BRCA1 mutations have been identified that increase the risk of developing hereditary breast and ovarian cancers. Genetic screening is now offered to patients with a family history of cancer, to adapt their treatment and the management of their relatives. However, a large number of BRCA1 variants of uncertain significance (VUS) are detected. To better understand the significance of these variants, a high-throughput structural and functional analysis was performed on a large set of BRCA1 VUS. Information on both cellular localization and homology-directed DNA repair (HR) capacity was obtained for 78 BRCT missense variants in the UMD-BRCA1 database and measurement of the structural stability and phosphopeptide-binding capacities was performed for 42 mutated BRCT domains. This extensive and systematic analysis revealed that most characterized causal variants affect BRCT-domain solubility in bacteria and all impair BRCA1 HR activity in cells. Furthermore, binding to a set of 5 different phosphopeptides was tested: all causal variants showed phosphopeptide-binding defects and no neutral variant showed such defects. A classification is presented on the basis of mutated BRCT domain solubility, phosphopeptide-binding properties, and VUS HR capacity. These data suggest that HR-defective variants, which present, in addition, BRCT domains either insoluble in bacteria or defective for phosphopeptide binding, lead to an increased cancer risk. Furthermore, the data suggest that variants with a WT HR activity and whose BRCT domains bind with a WT affinity to the 5 phosphopeptides are neutral. The case of variants with WT HR activity and defective phosphopeptide binding should be further characterized, as this last functional defect might be sufficient per se to lead to tumorigenesis.

Implications:

The analysis of the current study on BRCA1 structural and functional defects on cancer risk and classification presented may improve clinical interpretation and therapeutic selection.

This article is featured in Highlights of This Issue, p. 1

BRCA1 encodes a large, 1863-residue protein that functions as a hub, coordinating a large range of cellular pathways including DNA repair, transcriptional regulation, cell-cycle control, centrosome duplication, and apoptosis (1). Mutations in BRCA1 have been identified that predispose to breast and/or ovarian cancer. Further screening of patients revealed mutations throughout the whole BRCA1 gene sequence. The causal nature of mutations causing a premature stop is generally accepted. Other variants correspond to missense variations, deletions/insertions, or intronic variations preserving the reading frame. The causal or neutral nature of these variants is more difficult to establish. Therefore, they are called variants of uncertain significance (VUS). In France, they are detected in about 10% of the tested population in the absence of a causal mutation. Initial attempts to evaluate the clinical significance of VUS in BRCA1/2 were mainly based on family data including family history and cosegregation with the disease. From these data, VUS were classified as either neutral (class 1), likely neutral (class 2), associated to an unknown cancer risk (class 3), likely causal (class 4), and causal (class 5) (2). However, the majority of the VUS are rare, and therefore these family-based clinical analyses often lack statistical power. High-throughput experimental assays are then the most reliable way to determine the functional impact of variants with nontruncating changes in protein residue composition.

Our goal is to improve the understanding of cancer susceptibility caused by BRCA1 gene variations by developing a high-throughput and robust experimental approach for missense VUS functional characterization. At the present time and according to most frequently used BRCA1 genetic testing criteria, a causal mutation used for genetic counselling is found in approximately 10% of tested index cases (individuals for whom a complete study of both genes is performed; ref. 3). The carriers from such variants of class 5 affected with breast or ovarian cancer may now benefit from therapies based on platinum agents or PARP inhibitors that are specifically efficient against BRCA1/2-related tumors (4, 5). Moreover, geneticists can provide informative tests for relatives and adapt their management (reinsurance or preventive management) according to the test results (6). If the variant is classified as neutral (class 1), another molecular etiology for the familial risk should be found. However, today, most variants are assigned to class 3, which corresponds to a lack of knowledge on the VUS impact. Study of the impact of these VUS is based on different approaches (cosegregation, cooccurrence with a causal mutation, family history, RNA stability and quality, amino acid conservation, structural impact of the mutation, functional studies, loss of heterozygosity in tumors and tumor phenotype, and the use of multifactor risk models; refs. 7–11). A combination of structural and functional information is particularly well adapted to the characterization of rare missense variants, with the aim of evaluating the risk of cancer and taking clinical decisions.

Several groups have undertaken the description of the structure and function of a set of VUS, to predict whether the corresponding mutations or deletions/insertions are likely to affect biological pathways involving BRCA1 and lead to tumorigenesis. Most of the studies have analyzed the impact of mutations in the N-terminal and C-terminal globular domains, whose functions are largely described: the N-terminal RING domain associated to the protein BARD1 regulates BRCA1 localization and, through its E3 ubiquitin ligase activity, contributes to the G2–M cell-cycle checkpoint, whereas the C-terminal BRCT domains are responsible for transcription activation and binding to phosphorylated proteins involved in DNA double-strand break signaling and repair (12–14,53). The C-terminal region of BRCA1 encompasses two BRCT repeats (aa 1646–1736 and aa 1760–1855) that adopt similar structures and are packed together in a head-to-tail arrangement (15). These BRCT domains interact specifically with phosphorylated protein targets containing the sequence pSer-x-x-Phe (16–20, 52). The structural impact of the BRCT missense variations compiled in the BRCA1 Circos resource from the ENIGMA consortium was analyzed using limited proteolysis and phosphorylated BACH1–binding assays, and for a subset of these mutations, by expression in bacteria and characterization of the thermodynamic stability of the purified mutants (21–25). Impact of these mutations in cells was revealed using transcriptional activation assays (23), and for a small subset of these mutations, homology-directed DNA repair (or homologous recombination; HR) assays (26). However, a high-throughput approach that provides both structural and functional information on the same VUS and is recognized as sufficient to conclude in most cases about the impact of the VUS on tumorigenesis is still lacking.

To further understand the link between protein stability, phosphopeptide recognition, localization in cells, HR, and tumorigenesis, we decided to systematically analyze the 3D structure and function of the 78 BRCT missense variants deposited in the BRCAShare (ex-UMD-BRCA1 database; refs. 27, 28), including all VUS detected in France, except missense mutations with a splice impact (29, 30). Therefore, we developed high-throughput assays to (i) measure the capacity of the VUS to repair double-strand breaks by HR in cells, (ii) test their capacity to relocalize to the nucleus after addition of Mitomycin (MMC), (iii) measure the thermostability of the corresponding expressed/purified mutated BRCT domains, and (iv) investigate their binding to a large panel of phosphorylated peptides from different BRCA1 partners (ACC1, BACH1, CtiP, and Abraxas). Half of the selected VUS are also listed in other databases such as the BIC, KConfab, and ClinVar databases. For these VUS, we were able to compare our experimental data with previously published results obtained on subsets of VUS through either in vitro or in cell approaches, thus validating our results (23, 25, 26). Our large-scale analysis from both in vitro and in cell data provides a solid basis for discussing the impact of the different structural and functional defects on increased cancer risk.

Lentivirus-inducible BRCA1 shRNA system

An anti-BRCA1 shRNA construct was made targeting the 3′-untranslated region (3′-UTR) of the HsBRCA1 gene (Genbank accession number NC_000017.10). The target sense sequence is 5′-TATAAGACCTCTGGCATGAAT-3′ with a loop TTCAAGAGA. The sequence was cloned into AgeI and EcoRI of a pLKO-Tet-On vector purchased from Addgene (31). The pLKO-Tet-On expresses constitutively a tetracycline-controlled transcriptional suppressor Tet, which in turn controls expression of the anti-BRCA1 shRNA sequence inserted in the shRNA cloning site of the vector to be under the human H1 promoter. In the absence of doxycycline (a tetracycline derivative), expression of the anti-BRCA1 shRNA is blocked. When doxycycline is added to the culture medium, transcription of the anti-BRCA1 shRNA occurs, which results in the knockdown of the HsBRCA1 gene in a highly dose-dependent manner.

Lentivirus production

The constructed plasmid pLKO-Tet-On/Anti-BRCA1-shRNA, and the plasmids pMD2.G and pCMV-dR8.74 (kindly provided by Anne Galy, Genethon, France) were transfected into the HEK293T cells using Lipofectamine 2000 (Invitrogen) following the manufacturer's protocol. The plasmid pMD2.G encodes the envelope protein from the vesicular stomatitis virus (VSV-G) and the plasmid pCMV-dR8.74 encodes the proteins from the genes Gag and Pol from the human immunodeficiency virus type 1 (HIV-1). The media were changed 24 hours posttransfection. Then, the supernatants were collected at 48, 72, and 96 hours posttransfection. For each time point, after elimination of the cell debris, the collected supernatant was ultracentrifuged through a sucrose cushion (20%) at 20,000 rpm for 2 hours at 12°C. The supernatant was removed and the virus pellet was resuspended in 500 μL of complete media overnight at 4°C. The following day, the virus was pooled in the first tube. Finally, the volume was increased to 1 mL, and the solution was filtered, aliquoted, and stored at −80°C.

Design of the RG37 cell line containing the doxycycline-inducible shBRCA1

RG37 cells are SV40-transformed human fibroblasts containing a chromosomally integrated DR-GFP substrate that specifically monitors gene conversion (32). RG37 cells were infected with lentiviral particles containing a shRNA directed against the 3′UTR part of the BRCA1 messenger and the puromycin resistance gene. Puromycin was added 3 days after infection and cells were reseeded at low density to isolate individual clones. Clones were then screened for deficiency in the formation of BRCA1 foci after ionizing radiation, deficiency in HR (using the DR-GFP substrate; Fig. 2) and for the extinction of BRCA1 expression monitored by Western blot analysis. The new cell line was called RG37-shBRCA1.

Plasmid constructions for expression in mammalian cells

Cloning of the mutated BRCT1-BRCT2 gene into the pcDNA3 (modified)-full-length Brca1 was performed by nested PCR (33). First insert was generated by PCR from the gene coding for GST-BRCT1-BRCT2 using the primers: 5′TCAACAGAAAGGGTCAACAAAAGAAT3′ and 5′GCCGATATCATCGATTCAGTAGTGGCTGTG3′ (1). Second insert was generated by PCR from pcDNA3 (modified)-full-length BRCA1 gene using the primers: 5′ACACCCAGGATCCTTTCTTGATTG3′ (2) and 5′ATTCTTTTGTTGACCCTTTCTGTTGA3′. These inserts were associated by PCR using the primers 1 and 3. The new insert was cloned into the BamHI and EcoRV sites of the pcDNA3 (modified)-full-length BRCA1. All constructs were verified by DNA sequencing.

HR assays in human RG37-shBRCA1 cells

Cells were pretreated (or not) with 10 μg/mL doxycycline 3 days before plating. They were then seeded at 2 × 105 per well in 6-well plates. Twenty-four hours after plating, expression vectors encoding for BRCA1 (or its variant forms) and the HA-tagged meganuclease I-SceI were cotransfected using JetPEI reagent (PolyPlus, Ozyme). Forty-eight hours after transfection, cells were trypsinized and GFP+ cells were directly measured by flow cytometry.

Western blot analysis of BRCA1 protein

The expression of BRCA1 and I-SceI was monitored by Western blot analysis (Supplementary Fig. S1A). Therefore, protein extracts (25–50 μg) were resolved on 9% SDS-PAGE, then transferred to a nitrocellulose membrane (0.22 μm) and probed with the following specific antibodies: anti-BRCA1 (mouse, ab16780, Abcam), anti-HA (mouse, sc-7392, Santa Cruz Biotechnology Inc.), anti-αTubulin (mouse, T5168, Sigma-Aldrich). Immunoreactivity was visualized using an Enhanced Chemiluminescence Detection Kit (WesternBright ECL, Advansta Inc.).

Statistical analysis

Each batch of assays was run with positive and negative controls and each variant was tested at least in triplicate in at least four independent experiments. The statistical analysis was performed using the R language release 3.5.1 (34) running on a Linux server. We have compared the variant BRCA1 HR to the WT BRCA1 HR using a one-way paired Student t test. To account for the multiple testing, the P values were adjusted using the Benjamini–Hochberg method (35); we controlled the false discovery rates at a level α = 0.05.

Localization assays

RG37-shB1 cells were treated with doxycycline two days before transfection. Cells were seeded at a density of 1 × 105 per well in 6-well plates each containing a glass cover slide and transfected with various BRCA1 variants using JetPEI reagent. Twenty-four hours later, cells are treated with 2.5 μmol/L mitomycin for 2 hours and then the medium was changed. After 8 hours, cells were fixed (PBS, 2% paraformaldehyde, 10 minutes) and permeabilized (PBS, 0.1% Triton X-100, 5 minutes). Cells were then incubated with PBS containing 1% BSA, 0.05% Tween, and anti-BRCA1 antibody (ab16780, Abcam) for 45 minutes at 37°C. After washing three times, cells were then incubated with PBS containing 1% BSA, 0.05% Tween, and Alexa 568–conjugated secondary antibodies (Jackson ImmunoResearch). Cells were then stained with DAPI and examined with fluorescence microscope. Statistical significance was calculated using a two-tailed Student t test with GraphPad.

Plasmid construction for bacterial expression

The gene coding for the BRCA1 region BRCT1-BRCT2 (aa 1646–1863) was cloned using the LIC technology (Ligase Independent Cloning) into the pETM-10 and pETM-30 vectors (EMBL) using the following primers: forward 5′GGCAGGAGCAGCCTCGGAGAATCTTTATTTTCAGGGCGTCAACAAAAGAATGTCCATGGTGGT3′; and reverse 5′GCAAAGCACCGGCCTCGttaTCAGTAGTGGCTGTGGGGGATCT3′. All mutations were generated by Site-Directed Mutagenesis Kit (Stratagene) and verified by sequencing.

Peptides for binding studies

The following peptides were synthetized by GeneCust Europe: one control peptide CTRL-P (PTRV-pS-SPVFGAT), 4 monophosphorylated peptides CTiP-P (PTRVS-pS-PVFGAT), BACH1-P (ISRST-pS-PTFNKQTK), ACC1-P (DSPPQ-pS-PTFPEAGH), Abraxas-1P (or AB-1P) (GFGEYSR-pS-PTF), and 1 biphosphorylated peptide Abraxas-2P (or AB-2P) (GFGEY-pS-R-pS-PTF).

Protein expression and solubility assay

The BRCT variants cloned into the pETM-30 expression vector were introduced into the E. coli BL21(DE3) strain and expressed in 96-well microplates using a self-inducible medium. After lysis, soluble protein fractions were separated from insoluble fractions. Soluble fractions were transferred to new 96-well microplates for gel analysis and the rest of the soluble fraction was incubated with 20 μL Hi-Trap chelating beads, washed, and diluted into SB1x. Bacteria pellets were suspended in 50 μL 2% SDS. Protein expression and solubility were further analyzed by SDS-PAGE.

Purification of BRCA1 BRCT domain variants

The BRCTs variants cloned into the pETM-30 expression vector were introduced into the E. coli BL21(DE3) strain and were grown at 37°C in LB medium containing 75 μg/mL kanamycin to an OD600nm of 1.0. Protein overproduction was induced at 20°C with 1 mmol/L isopropyl β-D-thiogalactoside (IPTG) for 12 hours. The bacteria were then harvested by low-speed centrifugation at 6,000 rpm for 15 minutes. The bacterial pellet was suspended in 30 mL lysis buffer [100 mmol/L Tris pH 7.5, 150 mmol/L NaCl, 1% (w/v) Triton X-100, 10% glycerol, 1 mmol/L EDTA, 10 mmol/L DTT, 1 mmol/L phenylmethylsulfonyl fluoride (PMSF), 2 mmol/L ATP, 10 mmol/L MgSO4] and incubated with 1% lysozyme overnight at 4°C. Then, 1 μL benzonase (Sigma), 2 mmol/L ATP, 10 mmol/L MgSO4 and 10 mmol/L MgCl2 and 10 mmol/L DTT were added. After 30 minutes, the extract was centrifuged at 20,000 rpm for 30 minutes, and the soluble protein fraction was loaded on a 30 mL Glutathione sepharose 4FF column previously equilibrated in 50 mmol/L Tris, 150 mmol/L NaCl pH 7.5. The column was washed with a buffer containing 1 mol/L NaCl, then 150 mmol/L NaCl, and then with 50 mmol/L Tris, 150 mmol/L NaCl pH 7.5. The TEV protease was added and incubated overnight at 4°C. After TEV cleavage, the elution from the GST-trap column was loaded onto a HisTrap column equilibrated in 50 mmol/L Tris pH 7.5, 150 mmol/L NaCl. The protein was eluted with 50 mmol/L Tris pH 7.5, 150 mmol/L NaCl, and 10 mmol/L imidazole. Fractions highly enriched in BRCT variants were diluted 5-fold in 50 mmol/L Tris pH 7.5 and loaded on a 5 mL Q-Sepharose High Performance column equilibrated in 50 mmol/L Tris pH 7.5. Bound proteins were eluted with between 200 and 350 mmol/L NaCl. In both cases, the collected fractions were analyzed by 0.1% SDS-15% PAGE, using as a marker the broad range prestained protein marker (Bio-Rad). The purified protein was characterized by electrospray ionization mass spectroscopy.

Isothermal titration calorimetry

ITC was performed using a high-precision VP-ITC calorimetry instrument. To characterize interactions between the BRCT1-BRCT2 domains and peptides, all proteins were dialyzed against 50 mmol/L Tris-HCl pH 7.5, 150 mmol/L NaCl, 10 mmol/L β-mercaptoethanol, and protease inhibitors (Roche). BRCT1-BRCT2 domains (10–20 μmol/L) in the calorimetric cell at 30°C were titrated with the peptide (at a concentration of 100–200 μmol/L in the injection syringe). Analyses of the data were performed using the Origin software provided with the instrument.

Size-exclusion chromatography

Size-exclusion chromatography experiments aiming at identifying interactions between BRCT1-BRCT2 domains and peptides after ITC were performed using a Superdex-75 10/300 GL column (GE Healthcare) preequilibrated in the ITC buffer (50 mmol/L Tris-HCl pH 7.5, 150 mmol/L NaCl, 10 mmol/L β-mercaptoethanol and protease inhibitors; Roche). Proteins were concentrated after ITC (calorimetric cell) to obtain a volume of 500 μL and were loaded on the column at a flow rate of 0.5 mL/minute at 4°C.

Thermostability measurements by fluorescence-based thermal shift assay

The FBTSA method is used to monitor the BRCT domain thermal denaturation and changes induced by a mutation and/or a binding event. To measure the thermostability of WT and mutated BRCT domains, we mixed 1 μg of purified protein in 50 mmol/L Tris–HCl pH 7.4, 150 mmol/L NaCl, 2 mmol/L β-mercaptoethanol, and the SYPRO Orange dye (diluted 400-fold from a 5,000-fold stock solution, Invitrogen). The same experimental conditions were used to investigate the interaction of the BRCT domains and the phosphorylated peptides CTRL-P, ACC1-P, CTIP-P, BACH1-P, AB-1P (one phosphorylation), and AB-2P (two phosphorylations). The peptides were first diluted in the same buffer as the BRCT domains and then added to the reaction mixture with increasing concentrations up to 256 μmol/L. Reaction mixtures were made in duplicate in a MicroAmp Optical 384-well reaction plate at a final volume of 10 μL (4 μmol/L) and each experiment was repeated at least twice independently. Experiments were carried out in QuantStudio 12KFlex qPCR machine (Applied Biosystems) with a temperature gradient in the range of 15–95°C at 3°C/minute.

Selection of 78 BRCT VUS from the UMD-BRCA1 database

The VUS were selected from the French UMD-BRCA1 database, which gathered in November 2015, 2,020 distinct variants collected from 8,446 families, and included 1,028 distinct VUS of class 3 (http://www.umd.be/BRCA1/; ref. 27). A large majority of these VUS of class 3 were found in a unique family of patients. The exon containing the largest number of variants (exon 18) corresponded to the first BRCT domain (BRCT1; ref. 27). We focused our study on variants located in the C-terminal BRCT1 and BRCT2 domains of BRCA1 (aa 1646–1859). We selected the 65 VUS of class 3 that corresponded to a missense variation in the BRCT domains of BRCA1 (Fig. 1A; Supplementary Table S1). We also selected 6 variants of class 5 (V1665E, R1699W, G1706R, A1708E, S1715N in BRCT1 and M1775R in BRCT2; =causal variants) and 1 variant of class 4 (G1706E in BRCT1; =likely causal variant) as well as 3 variants of class 2 (M1652T in BRCT1, R1751Q in linker, M1783T in BRCT2; =likely neutral variants) and 3 variants of class 1 (M1652I, T1720A in BRCT1, V1804D in BRCT2; =neutral variants), to serve as controls for our experiments (Fig. 1A). The 78 selected mutations are distributed all along the BRCT domain sequence. They are also located on the whole 3D structure: in the BRCT domain hydrophobic cores, at the interface between the two BRCT domains, as well as in solvent-exposed regions and in particular the phosphopeptide-binding region (Fig. 1B). All 78 VUS were evaluated through a dedicated high-throughput workflow combining tests performed in cells (HR assay, localization after DNA damage) and in vitro (protein stability upon expression in bacteria and binding to a set of 5 phosphopeptides) to provide a comprehensive description of their function.

Figure 1.

Distribution of the 78 variations corresponding to the selected VUS. A, Position of the mutations along the BRCA1 sequence. On the top view, the whole BRCA1 sequence is displayed, with its 2 globular domains in red (N-terminal RING domain) and orange (C-terminal BRCT region), and its nuclear export/import signals in black (NES) and green (NLS), respectively. The lower view is a zoom of the BRCA1 BRCT region. It contains 2 BRCT domains indicated in gray. The 78 variations corresponding to the selected VUS are indicated on this zoom. The 7 causal and 6 neutral selected variants are displayed in red and blue, respectively. B, Position of the mutated residues in 3D structure of the BRCT domains (in dark gray) in complex with a phosphorylated BACH1 peptide (in light gray) (PDBcode 1T15). Mutated positions are found throughout the whole structure. Those corresponding to VUS of classes 1 and 2 are colored in blue, classes 4 and 5 in red, and class 3 in yellow. VUS of classes 1, 2, 4, and 5 are labeled.

Figure 1.

Distribution of the 78 variations corresponding to the selected VUS. A, Position of the mutations along the BRCA1 sequence. On the top view, the whole BRCA1 sequence is displayed, with its 2 globular domains in red (N-terminal RING domain) and orange (C-terminal BRCT region), and its nuclear export/import signals in black (NES) and green (NLS), respectively. The lower view is a zoom of the BRCA1 BRCT region. It contains 2 BRCT domains indicated in gray. The 78 variations corresponding to the selected VUS are indicated on this zoom. The 7 causal and 6 neutral selected variants are displayed in red and blue, respectively. B, Position of the mutated residues in 3D structure of the BRCT domains (in dark gray) in complex with a phosphorylated BACH1 peptide (in light gray) (PDBcode 1T15). Mutated positions are found throughout the whole structure. Those corresponding to VUS of classes 1 and 2 are colored in blue, classes 4 and 5 in red, and class 3 in yellow. VUS of classes 1, 2, 4, and 5 are labeled.

Close modal

Impact of the missense mutations on HR

To address this question, we designed a novel cell line RG37-shB1. This cell line was derived from RG37 cells, which are SV40-transformed human fibroblasts containing an integrated DR-GFP substrate to specifically monitor gene conversion upon expression of the meganuclease I-SceI (Fig. 2A; ref. 32). RG37 cells were infected with a retroviral construct containing a doxycycline-induced shRNA raised against the 3′-UTR of the endogenous BRCA1 gene (RG37-shB1cells). Therefore, supplying doxycycline specifically silenced the endogenous BRCA1 without affecting the expression of the exogenous transfected BRCA1 cDNA. Figure 2B shows that supplying doxycycline strongly decreased the expression of BRCA1 and reduced the efficiency of I-SceI–induced HR 2-fold. Cotransfection of a wild-type BRCA1 cDNA with the I-SceI–coding plasmid stimulated the frequency of HR in cells unexposed to doxycycline, as already described with RG37 cells, and, importantly, rescued HR frequency in doxycycline-treated cells (Fig. 2B). Interestingly, all tested BRCA1 variants reported as causal (V1665E, R1699W, G1706R, A1708E, S1715N, M1775R) or likely causal (G1706E) were unable to complement BRCA1 deficiency for HR (Fig. 2C).

Figure 2.

Description of the high-throughput cellular assay set up for measurement of VUS HR efficiency. A, Description of the HR substrate (DR-GFP). Two inactive GFP genes are organized into direct repeats. The 5′ GFP cassette is inactivated because of deletions in both the 5′ and the 3′ sequences. Expression of I-SceI generates a cleavage (DSB) targeted into the substrate. HR between the two GFP genes, with I-SceI, can generate a functional GFP gene through gene conversion without crossing over. Recombinant cells are thus GFP-positive (GFP+) and can be monitored by FACS (54). The DR-GFP substrate is stably integrated into a chromosome of SV40-transformed fibroblasts in the RG37 cell line (25). B, Inducible knockdown of endogeneous BRCA1. One cell line containing a doxycycline-inducible shRNA against the endogenous BRCA1 was derived from the RG37 cell lines (which bear the HR substrate): the RG37 cells have been transduced with a lentivirus coding for a shRNA against the 3′UTR of the endogenous BRCA1 mRNA. This allows to specifically targeting the endogenous BRCA1 and not the exogenous BRCA1 that will be then used. In addition, expression of this shRNA is inducible by the doxycycline. Supplying doxycycline (DOX) induces the expression of the shRNA leading to the silencing of the expression of the endogenous BRCA1 (left) and to reduced HR frequencies, monitored using the DR-GFP (right). This cell line (named RG37shB1) was then used to test all the BRCA1 variants. Because this shRNA does not affect the expression of exogenous BRCA1, expression of exogenous wild-type BRCA1 (WT-BRCA1) is able to stimulate HR frequency, as already shown for BRCA1 overexpression in RG37 cells (55), and to rescue decreased HR frequency in DOX-exposed cells, i.e., silenced for the expression of the endogenous BRCA1. The values correspond to at least 3 independent experiments. C, Impact of BRCA1 bearing causal mutations on the rescue of HR in cells silenced for the endogenous BRCA1. The values are shown normalized to WT-BRCA1 (in black). They correspond to four to eight independent experiments.

Figure 2.

Description of the high-throughput cellular assay set up for measurement of VUS HR efficiency. A, Description of the HR substrate (DR-GFP). Two inactive GFP genes are organized into direct repeats. The 5′ GFP cassette is inactivated because of deletions in both the 5′ and the 3′ sequences. Expression of I-SceI generates a cleavage (DSB) targeted into the substrate. HR between the two GFP genes, with I-SceI, can generate a functional GFP gene through gene conversion without crossing over. Recombinant cells are thus GFP-positive (GFP+) and can be monitored by FACS (54). The DR-GFP substrate is stably integrated into a chromosome of SV40-transformed fibroblasts in the RG37 cell line (25). B, Inducible knockdown of endogeneous BRCA1. One cell line containing a doxycycline-inducible shRNA against the endogenous BRCA1 was derived from the RG37 cell lines (which bear the HR substrate): the RG37 cells have been transduced with a lentivirus coding for a shRNA against the 3′UTR of the endogenous BRCA1 mRNA. This allows to specifically targeting the endogenous BRCA1 and not the exogenous BRCA1 that will be then used. In addition, expression of this shRNA is inducible by the doxycycline. Supplying doxycycline (DOX) induces the expression of the shRNA leading to the silencing of the expression of the endogenous BRCA1 (left) and to reduced HR frequencies, monitored using the DR-GFP (right). This cell line (named RG37shB1) was then used to test all the BRCA1 variants. Because this shRNA does not affect the expression of exogenous BRCA1, expression of exogenous wild-type BRCA1 (WT-BRCA1) is able to stimulate HR frequency, as already shown for BRCA1 overexpression in RG37 cells (55), and to rescue decreased HR frequency in DOX-exposed cells, i.e., silenced for the expression of the endogenous BRCA1. The values correspond to at least 3 independent experiments. C, Impact of BRCA1 bearing causal mutations on the rescue of HR in cells silenced for the endogenous BRCA1. The values are shown normalized to WT-BRCA1 (in black). They correspond to four to eight independent experiments.

Close modal

Using the same assay, we measured the capacity of the 78 missense variants (including the 7 causal mutations shown above) to complement BRCA1 silencing for I-SceI–induced HR. Either WT BRCA1 or the VUS were expressed in this system, as checked by Western blot analysis using anti-BRCA1 antibodies (Supplementary Fig. S1A). Variation of HR efficiency as a function of the position of the variation in the sequence did not reveal any functionally critical hotspot (Supplementary Fig. S1B). Expression of variants reported as neutral (M1652Ib due to G>A, T1720A, V1804D) or likely neutral (M1652T, R1751Q, M1783T) systematically complemented HR efficiency at least as much as WT-BRCA1 (Fig. 3A; Table 1). VUS M1652Ia (due to G>T; class 3) was tested, which also affects position 1652: it consistently showed a WT HR efficiency. On the basis of this unique assay, we could already confirm the link between causality and HR defect with a sensitivity of 100% (7/7 causal variants show an HR defect) and a specificity of 100% (6/6 neutral variants have no HR defect). This first analysis confirmed the essential role of HR defect in tumorigenesis.

Figure 3.

Impact of the BRCA1 missense mutations on HR. A, Plot of the HR efficiencies measured after expression of either WT BRCA1 or VUS normalized to the HR efficiency after expression of WT BRCA1. Statistical significance was calculated using a one-side paired Student t test. To account for the multiple testing, the P values were adjusted using the Benjamini–Hochberg method at a level α = 0.05. HR efficiencies significantly different from the WT value are marked by asterisks. They are indicated by * if P < 0.05 and ** if P < 0.01. All 7 (likely) causal (noted LC or C) variants (in red) cause a significantly decreased HR efficiency, whereas all 6 (likely) neutral (noted LN or N) variants (in blue) cause no significant HR difference. Twenty-one variants could not provide WT HR efficiency. B, Identification of the VUS defective for nuclear localization after addition of MMC, plotted as a function of increasing HR efficiencies as in A. For observing VUS nuclear localization defects, plates were examined after addition of mitomycin to RG37 cells transfected with VUS plasmids. Cells were immunostained with an anti-BRCA1 antibody followed by an Alexa Fluor 488–conjugated secondary antibody and the nucleus was stained with DAPI. BRCA1 localization was visualized by fluorescence microscopy. Statistical significance was calculated using a two-tailed Student t test with GraphPad. This analysis revealed that 34 VUS were not correctly addressed to the nucleus. The bars corresponding to these VUS are colored in green. A more quantitative report of the percentage of nuclear fluorescence measured for each VUS can be found in Supplementary Table S3. As an illustration, in the top part of B, images from control experiments showing that BRCA1 WT is correctly localized in the nucleus after mitomycin treatment (lane 1) as well as images revealing the localization of a neutral (M1652T; lane 2), and a causal (V1665E; lane 3) VUS are displayed.

Figure 3.

Impact of the BRCA1 missense mutations on HR. A, Plot of the HR efficiencies measured after expression of either WT BRCA1 or VUS normalized to the HR efficiency after expression of WT BRCA1. Statistical significance was calculated using a one-side paired Student t test. To account for the multiple testing, the P values were adjusted using the Benjamini–Hochberg method at a level α = 0.05. HR efficiencies significantly different from the WT value are marked by asterisks. They are indicated by * if P < 0.05 and ** if P < 0.01. All 7 (likely) causal (noted LC or C) variants (in red) cause a significantly decreased HR efficiency, whereas all 6 (likely) neutral (noted LN or N) variants (in blue) cause no significant HR difference. Twenty-one variants could not provide WT HR efficiency. B, Identification of the VUS defective for nuclear localization after addition of MMC, plotted as a function of increasing HR efficiencies as in A. For observing VUS nuclear localization defects, plates were examined after addition of mitomycin to RG37 cells transfected with VUS plasmids. Cells were immunostained with an anti-BRCA1 antibody followed by an Alexa Fluor 488–conjugated secondary antibody and the nucleus was stained with DAPI. BRCA1 localization was visualized by fluorescence microscopy. Statistical significance was calculated using a two-tailed Student t test with GraphPad. This analysis revealed that 34 VUS were not correctly addressed to the nucleus. The bars corresponding to these VUS are colored in green. A more quantitative report of the percentage of nuclear fluorescence measured for each VUS can be found in Supplementary Table S3. As an illustration, in the top part of B, images from control experiments showing that BRCA1 WT is correctly localized in the nucleus after mitomycin treatment (lane 1) as well as images revealing the localization of a neutral (M1652T; lane 2), and a causal (V1665E; lane 3) VUS are displayed.

Close modal
Table 1A.

Summary of the experimental results presented in this study

Summary of the experimental results presented in this study
Summary of the experimental results presented in this study
Table 1B.

Classification of the variants

Classification of the variants
Classification of the variants

Global analysis of the HR assay results performed on the 78 variants revealed that 31 VUS (40%) failed to complement BRCA1 deficiency for HR (Fig. 3A). These VUS include the 7 mutations of class 4 or 5, 24 VUS of class 3, and no VUS of class 1 or 2. Positions that are mutated in defective VUS are distributed all along the BRCA1 BRCT domains, from aa 1655 to aa 1837. However, 19 among the 31 defective VUS are mutated within BRCT1. Moreover, most mutated residues corresponding to defective VUS are buried in the 3D structure of the BRCT domains (25 within 31 are less than 20% solvent accessible), whereas only half of the positions mutated in variants characterized by a normal HR are buried in the hydrophobic cores of the BRCT domains or at the BRCT1/BRCT2 interface. This analysis shows that if a position is either in BRCT1 or buried, its mutation is more likely to negatively affect HR.

At given positions, several mutations were identified, corresponding to different VUS. For example, position 1665 is mutated in 3 VUS that showed either no defect (V1665M, V1665L) or a significant defect (V1665E, causal) in HR efficiency, suggesting that any hydrophobic residue is tolerated at this buried position, but the introduction of a charged amino acid causes a HR defect. At position 1699, which is solvent-exposed and is responsible for contacts with the phosphopeptide main chain (Fig. 5B), variation to Q had the same functional impact as the causal mutation to W. Impact of these mutations on BRCA1 structure and phosphorylated BACH1 binding are consistent with the analysis of Coquelle and colleagues (21), who showed that the R1699W mutation destabilized the BRCA1 protein structure and likely interfered with the docking of the peptide phenylalanine into the BRCT peptide–binding groove, whereas the R1699Q mutation caused a steric clash between the glutamine side-chain carbonyl and the main-chain carbonyl of the peptide phenylalanine residue. At position 1706, which is buried, variation to A did not affect HR, whereas the causal mutations to charged residues (G1706E and G1706R) did. Similarly, at position 1708, which is also buried, variation to V did not affect HR, whereas the causal mutation to R or E did. Such analysis underlines the consistency of our HR assay results and highlights the variability of the functional consequences of variations at a same position as a function of the physico-chemical properties of the WT versus variant amino acid. Clearly, at buried positions, mutation to a hydrophobic residue is tolerated but mutation to a charged residue impairs HR efficiency. At solvent-exposed positions, if the mutated residue is involved in the recognition of a functionally essential partner, then any type of replacement might impair recognition and increase cancer risk.

Impact of the missense mutations on BRCA1 nuclear localization after DNA damage in mammalian cells

A first explanation for the deleterious effect of a mutation on HR is the overall low concentration and/or the decreased nuclear localization after DNA damage of the corresponding variant. As no significant difference in protein amounts was observed throughout the VUS by Western blot analysis, we further revealed by immunofluorescence BRCA1 localization in RG37-shB1 cells after addition of mitomycin C. This molecule is a chemotherapeutic drug that upon reduction is converted into a highly reactive intermediate alkylating DNA. We observed that WT BRCA1 as well as a large set of VUS correctly localized at the nucleus after addition of this drug. However, 34 variants were not correctly localized in response to mitomycin C (Table 1; Fig. 3B and C). These variants include 4 causal variants (all except R1699W and A1708E), the likely causal G1706E, and 29 VUS of class 3. Within these 29 variants, 18 are defective for HR. The remaining 11 VUS are capable of compensating their localization defect to carry out normal DNA repair by HR. They might still show other functional defects. The 6 neutral or likely neutral VUS showed no localization defect. Finally, 6 VUS of class 3 showed HR defects but no localization defects, demonstrating that localization defect is not the only mechanism leading to impaired HR. In summary, this second assay enabled classification of 5 of 7 causal variants (sensitivity = 71.4%) and 6 of 6 neutral variants (specificity = 100%), but only 18 VUS of class 3 are identified as both HR-defective (within 24 VUS) and localization defective (within 29 VUS), suggesting that the two assays provide only partially overlapping conclusions.

Impact of the missense mutations on solubility in E. coli and thermostability of recombinant BRCT domains

Another explanation for the loss of HR activity of 40% of the VUS is the impact of BRCA1 variations on the BRCT domain thermostability leading to a global loss of function of these domains. We measured the structural impact of the 78 missense variations by expressing the mutated domains, purifying the soluble domains and measuring the thermostability of the resulting mutated BRCT domains by fluorescence-based thermal shift assay (FBTSA).

First, 23 mutated proteins were insoluble in E. coli (Table 1; Fig. 4A; Supplementary Fig. S2). We classified these proteins as highly unstable, and no further work was carried out on them. They corresponded to 4 variants of class 5, 1 variant of class 4, 17 VUS of class 3, and 1 variant of class 2. Interestingly, 21 of 23 (91%) of these VUS also showed significant HR defects. Only two of these VUS shows a normal HR function: M1652T that is classified as likely neutral and A1752E that belongs to class 3. Thus, a very large majority of highly unstable BRCT domains correspond to VUS with a defective HR activity. Reciprocally, only 4 VUS with impaired HR activities (13% of the HR-defective VUS) correspond to domains that could be purified in vitro: R1699W, R1699Q, M1775R, and V1833M. These results revealed a nice correlation between the HR activity of the whole BRCA1 VUS and the solubility of the mutated BRCT fragments in our conditions.

Figure 4.

Representation of the solubility in bacteria of the recombinant mutated BRCT domains. A, Bars of the HR plot of Fig. 3A are colored as a function of the solubility in E. coli of the corresponding mutated BRCT domains. Brown bars mark mutants that are completely insoluble as GST fusion proteins in E. coli. Yellow bars indicate mutants that are partially soluble in bacteria but aggregate during removal of the GST tag and purification. Only mutants corresponding to gray and black bars could be further characterized in vitro by fluorescence. The difference between these two last classes is based on their expression and purification yields, gray meaning low production yield and black WT-like production yield. B, 3D representation of the position of the residues mutated in VUS characterized by both (i) defective HR capacities and (ii) recombinant BRCT domains that could not be produced and purified from bacteria (in brown and yellow/green as in A).

Figure 4.

Representation of the solubility in bacteria of the recombinant mutated BRCT domains. A, Bars of the HR plot of Fig. 3A are colored as a function of the solubility in E. coli of the corresponding mutated BRCT domains. Brown bars mark mutants that are completely insoluble as GST fusion proteins in E. coli. Yellow bars indicate mutants that are partially soluble in bacteria but aggregate during removal of the GST tag and purification. Only mutants corresponding to gray and black bars could be further characterized in vitro by fluorescence. The difference between these two last classes is based on their expression and purification yields, gray meaning low production yield and black WT-like production yield. B, 3D representation of the position of the residues mutated in VUS characterized by both (i) defective HR capacities and (ii) recombinant BRCT domains that could not be produced and purified from bacteria (in brown and yellow/green as in A).

Close modal

When going further into this in vitro analysis, 13 poorly soluble mutated BRCT domains could not be purified because of aggregation, 26 could be expressed and purified but with low yields, and only 16 were obtained with a yield close to that of the WT-BRCT domains (Table 1; Fig. 4A). This classification seems to correlate with HR results, because 46% of the aggregating mutants, but only 12% of the poorly soluble mutants and 6% of the WT-like mutants, are HR-defective.

Finally, we measured the thermostability of the 42 purified mutated domains using a high-throughput fluorescence assay by FBTSA (Supplementary Fig. S3). The melting temperature measured for the WT is 52°C ± 0.3°C. No clear distinction could be observed between the results obtained for the two causal variants and the two likely neutral variants: causal variants M1775R and R1699W showed melting temperatures of 41.9°C and 43.8 ± 0.1°C, respectively, whereas likely neutral mutants R1751Q and M1783T showed melting temperatures of 42.7°C and 44.9 ± 0.1°C, respectively. However, more generally, the four HR-defective variants for which a melting temperature could be measured are distributed within the 45% variants with the lowest melting temperatures: their melting temperatures range between 40.4°C and 48.8°C. Also, the three neutral variants have significantly higher melting temperatures comprised between 50.5°C and 51.6°C.

In summary, the causality and HR assay results are strongly related to the observation of insolubility versus solubility of the BRCT domains in bacteria. We observed that five of seven causal variants (that are all HR-defective) exhibit BRCT domains insoluble in bacteria (sensitivity = 71.4%), whereas five of six neutral variants (that all show WT HR activity) have BRCT domains that are soluble in bacteria (specificity = 83%). Moreover, among the 24 HR-defective VUS of class 3, 16 correspond to insoluble domains, 6 to domains that aggregate during purification, 1 to a poorly soluble domain, and 1 to a WT-like domain and 1 to a WT-like domain. We have represented on the 3D structure of the BRCA1 BRCT domains the localization of the mutated positions corresponding to both HR-deficient VUS and insoluble (brown) or aggregating (yellow) BRCT domains (Fig. 4B). Clearly, these positions are mainly found in BRCT1 and in particular close to the phosphorylated serine of the BRCA1 partner, underlying the necessity for BRCA1 to interact with phosphorylated partners to promote HR. Fragment from residue 1685 to residue 1708, comprising a β-strand, a loop, and a α-helix interacting with the consensus motif Ser-X-X-Phe of BRCA1-phosphorylated partners (further named the phosphopeptide binding loop), concentrates 13 within the 27 identified positions colored in brown in Fig. 4B.

Impact of the missense variants on the binding of the BRCT domains to phosphorylated peptides

The BRCA1 BRCT domains bind to phosphoproteins such as Abraxas, ACC1, BACH1, and CtiP, which share a common Ser-X-X-Phe motif phosphorylated on the serine residue (17–19). Because several mutations leading to an increased cancer risk disrupt the binding surface of the BRCT domains to phosphorylated peptides, it was suggested that BRCA1 phosphopeptide binding is essential for BRCA1′s tumor-suppressing function (26, 36). Moreover, a previous study of mice carrying a BRCT mutant of BRCA1 that is defective in recognition of phosphorylated proteins also suggested that BRCT phosphoprotein recognition is required for BRCA1 tumor suppression (14). Using the same high-throughput fluorescence assay as for the thermostability measurements, revealing thermostability shifts due to peptide binding here, we tested binding of the 42 purified mutated BRCT domains to five different phosphopeptides. These are fragments of the DNA repair protein Abraxas (belonging to the so-called BRCA1-A complex), the acetyl-CoA carboxylase 1 (ACC1), the DNA helicase BACH1 (belonging to the so-called BRCA1-B complex), and the transcriptional corepressor CtiP (belonging to the so-called BRCA1-C complex). ACC1 is an enzyme essential for cancer cell survival that catalyzes fatty acid biosynthesis; phosphopeptide (ACC1-P) recognition is important for the regulation of fatty acid biosynthesis (37). BACH1 phosphopeptide (BACH1-P) recognition is involved in G2–M checkpoint control and DSB repair and CtiP phosphopeptide (CTIP-P) recognition is essential for DNA end resection of DSBs during HR. Abraxas and the BRCA1-A complex recruit BRCA1 to DNA double-strand break sites (DSB) through an ATM-dependent ubiquitin-mediated signaling pathway. The interaction of phosphorylated Abraxas (mono-phosphorylated peptide AB-1P and di-phosphorylated peptide AB-2P) with BRCA1 is critical for the function of Abraxas in DNA repair of DSBs and maintenance of genomic stability (23).

First, to confirm the binding capacity of the purified WT BRCT domains, we measured their affinity for a set of phosphorylated peptides using isothermal titration calorimetry (ITC). We measured the affinities of the interactions between the BRCT domains and either ACC1-P or BACH1-P. We obtained affinities of 2.1 ± 0.2 μmol/L and 0.19 ± 0.01 μmol/L, respectively, consistently with the literature (Supplementary Fig. S4A and S4B; refs. 37, 38). We also measured the affinities of the interactions between the BRCT domains and either AB-1P or AB-2P. We obtained affinities of 25 ± 2 nmol/L and 6.2 ± 1.2 nmol/L, respectively (Supplementary Fig. S4C and S4D). These last interactions are characterized by a large affinity increase compared with the interactions with ACC1-P and BACH1-P. Finally, we ran size-exclusion chromatography experiments on these complexes. The BRCA1 and BRCA1+BACH1-P elution volumes correspond to the molecular mass of a monomer, whereas that of BRCA1+AB-1P is indicative of a monomer–dimer equilibrium and that of BRCA1+AB-2P reveals a dimer, which is also consistent with the literature (Supplementary Fig. S5A and S5B; ref. 23).

Second, we verified using our FBTSA assay that binding of the WT BRCA1 BRCT domains to ACC1-P, BACH1-P, CtiP-P, AB-1P, and AB-2P induced a measurable increase in the BRCT thermostability that rose with peptide concentration (Supplementary Fig. S6A). Indeed, addition of ACC1-P and CtiP-P caused a maximal stability increase of 5.9°C and 6.2 ± 0.1°C at 256 μmol/L, respectively, whereas addition of the same amount of BACH1-P, AB-1P, and AB-2P increased the WT BRCT thermostability by 8.2°C, 14.1°C, and 15.1 ± 0.1°C, respectively. Control experiments showed that addition of a peptide with the same sequence as CtiP, but phosphorylated on the other serine residue (CTRL-P; see Fig. 5A) increased the thermostability by only 0.1°C.

Figure 5.

Evaluation of the impact of missense mutations on the BRCT phosphopeptide-binding capacities. A, Thermal shifts due to the addition of phosphopeptides at 256 μmol/L onto the WT and mutated BRCT domains. Mutations are ordered as a function of the mutated residue number. Likely neutral and neutral mutations are marked by blue LN and N letters, respectively, whereas causal mutations are marked by a red C. Black, brown, gray and yellow, light orange and dark orange bars correspond to the addition of a control peptide (CTRL-P), ACC1-P, CtiP-P, BACH1-P, AB-1P, and AB-2P, respectively. Bars boxed in green correspond to a reduction of the thermal shifts by at least a factor 2 (dark and light green when residual binding or no binding was observed, respectively; see B). The bars boxed in yellow/green mark a mutation that enhanced binding to CtiP-P, ACC1-P, AB-1P, and AB-2P but decreased binding to BACH1-P. B, 3D representation of the positions corresponding to the mutations identified by boxes in A. The BACH1-P peptide is displayed in red. Positions in dark and light green are in contact or far from the phosphopeptide-binding site, respectively. Position in yellow (1836) interacts with Lys995 from BACH1-P. As the corresponding residue is specific to BACH1 (see A), mutation of Glu1836 has different impacts on the binding of BRCA1 to BACH1-P compared with ACC1-P, CtiP-P, AB-1P, and AB-2P.

Figure 5.

Evaluation of the impact of missense mutations on the BRCT phosphopeptide-binding capacities. A, Thermal shifts due to the addition of phosphopeptides at 256 μmol/L onto the WT and mutated BRCT domains. Mutations are ordered as a function of the mutated residue number. Likely neutral and neutral mutations are marked by blue LN and N letters, respectively, whereas causal mutations are marked by a red C. Black, brown, gray and yellow, light orange and dark orange bars correspond to the addition of a control peptide (CTRL-P), ACC1-P, CtiP-P, BACH1-P, AB-1P, and AB-2P, respectively. Bars boxed in green correspond to a reduction of the thermal shifts by at least a factor 2 (dark and light green when residual binding or no binding was observed, respectively; see B). The bars boxed in yellow/green mark a mutation that enhanced binding to CtiP-P, ACC1-P, AB-1P, and AB-2P but decreased binding to BACH1-P. B, 3D representation of the positions corresponding to the mutations identified by boxes in A. The BACH1-P peptide is displayed in red. Positions in dark and light green are in contact or far from the phosphopeptide-binding site, respectively. Position in yellow (1836) interacts with Lys995 from BACH1-P. As the corresponding residue is specific to BACH1 (see A), mutation of Glu1836 has different impacts on the binding of BRCA1 to BACH1-P compared with ACC1-P, CtiP-P, AB-1P, and AB-2P.

Close modal

From this description of the WT BRCT domain–binding properties, it was possible to design a large high-throughput experiment aiming at identifying the VUS defective in phosphopeptides binding. Figure 5A shows the thermostability shifts caused by binding of the peptides CTRL-P, ACC1-P, BACH1-P, CtiP-P, AB-1P, and AB-2P to the BRCT domains of WT BRCA1 and the 42 mutants that could be expressed and purified. Clearly, addition of either ACC1-P, CtiP-P, BACH1-P, AB-1P, or AB-2P generally have the same impact on the thermostability of the mutated BRCT domains. This confirms that the five peptides bind to the same BRCA1 site. Five variants significantly lost affinity for the peptides: R1699W and M1775R (Supplementary Fig. S6B) are the two only causal variants that could be purified; mutants R1699Q, F1717Y and H1746N correspond to VUS of class 3. In the case of mutants F1717Y and H1746N, no binding of the peptides to the BRCT domains could be detected (Fig. 5A). Representation of the 4 positions mutated in these 5 variants on the 3D structure of the BRCT domains revealed that Arg1699 and Met1775 are directly involved in peptide binding, whereas Phe1717 and His1746 are further from the binding site and probably indirectly contribute to peptide recognition (Fig. 5B; ref. 20). These residues might be critical for correctly positioning the phosphopeptide binding loop (amino acids 1685 to 1708), as they interact with Lys1690 and Tyr1703, respectively. Altogether, the variants of classes 4 and 5 are all HR-deficient and when tested are phosphopeptide-binding deficient, whereas the variants of classes 1 and 2 have a WT HR activity and when tested show a WT phosphopeptide-binding capacity. By taking into account the results of the cellular HR assay and the in vitro binding assay, it is possible to classify 7 of 7 causal variants (sensibility = 100%) and 6 of 6 neutral variants (specificity = 100%).

In this study, we tested 78 BRCA1 variants using 4 assays that focus on different properties previously shown as related to BRCA1 oncogenic function. Our challenge was to set up a high-throughput protocol to (i) measure BRCA1 HR capacity in 6-well plates through a fluorescence reading, (ii) test bacterial expression of the BRCT domains in 96-well plates and read the stability and phosphopeptides binding of the domains using fluorescence-based thermal shift assays. Thus, it was possible to provide results for the 4 assays on a very large set of VUS from the UMD-BRCA1 database, corresponding to all VUS that are mutated in the BRCT domains (Table 1). We will now compare these results to the published data available on other VUS databases (Table 2). We will discuss the relationship existing between the measured properties of our VUS and predisposition to cancer and suggest related mechanisms of tumorigenesis.

Table 2.

Summary of the structural and functional data available on the 78 VUS, based on this study as well as on previously published studies

Summary of the structural and functional data available on the 78 VUS, based on this study as well as on previously published studies
Summary of the structural and functional data available on the 78 VUS, based on this study as well as on previously published studies

Comparison with the results obtained on the VUS from BIC, KConfab, and ClinVar databases

Over the past few years, several functional assays for classification of BRCA1 VUS have been developed. In particular, a fast cDNA-based functional assay was reported, based on the VUS ability to functionally complement BRCA1-deficient mouse embryonic stem cells, that is, restore the proliferation defect due to the absence of BRCA1 expression and survive to addition of cisplatin (39). Seventy-four VUS distributed all along the BRCA1 gene were tested using this assay. This analysis strongly suggested that causal missense variants are confined to BRCA1 RING and BRCT domains. Several other groups restricted their analyses to BRCA1 proteins mutated in the evolutionarily conserved RING or BRCT domains. They often aimed at understanding the impact of mutations on protein stability, phosphopeptide binding, and transcriptional activation (21, 23–25, 40–43). They suggested that phosphopeptide binding and transcription assays gave results that were correlated to cancer risk, with specificity and sensitivity higher than 80%. Additional studies reported the activity of a subset of VUS in HR, which was also closely associated with cancer risk (39, 44, 45). Finally, characterization of 12 BRCT domain variants performed both in vitro and in cells showed that all of the variants, regardless of how profound their destabilizing effects are in vitro, retained some DNA repair activity and thereby partially rescued the chicken BRCA1 knockout (26). In contrast, the variant R1699L, which disrupts the binding to phosphorylated proteins (but which is not destabilizing), was completely inactive. These studies raised the question of the link between protein stability, phosphopeptide binding, HR activity, and cancer risk.

We have compared our results to the data available from these studies (Supplementary Table SI). Our VUS list contains 42 VUS also present in the BIC, KConFab, or ClinVar databases, which were characterized by the teams of L.S. Itzhaki, J.N.M. Glover, J. Jonkers, and A.N. Monteiro (Table 2). In particular, the thermodynamic stability of 17 mutated BRCT domains and the effects of four of the corresponding variants on BRCA1-mediated DNA repair by HR were measured by Itzhaki's team and our team (25, 26). The resulting data are independent measurements obtained on a subset of VUS that can be used to evaluate the robustness of the approaches. The seven mutants described as insoluble by Itzhaki's team are mostly insoluble or aggregating during purification in our experimental set up. The 10 remaining mutants are always soluble and could be purified in both teams. Within the four VUS studied in cells by both teams, one is defective for HR in both assays (A1708E, causal), 2 VUS have a WT DNA repair activity in both assays (M1783T, likely neutral, and S1841N), and 1 VUS shows a moderate DNA repair activity in Itzhaki's assay and a WT in our assay (M1652Ib). This last variant, M1652Ib, is already classified as neutral. We can conclude that the results obtained by the two teams are in general consistent.

A large-scale analysis of the sensitivity to proteases, phosphorylated peptide binding, and in cell transcriptional activity of 117 VUS corresponding to mutations in the BRCT domains of BRCA1 was performed by Glover's team (21, 23). Most of the in vitro analysis was performed on BRCT domains produced by in vitro transcription/translation, which is an approach complementary to our study. Comparison of all these results provides essential arguments to guide further interpretation of our results (Table 2). Thirty-nine VUS were studied by both Glover's team and our team. Within these 39 mutants, 20 could be purified by us. Eighteen were resistant to degradation by proteases in Glover's team. The VUS H1746N and V1833M were sensitive to proteases in Lee and colleagues (23), and are significantly less stable (by more than 5°C) than the WT BRCT domains in our study. In the case of the 19 mutants that we could not be obtained using a bacterial expression system, 17 were sensitive to proteases, which confirm that these mutants were unstable. Only VUS V1713A and M1652T were not sensitive to proteases in Lee and colleagues (23). In summary, our results on solubility in bacteria and results of Lee and colleagues on protease sensitivity are mostly consistent; however, because of the different experimental set up used to produce the proteins, some differences are observed for a few VUS (23). Furthermore, Glover's team could measure the binding of 19 variants insoluble in bacteria to a phosphorylated BACH1 peptide. They revealed that 17 of these variants have a defective phosphorylated peptide–binding capacity (only mutants M1652T and V1833M bind to phospho-BACH1). This analysis shows that mutated BRCT domains insoluble in bacteria, when produced in other systems, are in general also defective in phosphopeptide recognition.

Relationship between phosphopeptide-binding defects and high cancer risk

From our data, it is particularly clear that the results from the phosphopeptide-binding assays are highly correlated with cancer risk. These assays generally gave similar results using any of the 5 tested peptides CTiP-P, BACH1-P, ACC1-P, AB-1P, and AB-2P and provided information on 42 mutants including five neutral variants that showed WT BRCT–binding properties and two causal variants that exhibited defective BRCT phosphopeptide-binding capacities. One VUS had a different behavior when tested for binding against the five phosphopeptides: E1836K showed improved binding to CtiP-P, ACC1-P, AB-1P, and AB-2P but reduced binding to BACH1-P (very close to that of the causal R1699W; Supplementary Fig. S6C; ref. 46). It had a normal HR activity and was correctly localized after addition of MMC. From our data, it is still unclear whether reduced binding to only BACH1-P is sufficient to significantly increase the cancer risk.

We could not test the binding properties of 36 mutated BRCT domains because there were insoluble in bacteria. Comparison of Glover's team results with our results suggests that most (89%) mutated BRCT domains insoluble in bacteria in our study are also defective in phosphopeptide binding. In particular, by producing proteins using in vitro transcription/translation, Glover's team was able to characterize the phosphopeptide-binding capacities of 5 of our 7 causal mutants. R1699W and M1775R (consistently with our study) as well as G1706E show large binding defects. A1708E and S1715N show smaller but still significant binding defects. Glover's team also tested the binding properties of our insoluble BRCT mutant corresponding to a likely neutral VUS (M1652T) and showed that this mutant binds phosphorylated BACH1 as the WT BRCT domains. This analysis strongly suggests that all causal mutations lead to binding defects and all neutral mutations do not impact BRCT phosphopeptide–binding properties. It was recently published from the study of mice carrying a BRCT mutant of BRCA1 that is defective in recognition of phosphorylated proteins that BRCT phosphoprotein recognition is required for BRCA1 tumor suppression (14).

It was also proposed that a significant defect in phosphopeptide binding could completely abolish HR (26). From Glover's team results and our data, we do not observe a complete correlation between the results of the binding and HR assays. R1699W (causal), R1699Q, and M1775R (causal) both show clear binding and HR defects (see consistently for M1775R reports of teams of S. Pellegrini and J.D. Parvin, and inconsistently report of the team of M.A. Caligo; refs. 23, 44, 47). However, E1836K is only defective for BACH1-P binding and exhibits a WT HR efficiency. More strikingly, F1717Y and H1746N show strong binding defects to all 5 phosphopeptides together with a WT HR activity. The functional consequences of these binding defects and their putative contribution to tumorigenesis are unclear. On the basis of the systematic binding defects measured for causal variants, impaired phosphopeptide binding might be sufficient per se to be associated with increased cancer risk.

Classification of variants

Most of the mutations responsible for familial breast or ovarian cancers affect genes that control HR and/or the replication/HR interface directly or indirectly (48, 49). The two most often mutated genes, BRCA1 and BRCA2, are major players in HR (50, 51). This overrepresentation of genes involved in the response to DNA damage and the communication between replication and recombination highlights the importance of these specific pathways in the etiology of breast cancer. We here present one of the first high-throughput structural and functional analyses of a large set of BRCA1 BRCTs VUS, which provides information on the cellular localization and DNA repair capacity of all variants, as well as the thermal stability and binding properties of a subset of variants. Analysis of these results together with the data published in the literature on overlapping subsets of variants enabled to correlate for the first time on a large-scale protein stability, phosphopeptide-binding, HR activity and cancer risk. From this analysis, three measurements seem essential to predict cancer risk: (i) BRCT phosphopeptide-binding affinities; (ii) BRCT capacity to be produced and purified from bacteria; and (iii) VUS HR activity. Combining these three measurements as well as results from the literature, we sorted the 65 BRCA1 VUS of class 3 into 3 groups (Table 2).

Thirty-three VUS show no HR and phosphopeptide-binding defect in our assays; they are proposed to have no impact on cancer risk (1P). They include all variants of classes 1–2 but one (M1652T, previously classified as likely neutral) and 28 VUS of class 3 (43% of this class). We cannot further classify the likely neutral VUS M1652T because its BRCT domains were insoluble in E. coli in our assays. Glover and colleagues reported that this variant binds to BACH1-P (23); however, its binding to other phosphopeptides has not been tested. Supporting our classification of 28 VUS of class 3 as 1P, it was recently shown from genetic data that G1706A and L1844P can be assigned to neutral (personal communication).

Thirty VUS exhibit several defects including systematic HR defect and either BRCT insolubility in bacteria or phosphopeptide-binding defect, they are proposed to have a severe impact on cancer risk (3P). They include all variants of classes 4–5 as well as 23 VUS of class 3 (35% of this class). In one case (S1655A), the VUS also showed a decreased binding to a library of phosphopeptides containing the Ser-X-X-Phe motif (46). Moreover, the corresponding mouse mutation caused a hypersensitivity to genotoxic stress and a defective HR DNA repair (14). Further supporting our classification, L1764P, which is HR-deficient and shows insoluble BRCT domains in bacteria, was classified as causal based on genetic data during our study (11). C1697R is also HR-deficient and exhibits insoluble BRCT domains in bacteria. We could recently calculate a causality score for C1697R from the French and literature data as described by D. Goldgar, and we classified this variant as causal (Supplementary Table SII; refs. 11, 53).

Only 15 VUS are identified as having an intermediate impact (2P). One of them (M1652T) was already classified as likely neutral and cannot be yet further classified, as explained above. For 3 other VUS of class 3, no HR defect was observed, and either we did not find the same binding defect as a function of the tested phosphopeptide (E1836K), or we measured a WT binding for the 5 phosphopeptides but a BACH1-P binding defect was reported by others (A1708V, A1789S; ref. 23). Here again, complementary experiments are needed to confirm that these VUS are neutral. In the case of the 11 other VUS, strong defects were observed by us, which suggests that these VUS could be causal. V1833M shows an HR defect, even if no binding defect could be measured by us and others (23). The mechanism leading to V1833M HR defect is yet unknown. G1770V, S1722Y, E1735K, and A1752E have insoluble BRCT domains but show no HR defect. G1770V was recently classified as causal based on genetic data (personal communication). D1739G, D1739V, W1837G, and S1841N have insoluble BRCT domains that were reported by others as defective in binding BACH1-P (23) but also show no HR defect. Finally, H1746N and F1717Y have soluble BRCT domains that poorly bind to the 5 tested phosphopeptides but show no HR defect. We propose that such VUS with a strong binding defect are likely causal and should systematically be monitored because of their association with high cancer risk.

Conclusion

Our analysis strongly suggests that VUS exhibiting WT HR efficiency as well as WT phosphopeptide-binding properties are neutral. Also, HR-defective VUS are generally causal. Association of HR deficiency with a binding defect is strongly associated to causality. Moreover, our study reveals that VUS with BRCT domains that are insoluble in bacteria are often causal. Their phosphopeptide-binding properties should systematically be characterized, and if a defect is identified, they should similarly be considered as increasing cancer risk. In particular, we identified two new variants that show a strong binding defect without being HR deficient and our analysis stresses that these variants are likely causal. From these new guidelines, we believe that it is possible to improve the clinical interpretation of BRCA1 variants, and select a larger group of patients for treatment with therapies based on platinum agents or PARP inhibitors that are specifically efficient against BRCA1/2-related tumors.

E. Rouleau has received speakers bureau honoraria from AstraZeneca and BMS and is a consultant/advisory board member for AstraZeneca, BMS, and Roche. No potential conflicts of interest were disclosed by the other authors.

Conception and design: E. Rouleau, B.S. Lopez, S. Zinn-Justin, S.M. Caputo

Development of methodology: A. Petitalot, E. Dardillac, E. Jacquet, N. Nhiri, J. Guirouilh-Barbat, I. Bouazzaoui, D. Bonte, B.S. Lopez, S. Zinn-Justin, S.M. Caputo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Petitalot, E. Dardillac, E. Jacquet, P. Julien, I. Bouazzaoui, J.A. Schnell, P. Lafitte, C. Nogues, R. Lidereau, B.S. Lopez, S. Zinn-Justin, S.M. Caputo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Petitalot, E. Dardillac, E. Jacquet, N. Nhiri, J.A. Schnell, J.-C. Aude, B.S. Lopez, S. Zinn-Justin, S.M. Caputo

Writing, review, and/or revision of the manuscript: A. Petitalot, D. Bonte, C. Nogues, E. Rouleau, B.S. Lopez, S. Zinn-Justin, S.M. Caputo

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Dardillac, E. Rouleau, S. Zinn-Justin, S.M. Caputo

Study supervision: S. Zinn-Justin, S.M. Caputo

Other (provided reagents): J. Feunteun

The authors thank the French oncogeneticists, the UNICANCER Genetic Group BRCA network led by Dr. Catherine Noguès and probands for their cooperation. We gratefully acknowledge Sylvie Mazoyer for the BRCA1 full-length plasmid; Carine Tellier and Sylvaine Gasparini for their help during the recombinant plasmid construction, Damarys Loew and Vanessa Masson for their help with mass spectroscopy, and Sophie Demontety for her help with causality scores. This work was supported by a joint translational research grant of the French National Cancer Institute (2011-1-PL BIO-09-IC-1; to A.Petitalot) and the ‘Direction Générale de l'Offre des Soins' (INCa-DGOS: PRTK2011-046 to A. Petitalot, E. Dardillac, P. Julien, I. Bouazzaoui, and P. Lafitte; and PRT-K 14 134, to P. Lafitte) and by the Association d'Aide à la Recherche Cancérologique de Saint-Cloud (ARCS). B.S. Lopez's team is labeled “Ligue 2014.”

UNICANCER Genetic Group BRCA network: Françoise Bonnet, Natalie Jones, Virginie Bubien, Michel Longy, Nicolas Sévenet: Institut Bergonié - Bordeaux; Sophie Krieger, Laurent Castera, Dominique Vaur: Centre François Baclesse - Caen; Nancy Uhrhammer, Yves Jean Bignon: Centre Jean Perrin - Clermont-Ferrand; Sarab Lizard: CHU de Dijon, Hôpital d'Enfants, Service de Génétique Médicale - Dijon; Aurélie Dumont, Françoise Revillion: Centre Oscar Lambret - Lille; Mélanie Léone, Nadia Boutry-Kryza, Olga Sinilnikova (deceased): Hospices Civils de Lyon and Centre Léon Bérard - Lyon; Audrey Remenieras, Violaine Bourdon, Tetsuro Noguchi, Hagay Sobol: Institut Paoli-Calmettes – Marseille, France; Pierre-Olivier Harmand, Paul Vilquin, Pascal Pujol: Laboratoire de Biologie Cellulaire et Hormonale (CHU Arnaud de Villeneuve) - Montpellier; Philippe Jonveaux, Myriam Bronner, Joanna Sokolowska: CHU de Nancy-Brabois - Vandoeuvre-lés-Nancy; Capucine Delnatte, Virginie Guibert, Céline Garrec, Stéphane Bézieau: CHU - Institut de Biologie - Hôtel Dieu - Nantes; Florent Soubrier, Erell Guillerm, Florence Coulet: Groupe hospitalier Pitié-Salpêtrière, Assistance Publique-Hôpitaux de Paris, Université Pierre et Marie Curie, Laboratoire d'Oncogénétique et Angiogénétique moléculaire - Paris; Cédrick Lefol, Virginie Caux-Moncoutier, Lisa Golmard, Claude Houdayer, Dominique Stoppa-Lyonnet: Institut Curie - Paris; Chantal Delvincourt, Olivia Beaudoux: Institut Jean Godinot - Reims; Danièle Muller: Centre Paul Strauss—Strasbourg; Christine Toulas: Institut Claudius Régaud - Toulouse; Marine Guillaud-Bataille, Brigitte Bressac-De Paillerets: Institut Gustave Roussy - Villejuif.

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.

1.
Wu
W
,
Koike
A
,
Takeshita
T
,
Ohta
T
. 
The ubiquitin E3 ligase activity of BRCA1 and its biological functions
.
Cell Div
2008
;
3
:
1
.
2.
Plon
SE
,
Eccles
DM
,
Easton
D
,
Foulkes
WD
,
Genuardi
M
,
Greenblatt
MS
, et al
Sequence variant classification and reporting: recommendations for improving the interpretation of cancer susceptibility genetic test results
.
Hum Mutat
2008
;
29
:
1282
91
.
3.
Institut National du Cancer
. 
Oncogénétique en 2014 [Internet]
; 
2016
.
Available from:
www.e-cancer.fr/Professionnels-de-sante/L-organisation-de-l-offre-de-soins/Oncogenetique
4.
Audeh
MW
,
Carmichael
J
,
Penson
RT
,
Friedlander
M
,
Powell
B
,
Bell-McGuinn
KM
, et al
Oral poly(ADP-ribose) polymerase inhibitor olaparib in patients with BRCA1 or BRCA2 mutations and recurrent ovarian cancer: a proof-of-concept trial
.
Lancet
2010
;
376
:
245
51
.
5.
Fong
PC
,
Boss
DS
,
Yap
TA
,
Tutt
A
,
Wu
P
,
Mergui-Roelvink
M
, et al
Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA mutation carriers
.
N Engl J Med
2009
;
361
:
123
34
.
6.
Robson
M
,
Offit
K
. 
Clinical practice. Management of an inherited predisposition to breast cancer
.
N Engl J Med
2007
;
357
:
154
62
.
7.
Goldgar
DE
,
Easton
DF
,
Byrnes
GB
,
Spurdle
AB
,
Iversen
ES
,
Greenblatt
MS
, et al
Genetic evidence and integration of various data sources for classifying uncertain variants into a single model
.
Hum Mutat
2008
;
29
:
1265
72
.
8.
Spurdle
AB
,
Whiley
PJ
,
Thompson
B
,
Feng
B
,
Healey
S
,
Brown
MA
, et al
BRCA1 R1699Q variant displaying ambiguous functional abrogation confers intermediate breast and ovarian cancer risk
.
J Med Genet
2012
;
49
:
525
32
.
9.
Moghadasi
S
,
Meeks
HD
,
Vreeswijk
MP
,
Janssen
LA
,
Borg
Å
,
Ehrencrona
H
, et al
The BRCA1 c. 5096G>A p.Arg1699Gln (R1699Q) intermediate risk variant: breast and ovarian cancer risk estimation and recommendations for clinical management from the ENIGMA consortium
.
J Med Genet
2018
;
55
:
15
20
.
10.
Dos Santos
ES
,
Caputo
SM
,
Castera
L
,
Gendrot
M
,
Briaux
A
,
Breault
M
, et al
Assessment of the functional impact of germline BRCA1/2 variants located in non-coding regions in families with breast and/or ovarian cancer predisposition
.
Breast Cancer Res Treat
2018
;
168
:
311
25
.
11.
Lindor
NM
,
Guidugli
L
,
Wang
X
,
Vallée
MP
,
Monteiro
ANA
,
Tavtigian
S
, et al
A review of a multifactorial probability-based model for classification of BRCA1 and BRCA2 variants of uncertain significance (VUS)
.
Hum Mutat
2012
;
33
:
8
21
.
12.
Jiang
Q
,
Greenberg
RA
. 
Deciphering the BRCA1 tumor suppressor network
.
J Biol Chem
2015
;
290
:
17724
32
.
13.
Savage
KI
,
Harkin
DP
. 
BRCA1, a “complex” protein involved in the maintenance of genomic stability
.
FEBS J
2015
;
282
:
630
46
.
14.
Shakya
R
,
Reid
LJ
,
Reczek
CR
,
Cole
F
,
Egli
D
,
Lin
CS
, et al
BRCA1 tumor suppression depends on BRCT phosphoprotein binding, but not its E3 ligase activity
.
Science
2011
;
334
:
525
8
.
15.
Williams
RS
,
Green
R
,
Glover
JN
. 
Crystal structure of the BRCT repeat region from the breast cancer-associated protein BRCA1
.
Nat Struct Biol
2001
;
8
:
838
42
.
16.
Shiozaki
EN
,
Gu
L
,
Yan
N
,
Shi
Y
. 
Structure of the BRCT repeats of BRCA1 bound to a BACH1 phosphopeptide: implications for signaling
.
Mol Cell
2004
;
14
:
405
12
.
17.
Varma
AK
,
Brown
RS
,
Birrane
G
,
Ladias
JA
. 
Structural basis for cell cycle checkpoint control by the BRCA1-CtIP complex
.
Biochemistry
2005
;
44
:
10941
6
.
18.
Clapperton
JA
,
Manke
IA
,
Lowery
DM
,
Ho
T
,
Haire
LF
,
Yaffe
MB
, et al
Structure and mechanism of BRCA1 BRCT domain recognition of phosphorylated BACH1 with implications for cancer
.
Nat Struct Mol Biol
2004
;
11
:
512
8
.
19.
Wu
Q
,
Paul
A
,
Su
D
,
Mehmood
S
,
Foo
TK
,
Ochi
T
, et al
Structure of BRCA1-BRCT/Abraxas complex reveals phosphorylation-dependent BRCT dimerization at DNA damage sites
.
Mol Cell
2016
;
61
:
434
48
.
20.
Wu
Q
,
Jubb
H
,
Blundell
TL
. 
Phosphopeptide interactions with BRCA1 BRCT domains: more than just a motif
.
Prog Biophys Mol Biol
2015
;
117
:
143
8
.
21.
Coquelle
N
,
Green
R
,
Glover
JN
. 
Impact of BRCA1 BRCT domain missense substitutions on phosphopeptide recognition
.
Biochemistry
2011
;
50
:
4579
89
.
22.
Jhuraney
A
,
Velkova
A
,
Johnson
RC
,
Kessing
B
,
Carvalho
RS
,
Whiley
P
, et al
BRCA1 Circos: a visualisation resource for functional analysis of missense variants
.
J Med Genet
2015
;
52
:
224
30
.
23.
Lee
MS
,
Green
R
,
Marsillac
SM
,
Coquelle
N
,
Williams
RS
,
Yeung
T
, et al
Comprehensive analysis of missense variations in the BRCT domain of BRCA1 by structural and functional assays
.
Cancer Res
2010
;
70
:
4880
90
.
24.
Millot
G
,
Carvalho
MA
,
Caputo
SM
,
Vreeswijk
MP
,
Brown
MA
,
Webb
M
, et al
A guide for functional analysis of BRCA1 variants of uncertain significance (VUS)
.
Hum Mutat
2012
;
33
:
1526
37
.
25.
Rowling
PJ
,
Cook
R
,
Itzhaki
LS
. 
Toward classification of BRCA1 missense variants using a biophysical approach
.
J Biol Chem
2010
;
285
:
20080
7
.
26.
Gaboriau
DC
,
Rowling
PJ
,
Morrison
CG
,
Itzhaki
LS
. 
Protein stability versus function: effects of destabilizing missense mutations on BRCA1 DNA repair activity
.
Biochem J
2015
;
466
:
613
24
.
27.
Caputo
S
,
Benboudjema
L
,
Sinilnikova
O
,
Rouleau
E
,
Béroud
C
,
Lidereau
R
. 
Description and analysis of genetic variants in French hereditary breast and ovarian cancer families recorded in the UMD-BRCA1/BRCA2 databases
.
Nucleic Acids Res
2012
;
40
:
D992
1002
.
28.
Béroud
C
,
Letovsky
SI
,
Braastad
CD
,
Caputo
SM
,
Beaudoux
O
,
Bignon
YJ
, et al
BRCA Share: a collection of clinical BRCA gene variants
.
Hum Mutat
2016
;
37
:
1318
28
.
29.
Gaildrat
P
,
Krieger
S
,
Giacomo
DD
,
Abdat
J
,
Révillion
F
,
Caputo
S
, et al
Multiple sequence variants of BRCA2 exon 7 alter splicing regulation
.
J Med Genet
2012
;
49
:
609
17
.
30.
Rouleau
E
,
Lefol
C
,
Moncoutier
V
,
Castera
L
,
Houdayer
C
,
Caputo
S
, et al
A missense variant within BRCA1 exon 23 causing exon skipping
.
Cancer Genet Cytogenet
2010
;
202
:
144
6
.
31.
Wiederschain
D
,
Wee
S
,
Chen
L
,
Loo
A
,
Yang
G
,
Huang
A
, et al
Single-vector inducible lentiviral RNAi system for oncology target validation
.
Cell Cycle
2009
;
8
:
498
504
.
32.
Dumay
A
,
Laulier
C
,
Bertrand
P
,
Saintigny
Y
,
Lebrun
F
,
Vayssière
JL
, et al
Bax and Bid, two proapoptotic Bcl-2 family members, inhibit homologous recombination, independently of apoptosis regulation
.
Oncogene
2006
;
25
:
3196
205
.
33.
Scully
R
,
Chen
J
,
Plug
A
,
Xiao
Y
,
Weaver
D
,
Feunteun
J
, et al
Association of BRCA1 with Rad51 in mitotic and meiotic cells
.
Cell
1997
;
88
:
265
75
.
34.
R Development Core Team
. 
R: A language and environment for statistical computing
; 
2013
.
Vienna, Austria: the R Foundatin for Stastical Computing.
Available from:
http://www.R-project.org/.
35.
Benjamini
Y
,
Hochberg
Y
. 
Controlling the false discovery rate: a practical and powerful approach to multiple testing
.
J Roy Stat Soc Ser B Methodol
1995
;
57
:
289
300
.
36.
Glover
JN
. 
Insights into the molecular basis of human hereditary breast cancer from studies of the BRCA1 BRCT domain
.
Fam Cancer
2006
;
5
:
89
93
.
37.
Shen
Y
,
Tong
L
. 
Structural evidence for direct interactions between the BRCT domains of human BRCA1 and a phospho-peptide from human ACC1
.
Biochemistry
2008
;
47
:
5767
73
.
38.
Botuyan
MV
,
Nominé
Y
,
Yu
X
,
Juranic
N
,
Macura
S
,
Chen
J
, et al
Structural basis of BACH1 phosphopeptide recognition by BRCA1 tandem BRCT domains
.
Structure
2004
;
12
:
1137
46
.
39.
Bouwman
P
,
van der Gulden
H
,
van der Heijden
I
,
Drost
R
,
Klijn
CN
,
Prasetyanti
P
, et al
A high-throughput functional complementation assay for classification of BRCA1 missense variants
.
Cancer Discov
2013
;
3
:
1142
55
.
40.
Williams
RS
,
Chasman
DI
,
Hau
DD
,
Hui
B
,
Lau
AY
,
Glover
JNM
. 
Detection of protein folding defects caused by BRCA1-BRCT truncation and missense mutations
.
J Biol Chem
2003
;
278
:
53007
16
.
41.
Carvalho
MA
,
Marsillac
SM
,
Karchin
R
,
Manoukian
S
,
Grist
S
,
Swaby
RF
, et al
Determination of cancer risk associated with germ line BRCA1 missense variants by functional analysis
.
Cancer Res
2007
;
67
:
1494
501
.
42.
Drikos
I
,
Nounesis
G
,
Vorgias
CE
. 
Characterization of cancer-linked BRCA1-BRCT missense variants and their interaction with phosphoprotein targets
.
Proteins
2009
;
77
:
464
76
.
43.
Woods
NT
,
Baskin
R
,
Golubeva
V
,
Jhuraney
A
,
De-Gregoriis
G
,
Vaclova
T
, et al
Functional assays provide a robust tool for the clinical annotation of genetic variants of uncertain significance
.
NPJ Genomic Med
2016
;
1
:
16001
.
44.
Towler
WI
,
Zhang
J
,
Ransburgh
DJ
,
Toland
AE
,
Ishioka
C
,
Chiba
N
, et al
Analysis of BRCA1 variants in double-strand break repair by homologous recombination and single-strand annealing
.
Hum Mutat
2013
;
34
:
439
45
.
45.
Chang
S
,
Biswas
K
,
Martin
BK
,
Stauffer
S
,
Sharan
SK
. 
Expression of human BRCA1 variants in mouse ES cells allows functional analysis of BRCA1 mutations
.
J Clin Invest
2009
;
119
:
3160
71
.
46.
Campbell
SJ
,
Edwards
RA
,
Glover
JN
. 
Comparison of the structures and peptide binding specificities of the BRCT domains of MDC1 and BRCA1
.
Structure
2010
;
18
:
167
76
.
47.
Guidugli
L
,
Rugani
C
,
Lombardi
G
,
Aretini
P
,
Galli
A
,
Caligo
MA
. 
A recombination-based method to characterize human BRCA1 missense variants
.
Breast Cancer Res Treat
2011
;
125
:
265
72
.
48.
Walsh
T
,
Casadei
S
,
Lee
MK
,
Pennil
CC
,
Nord
AS
,
Thornton
AM
, et al
Mutations in 12 genes for inherited ovarian, fallopian tube, and peritoneal carcinoma identified by massively parallel sequencing
.
Proc Natl Acad Sci U S A
2011
;
108
:
18032
7
.
49.
Castéra
L
,
Krieger
S
,
Rousselin
A
,
Legros
A
,
Baumann
JJ
,
Bruet
O
, et al
Next-generation sequencing for the diagnosis of hereditary breast and ovarian cancer using genomic capture targeting multiple candidate genes
.
Eur J Hum Genet
2014
;
22
:
1305
13
.
50.
Moynahan
ME
,
Chiu
JW
,
Koller
BH
,
Jasin
M
. 
Brca1 controls homology-directed DNA repair
.
Mol Cell
1999
;
4
:
511
8
.
51.
Moynahan
ME
,
Cui
TY
,
Jasin
M
. 
Homology-directed dna repair, mitomycin-c resistance, and chromosome stability is restored with correction of a Brca1 mutation
.
Cancer Res
2001
;
61
:
4842
50
.
52.
Williams
RS
,
Lee
MS
,
Hau
DD
,
Glover
JN
. 
Structural basis of phosphopeptide recognition by the BRCT domain of BRCA1
.
Nat Struct Mol Biol
2004
;
11
:
519
25
.
53.
Vallon-Christersson
J
,
Cayanan
C
,
Haraldsson
K
,
Loman
N
,
Bergthorsson
JT
,
Brøndum-Nielsen
K
, et al
Functional analysis of BRCA1 C-terminal missense mutations identified in breast and ovarian cancer families
.
Hum Mol Genet
2001
;
10
:
353
60
.
54.
Pierce
AJ
,
Johnson
RD
,
Thompson
LH
,
Jasin
M
. 
XRCC3 promotes homology-directed repair of DNA damage in mammalian cells
.
Genes Dev
1999
;
13
:
2633
8
.
55.
Plo
I
,
Laulier
C
,
Gauthier
L
,
Lebrun
F
,
Calvo
F
,
Lopez
BS
. 
AKT1 inhibits homologous recombination by inducing cytoplasmic retention of BRCA1 and RAD51
.
Cancer Res
2008
;
68
:
9404
12
.

Supplementary data