A BRCA1 Coiled-Coil Domain Variant Disrupting PALB2 Interaction Promotes the Development of Mammary Tumors and Confers a Targetable Defect in Homologous Recombination Repair

BRCA1 and BRCA2 are the main high-risk hereditary breast cancer suppressor genes (1). The functions of BRCA1 and BRCA2 in homologous recombination repair (HRR) appear requisite for tumor suppression, because all known pathogenic variants analyzed to-date result in HRR defects (2–8). Besides BRCA1 and BRCA2, several other HRR genes are associated with a high risk for breast cancer, including PALB2 (9). PALB2 binds to BRCA2 via its WD40 domain and is important for BRCA2 localization to chromatin and sites of DNA damage (10). The coiled-coil domains of PALB2 and BRCA1 can also bind, and several groups have shown that disruption of this interaction affects HRR (11–13). Together this suggests that an intact BRCA1–PALB2–BRCA2 complex is a key requirement for hereditary breast cancer suppression. Although BRCA1 sequence variants disrupting the interaction with PALB2 have been identified in patients with suspected hereditary breast cancer (11), their impact on BRCA1-mediated tumor suppression is not clear. Thus far, all known pathogenic BRCA1 missense variants are restricted to the evolutionary conserved N-terminal RING and C-terminal BRCT domains. However, three missense variants within the PALB2-binding coiled-coil domain of BRCA1 (residues 1364–1437) are known to confer functional defects in HRR: p.M1400V, p.L1407P, and p.M1411T (2, 3, 11). Of these VUS, p.L1407P appears to have the most adverse effect on BRCA1 function. BRCA1 p.L1407P severely compromises the interaction of BRCA1 with PALB2, impairs HRR (2, 11), and leads to increased sensitivity to cisplatin and PARP1 inhibition (2, 3). More recently, a mouse model of Fanconi anemia harboring the BRCA1 p.L1407P murine equivalent, Brca1 p.L1363P, exemplified the importance of BRCA1–PALB2 interaction in vivo (14). Constitutive expression of this missense mutation recapitulated an array of Fanconi anemia–associated clinical phenotypes like growth retardation, infertility, and bone marrow failure, and highlighted defects in HRR and interstrand cross-link repair pathways, which rendered hypersensitivity to DNA-damaging agents. However, the impact of this clinically relevant BRCA1-PALB2 coiled-coil domain on breast cancer development and therapy response remains to be determined. Therefore, we set out to model the BRCA1 coiled-coil VUS p.L1407P in mice using CRISPR/Cas9 germline modification. Our Brca1 p.L1363P mouse model was analyzed for HRR defects and crossed into a K14cre conditional Brca1;Trp53 knockout model of BRCA1-associated breast cancer (15–17) to investigate the importance of BRCA1–PALB2 interaction for breast cancer suppression and treatment response.

Brca1 p.L1363P mice were generated following a CRISPR/Cas9-mediated gene editing protocol (18). Zygotes isolated from FVB mice were coinjected with in vitro-transcribed Cas9 mRNA and Brca1 exon 13-targeting guide RNA (gRNA) from pX330 plasmid, along with oligonucleotide repair templates for the introduction of the p.L1363P variant (Supplementary Table S1). The Brca1 gRNA was designed using https://zlab.bio/guide-design-resources (19). Following implantation into pseudopregnant female FVB mice, offspring were screened for the desired missense mutation by Sanger sequencing, backcrossed once onto FVB mice, and maintained for the desired experimental genotype. To distinguish wild-type, heterozygous, and homozygous mutants, we established a three-primer PCR strategy (Supplementary Table S2). Brca1^L1363P/F;Trp53^F/F mice were crossed with a transgenic β-actin-cre germline deleter strain (FVB background; kindly provided by M. Vooijs and A. Berns, The Netherlands Cancer Institute, Amsterdam, the Netherlands) to obtain Brca1^L1363P/+;Trp53^Δ/+ mice. The β-actin-cre transgene was crossed out before compound mutant mice were used for experiments. Brca1^L1363P/+ mice were bred with K14cre;Brca1^F/F;Trp53^F/F (KB1P) animals (FVB background; ref. 17) to generate K14cre;Brca1^L1363P/F;Trp53^F/F (KB1(L1363P)P) mice. All mouse experiments were approved by an independent animal ethics committee of The Netherlands Cancer Institute and executed according to Dutch and European guidelines.

Timed matings were established and the presence of a copulation plug demarcated embryonic day 0.5 (E0.5). Uteri from pregnant females were isolated at various days after coitum and kept on ice in PBS. Individual embryos were then isolated from the maternal decidua, and yolk sac tissue was collected for genotyping. Images were collected using Olympus SC30 digital camera and cellSens software (Olympus).

Brca1^L1363P/+;Trp53^Δ/+ mice carrying Brca1^L1363P and Trp53^Δ alleles in cis were intercrossed for the isolation of mouse embryonic fibroblasts (MEF). Heads and organs of E13.5 embryos were removed, and the remaining tissue was minced, rinsed in PBS, and incubated for 30 minutes in 0.5 mL 0.05% (w/v) trypsin-EDTA (GIBCO, 25300054) at 37°C. Cell aggregates were dissociated in DMEM (GIBCO, 11880028) supplemented with 10% volume for volume (v/v) FBS (Serana, S-FBSP-EU-015) and penicillin-streptomycin (P/S; GIBCO, 15070063). Cells were plated on 10 cm dishes and cultured at 3% O₂.

Tumor-derived organoids were established following an adapted protocol (20). Cryopreserved tumor pieces were digested in Advanced DMEM/Ham's F-12 (AdDMEM/F12; GIBCO, 12634028) with 2 mg/mL collagenase type IV (GIBCO, 17104–019) for 30 minutes at 37°C while gently shaking. Suspensions were centrifuged at 1,500 rpm for 10 minutes and pellets were resuspended in ADDF+++ [AdDMEM/F12, 100× GlutaMAX (GIBCO, 35050061), 1 mol/L HEPES (Sigma-Aldrich, H4034), 100× P/S]. After centrifugation at 1,500 rpm for 10 minutes, pellets were resuspended in ADDF+++ and filtered (70 μm). Filtered organoids were pelleted (1,500 rpm, 10 minutes) and suspended in ENR growth medium [ADDF+++, 50x B-27 supplement (GIBCO, 17504044), 20 μg/mL EGF (Sigma-Aldrich, E4127), 500 mmol/L N-acetyl-L-cysteine (Sigma-Aldrich, A9165)] mixed 1:1 with Cultrex Reduced Growth Factor Basement Membrane Extract Type 2 (BME; Trevigen, 3533-001-02). Embedded organoids were seeded, incubated for 30 minutes at 37°C, and cultured in ENR medium supplemented with 10 μmol/L Nutlin-3a (Sigma-Aldrich, SML0580) under standard conditions (37°C, 5% CO₂, 21% O₂).

Mouse tumor cell lines were subsequently derived from respective early-passage organoid cultures. Organoids were dissociated by triturating in TrypLE (GIBCO, 12605010) and centrifuged at 800 rpm for 5 minutes. Pellets were suspended in DMEM/Nutrient Mixture F-12 media (DMEM/F12; GIBCO, 11320033) supplemented with 10% FBS, 100× P/S, 5 μg/mL insulin (Sigma-Aldrich, I0516), 5 ng/mL EGF, and 5 ng/mL cholera toxin (Gentaur, INABA 569B). Cell lines were cultured under low oxygen conditions (3% O₂).

Cultured cells were trypsinized, counted, washed twice with PBS, and lysed in NETNG-300 lysis buffer (300 mmol/L NaCl, 20 mmol/L Tris-HCl pH 7.5, 1 mmol/L EDTA, 0.5% NP40, 10% glycerol) complemented with 1 mg/mL Pefabloc (Roche, 11429868001) and cOmplete Mini Protease Inhibitor Cocktail (Roche, 11836153001). Protein concentrations were determined with Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, 23225). For BRCA1 coimmunoprecipitation, lysates were incubated overnight with mouse anti-BRCA1 antibody (21) at 4°C. Subsequently, lysates were incubated with Pierce agarose beads (Thermo Fisher Scientific, 20423) for 2 hours at 4°C and spin column-purified complexes were resolved on NuPAGE 4%–12% Bis-Tris SDS-MOPS polyacrylamide gels (Invitrogen, NP0321). Western blot protein analysis of input and immunoprecipitated samples used the above mouse anti-BRCA1 (1:10) and rabbit polyclonal anti-PALB2 (1:2,000) primary antibodies and rabbit anti-mouse IgG/HRP (1:2,000; Dako, P0260, RID: AB_2636929) and goat anti-rabbit IgG/HRP (1:2,000; Dako, P0448, RID: AB_2617138) secondary antibodies. Protein was visualized using ECL (Bio-Rad, 1705060) and imaged with Fusion FX7 imaging system (Vilber).

Experiments were performed as described previously (22) with small modifications. MEFs were seeded on coverslips overnight, washed with PBS, and covered with a mylar foil (Birkelbach Kondensatortechnik), allowing α-particle irradiation from below by turning over the coverslip. Irradiation was done using a 241 Am point source by moving the source underneath three areas of the coverslip for 30 seconds. Cells were incubated for 1 hour at 37°C and washed with ice-cold PBS. Subsequently, cells were extracted with cold CSK buffer [10 mmol/L HEPES-KOH pH 7.9, 100 mmol/L NaCl, 300 mmol/L sucrose, 3 mmol/L MgCl₂, 1 mmol/L EGTA, 0.5% (v/v) Triton X-100] and cold CSS buffer [10 mmol/L Tris pH 7.4, 10 mmol/L NaCl, 3 mmol/L MgCl₂, 1% (v/v) Tween 20, 0.5% (w/v) sodium deoxycholate] for 5 minutes each before fixation with 4% paraformaldehyde for 30 minutes at room temperature. Fixed cells were washed twice for 10 minutes (0.1% Triton X-100 in PBS) and washed for 30 minutes in blocking solution (0.5% BSA and 0.15% glycine in PBS). Primary antibodies were diluted in blocking buffer and incubated overnight at 4°C. Cells were washed twice for 10 minutes (0.1% Triton X-100 in PBS) and once quickly in blocking buffer. Secondary antibodies were diluted in blocking buffer and cells were incubated for at least 1 hour at room temperature in the dark. Finally, cells were washed twice with PBS and coverslips were mounted using Vectashield with DAPI. Quantification was done as described previously. Primary antibodies include: rabbit anti-53BP1 (1:1,000; Novus Biologicals, NB100-304, RID: AB_10003037) and mouse anti-RPA34 (1:1,000; Abcam, Ab2175, RID: AB_302873). Secondary antibodies include: Alexa Fluor 594 goat anti-rabbit IgG (Invitrogen, A-31631) and Alexa Fluor 488 goat anti-mouse IgG (1:1,000; Invitrogen, A-11001, RID: AB_2534069).

Cells were grown overnight on 12 mm glass coverslips and γ-irradiated (Best Theratronics) the following day with 10 Gy. EdU was added to refreshed medium after 2 hours of recovery, and 1 hour later cells were washed with PBS++ (PBS with 0.5 mmol/L MgCl₂ and 1 mmol/L CaCl₂). Ice-cold 0.5% Triton X-100 (Sigma-Aldrich, X100) in PBS was used for pre-extraction, followed by a PBS++ wash and fixation with 4% formaldehyde solution for 10 minutes at room temperature. Fixed cells were washed once with PBS and permeabilized in ice-cold 1:1 methanol:acetone for 20 minutes. Click-iT EdU Alexa Fluor 647 imaging kit (Invitrogen, C10340) was used for EdU detection. Staining buffer [SB; 1% BSA (Sigma-Aldrich, A8022), 0.15% glycine (Sigma-Aldrich, 15527), 0.1% Triton X-100, 0.5% FBS, in PBS] was used for blocking and antibody dilutions. Cells were incubated in SB for 15 minutes prior to the primary antibody incubation of 2 hours at room temperature. Secondary antibody incubation was for 1 hour at room temperature. Coverslips were mounted with VECTASHIELD Hardset Antifade Mounting Medium with DAPI (Vector Laboratories, H-1500). Antibodies included recombinant rabbit monoclonal anti-RAD51 primary antibody (1:2,000; Abcam, ab133534, RID: AB_2722613) and goat anti-rabbit Alexa Fluor 488 secondary antibody (1:400; Invitrogen, A-11008, RID: AB_143165). Stacked images (∼5 μm Z-volume) were taken with Leica TCS SP5 confocal microscope at 63× oil objective, and a minimum of 50 EdU+ cells per line was analyzed in ImageJ using an in-house macro. Automatic segmentation of nuclei based on DAPI signal thresholds was used to detect nuclei. Foci were detected using maximum intensity Z-stack projections with background-subtraction correction, in addition to filtering parameters for discriminating foci diameter size and adjusting signal to noise. The mean intensity EdU signal was used to select S phase nuclei.

MEFs were plated in triplicate at a density of 1,000 cells per well in a 12-well plate. The following day, cisplatin (in saline-mannitol, Mayne Pharma) or AZD2461 PARP inhibitor (in DMSO, Syncom) was added to refreshed medium at varying concentrations. After 6 days, MEFs were fixed with 4% formaldehyde for 15 minutes, stained with crystal violet (Supelco, 101408) for 30 minutes, washed three times with demineralized water, and dried overnight. Plate images were taken with Gel Count (Oxford Optronix). For quantification, crystal violet was dissolved in 10% acetic acid for 1 hour while shaking, and absorbance was measured at 560 nm using Infinite 200 microplate reader (Tecan).

DNA fiber analysis was carried out according to the previously defined standard protocol (23, 24). Briefly, cells were sequentially pulse labeled with 30 μmol/L 5-Chloro-2′-deoxyuridine (CldU; Sigma-Aldrich, C6891) and 250 μmol/L idoxuridine (IdU; European Pharmacopoeia, I0050000) for 15 minutes each. For assessing fork degradation, cells were treated with 4 mmol/L hydroxyurea (Sigma-Aldrich, H8627) for 3 hours. After incubation, cells were collected and resuspended in PBS at 2.5 × 10⁵ cells/mL. Labeled cells were mixed 1:1 with unlabeled cells, and 2.5 μL of cells were added to 7.5 μL of lysis buffer [200 mmol/L Tris-HCl pH 7.5, 50 mmol/L EDTA, 0.5% (w/v) SDS] on a glass slide. After 8 minutes, the slides were tilted at 15°–45° and the resulting DNA spreads were air dried and fixed in 3:1 methanol:acetic acid overnight at 4°C. The fibers were denatured with 2.5 mol/L HCl for 1 hour, washed with PBS, and blocked with 0.2% Tween 20 in 1% BSA/PBS for 40 minutes. The newly replicated CldU and IdU tracks were incubated for 2.5 hours in the dark at room temperature with anti-BrdU antibodies recognizing CldU (1:500; Abcam, ab6326, RID: AB_305426) and IdU (1:100; B44; BD Biosciences, 347580, RID: AB_400326), followed by a 1-hour incubation with secondary antibodies at room temperature in the dark: Alexa Fluor 488 goat anti-mouse IgG (1:300; Invitrogen, A11001, RID: AB_2534069) and Cy3 donkey anti-rat IgG (1:150; Jackson Immuno-Research Laboratories, 712-166-153, RID: AB_2340669). Fibers were visualized and imaged with Carl Zeiss Axio Imager D2 microscope using 63× Plan Apo 1.4 NA oil immersion objective. Data analysis was carried out with ImageJ software64 and statistical analysis using GraphPad Prism.

Metaphase spreads were carried out according to the standard protocol described previously (24). Briefly, cells were treated with 500 nmol/L olaparib (AZD2281; Selleck Chemicals, S1060) for 24 hours before preparing metaphase spreads (final confluence of 50%–80%). Cells were arrested in metaphase by incubating with colcemid (N-methyl-N-deacetyl-colchicine; Roche, 10295892001) for the last 16 hours before harvesting the cells. Collected cells were then treated with hypotonic solution (0.075 mol/L KCl) for 30 minutes at 37 °C and fixed with 3:1 methanol:acetic acid. Telomere-FISH was further carried out to study chromosomal aberration. Metaphases were hybridized with telomere repeat-specific peptide nucleic acid probes (PNA Bio, F1002) as described previously (25) to label the telomeres. A minimum of 60 metaphase images were obtained using Carl Zeiss Axio Imager D2 microscope with 63x Plan Apo 1.4 NA oil immersion objective and analyzed with ImageJ software64.

llumina HiSeq 2500 using V4 chemistry (Illumina Inc.) was performed as previously described with 65 base single reads (26), and aligned and summarized as done before (27). The entire analysis was implemented by Julian de Ruiter using Snakemake (Snakemake version 5.19.2; wrapper version 0.60.0; ref. 28) and is freely available on GitHub (https://github.com/jrderuiter/snakemake-cnvseq). An unsupervised clustering was performed on the segmented copy number data. Copy number instability was scored by calculating the fraction of bins with copy number values above or below a threshold of 2.5 and 1.5, respectively, in the segmented copy number data. Copy number variation sequencing (CNV-seq) data generated in this study have been deposited in the NCBI's Gene Expression Omnibus (GEO) database (GEO GSE182449).

llumina TruSeq mRNA libraries were generated and sequenced with 65 base single reads on HiSeq 2500 using V4 chemistry (Illumina Inc.) as described previously (26), and reads were aligned and expression counts generated as before (27). The analysis was implemented by Julian de Ruiter using Snakemake (Snakemake version 5.19.2; wrapper version 0.63.0; ref. 28) and is freely available on GitHub (https://github.com/jrderuiter/snakemake-rnaseq). Singscore package (29) was used to determine epithelial–mesenchymal transition (EMT) scores based on the EMT signature in Supplementary Table S2B of Huang and colleagues (30). RNA sequencing data generated in this study have been deposited in the NCBI's GEO database (GEO GSE182448).

Cryopreserved tumor pieces (1–2 mm diameter) from KP, KB1P, and KB1(L1363P)P spontaneous tumor donors were transplanted orthotopically into the fourth mammary gland of recipient female FVB mice, ages 8–12 weeks. Once outgrown tumors reached 200 mm³ (day 0), animals were randomly apportioned among three treatment arms (cisplatin, PARP inhibitor AZD2461, or no treatment). Cisplatin (6 mg/kg) was administered intravenously into the tail on day 0. Repeat doses were given after 14 days to unresponsive tumors (>100 mm³); responsive tumors (<100 mm³) received a repeat dose once returning to 200 mm³. A maximum of four doses was administered. AZD2461 (100 mg/kg) was administered by oral gavage for 29 consecutive days (days 0–28). Animals of all treatment arms were sacrificed when tumor burden reached 1,200 mm³ or if they showed signs of distress outlined in the Code of Practice for Animal Experimentation in Cancer Research.

Details for statistical methods are explained in each figure legend. All multiple comparison follow-up tests (Figs. 2B, 2C, 2F, 2G, and 4I) were performed relative to the indicated wild-type Brca1 MEF and tumor cell line controls. Statistical analysis for RAD51 IRIF (Figs. 2C and 4I) was performed on only the γ-irradiated groups, non-irradiated lines were excluded from comparison. Log-rank (Mantel–Cox) was used for Kaplan–Meier survival curves while applying Bonferroni multiple comparisons correction with a significance threshold of P = 0.05. All statistical significance was determined using GraphPad Prism.

We used CRISPR/Cas9-mediated genome editing in FVB mouse zygotes to model the BRCA1 coiled-coil domain VUS c.4220T>C p.L1407P, which disrupts the interaction of BRCA1 with PALB2. The BRCA1 coiled-coil domain is well conserved between humans and mice (Fig. 1A), where the murine equivalent of human p.L1407P is p.L1363P. As expected, the COILS algorithm (31) predicts that both human BRCA1 p.L1407P and mouse BRCA1 p.L1363P disable the alpha-helical structure of the coiled-coil domain (Fig. 1B). Because CRISPR/Cas9 often modifies both alleles of a target gene and Brca1 is essential for embryonic development (32, 33), we combined the homology-dependent recombination template for introduction of the p.L1363P variant with a second template that only disrupts gRNA recognition (Supplementary Fig. S1A). Through this approach, we obtained a Brca1 p.L1363P (Brca1^LP; Supplementary Fig. S1A and S1B) founder, which was crossed with wild-type FVB mice for subsequent experiments. Heterozygous Brca1^LP mice are viable without any apparent aberrant phenotypes.

For a first functional analysis of Brca1 p.L1363P in vivo, heterozygous Brca1^LP mice were intercrossed and their offspring was genotyped. No Brca1^LP/LP mice were born; therefore, embryos were analyzed at different stages of embryonic development. Like Brca1-null mice, Brca1^LP/LP mice showed a general growth defect at mid-gestation (Fig. 1C). Although the phenotype was less severe than those of Brca1-null mice (32, 33) or mice carrying pathogenic mutations in the Brca1 RING or BRCT domain-encoding sequences (15, 16), no living embryos were observed after E11.5 (Table 1).

Loss of Trp53 is known to delay embryonic lethality in Brca1- and Palb2-null mice (33–35), so we questioned whether this phenotype would likewise occur in Brca1^LP/LP mice on a Trp53-deficient background. Brca1 and Trp53 are both located on chromosome 11 (at 65.18 and 42.83 cM, respectively), so we bred male mice with Brca1^LP and Trp53^Δ2–10 (Trp53^Δ; ref. 36) in cis (as determined from genotyped offspring) and crossed these with Brca1^LP/+ or Brca1^LP/+;Trp53^Δ/+ female mice. While heterozygous inactivation of Trp53 delayed embryonic lethality in Brca1^LP/LP mice, it did not result in viable offspring (Table 2), Brca1^LP/LP;Trp53^Δ/+ embryos were smaller than Brca1-proficient littermates (Fig. 1C) but continued to develop until at least E13.5. For a molecular explanation to the growth defect of Brca1^LP/LP embryos, we checked markers of proliferation, apoptosis, and cell-cycle regulation (Fig. 1D and E; Supplementary Fig. S1C). In line with the partial rescue of embryonic development, heterozygous deletion of Trp53 allowed most E13.5 Brca1^LP/LP cells to proliferate, as indicated by the expression of Ki-67, and we observed only few cleaved caspase 3–positive apoptotic cells (Supplementary Fig. S1C). However, similar to Brca1-knockout embryos (32), E13.5 Brca1^LP/LP;Trp53^Δ/+ embryos showed increased expression of the TP53 effector protein Cyclin-dependent kinase inhibitor 1A (P21) in several tissues including the embryonic heart and liver (Fig. 1D and E). P21-mediated G₁ phase cell cycle arrest can explain the marked growth suppression of critical organs such as heart and liver and likely contributes to the embryonic lethality of Brca1^LP/LP and Brca1^LP/LP;Trp53^Δ/+ mice.

In the complete absence of TP53, Brca1^LP/LP mice developed apparently normal until at least E13.5, although no postnatal survival was observed upon compound heterozygous intercrosses (Table 3). This allowed us to isolate E13.5 MEFs from Brca1^LP/LP;Trp53^Δ/Δ embryos and Brca1^LP/+;Trp53^Δ/Δ littermate controls for further evaluation of the functional consequences of Brca1 p.L1363P.

To verify whether mouse Brca1 p.L1363P phenocopies human BRCA1 p.L1407P, we analyzed Brca1^LP/LP;Trp53^Δ/Δ (LP/LP) mutant and Brca1^LP/+;Trp53^Δ/Δ (LP/+) control MEFs for BRCA1–PALB2 interaction and HRR defects. As expected, based on sequence conservation and the predicted effect of leucine to proline substitution on the coiled-coil structure (Fig. 1A and B), Brca1 p.L1363P severely attenuated BRCA1-PALB2 binding. While PALB2 readily coimmunoprecipitated with BRCA1 in whole-cell extracts of Brca1^LP/+;Trp53^Δ/Δ MEFs, this was not possible for Brca1^LP/LP;Trp53^Δ/Δ MEFs (Fig. 2A). Although the abrogation of BRCA1-PALB2 binding did not affect the initial DNA end-resection step of HRR, as measured by recruitment of RPA to 53BP1-positive alpha tracks (Fig. 2B), it still led to HRR defects. Ionizing radiation–induced foci (IRIF) of RAD51, a marker for HRR proficiency, were significantly decreased in Brca1^LP/LP;Trp53^Δ/Δ MEFs (Fig. 2C; Supplementary Fig. S2). This correlated with an increased sensitivity to cisplatin and PARP1 inhibition (Fig. 2D and E; Supplementary Fig. S3A–S3D) and accumulation of DNA chromosomal abnormalities upon PARP1 inhibitor treatment (Fig. 2F). Abrogation of BRCA1-PALB2 binding may affect functions of both BRCA1 and BRCA2, including those outside of HRR such as their shared role in protecting stalled replication forks (37, 38). Nevertheless, and in line with previously published work (39, 40), DNA fiber analysis of Brca1^LP/LP;Trp53^Δ/Δ MEFs did not reveal defects in replication fork protection (Fig. 2G). Taken together, these data show that the abrogation of BRCA1–PALB2 interaction by Brca1 p.L1363P impairs HRR, without affecting the roles of BRCA1 in DNA end-resection and replication fork protection.

The embryonic lethality of Brca1^LP/LP mice indicates that an intact BRCA1 coiled-coil domain is functionally important in vivo, in line with its requirement for BRCA1-mediated HRR (2, 11). To analyze whether the functional defect of Brca1 p.L1363P also compromises mammary tumor suppression, the Brca1^LP allele was crossed into a K14cre-driven (FVB) mouse model for conditional inactivation of Brca1 and Trp53 in epithelial cells (17). Resulting K14cre;Brca1^LP/F;Trp53^F/F [KB1(L1363P)P] female mice were monitored for tumor formation and compared with K14cre;Brca1^F/F;Trp53^F/F (KB1P), K14cre;Brca1^F/+;Trp53^F/F (KB1hetP), and K14cre;Brca1^LP/+;Trp53^F/F [KB1(L1363P)hetP] control cohorts. KB1(L1363P)P and KB1P mice developed tumors at similar median latencies of 166.5 and 173 days, respectively (Fig. 3A; Supplementary Fig. S4A). These median latencies are significantly shorter than the 203.5 days for KB1hetP controls. Thus, in our model, Brca1 p.L1363P causes a similar acceleration of Trp53-deficient tumor formation as complete loss of Brca1. The median tumor-free latency of 228 days for KB1(L1363P)hetP mice did not differ from that of KB1hetP mice, showing that Brca1 p.L1363P does not have any dominant negative effects on tumor suppression in this mouse model. Although K14cre-mediated inactivation of Trp53 results in a range of epithelial tumors, including skin and mammary tumors, concomitant loss of Brca1 is known to skew the tumor type spectrum toward mammary tumors (17). Strikingly, the percentage of mammary tumors is even higher in KB1(L1363P)P mice than in KB1P mice, coinciding with a moderate but significantly shorter median mammary tumor-free survival (Fig. 3B and C). If and how this skew toward mammary tumor formation is caused by hypomorphic Brca1 p.L1363P activity remains to be determined. Regardless, the combined data from our tumor observation cohorts clearly show that Brca1 p.L1363P abolishes BRCA1-mediated mammary tumor suppression.

KB1P mammary tumors are mainly adenocarcinomas, defined by their epithelial nature and solid growth pattern (Fig. 3D; Supplementary Fig. S4B; refs. 15–17). In contrast, KB1(L1363P)P mammary tumors are predominantly carcinosarcomas with both epithelial and mesenchymal components or wholly mesenchymal sarcomatoid tumors (Fig. 3D; Supplementary Fig. S4B–S4D). As previously observed for K14cre;Trp53^F/F (KP), KB1P, and K14cre;Brca2^F/F;Trp53^F/F (KB2P) mammary tumors, tumors with mesenchymal phenotypes show more stable DNA copy number profiles than adenocarcinomas (Fig. 3E; ref. 41). This dampening can partially be explained by a larger contribution of stromal cells in which Brca1 and Trp53 are not switched off (Supplementary Figs. S4E-S4G and S5A). The three KB1(L1363P)P adenocarcinomas available for sequencing showed profound copy number changes, but this sample size is too small to draw robust conclusions on the effect of Brca1 p.L1363P on mammary tumor genome stability. Averaged over all histology types, KB1(L1363P)P mammary tumors do not show more copy number gains or losses than KP tumors (Fig. 3E; Supplementary Fig. S5B), which seems largely dependent on histology type. The same applies to differences in gene expression, which mainly confirm the histological classifications of adenocarcinoma, carcinosarcoma, and sarcomatoid tumors, and show that most KB1(L1363P)P mammary tumors have an EMT-like phenotype (Fig. 3F; Supplementary Fig. S5C and S5D).

To analyze the response of KB1(L1363P)P mammary tumors to HRR deficiency–targeted therapy, we performed orthotopic transplantations with spontaneous donor tumors as described previously (15, 16, 42). To capture the heterogeneity of KB1(L1363P)P tumors, we transplanted five different donor tumors: one adenocarcinoma, three carcinosarcomas, and one sarcomatoid (Supplementary Fig. S6A and S6B). For comparison of treatment responses, we transplanted three KB1P adenocarcinomas, two KP adenocarcinomas, and one KP carcinosarcoma. Once outgrowths of transplanted tumor pieces reached a size of 200 mm³, mice were treated with cisplatin, the PARP inhibitor AZD2461, or left untreated (Fig. 4A). KB1(L1363P)P mammary tumors responded significantly better than KP tumors to both cisplatin and AZD2461 and showed an overall intermediate response to both DNA repair-targeting drugs compared with Brca1-proficient and Brca1-deficient controls (Fig. 4B–4E; Supplementary Fig. S6C). While a faster growth rate may partially explain the decreased sensitivity of KB1(L1363P)P tumors to these targeted therapies (Fig. 4C and 4F–H; Supplementary Fig. S6D–S6I), it seems likely that the HRR defect of tumor cells expressing hypomorphic BRCA1 p.L1363P—with intact RING and BRCT domains—is less severe than that of Brca1-null tumor cells. To establish whether this intermediate response is explained by a partial HRR defect in tumor cells expressing hypomorphic BRCA1 p.L1363P, we performed RAD51 IRIF analysis of KB1(L1363P)P mammary tumor-derived cell cultures. In contrast to KB1P, KB1(L1363P)P tumor cells were able to induce RAD51 foci in response to γ-radiation; however, RAD51 formation was clearly compromised compared with KP (Fig. 4I and J). Collectively, the results of our study suggest that BRCA1 coiled-coil variants that attenuate BRCA1–PALB2 interaction increase the risk of developing breast cancer but leave sufficient functionality to partially withstand HRR deficiency–targeted therapy.

In this study, we present the first direct evidence that a BRCA1 missense variant outside of the RING and BRCT domains increases the risk to develop breast cancer. In line with previous observations (9, 11), we show that mouse Brca1 p.L1363P coiled-coil domain variant (equivalent to human BRCA1 p.L1407P) disrupts the interaction of BRCA1 with PALB2 and causes a defect in RAD51 recruitment to DNA double-strand breaks (DSB). Although this correlates with increased sensitivity to cisplatin and PARP inhibition, the functions of BRCA1 p.L1363P in DNA DSB end-resection and replication fork protection remain unaffected. Thus, the BRCA1 p.L1363P HRR defect differs from those caused by BRCA1 loss-of-function mutations in the RING and BRCT domains. In addition, we make a number of in vivo observations that clearly separate the defects caused by this coiled-coil domain mutation from those of previously generated Brca1-mutant mouse models.

Brca1 p.L1363P homozygous embryos show a milder embryonic phenotype than homozygous Brca1-null (17, 32), RING mutant (15, 16), or BRCT mutant (16) mice, although they still show a general growth retardation and die before day E13.5 of embryonic development. The observed embryonic lethality of homozygous Brca1 p.L1363P mice seems at odds with the viability of the reciprocal Palb2 coiled-coil domain mouse mutant. The Palb2 p.(L24_K26delinsAAA) mutation of Palb2^CC6 mice also disrupts the interaction of PALB2 with BRCA1, yet homozygous Palb2^CC6/CC6 mice are viable and only show a defect in male fertility that is linked to impaired meiosis (43). Palb2^CC6/CC6 MEFs, however, have a growth defect and their premature senescence and mitomycin C sensitivity (43) are in line with the phenotypes we observed in Brca1^LP/LP embryos and fibroblasts. It cannot be excluded that the consequences of disrupting BRCA1–PALB2 interaction are more or less severe depending on genetic background (mixed for Palb2^CC6 mice vs. FVB for Brca1^LP mice). Genetic background effects on the embryonic phenotypes of Brca1-mutant mice have been observed before (44) and are in fact supported by the prolonged survival of Brca1^LP/LP embryos upon reduction of Trp53 gene dosage. Alternatively, inactivation of the BRCA1 coiled-coil domain may affect other interactions or processes, or PALB2 p.(L24_K26delinsAAA) may be less disruptive for BRCA1–PALB2 interaction than BRCA1 p.L1363P. Of note, a recently published C57BL/6J x C3H/HeJ mouse model carrying an in-frame deletion in the BRCA1 coiled-coil region also shows embryonic lethality, albeit not with complete penetrance (40), while Brca1 p.L1363P homozygosity in mice with a mixed C57BL/6J x 129/Sv genetic background showed growth defects at E13.5 but were nonetheless born at expected Mendelian frequencies (14).

In the context of K14cre-driven conditional Brca1;Trp53 knockout mouse breast cancer models, the Brca1^LP allele has a similar effect on overall tumor latency as other Brca1 loss-of-function alleles (15–17); however, the tumor spectrum of KB1(L1363P)P mice differs from these other models by several aspects. K14cre;Trp53^F/F-based models can develop skin tumors, yet KB1(L1363P)P mice show a remarkable skew toward mammary tumor formation. In addition, whereas Brca1-null, RING mutant, or BRCT-mutant mammary tumors are mainly basal-like adenocarcinomas, KB1(L1363P)P mammary tumors show EMT-like characteristics that also associate with relatively low genomic instability, as assessed by copy number aberrations. Despite these phenotypes, mammary tumors of KB1(L1363P)P mice are HRR-deficient and show an intermediate response to cisplatin and PARP inhibition compared with Brca1 wild-type KP and Brca1-null KB1P mammary tumors.

Taken together, our data suggest that BRCA1 coiled-coil mutations disabling PALB2 binding confer a similarly increased risk to develop breast cancer as known pathogenic BRCA1 RING, BRCT, and frameshift mutations. Similar to what we observed in our mouse model, human BRCA1 coiled-coil mutant breast tumors may show a difference in histology compared with tumors lacking a functional RING or BRCT domain. BRCA1-associated breast cancers are frequently of the basal-like subtype, in contrast to BRCA2- or PALB2-associated breast cancers, which show similar histology types as sporadic cases (45, 46). Because BRCA1 coiled-coil mutations affect the interaction with PALB2 but leave other BRCA1 domain functions intact, BRCA1 coiled-coil mutant tumors may well lack a bias toward the basal-like subtype. In addition, human BRCA1 coiled-coil mutant breast tumors are expected to be sensitive to HRR deficiency–targeted treatment, although response and resistance mechanisms may differ from other BRCA1-deficient tumors.

No disclosures were reported.

E.M. Pulver: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. C. Mukherjee: Data curation, formal analysis, validation, investigation, methodology. G. van de Kamp: Data curation, formal analysis, validation, investigation, methodology. S.J. Roobol: Data curation, formal analysis, validation, investigation, methodology. M.B. Rother: Data curation, formal analysis, validation, investigation, methodology. H. van der Gulden: Data curation, investigation, writing–review and editing. R. de Bruijn: Data curation, software, formal analysis, investigation, methodology, writing–review and editing. M.V. Lattanzio: Investigation. E. van der Burg: Data curation, investigation. A.P. Drenth: Data curation, investigation. N.S. Verkaik: Investigation. K. Hahn: Funding acquisition, investigation, methodology. S. Klarenbeek: Investigation, methodology, writing–review and editing. R. de Korte–Grimmerink: Data curation, investigation. M. van de Ven: Supervision. C.E.J. Pritchard: Data curation, validation, investigation, methodology. I.J. Huijbers: Supervision. B. Xia: Resources, writing–review and editing. D.C. van Gent: Supervision, funding acquisition, validation. J. Essers: Supervision, funding acquisition, validation, writing–review and editing. H. van Attikum: Supervision, funding acquisition, validation. A. Ray Chaudhuri: Resources, supervision, funding acquisition, validation. P. Bouwman: Conceptualization, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. J. Jonkers: Conceptualization, supervision, funding acquisition, validation, visualization, writing–original draft, project administration, writing–review and editing.

The authors thank Jinhyuk Bhin for his assistance with bioinformatics analyses. The authors thank the NKI animal laboratory facilities and pathology unit, the NKI genomics core facility, and the NKI bioimaging facility for expert technical support. This work was supported by grants from the Dutch Cancer Society (KWF; NKI 2015-7877 to P. Bouwman, J. Jonkers, and Maaike P.G. Vreeswijk; 11008/2017-1 to A. Ray Chaudhuri; and 12092/2018 to J. Essers); the Dutch Research Council (NWO; VIDI VI. 193.131 to A. Ray Chaudhuri; VICI 91814643 to J. Jonkers; and VICI VI.C.182.052 to H. van Attikum); Oncode Institute, which is partly financed by KWF, NWO research program Mosaic (grant 017.008.022 to J. Jonkers); the Lundbeck Foundation (grant R223-2016-956 to J. Jonkers, Claus Storgaard Sørensen, and Finn Cilius Nielsen); Technologie stichting STW (project number 13577 to D.C. van Gent and J. Essers); and the Swiss National Science Foundation to K. Hahn.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.