived using the NE-PER Nuclear and Cytoplasmic Extraction Kit following the manufacturer’s instructions (Thermo Scientific). Protease inhibitor and phosphatase inhibitors were added (Millipore, Cat#524625 and #539131). Lysates were separated by SDS-PAGE and transferred to a PDVF membrane (ImmobilonP, Millipore). Membranes were blocked for 1 hr with 5% nonfat milk in PBS. Primary Antibodies were incubated overnight at 4 degrees directed against 53BP1 (Millipore 1:1000), BARD1 (Bethyl 1:2000), BRCA1 (Millipore MS110, 1:500), BRCA2 (Bethyl 1:2000), CtIP (Bethyl 1:2000), PALB2 (Bethyl 1:2000), RAD51 (Genetex 1:1000), RPA32 (Santa Cruz 1:1000), Tubulin (Cell signaling 1:2000) and Histone 3 (Cell signaling 1:1000). Brca1 (1:1000 recognizing mouse species Brca1, gift from A. Nussenzweig), HRP conjugated secondary antibodies were used. Immunoprecipita-tion of BRCA1 complexes were carried out using the Pierce Classic IP Kit (Thermo Fischer). Briefly, 1 mg of nuclear extract was precleared with Control Agarose Resin Slurry for 1 hr at 4 degrees. Each lysate was incubated with 10 μg of anti-HA antibody (Cell Signaling) and incubated at 4 degrees overnight. To retrieve the com plex 20 μl of Protein A/G agarose beads was incubated with the antibody/lysate sample for 3 hr. The beads were washed three times and eluted with 50 μl of acidic Elution Buffer provided by the manufacturer. To neutralize the low pH of the buffer, 5μl of 1 M Tris pH 9.5 was added.

Metaphase spreads

Assay was performed by the Fox Chase Cancer Center Cytogenetics Core Facility. Briefly, cells were exposed to colcemid (0.01 μg/ml) for 12 h, harvested with 0.075 M KCl and fixed with methanol-acetic acid (3:1) (Sementino et al., 2018).

Homologous recombination reporter assays

The HR reporter assay was previously described (W yatt et al., 2016). Briefly, 2×10⁶ cells were transfected with 5 μg pGL3-U6-sgRNA-PGK-puromycin (gift of Xingxu Huang, Addgene # 51133), 5 μg Flag-Cas9 (gift of Xingxu Huang, Addgene # 44758), 10 μg HR long donor (Wyatt et al., 2016) and 1 μg pEGFP-N2 (Clontech) by Neon transfection kit (Invitrogen) using a 1350V, 30 ms pulse in a 100 μL chamber. The Rosa26-specificsgRNA sequence used here was: 5’-ACTCCAGTCTTTCTAGAAGA. See Table S1 for additional primer sequences. 48 hr post transfection, a portion of the cells were analyzed by flow cytom etry to quantify the transfection efficiency, and the remaining cells were harvested for genomic DNA extraction (QIAGEN). Digital PCR was performed to quantify the frequency of gene conversion events using the primers and Taqman probe listed below. The HR signal was normalized to 5000 copies of genomic DNA, measured using a Chromosome 6 control dPCR assay, using primers/probes sequences listed below. Analysis of dPCR data were performed using QuantaSoft (Bio-Rad).

Generation of Brca1^ΔC mutant mice

Candidate guide RNAs targeting the CC region of Brca1 were selected using crispr.mit.edu and tested in vitro using Guide-it Complete sgRNA Screening System (Takara), see Table S1 for sequences. The selected guide for the mouse generation was the one that targeted closer to the Leucine 1363 residue, had the highest cut efficiency and low off-target score. The sgRNA was transcribed in vitro using the Guide-it Complete sgRNA Screening System (Takara) following manufacture instructions. RNA was purified using the MEGAclear Transcription Clean-Up (Invitrogen). The transcribed RNA together with Alt-R Cas9 (IDT) were resuspended (20 ng/μl) in injection buffer (1 mM Tris HCl, pH 7.5; 0.1 mM EDTA) and 1–2μl injected in B6C3/F1 single cell 0.5 day embryos (pronuclei). The embryos were transferred to Swiss Webster recipient females (Taconic Tac:SW).

Mouse Genotyping

DNA was extracted using 240 μl of 50 mM NaOH. Samples were incubated at 97° C degrees for 30 min, followed by addition of 60 μl of Tris-HCl 1 M pH 8. After centrifugation 1 μl was used for each PCR reaction. To evaluate Brca1 status a 2kb fragment was generated using Brca1 Forward and Brca1 Reverse primers (Table S1). Following amplification fragments were incubated at 37°C degrees for 2 hr with the Bpu10I restriction enzyme. The mutant copy sequence displays the restriction site whereas the WT does not. After digestion samples were run on a 1% agarose gel. WT animals showed 1 band, heterozygous 3 bands and homozygous mutant 2 bands. The Trp53bp1 KO and Brca1-Δ11 deletion mice were genotyped as previously described (Ward et al., 2003; Xu et al., 2001).

PARPi treatm ent and peripheral blood analysis

To evaluate tolerability to PARPi, 6 to 8 weeks old, sex matched males and females littermates were randomly selected to receive either 200 mg/kg rucaparib or vehicle orally bi-daily for 5 days. Mice were euthanized in accordance with institutional guidelines. Peripheral blood was also collected from rucaparib and vehicle treated animals and analyzed using an automated analyzer using mouse specific parameters (Abaxis VetScan5).

MEF generation

Mouse embryos (12.5–13.5 days postcoitum) were dissected and internal organs removed. Dissociation was performed by aspiring embryos through a 16-G needle followed by 5–10 min incubation with 0.25%Trypsin at 37°C. The solution was mixed again by trituration and incubated for an additional 5–10 min. After centrifugation cell pellet was resuspended in 75 cm² flasks. Experiments with MEFs were carried out within 3 passages of their derivation to avoid senescence. All MEFs were also cultured in hypoxic incubator prior to being set up for experiments to increase proliferative capacity.

Embryo dissection

Pregnant mice were euthanized 9.5 days postcoitum, uterus was removed and separated from the mesometrium. Individual embryos were dissected under microscope vision (Nikon SMZ1500).

Xenograft studies

MDA-MB-436 cell lines were subcutaneously implanted in 6–8 weeks-old female NSG mice. When tum or size reached between 150 and 180 mm³ animals were randomly selected for rucaparib (Clovis Oncology) or vehicle (PBS) treatment. Rucaparib was administered by oral gavage 200 mg/kg, twice a day for 10 days with a 2-day break after the first 5 days. Tumors were measured twice a week with digital calipers and volume was calculate using the formula (length × width²)/2. When tumors reached 1500 mm³ mice were euthanized in accordance with institutional guidelines.

Histologic and immunohistochemical staining

Slides were deparaffinized with xylene and hydrated with ethanol. Antigen retrieval was performed using Tris/EDTA buffer (DAKO Target Retrieval Solution) and endogenous peroxidase was quenched by immersing slides in 3% hydrogen peroxide solution (30% H2O2, Fisher BP2633–500) diluted in methanol. The following primary antibodies were used: BRCA1 N-terminal (MS110 Millipore 1:400), BRCA1 C-terminal (EMD Millipore #07–434,1:7,500) and 53BP1 (Millipore 1:4000). Antibodies were diluted with Da-Vinci Green Diluent (Biocare) and incubated on slides overnight at 4°C in a humidified slide chamber. Slides were then washed and incubated with EnVision+System HRP Labeled Polymer Anti-Rabbit or Anti-Mouse for 1 hr at RT. Specimens were washed, then developed with DAB solution (Dako) and counterstained in Meyer’s Hematoxylin (Sigma-Aldrich).

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis was performed using GraphPad Prism software. Unpaired t test was used in Figures 3, 4, 5, and 6, one-way ANOVAwas used in Figure 2 and two-way ANOVA in Figure 1. Mean and SEM values are shown, p <0.05 was considered statistically significant. Statistically significant p values and number of replicates are indicated in the Figure legends. There were similar variances between statistical groups compared. For the PARPi sensitivity experiments (Figure 3), 3–5 animals were used for each condition. In xenograft experiments (Figure 6), 5 animals were used for each condition. Mouse tumor growth curve data were analyzed using linear mixed effects models. For each experiment, the differences in growth rates of log-transformed tumor volumes between vehicle and rucaparib treatments were calculated. We fit linear mixed-effects models with random intercepts and slopes to account for within-mouse correlation, and tested the interaction effects between treatment and time. We assessed whether there was a non-linear effect of time by examining residual plots, and tested models which include a quadratic effect of time for improved fit using likelihood ratio tests. Reported p values are from tests of the interaction terms, interpreted as differences in growth rates between vehicle and rucaparib treatments.

Highlights.

• The embryonic viability of Brca1^ΔC/ΔC mice can be rescued with 53bp1 knockout

• Brca1^ΔC/ΔC;5 3 bp 1^–/– mice are tumor prone and sensitive to PARP inhibitor treatment

• 53BP1 loss provides varying levels of PARPi resistance in BRCA1 mutant cell lines

• Efficient RAD51 loading and HR requires BRCA1 hypomorphs that interact with PALB2