The BRCA1 coiled-coil domain is dispensable for suppression of tandem duplications and tolerance of FANCM loss Authors: Namrata M Nilavar, Alberto Marin-Gonzalez, Francesca Menghi, Daniel Nguyen, Nicholas A Willis, Ellen Wientjens, Bing Xia, Jos Jonkers, Edison T Liu, Ralph Scully Published: 2026 ABSTRACT AbstractBRCA1-linked cancers contain abundant ~10 kb ‘Group 1’ tandem duplications (TDs). Group 1 TDs form at a Tus/Ter replication-fork barrier in DNA-end resection-defective mouse embryonic stem (mES) cells lacking Brca1 exon 11. To elucidate how BRCA1 suppresses Group 1 TDs, we analyzed Brca1 coiled-coil (CC)-domain mutants—separation-of-function alleles impaired for homologous recombination (HR) through loss of PALB2-binding and RAD51-loading functions but competent for DNA-end resection. Notably, Brca1 CC mutants retain the ability to suppress Group 1 TDs in the Tus/Ter system and in a mouse model of Brca1-linked mammary tumorigenesis. These data suggest that Brca1 CC domain-mutant cancers follow a path of tumorigenesis distinct from that of other Brca1-linked cancers. FANCM is a TD co-suppressor, loss of which is synthetic lethal/sick in Brca1 exon 11-deleted cells. In contrast, Fancm deletion unexpectedly improves the growth of Brca1 CC mutant and Brca2 mutant mES cells. Thus, Group 1 tandem duplication formation and Fancm synthetic lethality are linked phenotypes, potentially related to defective BRCA1-mediated DNA end resection but genetically separable from BRCA1/PALB2-mediated RAD51 loading.Subject terms: Cancer; DNA Replication, Recombination & Repair Introduction The hereditary breast/ovarian cancer (HBOC) predisposition genes BRCA1 and BRCA2 regulate homologous recombination (HR), a DNA repair pathway involved in the error-free repair of double-strand breaks (DSBs), protection of replication forks, and the prevention of genomic instability (Chen et al, 2018; Scully et al, 2019). Several features of BRCA1- and BRCA2-linked cancers are shared, reflecting their roles as key HR genes. These features include elevated levels of genomic instability, a specific mutational signature (SBS3), and sensitivity to inhibitors of poly(ADP-ribose) polymerase (PARP) (Lord and Ashworth, 2017; Nik-Zainal et al, 2016). However, the genomes of BRCA1-linked breast and ovarian cancers uniquely reveal abundant ~10 kb (“Group 1”) non-homologous tandem duplications (TDs), distributed widely across the genome (Menghi et al, 2016; Nik-Zainal et al, 2016; Willis et al, 2017). “Non-homologous” here refers to the absence of sequence homology between the two TD breakpoints, which define the edges of the duplicated segment and are apposed to form the non-homologous TD breakpoint junction (Scully et al, 2025). BRCA1-linked Group 1 TDs differ from the more common types of cancer-related TDs, which are typically >100 kb in size. Strong (~ 95%) mutual concordance between BRCA1 loss and Group 1 ‘TD phenotype’ (TDP) in human breast and ovarian cancer suggests a causal relationship between BRCA1 loss and Group 1 TD formation (Willis et al, 2017). Indeed, conditional inactivation of Brca1 in the mouse mammary epithelium recapitulates the Group 1 TD phenotype (Menghi et al, 2018). Loss of BRCA2 is not associated with Group 1 TDP in either human or mouse tumorigenesis, nor is Brca2 required for the formation of the Group 1 TD in the mouse mammary tumor model. These observations raise several questions: what mechanisms normally suppress Group 1 TDs, and what functions specific to BRCA1 mediate Group 1 TD suppression? BRCA1 and BRCA2 operate at distinct steps of the HR pathway (Chen et al, 2018; Scully et al, 2019; Zhao et al, 2019). The BRCA1-BARD1 heterodimer promotes short-range DNA end resection by MRE11/RAD50/NBS1 through interactions with the scaffolding protein, CtIP (Cejka and Symington, 2021; Cruz-Garcia et al, 2014; Yu et al, 1998). BRCA1-BARD1 also activates long-range resection mediated by EXO1 or by the Bloom’s (BLM) or Werner’s (WRN) syndrome helicases in association with DNA2 (Ceppi et al, 2024; Salunkhe et al, 2024). These resection activities convert the blunt or near-blunt ends of a DSB to single-stranded (ss)DNA with long 3’ tails, which are then rapidly bound by the ssDNA binding protein complex RPA. BRCA2 displaces RPA from ssDNA and loads RAD51 in its place, forming the RAD51 nucleoprotein filament, which then executes the search for a homologous template (Bell et al, 2023; Jensen et al, 2010; Shahid et al, 2014; Thorslund et al, 2010). In addition to its role in DNA end resection, BRCA1-BARD1 interacts with the product of another HBOC gene, PALB2, facilitating BRCA2 recruitment and RAD51 loading (Anantha et al, 2017; Dray et al, 2010; Sy et al, 2009; Xia et al, 2006; Zhang et al, 2009). BRCA1 also acts at later stages of HR, facilitating strand exchange with a homologous template by the RAD51 nucleoprotein filament (Zhao et al, 2017). BRCA1 binds PALB2 directly via its coiled coil (CC) domain, mutations in which disrupt the interaction (Sy et al, 2009; Zhang et al, 2009). Consequently, BRCA1 CC domain mutants are defective for HR but competent for DNA end resection. Human germline BRCA1 CC missense alleles are exceptionally rare, and these alleles are classified as variants of uncertain significance (VUS). However, a coiled coil missense mutation in the mouse Brca1 CC domain, p.L1363P, was found to promote mouse mammary tumorigenesis, suggesting that its human counterpart p.L1407P, a patient-derived variant, is pathogenic (Pulver et al, 2021). Consistent with the separation-of-function properties of BRCA1 CC domain mutants, mice that are compound heterozygotes of Brca1 ∆11 and Brca1 ∆CC display interallelic complementation in mouse development (Nacson et al, 2020). BRCA1’s function in DNA end resection may explain its role in single-strand annealing (SSA)—a BRCA2/RAD51-independent DSB repair mechanism whereby two neighboring homologous repeats, either side of a DSB, can be collapsed to one copy by annealing of the complementary sequences exposed by DNA end resection (Chen et al, 2018; Scully et al, 2019). To study mechanisms of stalled fork repair in mammalian cells, we previously adapted the Escherichia coli Tus/Ter replication fork barrier (RFB) to provoke bidirectional site-specific replication fork stalling on a mammalian chromosome (Willis et al, 2014). By targeting an HR reporter containing an array of six 23 bp Ter sequences to a defined chromosomal locus, we were able to quantify HR and other repair outcomes triggered by fork stalling at the Tus/Ter RFB. We found that conservative HR by short tract gene conversion (STGC) at Tus/Ter is mediated by the canonical BRCA1-BRCA2-RAD51 pathway, whereas replicative long tract gene conversion (LTGC) at Tus/Ter is independent of BRCA1 and BRCA2 (Willis et al, 2014). In addition, non-homologous Group 1 TDs form at Tus/Ter in Brca1 mutant mouse embryonic stem (mES) cells, but not in Brca2 mutants (Willis et al, 2017). This pattern of genetic dependency for TD formation suggests that the Tus/Ter system recapitulates the BRCA1-linked Group 1 TD mechanism observed in cancer. Analysis of Tus/Ter-induced Group 1 TD formation in mES cells identified BRCA1, BARD1, and CtIP as major TD suppressors in wild-type (WT) cells (Willis et al, 2017). The CtIP connection suggested a possible role for DNA end resection in TD suppression. Notably, Tus/Ter-induced TDs in DNA end resection-defective Brca1 mutants lacking the large central in-frame exon 11 (Brca1∆11) are elevated ~15-fold by loss of either of two stalled fork motor proteins, FANCM or BLM, whereas loss of FANCM or BLM has minimal impact on TD formation in cells expressing WT Brca1 (Panday et al, 2021; Willis et al, 2017). FANCM and BLM act in concert as TD suppressors, and loss of Fancm is synthetic lethal/synthetic sick with Brca1∆11 mutation (Panday et al, 2021). Indeed, specific inactivation of FANCM ATPase function is sufficient to disable Brca1∆11/– cell growth. Whole-genome CRISPR/Cas9 screens in primary human and mouse cells also identified specific synthetic lethality following combined loss of BRCA1 and FANCM, but not in cells lacking BRCA2 and FANCM (Adam et al, 2021; preprint: Dossin et al, 2026). These findings, if verified in cancer models, identify FANCM as a potential target for therapy in BRCA1-linked cancer. We proposed that TDs form by a replication restart/replication bypass mechanism, in which FANCM/BLM suppresses fork restart at the Tus/Ter RFB, while BRCA1 promotes SSA-mediated collapse of a nascent TD back to single copy (Scully et al, 2025; Scully et al, 2019). Such a mechanism could explain how the combined inactivation of FANCM/BLM and BRCA1-mediated DNA end resection disables different steps of the TD suppression mechanism and thereby synergizes to induce very high levels of TDs and, potentially, other genomic instability outcomes (Panday et al, 2021; Setton et al, 2023). However, the explicit relationships between BRCA1-mediated DNA end resection, Group 1 TD formation, and FANCM synthetic lethality remain to be rigorously tested. Here, we study these relationships by analyzing the stalled fork repair functions of Brca1 CC domain mutants—separation-of-function mutants that disrupt HR without affecting BRCA1-mediated DNA end resection. Our findings suggest that Group 1 TD formation and Fancm synthetic lethality are linked phenotypes associated with defective BRCA1-mediated DNA end resection but are genetically separable from BRCA1/PALB2-mediated RAD51 loading functions. Results A haploid system for genetic analysis of Brca1 in stalled fork repair To support a genetic analysis of Brca1 in stalled fork repair, we generated a system that enables targeted mutagenesis of Brca1 as a haploid genetic element in mES cells. This approach guarantees that subsequent CRISPR/Cas9-mediated mutagenesis of Brca1 cannot induce unplanned changes in a second Brca1 allele that might produce unpredictable alterations in Brca1 function (Egli et al, 2018; Panday et al, 2021). We first determined whether Brca1+/– mES cells display an overt haploinsufficiency phenotype in DSB repair or stalled fork repair. We targeted a previously described Ter-HR reporter (Willis et al, 2014) to the Rosa26 locus of Brca1fl5-13/+ cells, in which Brca1 exons 5–13 are flanked by loxP sites (Fig. EV1A; Appendix Fig. S1A) (Liu et al, 2007). The Brca1∆5-13 allele is functionally null. We then generated Cre-treated derivatives in which Brca1 is either haploid (Brca1∆5-13/+) or diploid (Brca1fl5-13/+) (Appendix Fig. S1B). In assays of DSB-induced HR or Tus/Ter-induced HR, Brca1fl5-13/+ and Brca1∆5-13/+ HR reporter cells were indistinguishable (Appendix Fig. S1C,D). Thus, Brca1 is not overtly haploinsufficient for DSB-induced or stalled fork HR in mES cells. This conclusion is restricted to the cells, DNA repair functions, and genotypes studied here. A haploid system for genetic analysis of Brca1 in stalled fork repair. (A) Schematic of Ter-HR reporter for quantitation of Tus/Ter-induced and DSB-induced HR. Tr-GFP: 5’ truncated GFP allele acts as a donor during HR. RFP cassette elements A and B: artificial exons encoding WT RFP, placed in unproductive orientation. Promoters and polyadenylation signals not shown. STGC short tract gene conversion, LTGC long tract gene conversion, TD Group 1 tandem duplication. Each repair product is associated with fluorescent products, as shown. Red line in TD product: non-homologous TD breakpoint junction. (B) Outline of method for generating haploid mES cells expressing Brca1-HA. Black triangles: LoxP sites flanking Brca1 exon 11 (NCBI exon 10). The deleted Brca1 allele is not shown in the cartoon of Brca1fl11-HA/– allele. (C) Identification of HA-tagged gene product in Brca1fl11-HA/– cells. Upper panel: anti-HA immunoblotting of whole cell extracts. Note the abolition of the BRCA1-HA signal by Brca1 siRNA but not by control Luciferase siRNA (siLuc). FL full-length Brca1 gene product. Lower panel: Vinculin loading control. We next engineered haploid derivatives of Ter-HR reporter mES cells that carry one conditional allele of Brca1, Brca1fl11, in which the large central in-frame exon 11, corresponding to NCBI exon 10 (RefSeq NM_009764), can be conditionally deleted using Cre-mediated recombination (Xu et al, 1999). The Cre-deleted allele, Brca1∆11, is a viable hypomorph that is defective for DNA end resection (Bunting et al, 2010). The second allele, Brca1hyg, contains a hygromycin resistance gene that disrupts Brca1 gene expression (Fig. EV1B). To delete the Brca1hyg allele, we used two sgRNAs to target Cas9 to the extreme 5’ and 3’ ends of the Brca1 open reading frame (ORF) and screened for clones in which one full allele had been deleted (Fig. EV1B). All derived clones were Brca1fl11/–, likely reflecting the inviability of Brca1hyg/– cells. We selected a Brca1fl11/– clone that contained no indels in the retained Brca1fl11 allele at the 5’ and 3’ sgRNA target sites. We next used CRISPR/Cas9-assisted gene targeting to introduce an in-frame linker peptide, followed by an 8x influenza hemagglutinin (HA) epitope tag sequence to the 3’ end of the Brca1 ORF, together with a SMASh degron (Chung et al, 2015) and a neomycin expression cassette fused to a self-cleaving peptide T2A, to generate Brca1fl11-HA/– cells (Fig. EV1B). The SMASh degron is also self-cleaving so that, in the absence of the degron-activating drug asuprenavir, the C-terminus of BRCA1 is fused to only the linker peptide and the 8xHA epitope tag. The full-length ~220 kDa gene product, BRCA1-8xHA, was detectable by western blotting and was abolished by siRNA-mediated depletion of Brca1 mRNA (Fig. EV1C). Anti-HA immunoblotting also revealed lower molecular weight Brca1 products, reflecting products of alternative splicing or proteolytic cleavage of full-length BRCA1. We tested the HR functions of parental Brca1fl11/hyg, Brca1fl11/–, and Brca1fl11-HA/– cells and found them to be equivalent (Appendix Fig. S1E). Generation of Brca1 coiled coil domain mutants To engineer Brca1 CC domain mutants, we used CRISPR/Cas9-assisted gene targeting to introduce the missense mutation Brca1 p.L1363P into Brca1fl11-HA/– cells (See Methods). This allele is equivalent to the human BRCA1 p.L1407P missense variant, which shows impaired PALB2 binding (Sy et al, 2009). Because it is a rare allele in the human population, BRCA1 p.L1407P is classified in ClinVar as a variant of uncertain significance (VUS). However, Brca1 p.L1363P promotes tumorigenesis in a mouse model of mammary carcinoma (Pulver et al, 2021). We identified several independent Brca1L1363P-HA/– clones that express the gene product at levels equivalent to WT BRCA1, as well as two non-targeted clones in which small, in-frame deletions of 9 or 12 bp had been introduced at the sgRNA target site within the CC-coding region (Figs. 1A,B and EV2A). These deletion alleles were termed Brca1CC∆9bp-HA/– (abbreviated to ‘∆9 bp’ in some figures) and Brca1CC∆12bp-HA/–, respectively. Generation of Brca1 coiled coil domain mutants. (A) Schematic of Brca1 mutant alleles generated by Cre-mediated recombination (Brca1∆11-HA/–) or by CRISPR/Cas9-mediated gene editing. Brca1 allele names are shown, together with abbreviations used in this and subsequent figures and corresponding amino acid residues in mouse and human BRCA1. (B) Identification of HA-tagged gene product in Brca1HA/– cells with mutations as shown. *Parental Brca1fl11/– clone, lacking HA-tag. Upper panel: anti-HA immunoblotting of whole cell extracts. FL: arrow shows full-length BRCA1. ∆CC: arrow shows Brca1∆CC-HA/– gene product. Lower panel: vinculin loading control. (C) Detection of PALB2 association with products of different Brca1HA/– alleles. Left panels: whole cell extracts. Right panels: anti-HA immunoprecipitates. Immunoblots were probed with the antibodies shown. Lower panel: Vinculin loading control in whole cell extracts. Source data are available online for this figure. Sequence analysis of Brca1 CC mutant clones and exon skipping in Brca1∆CC-HA/– cells. Brca1 allele abbreviations as in Fig. 1A. Nucleotide sequence is shown in black and amino acids in orange. (A) Nucleotide sequence of mouse Brca1 NCBI exon 12 encoding amino acids 1357-1376 (equivalent residues in human BRCA1: 1401-1420). The red arrow shows the predicted site of Cas9-induced DSB for the generation of CC mutants. The red dotted line denotes deletion of nucleotides. Sequences of Brca1 p.L1363P, Brca1 CC∆9 bp, and Brca1 CC∆12 bp alleles are shown. L to P mutation in Brca1 p.L1363P mutants is highlighted in red. Engineered silent mutations in potential sgRNA PAM sites are highlighted in green. (B) Brca1 NCBI exon 11 and 12 nucleotide sequence encoding amino acids 1335 to 1408. ∆CC clone #59 has a frame-shift mutation, whereas clone #70 is in-frame. (C) RT-qPCR results showing amplification across three different exon-exon junctions (normalized to parental Brca1fl11/hyg clone #109). Note the absence of NCBI exons 10–11 amplification in ∆11 mutant clones and over-representation of NCBI exon 10–13 RT-qPCR product in ∆CC clones, reflecting exon skipping. Data show mean and s.e.m. of three independent biological repeats (n = 3). (D) Sequence of WT Brca1 cDNA and Brca1 ∆CC cDNA encompassing NCBI exons 10–13. The black dashed line represents the continuation of the sequence. The red dotted line represents the deletion of nucleotides. ∆CC clones #59 and #70 revealed identical cDNA sequences, consistent with skipping of NCBI exons 11 and 12 in the mRNA and cDNA. (E) BRCA1 amino acid sequence region encoded by the 3’ end of Brca1 NCBI exon 10 to the 5’ end of NCBI exon 13. Amino acids encoded by NCBI exons 10 and 13 are highlighted in gray. Amino acids encoded by NCBI exons 11 and 12 are highlighted in orange and yellow, respectively. Predicted amino acid sequence based on sequencing of Brca1 gDNA is shown for each ∆CC clone. Actual amino acid sequence based on Brca1 ∆CC cDNA sequencing shows in-frame deletion resulting from skipping of NCBI exons 11–12. We also used CRISPR/Cas9 dual incisions to generate larger deletions within the Brca1 CC-encoding region (Brca1∆CC-HA/–), spanning the two exons (NCBI exons 11 and 12) that encode this domain. We retrieved two clones, one of which (#70) contained a predicted in-frame deletion and one (#59) a frame-shifting deletion (Fig. EV2B). Despite these genetic differences, the two alleles encoded HA-tagged Brca1 gene products of equal size that were slightly smaller than the WT BRCA1 protein (Fig. 1B). Analysis of Brca1 mRNA by RT-qPCR across specific exon-exon junctions and by direct sequencing of Brca1 cDNA revealed that NCBI exons 11 and 12 were skipped in both ∆CC clones, resulting in an in-frame deletion of 81 amino acids, effectively deleting the entire coiled coil domain (Fig. EV2C–E). This exon-skipping explains why ∆CC allele #59 does not truncate the protein, and why the two different ∆CC alleles encode similar gene products. We presume that the emergence of the exon-skipping adaptation is achieved through selection against cells that fail to skip NCBI exons 11 and 12, since the exon-skipped mRNA is virtually undetectable in WT cells (Fig. EV2C). It is not clear how reconfigurations of the splicing apparatus enable the observed exon skipping to occur. We used Cre-mediated recombination to delete exon 11 in Brca1fl11-HA/– cells (see “Methods”) and identified the Brca1∆11-HA/– gene product by western blotting (Fig. 1A,B). We used immunostaining and immunofluorescence microscopy to assess the subcellular localization of each Brca1 gene product (Appendix Fig. S2). Consistent with previous reports, BRCA1∆11-HA and all other BRCA1 proteins localized to characteristic nuclear foci (Huber et al, 2001; Scully et al, 1996). The anti-HA signal of WT and CC mutant proteins co-localized with foci detected by an antibody raised against a polypeptide encoded by Brca1 exon 11, whereas this exon 11-specific BRCA1 signal was absent in Brca1∆11-HA/– cells (Appendix Fig. S2). The anti-HA signal was absent in Brca1fl11/– cells, which lack the HA epitope tag. The mean numbers of HA-positive foci per cell were similar across all Brca1HA/– genotypes (Appendix Fig. S3A). We next assessed PALB2 binding by BRCA1 in each genetic background, using anti-HA immunoprecipitation (IP) followed by immunoblotting for endogenous PALB2 (Fig. 1C). Despite equivalent levels of PALB2 in cell extracts, IPs from BRCA1 CC mutants revealed a severely attenuated PALB2 signal, with BRCA1∆CC-HA revealing the most severe PALB2 binding defect. Interestingly, BRCA1∆11-HA also revealed reduced binding to PALB2 in comparison to WT BRCA1-HA, as was noted previously (Nacson et al, 2020). These results confirm that the Brca1 CC mutants generated here are defective for PALB2 interaction and, hence, lack the contribution to RAD51 loading that this interaction normally contributes to the HR process. Defective HR and intact DNA end resection in Brca1 coiled coil mutants Brca1 CC mutants are known to be defective for DSB-induced HR, but their function in stalled fork HR has not yet been studied. We showed previously that Brca1∆11/hyg cells exhibit reduced STGC in response to a Tus/Ter RFB, although less markedly than in response to a replication-independent DSB (Willis et al, 2014). Tus/Ter-induced LTGC is independent of Brca1 and Brca2. To study stalled fork HR in Brca1 CC mutant cells, we quantified Tus/Ter-induced and I-SceI-induced HR in Brca1HA/– mutant clones vs. isogenic WT controls, which had been exposed to the same CRISPR/Cas9 or Cre treatments but had retained WT Brca1. Mean values of Tus/Ter-induced STGC (averaged across the clones shown) were 64% of WT in Brca1∆11-HA/– cells, 23% in Brca1∆CC-HA/– cells, and 37% in Brca1L1363P-HA/– cells (Fig. 2A–C). Equivalent mean values of I-SceI-induced STGC were 36% of WT in Brca1∆11-HA/– cells, 27% in Brca1∆CC-HA/– cells, and 50% in Brca1L1363P-HA/– cells (Fig. EV3A–C). STGC defects in the Brca1CC∆9bp-HA/– clone resembled those of Brca1L1363P-HA/– clones. There was no consistent impact of Brca1 CC domain mutation on Tus/Ter-induced LTGC (Fig. 2B,C). These results confirm our previous finding that defects in Tus/Ter-induced STGC in Brca1 ∆11 cells and Brca1 BRCT mutant cells are less pronounced than DSB-induced STGC (Willis et al, 2014). Interestingly, Brca1 CC mutants behave differently, since they reveal approximately equivalent STGC defects in response to the two stimuli. Our findings suggest that the RAD51 loading function of BRCA1 (defective in Brca1 ∆CC mutants) is equally important for DSB-induced and stalled fork HR, while the DNA end resection function of BRCA1 (defective in both Brca1 exon 11 and BRCT mutants) plays a more important role in DSB-induced HR than in stalled fork HR. Tus/Ter-induced HR and PARPi sensitivity of Brca1 coiled coil mutants. HR was quantified in independent Brca1 mutant (gray) or isogenic Brca1 WT (white) clones as described in “Methods”. Brca1 allele abbreviations as in Fig. 1A. Data show mean and s.e.m. of n independent experiments. (A) Brca1∆11-HA/– (n = 4). (P values: STGC 0.0469; LTGC 0.0702; LTGC:Total HR 0.0120). (B) Brca1∆CC-HA/– cells (n = 3). (P values: STGC 0.0080; LTGC 0.2764; LTGC:Total HR 0.0935). (C) Brca1L1363P-HA/– and Brca1CC∆9bp-HA/– (n = 5). (P values: STGC < 0.0001; LTGC 0.0551, LTGC:Total HR 0.0023). Statistical analysis: RM-one way ANOVA with Geisser–Greenhouse correction: ns: not significant, *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.0001. (D) Colony formation of different Brca1HA/– genotypes in response to concentrations of Olaparib as shown (see “Methods”). Data show the mean of five independent biological replicates normalized to colony numbers in untreated samples of the same genotype (n = 5). Source data are available online for this figure. DSB-induced HR defects in Brca1 mutants. (A, C) DSB-induced HR was quantified in Brca1 mutants vs. isogenic Brca1fl11-HA/– clones. Brca1 allele abbreviations as in Fig. 1A. Data show mean and s.e.m. of n independent experiments. (A) Brca1∆11-HA/– cells (n = 4). (P values: STGC 0.0004; LTGC 0.0635; LTGC:Total HR 0.1513). (B) Brca1∆CC-HA/– cells (n = 3) (P values: STGC 0.0081; LTGC 0.0378; LTGC:Total HR 0.5448). (C) Brca1L1363P-HA/– and Brca1CC∆9bp-HA/– cells (n = 5). (P values: STGC < 0.0001; LTGC 0.0352; LTGC:Total HR 0.0802). Statistical analysis: RM-one way ANOVA with Geisser–Greenhouse correction: ns: not significant. To further characterize the HR functions of the different Brca1 mutant clones, we assessed sensitivity to the PARP inhibitor Olaparib by colony formation assay (Fig. 2D). IC50 values were ~150 nM for WT cells, 2 nM for Brca1∆11-HA/– cells, 3.5 nM for Brca1∆CC-HA/– cells, and ~15 nM for both Brca1L1363P-HA/– and Brca1CC∆9bp-HA/– clones. Notably, Brca1 ∆11 mutants were among the most sensitive to Olaparib and the most impaired in I-SceI-induced STGC, but revealed the mildest defect in Tus/Ter-induced STGC. This discordance suggests that an individual Brca1 mutant’s PARPi sensitivity correlates better with its function in DSB-induced HR than in stalled fork HR. To determine the DNA end resection phenotypes of Brca1 exon 11-deleted and CC domain mutants, we quantified RPA accumulation at a defined chromosomal DSB site using chromatin-immunoprecipitation-sequencing (ChIP-seq; see “Methods”) (Fig. 3A). Three hours after electroporation of CRISPR/Cas9 ribonucleoprotein (RNP) complexes targeting the I-SceI site within the Ter-HR reporter, we observed strong RPA enrichment in regions flanking the DSB site in Brca1fl11-HA/– cells (Fig. 3B). In contrast, the RPA ChIP-seq signal was greatly diminished in Brca1∆11-HA/– cells, consistent with the expected defect in DNA end resection. Brca1 CC domain mutants Brca1L1363P-HA/–, Brca1CC∆9bp-HA/– and Brca1∆CC-HA/– each revealed levels of RPA accumulation similar to those of Brca1fl11-HA/– cells (Fig. 3B,C). Thus, each of the Brca1 CC domain mutants studied here supports WT levels of DNA end resection. Brca1 coiled coil mutants are competent for DNA end resection. (A) Schematic of RPA accumulation on ssDNA in response to DNA end resection of a Cas9-induced DSB, targeted to the I-SceI site within the HR reporter at Rosa26. Intensity of RPA ChIP-seq signal reflects the efficiency of DNA end resection. (B) Cas9 DSB-induced RPA ChIP-seq signals in one representative experiment of 3 independent biological replicates using Brca1 genotypes shown. Brca1 allele abbreviations as in Fig. 1A. See “Methods” for details. (C) Quantitation of RPA enrichment averaged over 30 kb window flanking I-SceI site from three independent experiments (n = 3; data show mean and S.D.). Signals were normalized to WT for each of the replicates. Statistical analysis: unpaired t test with Welch’s correction. ns not significant. (P values WT vs ∆11 0.0034; WT vs LP 0.1359; WT vs ∆9 bp 0.1626; WT vs ∆CC 0.2738). Source data are available online for this figure. Brca1 CC domain mutants are competent for suppression of Tus/Ter-induced TDs We showed previously that FANCM depletion strongly enhances Tus/Ter-induced TD formation in Brca1∆11/hyg cells (Panday et al, 2021; Willis et al, 2017). To test whether Brca1 CC domain mutants are competent for TD suppression, we quantified Tus/Ter-induced TD frequencies in cultures of Brca1fl11-HA/–, Brca1L1363P-HA/–, Brca1CC∆9bp-HA/–, Brca1∆CC-HA/–, and Brca1∆11-HA/– cells that had been depleted of FANCM using siRNA. Controls in this experiment included non-targeting luciferase siRNA (siLuc, negative control) and combined siRNA directed to Brca1 and Fancm (positive control). The former provides the baseline level of Tus/Ter-induced TDs in each clone; the latter tests whether each clone can, in principle, form TDs when BRCA1 and FANCM are co-depleted. Indeed, each cell line revealed induction of Tus/Ter-induced GFP–RFP+ products (i.e., TDs) following siRNA-mediated co-depletion of BRCA1 and FANCM (Fig. 4; Appendix Fig. S3B,C). When depleted of FANCM alone, Brca1fl11-HA/– cells revealed minimal TD induction, whereas Brca1∆11-HA/– cells revealed high levels of Tus/Ter-induced TDs, as expected. Strikingly, unlike in Brca1∆11-HA/– cells, depletion of FANCM alone failed to stimulate Tus/Ter-induced TDs in Brca1L1363P-HA/– or Brca1CC∆9bp-HA/– cells, with TD frequencies equivalent to those observed in WT Brca1fl11-HA/– cells (Fig. 4A; Appendix Fig. S3C). Brca1∆CC-HA/– cells revealed a modest increase in Tus/Ter-induced TDs, but significantly less than in Brca1∆11-HA/– cells (Fig. 4B). Thus, Brca1 CC domain mutants retain the ability to suppress Group 1 TD formation at a Tus/Ter RFB. We showed previously that I-SceI-induced GFP–RFP+ products are not non-homologous TDs, but represent a rare, aberrant outcome of sister chromatid recombination (Elango et al, 2025). Fancm depletion in the context of I-SceI-induced repair failed to generate a significant increase in GFP–RFP+ products in any of the Brca1 mutant clones studied (Appendix Fig. S3D–F). In summary, the data show that the BRCA1 coiled coil domain and, hence, BRCA1/PALB2-mediated RAD51 loading is dispensable for Tus/Ter-induced Group 1 TD suppression. Since Brca1 CC domain mutants are competent for DNA end resection, this result establishes a correlation (albeit a limited one, based on limited numbers of Brca1 alleles tested) between BRCA1-mediated DNA end resection competence and Group 1 TD suppression at a Tus/Ter RFB. Brca1 coiled coil mutants suppress Tus/Ter-induced tandem duplications. Tus/Ter-induced GFP–RFP+ repair products in Brca1 WT and mutant genotypes, in the presence of siRNAs against Fancm and Brca1. Brca1 allele abbreviations as in Fig. 1A. See “Methods” for details. siRNA against luciferase (siLuc): non-specific siRNA negative control. Gray columns: combined siFancm and siBrca1 induces TDs in all cell lines (positive control). Red columns: siFancm and siLuc test TD suppression in the Brca1 genotypes shown. (A) Brca1L1363P-HA/– and Brca1CC∆9bp-HA/– mutants are compared with isogenic WT Brca1fl11-HA/– cells and with Brca1∆11-HA/– cells (n = 4; data shows mean and s.e.m. of 4 independent biological repeats). Note induction of TDs in Brca1∆11-HA/– cells but not in CC mutants in samples treated with siFancm alone. (B) Brca1∆CC-HA/– mutants are compared with isogenic WT cells and Brca1∆11-HA/– cells (n = 4; data show mean and s.e.m. of four independent biological repeats). Statistical analysis: Unpaired t test with Welch’s correction. ns not significant, **P < 0.005, ***P < 0.0005. Source data are available online for this figure. Brca1 p.L1363P mouse mammary tumors do not display a Group 1 TD phenotype Our findings that Brca1 CC domain mutants are competent for suppression of Tus/Ter-induced TDs raised the question of whether cancers driven by Brca1 CC mutation will accumulate Group 1 TDs genome-wide during tumorigenesis and whether the Brca1 CC mutant cancer genome will register as Group 1 TDP. While the Tus/Ter system is an indication of TD formation at one specific euchromatic site, analysis of TDs in the Brca1 CC mutant cancer genome would enable us to test how well the Tus/Ter system predicts Group 1 TD formation in cancer. We were unable to obtain tissue samples from human BRCA1 CC mutant cancers, reflecting their rarity in the human population. We therefore analyzed a well-established mouse model of mammary tumorigenesis, in which K14-Cre or WAP-Cre expression in the mammary epithelium drives biallelic deletion of Trp53 by converting Trp53fl2-10/fl2-10 to null (Trp53∆2-10/∆2-10) together with deletion of conditional alleles of Brca1 (Brca1fl5-13/fl5-13) or Brca2 (Brca2fl11/fl11) (Jonkers et al, 2001; Liu et al, 2007; Menghi et al, 2018). Pulver et al developed this model further, using the Brca1 p.L1363P allele as a cancer driver (Pulver et al, 2021). This is the only tractable mouse model for studying the impact of Brca1 CC mutation on mammary tumorigenesis. We analyzed TDs in the genomes of mammary tumors arising from Cre-expressing Trp53fl2-10/fl2-10 mice in three different Brca genotypes: first, Brca1fl5-13/fl5-13, in which Cre converts both Trp53 and Brca1 to null (Brca1∆5-13/∆5-13;Trp53∆2-10/∆2-10, hereafter termed ‘B1P’); second, Brca2fl11/fl11, in which Cre converts Trp53 and Brca2 to null (Brca2∆11/∆11;Trp53∆2-10/∆2-10, hereafter ‘B2P’); third, Brca1fl5-13/L1363P, in which Cre converts Trp53 and the Brca1fl5-13 allele to null, resulting in functional expression of Brca1L1363P alone (Brca1∆5-13/L1363P;Trp53∆2-10/∆2-10, hereafter ‘Br1L1363P’). We used whole-genome sequencing (WGS) to analyze 5 independent B1P and B2P tumors and 10 Br1L1363P tumors, using WGS of spleen tissue from each mouse to define the reference genome. For some of the analyses reported here, we included samples from our previous analysis of Cre-driven Brca mutant mouse mammary tumors, including controls in which Trp53 was the only conditional allele (Menghi et al, 2018). To confirm loss of the floxed regions via Cre-mediated recombination in each tumor genome, we first manually inspected the Cre-targeted regions using the Integrative Genomic Viewer (IGV), for evidence of (i) a decrease in coverage affecting the Cre-targeted segments of the Δ alleles of Trp53, Brca1 or Brca2 (ii) a deletion rearrangement supporting the recombination of targeted regions, and/or (iii) evidence of the p.L1363P mutation (examples in Appendix Fig. 4A). Br1L1363P cancers contain a single copy of the Brca1∆5-13 allele in their genomes, and generally exhibit high stromal infiltration, as previously described (Pulver et al, 2021). Elevated stromal content can dilute tumor cell-derived genotype signals, making it challenging to determine the true extent of Cre-mediated recombination of floxed alleles in tumor cells (i.e., partial vs. complete). To address this potential confounder, we compared the ratio between the Δ allele frequency (DeltaFr) of the Trp53 gene vs. that of the Brca1 gene in both the Br1L1363P and B1P cancer genomes. A DeltaFr = 1 corresponds to 100% decrease in coverage across the Δ region compared to its flanking regions (i.e., no stromal contamination and complete recombination); whereas a DeltaFr = 0 corresponds to equal coverage between flanking and Δ regions (i.e., elevated stromal contamination and/or ineffective recombination). Finally, we computed the ratio between the DeltaFr relative to the Trp53 and Brca1 loci, working under the expectation that a complete recombination would result in Trp53-DeltaFr:Brca1-DeltaFr ratios of ~1 for the B1P models and ~2 for the Br1L1363P models (example in Appendix Fig. 4B; see also Methods). Indeed, this ratio was ~1 in B1P, but ~2 in Br1L1363P tumors, reflecting the loss of one Brca1 allele and two Trp53 alleles, with retention of the Brca1 p.L1363P allele (Appendix Fig. 4C). In one Br1L1363P tumor (T7), loss of Brca1fl5-13 was accompanied by a much larger deletion that included the flanking sequences, making calculation of this ratio unreliable. This larger deletion resulted in the loss of the entire Brca1fl5-13 allele, as reflected by the p.L1363P mutation having a variant allele frequency of ~1 in this sample (not shown). In summary, this analysis showed that each tumor analyzed had undergone the planned cancer-promoting Trp53, Brca1, and/or Brca2 deletions during tumorigenesis. We assessed the TD configuration of each tumor genome on a whole-genome scale. For the definition of the TDP score formula, see (Menghi et al, 2016). We noted abundant Group 1 TDs in 5/5 B1P tumor genomes in sufficient numbers to be scored as TDP Group 1, in agreement with Menghi et al (Menghi et al, 2018) (Fig. 5A,B). In contrast, 10/10 Br1L1363P and 5/5 B2P tumor genomes revealed no enrichment of Group 1 TDs, failed to score as TDP, and revealed statistically similar distributions of TD span sizes (Fig. 5A,B). One B2P tumor sample, B2P-T2, showed an exceptional rearrangement pattern, with hyperabundance of TDs, mostly of span size ≤1 kb or >100 kb (Fig. 5A). However, the B2P-T2 tumor failed to score as TDP by the algorithm of Menghi et al (Menghi et al, 2016), because the regions of intense instability were not dispersed genome-wide, as is characteristic of TDP cancer genomes. Furthermore, the B2P-T2 genome revealed single-nucleotide variant frequencies ~100-fold higher than all other samples (Appendix Fig. S4D), suggesting a complex origin of this excessive genomic instability. The 5 B2P tumors and 10 Br1L1363P tumors revealed statistically similar TD spectra mostly devoid of the Brca1-deficiency-linked Group 1 TDs. Most TDs in B2P and Br1L1363P tumor genomes were >100 kb, as is typical of the generic cancer genome (Li et al, 2020; Scully et al, 2025). Comparison to historical Trp53∆2-10/∆2-10 controls showed that tumor genomes from all groups in which a Brca gene had been targeted revealed statistically significant elevations in the abundance of SBS3, a well-characterized single-base substitution mutational signature widely recognized as a hallmark of HR deficiency (Fig. 5C) (Nik-Zainal et al, 2016). The findings that Brca1 p.L1363P drives accelerated tumorigenesis in the mouse (Pulver et al, 2021) and that Br1L1363P tumor genomes show the distinctive SBS3 mutation signature strongly suggest that Brca1 p.L1363P is a pathogenic variant. Analysis of tandem duplications in Brca p.L1363P mouse mammary tumors. (A) Left panel: Distribution of TD span sizes in mouse mammary tumor genomes of different Brca genotypes. B1P (n = 5): Brca1∆5-13/∆5-13;Trp53∆2-10/∆2-10. B2P (n = 5): Brca2∆11/∆11;Trp53∆2-10/∆2-10. Br1L1363P (n = 10): Brca1∆5-13/L1363P;Trp53∆2-10/∆2-10. The dotted horizontal lines denote the TD span size range for Group 1 TDs. Note abundant Group 1 TDs in B1P tumor genomes but not in B2P or Br1L1363P tumor genomes. The center line represents the median (50th percentile), the lower and upper bounds of the box represent the first (25th percentile) and third (75th percentile) quartiles, respectively, and the whiskers extend to the minimum and maximum values within 1.5 times the interquartile range from the first and third quartiles; values beyond the whiskers are plotted individually as outliers. Right panel: TDP scores for each tumor genome. 5/5 B1P tumors score as Group 1 TDP but no B2P or Br1L1363P tumors score as TDP. (B) TD span size distribution in pooled tumors of each Brca genotype. Mann–Whitney U test of TD size distributions: B1P vs B2P: P = 1.4E-22. B1P vs Br1L1363P: P = 5.7E-32. B2P vs Br1L1363P: not significant (ns). (C) Quantitation of SBS3 mutation signature in the genotypes shown. These data include both current data (gray data points) and historical data from a previous study (black data points) that includes negative control Trp53∆2-10/∆2-10 tumors (P53) in which Brca alleles are wild-type. Statistical comparisons by Student’s t test (two-tailed); P values as shown. B1P vs. B2P: ns. B1P vs. Br1L1363P: ns. B2P vs. Br1L1363P: ns. The median (50th percentile) is represented by the center line, while the first (25th percentile) and third (75th percentile) are represented by the lower and upper bounds of the box, respectively. Whiskers represent the extent of minima and maxima values within 1.5 times the interquartile range from the first and third quartiles; individual outliers are plotted beyond the whiskers. Quantitation of the SBS3 mutation signature in the genotypes shown. This data includes both current data (gray data points, n = 19) and historical data from a previous study (black data points, n = 15) that includes negative control Trp53∆2-10/∆2-10 tumors (P53, n = 6) in which Brca alleles are wild-type. Taken together, despite being defective for HR, Brca1 p.L1363P remains competent for suppression of genome-wide Group 1 TD formation during tumorigenesis. The Tus/Ter system accurately predicted this separation-of-function phenotype of Brca1 p.L1363P-driven tumors. Brca1 CC domain mutants tolerate Fancm deletion DNA end resection-defective Brca1∆11/hyg mES cells reveal abundant TDs when depleted of FANCM and are synthetic lethal/synthetic sick on a Fancm null background (Panday et al, 2021). To better understand the relationships between Group 1 TD formation, DNA end resection, and Fancm synthetic lethality, we asked whether Brca1 CC domain mutants tolerate Fancm deletion. To this end, we used CRISPR/Cas9-mediated dual DSB incisions to delete the DEAH (ATP binding) motif of the FANCM ATPase domain and to introduce an 85 bp frame-shift within Fancm exon 2, generating a functionally null allele of Fancm (Panday et al, 2021). See “Methods” for details. We generated clones of Fancm+/– and Fancm–/– mES cells on different Brca genetic backgrounds: Brca1fl11-HA/–, Brca1L1363P-HA/–, Brca1CC∆9bp-HA/–, Brca1∆CC-HA/–, Brca1∆11-HA/–, and Brca2∆27/∆27 (Morimatsu et al, 1998) (Appendix Fig. S5A). Each Fancm–/–;Brca1∆11-HA/– clone was extremely slow-growing, indicating a synthetic sick phenotype. We quantified Tus/Ter-induced TDs in the Fancm–/–;Brca mutants derived above. Fancm–/–;Brca1∆11-HA/– clones revealed robust TD induction. In contrast, Fancm–/–;Brca1L1363P-HA/– and Fancm–/–;Brca2∆27/∆27 clones revealed frequencies of Tus/Ter-induced TDs similar to or lower than those seen in Fancm–/–;Brca1WT-HA/– cells (Appendix Fig. S5B). We quantified the growth properties of selected Fancm–/–;Brca mutants using two independent assays: plating efficiency/colony formation assays and competitive growth assays. We noted a dramatic loss of plating efficiency in two independent clones of Fancm–/–;Brca1∆11-HA/– cells, in comparison to Fancm+/–;Brca1∆11-HA/– isogenic controls (Fig. 6A,B). In contrast, two independent clones of Fancm–/–;Brca1L1363P-HA/– cells revealed no loss of plating efficiency in comparison to isogenic Fancm+/–;Brca1L1363P-HA/– cells. There was a modest reduction in plating efficiency in two independent clones of Fancm–/–;Brca2∆27/∆27 cells. To perform competitive growth assays, we mixed ~10% GFP+ Fancm–/–;Brca1fl11-HA/– cells (WT controls) with ~90% GFP– Fancm–/–cells of the test Brca genotypes. We then cultivated the mixed populations over a 2-week period, harvesting aliquots at defined timepoints for FACS quantitation of the GFP+ fraction (Fig. 6C,D). As expected, the Brca1 WT (GFP–) Fancm–/–;Brca1fl11-HA/– sample produced GFP fractions that were stable over the 2-week period. In contrast, Fancm–/–;Brca1∆11-HA/– cultures revealed progressive increases in the GFP fraction over time, indicative of a growth defect in the GFP– Fancm–/–;Brca1∆11-HA/– cells in comparison to the GFP+ Fancm–/–;Brca1fl11-HA/– co-cultured cells. We noted no competitive growth deficit in the Fancm–/–;Brca2∆27/∆27 cells. Two independent Fancm–/–;Brca1L1363P-HA/– clones and two independent Fancm–/–;Brca1CC∆9bp-HA/– clones revealed a modest reduction in the GFP fraction during the 2-week co-culture period, suggestive of a modest growth advantage over the GFP+ Fancm–/–;Brca1fl11-HA/– co-cultured cells. Fancm deletion is tolerated by Brca1 p.L1363P cells. (A) Representative images of colony formation in the genotypes shown. Brca1 allele abbreviations as in Fig. 1A. Note severe loss of colony-forming ability of two independent Fancm–/–;Brca1∆11-HA/– clones. (B) Quantitation of colony formation assays in five independent biological replicates (n = 5; data shows mean and s.e.m. of five independent biological repeats). Unpaired t test with Welch’s correction. ns: not significant, **P < 0.005. (C) Schematic outline of competitive growth assay. (D) Competitive growth assays of the Fancm–/–;Brca genotypes shown. Bars show mean and s.e.m. of GFP+ % at the timepoints shown in four independent experiments (n = 4). Note progressive enrichment of GFP+-marked Fancm–/–;Brca1fl11-HA/– cells when mixed with Fancm–/–;Brca1∆11-HA/– cells. Source data are available online for this figure. To determine the extent to which Fancm deletion had caused the observed changes in growth, we repeated the competitive growth assays on parental (i.e., Fancm WT) clones of the above-noted Brca genotypes, using GFP+ Fancm+/+;Brca1fl11-HA/– cells as the co-cultured competitors (Fig. EV4). In this setting, Brca1∆CC-HA/– and Brca2∆27/∆27 cells revealed the most severe growth deficit, whereas Brca1∆11-HA/–, Brca1L1363P-HA/–, and Brca1CC∆9bp-HA/– cells revealed only a modest growth deficit in comparison to the GFP+ Brca1fl11-HA/– co-cultured cells. These results show that Fancm deletion had conferred a severe growth defect on Brca1 ∆11 mutants, as reflected in both colony formation and competitive growth assays. In contrast, Brca1 CC domain mutants and Brca2∆27/∆27 cells acquire modestly improved competitive growth characteristics in the absence of Fancm. These divergent effects of Fancm deletion match those of a recent CRISPR screen in mES cells (preprint: Dossin et al, 2026). Our data, although limited to a small number of Brca1 alleles, suggest that synthetic lethality/sickness following Fancm loss in Brca1 mutants is correlated with defective DNA end resection function but not with overall HR capacity (Table 1). Further testing of this hypothesis will require analysis of additional Brca1 mutant alleles. Controls for Fig. 6: competitive growth of Brca mutant cell lines on Fancm WT background. (A) Schematic outline of competitive growth assay; all cells are Fancm+/+. (B) Competitive growth assays of different Fancm+/+;Brca genotypes. Bars show mean and s.e.m. of GFP+ % at the timepoints shown across 6 independent experiments (n = 6). Note degree of enrichment of GFP+-marked Fancm+/+;Brca1fl11-HA/– cells when mixed with different Fancm+/+;Brca mutant cells. Comparison with Fig. 6 shows that deletion of Fancm alters competitive growth properties, disadvantaging Brca1∆11-HA/– cells but improving the relative growth of Brca1 CC mutants and Brca2∆27/∆27 cells. Table 1Summary of relationships between Brca genotype, DNA end resection function, DSB-induced HR, Group 1 TD suppression, and Fancm synthetic lethality.Brca1Brca2DNA end resectionDSB-induced HRGroup 1 TD suppressionImpact of Fancm deletionWTWTCompetentWTCompetentViable∆11WTDefectiveImpairedDefectiveSynthetic sick/lethalCC mutantWTCompetentImpairedCompetentViableWT∆27CompetentImpairedCompetentViable Summary of relationships between Brca genotype, DNA end resection function, DSB-induced HR, Group 1 TD suppression, and Fancm synthetic lethality. Discussion Here, we compare two classes of Brca1 mutant alleles in stalled-fork HR and in Group 1 TD formation. BRCA1 CC mutant proteins are defective for interaction with PALB2, leading to a defect in RAD51-mediated HR (Anantha et al, 2017; Dray et al, 2010; Sy et al, 2009; Xia et al, 2006; Zhang et al, 2009). However, unlike the majority of cancer-associated BRCA1 mutants, BRCA1 CC mutants remain competent for DNA end resection because they retain the key BRCA1 domains (RING, tandem BRCT, and exon 11-encoded region) required for this function (Ceppi et al, 2024; Salunkhe et al, 2024). Brca1 ∆11 and Brca1 CC mutants are complementary loss-of-function alleles, as shown by interallelic complementation in mouse development (Nacson et al, 2020). A likely explanation of this phenomenon is that Brca1 CC mutants provide DNA end resection capability, while Brca1 ∆11 mutants facilitate PALB2 recruitment for HR. In the mES cell system studied here, we also observed reduced PALB2 binding by endogenous BRCA1 ∆11 in comparison to WT BRCA1. This defect was more severe than that noted previously in human cancer cells expressing exogenous BRCA1 alleles (Nacson et al, 2020). Thus, although Brca1 ∆11 mutants are DNA end resection-defective, we cannot exclude the possibility that coordination between BRCA1 ∆11 and PALB2-BRCA2-RAD51 is also somewhat impaired. The minimal PALB2-binding region of mouse BRCA1 comprises residues 1350-1377 (PDB: 7K3S). Conceivably, the full interface between BRCA1 and PALB2 might extend to Brca1 exon 11-encoded regions neighboring the CC domain. The effects of Brca1 exon 11 deletion and CC mutation on stalled fork HR, DSB-induced HR, and PARPi sensitivity were qualitatively similar but revealed interesting quantitative differences. Brca1 ∆CC mutant cells, in which the entire CC domain is deleted, revealed greater sensitivity to the PARP inhibitor Olaparib and more severe STGC defects than Brca1 L1363P or ∆9 bp cells, in which the BRCA1 coiled coil domain is disrupted but not fully deleted. This difference may reflect a more complete loss of PALB2 binding by BRCA1 ∆CC than by BRCA1 L1363P or BRCA1 ∆9 bp. Interestingly, stalled fork (Tus/Ter-induced) STGC failed to accurately predict PARPi sensitivity. This is most strikingly seen in Brca1 ∆11 cells, which were among the most PARPi-sensitive but revealed the mildest defects in Tus/Ter-induced STGC of all Brca1 mutants studied here. Similarly, we previously noted that Brca1 BRCT domain mutant mES cells reveal no defect in Tus/Ter-induced STGC, despite being PARPi sensitive and defective for DSB-induced STGC (Chandramouly et al, 2013; Farmer et al, 2005; Willis et al, 2014). This discrepancy suggests that the DNA end resection function of BRCA1 is less critical for Tus/Ter-induced STGC than it is for I-SceI-induced STGC or for PARPi resistance. Similar effects were noted in recent studies of nickase-induced repair, where Brca1 was found to be dispensable for DNA end resection at nickase-broken forks, and Brca1 ∆11 mutants revealed wild-type levels of nickase-induced STGC (Elango et al, 2025; Pavani et al, 2024; Scully et al, 2024; Zong et al, 2026). Taken together, current evidence suggests that DNA end resection at stalled or broken forks is less tightly regulated than at a replication-independent DSB. The avid engagement of DNA end resection at stalled or broken forks may also explain why classical nonhomologous end joining (cNHEJ) does not compete with HR in nickase-induced broken fork repair or in stalled fork repair (Elango et al, 2025; Scully et al, 2019; Vriend et al, 2016; Willis et al, 2018). Tandem duplications are common structural variants in the cancer genome, but the mechanisms that form them remain poorly understood (Li et al, 2020; Scully et al, 2025). The ability to recapitulate BRCA1-linked Group 1 TD formation in the Tus/Ter system provides an opportunity to decipher mechanisms underlying this class of TDs. Our previous work led us to propose that Group 1 TDs form at a Tus/Ter RFB in cells lacking BRCA1 by a replication restart/bypass mechanism, resulting in re-replication of the TD segment, with completion of the TD by means of an end joining step (Scully et al, 2025; Scully et al, 2019; Willis et al, 2017) (Fig. EV5). Our initial findings implicating the DNA end resection mediator CtIP as a Group 1 TD suppressor suggested that DNA end resection might play a role in Group 1 TD suppression (Willis et al, 2017). In the current study, we approached this hypothesis by studying Tus/Ter-induced TD formation in different Brca1 allelic backgrounds, in order to relate BRCA1-mediated Group 1 TD suppression to the known functions of BRCA1 in DSB repair. We find that, in contrast to Brca1 ∆11 mutants, Brca1 CC mutants are competent for suppression of Tus/Ter-induced TDs. This result supports the hypothesis that BRCA1-mediated DNA end resection, but not BRCA1/PALB2-mediated RAD51 loading, is required for Group 1 TD suppression. We tested the predictive power of the Tus/Ter TD assay in a mouse model of Brca-linked mammary tumorigenesis (Pulver et al, 2021). Consistent with the predictions of the Tus/Ter system, we find that Brca1 p.L1363P mutant-driven mammary tumors, although defective for HR, do not display a Group 1 TD phenotype. In the same experiment, confirming our previous work, Brca1 null mammary tumors displayed a robust Group 1 TD phenotype (Menghi et al, 2018). In this regard, Brca1 p.L1363P mutant cancers more closely resemble Brca2 mutant cancers, which do not register as TDP (Menghi et al, 2018) (and this study). Thus, the Tus/Ter system accurately predicted the absence of a Group 1 TD phenotype in Brca1 p.L1363P-driven cancers. Proposed mechanisms of action of FANCM and BRCA1 in suppression of Tus/Ter-induced Group 1 TDs. (A) Interaction of converging replication forks with Tus/Ter replication fork barrier (RFB) sets up the Group 1 TD intermediate. The Tus/Ter-induced TD breakpoint junction is formed from one breakpoint in close proximity to the Ter array (the ‘Ter-proximal’ breakpoint) and one ‘upstream’ breakpoint several kilobases from the Ter array (Willis et al, 2017). We infer that for one fork (the rightward fork shown here), replication extends beyond the Tus/Ter RFB by an aberrant replication restart mechanism, while the second fork (here, the leftward fork) stalls and is processed at the Tus/Ter RFB site. We propose that the rightward fork stalls at the RFB and is restarted as a migrating bubble (i.e., BIR-type conservative DNA synthesis; red). This process could be RAD51-dependent or -independent (Panday et al, 2021; Scully et al, 2021). FANCM, in partnership with BLM, suppresses this form of replication restart by disassembling migrating bubble BIR intermediates. Eventual termination of BIR generates a displaced nascent strand that spans the potentially duplicated segment (a). The 3’ end of the displaced nascent strand forms the potential upstream breakpoint of the TD. The leftward fork stalls at the RFB and either undergoes fork breakage as shown or fork reversal (not shown). The resulting solitary DNA end forms the potential Ter-proximal breakpoint. Completion of DNA replication through the RFB generates one intact sister chromatid and one broken sister chromatid that contains two copies of the TD segment a, flanking the break site. (B) BRCA1 acts on the broken sister chromatid to suppress Group 1 TD formation. If BRCA1-mediated DNA end resection and SSA functions are intact, the two copies of the duplicated segment a flanking the break site are collapsed to a single copy by BRCA1-mediated SSA. If BRCA1-mediated SSA is disabled, end joining opportunistically repairs the break, generating a Group 1 TD. The breakpoint junction is formed by end joining between the upstream breakpoint and the Ter-proximal breakpoint. The synergistic effects of Brca1 exon 11 deletion and FANCM depletion on Tus/Ter-induced TD formation led us to study the genetic relationship between these two genes (Panday et al, 2021; Willis et al, 2017). We found that Brca1 ∆11 mES cells are synthetic lethal/sick on a Fancm null or FANCM ATP hydrolysis-defective genetic background. If this relationship were to hold for BRCA1-linked human cancers, small-molecule inhibitors of FANCM ATPase function might be effective as therapeutics in BRCA1-linked cancer. These findings raised the possibility that Group 1 TD formation and Fancm synthetic lethality might be mechanistically linked. Here, we tested this hypothesis by comparing the impact of Fancm deletion on the growth properties of Brca1 ∆11 mutants, Brca1 L1363P CC mutants, and Brca2 ∆27 mutant cells. By use of two different assays—colony plating efficiency and competitive growth assays—we confirm the synthetic sick phenotype of Brca1 ∆11 Fancm null cells. In contrast, Fancm deletion in Brca1 L1363P CC mutants was well-tolerated, with no defect in colony growth in comparison to Brca1 WT Fancm null cells. Indeed, in competitive growth assays, Brca1 L1363P Fancm null cells displayed a modest growth advantage over Fancm null cells that harbor WT Brca1. We found that Fancm deletion modestly suppressed colony formation but normalized the relative growth of Brca2 ∆27 cells in comparison to Fancm null cells with no Brca mutation. Similar differential impacts of Fancm loss on different Brca1 allelic backgrounds and on Brca2 defective cells have been reported recently (preprint: Dossin et al, 2026). Brca2 ∆27 mutants disrupt a C-terminal RAD51 binding domain and are defective for classical HR, fork protection and gap suppression, but are competent for suppression of Tus/Ter-induced TDs (Lim et al, 2024; Longo et al, 2025; Willis et al, 2017) (Appendix Fig. S5B). It will be interesting to study the impact of Fancm deletion in other Brca2 allelic backgrounds. If Group 1 TD formation and Fancm synthetic lethality are different consequences of defective BRCA1-mediated DNA end resection, how might these processes be linked? The data suggest that FANCM and BRCA1 act as independent barriers to Group 1 TD formation. We proposed that FANCM (in association with BLM (Panday et al, 2021; Willis et al, 2017)) suppresses break-induced replication (BIR)-type replication restart at stalled forks; BRCA1, via its SSA function, provides a ‘failsafe’ mechanism for collapsing back to single copy any nascent Group 1 TDs that escape the action of FANCM/BLM (Fig. EV5A). SSA is dependent on BRCA1-mediated DNA end resection but not on RAD51 loading—it is increased in Brca2 mutant and Brca1 CC mutant cells (Anantha et al, 2017; Tutt et al, 2001). Although traditionally considered an error-prone DSB repair pathway, in the context of a nascent Group 1 TD, SSA would perform a conservative (genome-stabilizing) role by preventing TD formation (Fig. EV5B). When both BRCA1 and FANCM are inactivated, two important, independent checks on Group 1 TD formation are lost, unleashing intolerable levels of genomic instability at sites of fork stalling (Fig. 7). One manifestation of this heightened genomic instability is an enhanced Group 1 TD phenotype; the other is the loss of cell fitness, i.e., synthetic lethality. Hypothesis linking defective BRCA1-mediated DNA end resection and SSA to Fancm synthetic lethality. Left: FANCM loss alone permits BIR-type replication restart at sites of stalled replication. BRCA1-mediated SSA collapses most nascent Group TDs back to a single copy. Center: Loss of BRCA1-mediated DNSA end resection abrogates SSA and permits occasional BIR-restarted replication forks to resolve as Group 1 TDs. Right: combined loss of BRCA1 and FANCM increases BIR-type restart at stalled forks and combines with SSA defect to massively increase the probability of Group 1 TD formation at sites of replication stalling. The increases in Group 1 TDs and related chromosome rearrangements impair cell fitness, resulting in synthetic sickness/lethality in the double mutant background. Limitations of the study This study is limited to mouse ES cells and mouse mammary tumors. Two other studies—one in somatic human cells and one in mES cells—have also identified a synthetic lethal relationship between BRCA1 and FANCM mutations (Adam et al, 2021; preprint: Dossin et al, 2026). However, it is not yet clear to what extent the synthetic lethal relationship between BRCA1 and FANCM translates to the cancer setting. The development of therapeutics that specifically target FANCM would facilitate such studies and might ultimately prove effective in the treatment of BRCA1-linked cancers. We have limited this study to mutations affecting Brca1 exon 11 and the CC domain. It will be instructive to study additional domain mutants of BRCA1 and its associated proteins to clarify whether tolerance of FANCM loss is genetically separable from BRCA1-mediated DNA end resection (Tsukada et al, 2024). As regards the mechanism of Group 1 TD formation, the link between BRCA1-mediated DNA end resection and TD suppression suggested by our work is correlative; it is possible that some other as yet unidentified function of BRCA1, which maps to the same domains as those required for DNA end resection, is the actual mediator of Group 1 TD suppression. The hypothesis that SSA specifically suppresses Group 1 TD formation is similarly unproven. The mechanisms discussed here apply only to Group 1 TDs, which have much smaller span sizes than most cancer-related TDs. Larger Group 2 or Group 3 TDs, which dominate the TD spectrum of non-BRCA1-linked cancers, likely arise by fundamentally different mechanisms, as reflected in their distinct genetic dependencies (Menghi et al, 2018; Menghi et al, 2016; Scully et al, 2025). The naturally occurring triggers of Group 1 TD formation in the cancer genome remain unknown. Although the Tus/Ter fork stalling system has thus far made accurate predictions regarding TD formation in BRCA1-linked cancer, additional types of lesions likely contribute to Group 1 TD formation in cancer. Huang et al recently showed that Group 1 TDs can form at a site-specific nick, implying that replication fork breakage can also trigger Group 1 TD formation (Huang et al, 2025). This finding raises the question: which form of replication stress—fork stalling or fork breakage—accounts for the major fraction of Group 1 TDs in a developing BRCA1-linked cancer? It will be instructive to compare the genetic regulation of Tus/Ter-induced vs. nickase-induced Group 1 TD formation and to relate these findings to the genetics of BRCA1-linked cancer-associated Group 1 TDs. Methods Reagents and tools tableReagent/resourceReference or sourceIdentifier or catalog number Experimental models 6x Ter-HR reporter containing TC1 mouse Embryonic Stem cells (mES)Willis et al, 20146x Ter-HR reporter TC1 (mES) Brca1 mutant clonesThis studyFVB micePulver et al, 2021Brca1L1363P/F;Trp53F/F mice Recombinant DNA pcDNA3β-myc NLS-TusWillis et al, 2014pcDNA3β -myc NLS-I-SceIWillis et al, 2014 Antibodies Anti-HA (for immunoblot)Abcamab9110Anti-HA (for immunofluorescence)Abcamab18181Anti-vinculinCell Signalling Technologies#4650Anti-mouse BRCA1 exon 11Scully labAnti-RPA (for ChIP)Abcamab10359Anti-mouse PALB2Xia labAnti-mouse Alexa Fluor 647Jackson ImmunoResearch#115-605-003Anti-rabbit Alexa Fluor 488Jackson ImmunoResearch#111-545-003 Oligonucleotides and other sequence-based reagents sgRNA oligomer sequences with PAM Brca1 Exon 2 sgRNA cttctgttgacttaccagatcgg Brca1 Exon 23 sgRNA gtctatccagctaccggtgtcgg Brca1 CC sgRNA 1 tctgcgagcagtcttcagaaagg Brca1 CC sgRNA 4 gatctggaaccaaacatgtcagg Brca1 CC sgRNA PE3 gcagcaggaaatggctcacctgg Fancm del85 sgRNA5' ggttcttttcctgacaccgcagg Fancm del85 sgRNA3' gatgaagctcataaggcacttgg Primers Brca1 hemizygous clone generationDeletion junction PCR Brca1 Exon 2 FWD Primer 2 tgggggtgacttaccttcag Brca1 Exon 23 REV Primer 1 acaaatcaggaccactagct 5’ region PCR Brca1 Exon 2 FWD Primer 2 tgggggtgacttaccttcag Brca1 Exon 2 REV Primer 4 atcccacactggacttgcta 3’ region PCR Brca1 Exon 23 FWD Primer 6 gggcttttaccacgctcact Brca1 Exon 23 REV Primer 1 acaaatcaggaccactagct Deletion of Floxed Brca1 exon 11 FLOX BRCA1 (P13.63 FWD) CTGGGTAGTTTGTAAGCATCC OUT-FLOX Brca1 (P14.02 REV) CTGCGAGCAGTCTTCAGAAAG HA-tagging Brca15’ region PCR B1 tag F3 ggatctctgccagcttaagtacc B1 tag R1 CGTAAGGGTAcccACTAGCG 3’ region PCR B1 tag F5 gtgtctctcactcggaaggac B1 tag R5 ggaattagaaggctctccttgacac Brca1 ∆CCDeletion junction PCR B1-CC F6 ccagtgtcccaaatgtagcc B1-CC R6 tgcactgagcaaagatcacac 5’ region PCR B1-CC F6 ccagtgtcccaaatgtagcc B1-CC R4 gtaggatttttagatctggctggg 3’ region PCR B1-CC F7 cagctcccatgctaaaattaattgc B1-CC R6 tgcactgagcaaagatcacac Brca1 L1363P B1-pt primer Fw1 tatctggaagagctcgaggg L63P_Rev2 tgagccatttcctgctgcag Fancm exon 2 Fancm_Fwd_1 ctacctcaagctccagagtcctgg Fancm_Rev_1 agttcccatcactgagacttattcc Chemicals, enzymes, and other reagents Spy NLS Cas9New England BiolabsM0646Tanti-HA beadsPierce/Invitrogen88836P3 solution with supplementLonzaV4XP-3032Alt-R™ CRISPR-Cas9 crRNABiosynthesis, IDTAlt-R™ CRISPR-Cas9 tracrRNAIDT1072533Alt-R™ S.p. Cas9 Nuclease V3IDT1081059Lipofectamine 2000Invitrogen11668019NEB Ultra II DNA Library Prep KitNEBE7645KAPA Hyper Prep KitKAPA Biosystems/RocheKK8502 Software ImageJ https://imagej.net/software/fiji/#publication Keyence BZ-X800 AnalyzerKeyence Corporation, Osaka, JapanFlowJoBecton Dickinson & Company (BD)FlowJo 10.10.0bcl2fastq https://support.illumina.com/sequencing/sequencing_software/bcl2fastq-conversion-software.html RRID:SCR_015058samtools https://github.com/samtools/samtools RRID:SCR_002105deeptools https://github.com/deeptools/deepTools PyGenomeTracks https://github.com/deeptools/pyGenomeTracks Burrows-Wheeler AlignerLi and Durbin, 2009 Other CytoFlexBeckman CoulterCeligo imaging cytometerRevvityNextSeq 1000IlluminaR230 ultrasonicatorCovarisNovaseq X Plus SeqIllumina Reagents and tools table Molecular biology, siRNA, and sgRNA oligos 6xTer HR reporters were engineered as previously described Using conventional cloning methods(Willis et al, 2014; Willis et al, 2017). Plasmids used for transfection were prepared using endotoxin-free maxiprep (QIAGEN, 12362) kit and have been described previously (Willis et al, 2014). Brca1 exon 12 (L1363P) targeting construct was generated using large double-stranded DNA gBlocks (Integrated DNA Technologies). Oligomers and primers used for genotyping, RT-qPCR, and sgRNA synthesis were purchased from Thermo-Fisher Scientific and GeneWiz (Azenta). siRNA SMARTpools were purchased from Horizon Scientific/Dharmacon. Cas9-sgRNA target sites in the mouse genome sequence were identified using the Heidelberg CCtop tool: https://cctop.cos.uni-heidelberg.de. sgRNA in vitro synthesis was performed using the Engen sgRNA Synthesis Kit (New England Biolabs, E3322S) and purified using the Clean and Concentrate Kit (Zymo Research, R1017). sgRNA was electrophoresed on a denaturing gel (Novex 10% TBE-Urea; ThermoFisher Scientific, EC6875BOX), to assess quality prior to transfection. Cell culture and cell lines mES cells are thawed onto plates coated with mouse embryonic fibroblast (MEF) feeders, maintained in ES medium (DMEM with 15% serum, 20 mM HEPES pH 7.6, 0.1 mM bME, 500 U/mL rLIF, 1× MEM NEAA, 2 mM L-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, 0.001 M sodium pyruvate) on gelatinized plates, and regularly tested for mycoplasma infection by Myco Alert assay (Lonza, LT07-318). Brca1fl5-13/+ cells were targeted with a 6xTer-HR reporter as a single copy to the Rosa26 locus using standard methods described previously (Willis et al, 2014). All cell lines for analysis of Brca1 separation-of-function mutant alleles were derived from a Brca1fl11/hyg founder cell line carrying one Brca1 allele in which most of the conventionally termed exon 11 (NCBI exon 10) is replaced by a hygromycin resistance cassette, and one functionally wild-type conditional Brca1 allele in which exon 11 is flanked by LoxP elements (Xu et al, 1999). The founder cell line contains a single copy of the 6xTer-HR reporter, targeted as a single copy to the Rosa26 locus (Willis et al, 2014). All cell lines were regularly tested for mycoplasma contamination and were found to be mycoplasma-free in all tests. Generation of Brca1 mutant cell lines To delete floxed alleles of Brca1, cells were treated with adeno-Cre for 24 h and then plated on 60-mm dishes with MEF feeder layer without any selection. Clones were picked after 4–5 days and expanded for another 5 days. Genomic DNA from clones was prepared using Qiagen Puregene buffers and protocol. The clones were then assessed to check for recombined LoxP sites using primers specific for the deleted or undeleted conditional allele (see “Reagents table”). To generate Brca1 hemizygous cell lines, sgRNA against NCBI exon 2 and 23 (see “Reagents table”) were designed and synthesized as described above. Cas9-sgRNA RNP was pre-assembled in vitro in OptiMEM by mixing 3 µM Spy NLS Cas9 (New England Biolabs, M0646T) and 3 µM purified sgRNAs. Cells were transfected using Lipofectamine 2000 mixed with RNP complexes. 48 h after transfection, cells were plated onto 60 mm dishes containing feeder MEFs without selection. Individual clones were picked for expansion between 4 and 5 days and then expanded on MEFs for another 5 days. Brca1 hemizygous clones were identified by PCR and confirmed by sequencing using primers spanning the deleted Brca1 allele (see “Reagents table”). For in-frame Brca1 HA-tagging, we generated a targeting construct that positions 8xHA tag sequences in-frame with the 3’ end of the Brca1 open reading frame, followed by a SMASh degron and a neomycin resistance gene downstream of self-cleaving peptide T2A sequences. Cas9 RNP complexes targeting Brca1 NCBI Exon 23 (see “Reagents table”) were co-transfected with a linearized targeting construct using Lipofectamine 2000. 48 h after transfection, cells were plated onto 60-mm dishes coated with a feeder layer of MEFs that express the neomycin resistance gene. Twenty-four hours after seeding, 4 µM G418-containing medium was added to the plates. Five to seven days later, G418-resistant colonies were isolated and expanded. Colonies were confirmed using PCR and Sanger sequencing using primers to detect successful integration of the HA-SMASh elements (see “Reagents table”). The clones were also assessed for full-length Brca1 gene product expression using immunoblotting against HA, as described in the “Results” section. Domain mutations in the BRCA1 coiled coil were also generated using CRISPR/Cas9 gene editing. To generate ∆CC clones, sgRNAs targeting NCBI exon 11 and 12 (see “Reagents table”) were designed, synthesized in vitro, and transfected as RNP complexes using Lipofectamine 2000. Clones were generated as mentioned above, were grown without selection, and were screened by PCR followed by sequencing using primers for ∆CC Brca1 (see “Reagents table”). Brca1 p.L1363P point mutant clones were generated by transfecting a template containing specific mutations in a region of Brca1 NCBI exon 12. For the p.L1363P mutation, ctg was mutated to ccg, along with silent mutations in the PAM site of the sgRNA used (see “Reagents table”). Primers Brca1 L63P (see “Reagents table”) were used to identify clones through PCR and sequencing. Generation of mouse Fancm hemizygous and null clones Characterized Brca1HA/– (WT, L1363P and ∆11) and Brca2∆27/∆27 (deleted for exon 27) mutant clones were subjected to CRISPR/Cas9 gene editing to generate Fancm + /– and Fancm–/– clones. sgRNAs targeting exon 2 (see “Reagents table”) of the Fancm gene were designed to generate an 85 bp frame-shift deletion in the coding region of Fancm. We used CRISPR/Cas9-mediated dual DSB incisions to delete the DEAH (ATP binding) motif of the FANCM ATPase domain and to introduce an 85 bp frame-shift within Fancm exon 2, generating clones of Fancm+/– mES cells on different Brca genetic backgrounds: Brca1fl11-HA/–, Brca1L1363P-HA/–, Brca1∆11-HA/–, and Brca2∆27/∆27. We then repeated the CRISPR/Cas9 transfection to delete the residual WT Fancm allele. This two-step method was necessary because the efficiency of biallelic Fancm deletion in Fancm+/+ cells was low. We retrieved viable Fancm+/– or Fancm–/– derivatives on all Brca genetic backgrounds. Clones were grown without selection, isolated, and expanded as described above. Fancm mutant clones and isogenic controls that retained WT Fancm were identified through PCR (see “Reagents table”) followed by Sanger sequencing the region of interest after TA cloning. Recombination assays In total, 1.6 × 105 mES cells were transfected in suspension with 0.5 μg empty vector or pcDNA3β-myc NLS-Tus, or pcDNA3β -myc NLS-I-SceI using Lipofectamine 2000 (Invitrogen) (Puget et al, 2005; Willis et al, 2014). Transfection efficiency was measured in parallel by transfecting 0.05 μg wild-type GFP expression vector and 0.45 μg control vector. For experiments with siRNAs, 0.3 µg of plasmid vectors were transfected along with 10 pmol siRNA for each of the target gene or along with 10 pmol of luciferase siRNA. 72 h after transfection, 3 × 105–6 × 105 total events were scored for each sample, and GFP + RFP–, GFP + RFP+, and GFP–RFP+ frequencies were analyzed through flow cytometry using a Beckman Coulter CytoFlex, in duplicates. Reported repair frequencies were corrected for background events and for transfection efficiency (50–85%). Colony formation assays To study the effect of PARPi on different Brca1 mutants, approximately 800 cells were seeded onto six-well tissue culture plates after coating them with 0.1% gelatin. 24 h after seeding the cells, medium was replaced, and the plates were replenished with PARPi-containing medium. Six days after plating the cells, colonies were fixed using Carnoy’s fixative (75% methanol and 25% glacial acetic acid) and stained with crystal violet. The colonies were counted using a Celigo imaging cytometer (Revvity). For plating assays of Fancm–/– clones, an equal number (400) of cells were plated onto six-well cell culture dishes and allowed to grow for 6 days, after which the colonies were processed as described earlier. Competitive growth assays In total, 0.04 × 106 GFP+ Brca1fl-HA/–;Fancm–/– cells were mixed with 0.36 × 106 GFP– Brca mutant Fancm–/– cells. Approximately 0.02 × 106 cells were seeded onto a six-well plate for growth over 2 weeks, and the GFP+ fractions were quantified by flow cytometry on the days shown in Figs. 6 and EV4. Immunoprecipitation In total, 500 µL of NETN300 buffer with protease inhibitors was added onto a 60–70% confluent 100 mm cell culture dish and incubated on ice for 30 min. The lysate was then scraped and collected in a microfuge tube and spun down at 14 K RPM in a refrigerated microcentrifuge at 4 °C for 20 min. The supernatant was collected for immunoprecipitation. Approximately 600 µg of protein was added to 25 µL of anti-HA beads (Pierce) and rotated on an end-to-end rotator for 16–20 h at 4 °C. The beads were then washed twice with lysis buffer, followed by one wash with 1xTBST (Tween20) and a final wash with ultrapure nuclease-free water. The eluate was obtained by incubating the beads with 60 µL of 1× non-denaturing SDS loading buffer at 95 °C for 5 min. Samples were electrophoresed on a SDS-TAE 3–8% gradient gel overnight, and the protein was transferred onto a nitrocellulose membrane for analysis by immunoblotting. Western blotting RIPA buffer was added to the cell pellet and incubated on ice for 30 min, following which the lysate was sonicated and spun down at 14 K RPM in a refrigerated microcentrifuge at 4 °C for 20 min. Protein was estimated using the Thermo Rapid BCA Gold kit. Approximately 30 µg of protein was used for electrophoresis in a TAE 3–8% gradient gel for 16–20 h overnight at 50 V in a cooled chamber. Protein was transferred onto a nitrocellulose membrane using wet transfer in a transfer buffer containing 20% methanol. Following transfer, the membrane was blocked in 5% skimmed milk for anti-HA and anti-vinculin immunoblots, whereas 5% BSA was used for anti-PALB2 immunoblot. After incubating the membranes with the appropriate secondary antibody, Pico PLUS Chemiluminescent Substrate was used to visualize the immunoblot signals using a BioRad ChemiDoc MP. Immunofluorescence staining and microscopy In all, 22 × 22 mm coverslips were disinfected using 70% ethanol and UV-C radiation in the biosafety hood. Coverslips were then coated with 20 µg/mL of fibronectin in 1× PBS and incubated in the biosafety hood at RT. After 30 min fibronectin was aspirated and the coverslips were dried for another 30 min. Coverslips were then rinsed with 1× PBS and ~1 × 106 cells were plated for staining the next day. On the day of staining, coverslips were washed twice with 1× PBS, and the cells were fixed with fixative solution (1× PBS, 3% paraformaldehyde, 2% sucrose) for 10 min at RT. After rinsing the coverslips thrice with 1xPBS, the cells were permeabilized for 2 min using ice-cold TX solution (0.5% Triton-X-100, 20 mM HEPES pH 7.4, 50 mM NaCl, 3 mM MgCl2, 300 mM sucrose) while keeping the plates with coverslips on ice. It is critical to wash the coverslips five times with 1× PBS after permeabilization. In total, 1:500 dilution of anti-HA and anti-BRCA1”E” antibodies was added to the hybridization buffer (5% goat serum, 1× PBS, 0.5% NaN3), and ~100 µL was added onto each coverslip, followed by incubation at 37 °C for 20 min. Coverslips were washed twice with 1× PBS. In total, ~100 µL of 1:400 dilution of secondary antibodies in the hybridization buffer was added to the coverslips dropwise and followed with another incubation at 37 °C for 20 min. Coverslips were washed twice with 1× PBS and mounted onto slides using Prolong Gold containing DAPI (Invitrogen). Anti-HA Foci were counted using the “Find Maxima” function in ImageJ software. Background subtraction for all images was set at 50 pixels, and prominence for maxima was set at 10,000. The number of cells in the field was counted using DAPI staining, and the mean number of foci per nucleus is reported (Fig. S3A). Nucleofection of CRISPR-Cas9 RNP Equal volumes of 100 μM crRNA (Biosynthesis, IDT) and 100 uM tracrRNA (IDT) were mixed and incubated on a 95 °C heat block for 3 min, then cooled on a benchtop for 5 min. Dialysis buffer (20 mM HEPES pH 7.5, 500 mM KCl, 20% glycerol), Cas9 (IDT, 10 μg/μL), and annealed crRNA/tracrRNA were mixed to form RNP and allowed to sit at room temperature for at least 20 min. P3 solution with supplement (Lonza) were freshly mixed and used for nucleofection of trypsinized cells. In total, 2 μL of 100 μM electroporation enhancer (IDT) was added to each nucleofection reaction. Approximately 12 × 106 cells were used for each nucleofection. Chromatin-immunoprecipitation (ChIP) Nucleofected cells were collected via cell scraping in plain DMEM. Cells were fixed with 1% formaldehyde for 10 min and gently agitated by rotation at room temperature, and then quenched with 750 µL of 2 M Glycine for 4 min with rotation at room temperature. Cells were spun at 1200 × g for 3 min at 4 °C, washed with PBS, spun, and flash frozen and stored at −80 °C. Protein A magnetic beads (50 μL per sample) were washed with 0.5% BSA three times, then incubated with RPA antibody (abcam 10359, 5 μL per sample). The antibody was allowed to bind to the magnetic beads at room temperature with rotation for 1 to 3 h. After incubation with the antibody, the beads were washed three times with 0.5% BSA in PBS. Cell pellets were removed from −80 °C and resuspended in lysis buffer (50 mM HEPES, 140 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% Igepal CA-630, 0.25% Triton-X; with pH adjusted to 7.5 using KOH and with 1× protease inhibitors immediately before use (ThermoFisher, 78429) for 10 min at 4 °C with rotation. The lysate was pelleted at 2000 × g for 3 min at 4 °C, then resuspended in nuclear lysis buffer (10 mM Tris-HCl pH 8, 200 mM NaCl, 1 M EDTA, and 0.5 mM EGTA; with pH adjusted to 8 using HCl and with 1x protease inhibitors added before use) and allowed to rotate for 5 min at 4 °C. The nuclear lysate was pelleted at 2000 × g for 3 min at 4 °C, then resuspended in sonication buffer (10 mM Tris-HCl pH 8, 100 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 0.1% Na-Deoxycholate, 0.5% N-lauroylsarcosine; with pH adjusted to 8.0 using HCl, and adding 1× protease inhibitor right before use) and sonicated using (Qsonica Q125)) with 30 s of pulse, 30 s off, 55% amplitude for 6 min at 4 °C. Chromatin was spun at 20,000 × g at 4 °C. The supernatant was added to a fresh 5-ml LoBind tube, and 1% final volume of TritonX-100 and antibody-bound beads were added. The final volume was increased to 3 mL using the sonication buffer. The mix was allowed to rotate for 8–20 h at 4 °C. After 8–20 h of rotation, beads were collected and washed 6 times with RIPA solution (50 mM HEPES, 500 mM LiCl, 1 mM EDTA, 1% Igepal CA-630, 0.7% Na-Deoxycholate; pH adjusted to 7.5 using KOH), then once with TBS wash buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl) and resuspended in elution buffer (50 mM Tris-HCl pH 8, 10 mM EDTA, 1% SDS). Beads and elution buffer were incubated for at least 8 h with shaking at 65 °C for reverse crosslinking. RNaseA was added to the beads and incubated for 15 min at 37 °C, followed by Proteinase K at 55 °C for 1 h. Beads were separated from the eluate using magnetic separation, and the DNA was purified using MinElute column purification. The ChIP libraries were prepared according to the NEB Ultra II DNA Library Prep Kit (NEB, #E7645). End-repair and A-tailing were performed, followed by adapter ligation (NEB 7600), and 13 cycles of PCR amplification. Samples were then pooled, and DNA was quantified using a Qubit (ThermoFisher) and sequenced on an Illumina NextSeq 1000 machine using paired-end 2×60 bp read length. ChIP-Seq data analysis ChIP-Seq data were demultiplexed using the bcl2fastq software, and the resulting fastq files were aligned to a mouse genome (mm10) containing the HR reporter using bowtie2. Alignments were converted to binary, filtered by quality, and sorted using samtools. Enrichment bigwig files were obtained using deeptools by subtracting the signal from the undamaged sample (No Cas9) from the damaged one (window size of 50 bp). Final plots (shown in Fig. 3B) were generated using the PyGenomeTracks software. ChIP-Seq enrichments (Fig. 3C) were computed using a custom Python code that subtracts the number of normalized (RPM) reads in the damaged and undamaged samples in a window of 30 kb centered at the cut site. Analysis code is available on GitHub (https://github.com/AlbertoMarinG/BRCA1_SoF). Raw WGS data for ChIP-seq experiments are on the Sequence Read Archive (SRA: https://www.ncbi.nlm.nih.gov/sra), BioProject: PRJNA1378990; BioSamples: SAMN53860886. URL for access (no reviewer’s access code, since data has been publicly released): https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1378990. Mouse mammary tumor models and WGS Mouse mammary tumors were generated in K14-Cre;Trp53fl2-10/fl2-10;Brca1fl5-13/fl5-13 (B1P), K14-Cre;Trp53fl2-10/fl2-10;Brca2fl11/fl11 (B2P) and K14-cre;Trp53fl2-10/fl2-10;Brca1L1363P/fl5-13 (Br1L1363P) female mice as described previously (Jonkers et al, 2001; Liu et al, 2007; Menghi et al, 2018; Pulver et al, 2021). Genomic libraries were derived from 20 mouse tumor tissues and their matched mouse spleen tissues (normal controls), by shearing 200 ng of genomic DNA using the R230 ultrasonicator (Covaris), followed by size selection using Ampure XP Beads. DNA fragments were then treated with end-repair, A-tailing, and ligation of Illumina unique adapters (Illumina) using the KAPA Hyper Prep Kit (Illumina) (KAPA Biosystems/Roche), after which five cycles of PCR amplification were performed. Library quantification was performed using real-time qPCR (Thermo Fisher) before sequencing on a Novaseq X Plus Seq (Illumina) platform to generate 151-bp-long paired-end reads, aiming for 30X coverage. Sequencing reads were aligned to the mouse GRCm39 reference genome using the Burrows-Wheeler Aligner (Li and Durbin, 2009). For each cancer genome, we calculated the sequencing coverage across the Trp53 and Brca1 loci, extending each region by 50% of its length upstream and downstream, using non-overlapping 100-bp bins. Coverage values were then normalized across samples, using the edgeR::calcNormFactors function in R. A rolling average of the normalized coverage was computed for each sample, using windows of 25 and 50 consecutive 100-bp bins for Trp53 and Brca1, respectively, as represented in Appendix Fig. S4B. We then averaged the values corresponding to the windows mapping to the delta region as well as the left and right flanking regions, for either gene, separately, and computed the delta allele frequency (DeltaFr) as 1− (coverage in the flanking windows/coverage in the delta windows). Structural variant calls were generated using four different tools (Manta (Chen et al, 2016), Delly (Rausch et al, 2012), LUMPY (Layer et al, 2014), and SvABA (Wala et al, 2018), and high-confidence events were selected when called by at least two tools and by requiring split-read support. Matched normal DNA samples were used as controls to remove germline variants. A TDP score was computed for each tumor sample based on the number and chromosomal distribution of its somatic tandem duplications (TDs), and TDP Group 1 was assigned to genomes exhibiting a TD span distribution peak between 1.64 and 51 kb, as previously described (Menghi et al, 2018; Menghi et al, 2016). Single-nucleotide variants were called by MuTect2 (preprint: Benjamin et al, 2019), Strelka2 (Kim et al, 2018), and Lancet (Narzisi et al, 2018). High-confidence calls were defined as those that are either called by at least two variant callers or called by one caller and seen in the Lancet validation calls, and that occur at a variant allele frequency of 0.07 or above. Putative germline variants were removed by only keeping mutations unique to each cancer genome. In addition to the 15 tumor genomes sequenced as part of this study, we also applied the same pipeline to reanalyze 15 previously published mouse breast cancer genomes from the following genotypes: K14-cre;Trp53fl2-10/fl2-10 (KP, n = 3), WAP-Cre;Trp53fl2-10/fl2-10 (WP, n = 3), K14-Cre;Trp53fl2-10/fl2-10;Brca1fl5-13/fl5-13 (KB1P, n = 3), WAP-Cre;Trp53fl2-10/fl2-10;Brca1fl5-13/fl5-13 (WB1P, n = 3), K14-Cre;Trp53fl2-10/fl2-10;Brca2fl11/fl11 (KB2P, n = 3) (Menghi et al, 2018; Menghi et al, 2016). Single-base substitution (SBS) signatures were extracted with the YAPSA R package (Hübschmann et al, 2021) using the original 22 high-confidence COSMIC signatures described by Alexandrov et al as the mutational process matrix (Alexandrov et al, 2013). WGS data relative to the new mouse breast cancers sequenced as part of this study (i.e., 20 pairs of spleen and tumor DNA samples) are available from the Sequence Read Archive database (www.ncbi.nlm.nih.gov/sra), SRA: PRJNA1393414. Previously published mouse cancer WGS datasets (Menghi et al, 2018) are also available from SRA: PRJNA430898. Supplementary information Appendix Peer Review File Source data Fig. 1 Source data Fig. 2 Source data Fig. 3 Source data Fig. 4 Source data Fig. 6 Expanded View Figures Acknowledgements We gratefully acknowledge helpful discussions with other Scully lab members and the contributions of the flow cytometry core facility at BIDMC. We also thank the Genome Technologies Service and Data Science at The Jackson Laboratory for expert assistance with the WGS work described herein. This work was supported by NCI grants R35CA263813 to RS, P30CA034196 and R01CA255705 to ETL, R01CA138804, R01CA262227, and P01CA250957 to BX, grants from Oncode Institute, the Dutch Cancer Society (NKI 2015-7877), and the Lundbeck Foundation (R223-2016-8) to JJ and American Cancer Society Postdoctoral Fellowship PF-24-1321213-01-DMC to NMN. Author contributions Namrata M Nilavar: Conceptualization; Data curation; Investigation; Visualization; Methodology; Writing—original draft; Writing—review and editing. Alberto Marin-Gonzalez: Conceptualization; Software; Investigation; Visualization; Methodology; Writing—review and editing. Francesca Menghi: Conceptualization; Formal analysis; Investigation; Visualization; Methodology; Writing—review and editing. Daniel Nguyen: Investigation; Methodology; Writing—review and editing. Nicholas A Willis: Conceptualization; Investigation; Methodology; Writing—review and editing. Ellen Wientjens: Resources; Data curation; Investigation; Writing—review and editing. Bing Xia: Resources; Investigation; Writing—review and editing. Jos Jonkers: Resources; Supervision; Methodology; Writing—review and editing. Edison T Liu: Conceptualization; Data curation; Software; Supervision; Methodology; Writing—review and editing. Ralph Scully: Conceptualization; Supervision; Funding acquisition; Writing—original draft; Project administration; Writing—review and editing. Source data underlying figure panels in this paper may have individual authorship assigned. Where available, figure panel/source data authorship is listed in the following database record: biostudies:S-SCDT-10_1038-S44318-026-00913-x. Data availability ChIP-seq data: Analysis code is available on GitHub (https://github.com/AlbertoMarinG/BRCA1_SoF). Raw WGS data for ChIP-seq experiments are on the Sequence Read Archive (SRA: https://www.ncbi.nlm.nih.gov/sra), BioProject: PRJNA1378990; BioSamples: SAMN53860886. WGS data: WGS data relative to the new mouse mammary tumors sequenced as part of this study (i.e., 20 pairs of spleen and tumor DNA samples) are available from the Sequence Read Archive database (www.ncbi.nlm.nih.gov/sra), SRA: PRJNA1393414. The source data of this paper are collected in the following database record: biostudies:S-SCDT-10_1038-S44318-026-00913-x. Disclosure and competing interests statement RS is a member of the Scientific Advisory Board of MoMa Therapeutics, Inc. Supplementary information Expanded view data, supplementary information, appendices are available for this paper at https://doi.org/10.1038/s44318-026-00913-x. References Associated Data Supplementary Materials Appendix Peer Review File Source data Fig. 1 Source data Fig. 2 Source data Fig. 3 Source data Fig. 4 Source data Fig. 6 Expanded View Figures Data Availability Statement ChIP-seq data: Analysis code is available on GitHub (https://github.com/AlbertoMarinG/BRCA1_SoF). Raw WGS data for ChIP-seq experiments are on the Sequence Read Archive (SRA: https://www.ncbi.nlm.nih.gov/sra), BioProject: PRJNA1378990; BioSamples: SAMN53860886. WGS data: WGS data relative to the new mouse mammary tumors sequenced as part of this study (i.e., 20 pairs of spleen and tumor DNA samples) are available from the Sequence Read Archive database (www.ncbi.nlm.nih.gov/sra), SRA: PRJNA1393414. The source data of this paper are collected in the following database record: biostudies:S-SCDT-10_1038-S44318-026-00913-x.