Main

Mutations in BRCA1 (breast cancer susceptibility gene 1) are linked to familial breast and ovarian cancers, and also to Fanconi anaemia1,2,3,4,5. Since its discovery over twenty years ago6,7,8, BRCA1 has been implicated in various biological processes, including mRNA splicing and microRNA biogenesis9,10,11,12,13, DNA damage signalling, cell cycle checkpoints2,14, the avoidance of replication–transcription conflicts15,16, and DNA double-strand break (DSB) repair by homologous recombination1,3,17,18,19. The role of BRCA1 in these processes has remained mostly undefined, largely because of difficulties in obtaining high-quality protein preparations for biochemical analyses. BRCA1 (1,863 residues) forms a stable complex with BARD1 (BRCA1-associated RING domain protein 1; 777 residues)20,21. Depletion of BARD1 engenders DNA damage sensitivity, homologous recombination deficiency, and genome destabilization21,22,23,24,25. The ablation of BARD1 in mice leads to cancer susceptibility22, and probable disease-causing mutations are found in patients with cancer26,27,28,29,30.

We have investigated the multifaceted role of BRCA1–BARD1 in homologous recombination-mediated DSB repair. During the repair process, the DSB ends are resected to yield 3′ single-stranded DNA tails31. These DNA tails become coated with replication protein A (RPA), which is subsequently displaced by the recombinase protein RAD51 to form a nucleoprotein complex termed the presynaptic filament. The presynaptic filament searches for, engages, and then invades a homologous duplex target to form a nascent heteroduplex DNA joint, the displacement loop or D-loop. This is followed by DNA synthesis and resolution of DNA intermediates to complete repair32. There is evidence that BRCA1 promotes DNA end resection by acting as an antagonist of 53BP1 and regulating the MRE11–RAD50–NBS1–CtIP resection nuclease complex, and also participates in RAD51-mediated presynaptic filament formation with the tumour suppressors BRCA2 and PALB24,33. To gain mechanistic insights into how BRCA1–BARD1 promotes homologous recombination, we have developed a robust system for co-expressing BRCA1 and BARD1 in insect cells and a protocol to obtain BRCA1–BARD1 for biochemical testing. Our results reveal novel attributes of BRCA1–BARD1 and a role for this protein complex in the DNA strand invasion step of homologous recombination-mediated chromosome damage repair.

DNA binding by BRCA1 and BARD1

BRCA1–BARD1 was expressed in insect cells and purified to near homogeneity (Extended Data Fig. 1a–d). We used the DNA electrophoretic mobility shift assay (EMSA) to test the ability of BRCA1–BARD1 to bind radiolabelled single-stranded DNA (ssDNA), double-stranded DNA (dsDNA), the replication fork, the D-loop and DNA bubble (a short section of unwound DNA that forms during biological processes such as transcription). We also performed competition experiments in which the nucleoprotein complex of BRCA1–BARD1 and radiolabelled D-loop was challenged with an unlabelled DNA species. The results revealed that BRCA1–BARD1 has the highest affinity for the D-loop and DNA bubble, followed by the replication fork, dsDNA and ssDNA (Fig. 1a, b and Extended Data Figs 1e–h, 2a, b).

Figure 1: DNA-binding and RAD51-interaction attributes of BRCA1–BARD1.
figure 1

a, Binding of D-loop, DNA bubble (Bubble), replication fork (RF), dsDNA and ssDNA. b, Quantification of interaction shown in a. Data are means ± s.d., n = 3 (bubble, RF, dsDNA and ssDNA) or 5 (D-loop). c, Southwestern analysis to test D-loop binding. Bovine serum albumin (BSA) was the negative control. d, Pull-down analysis for interaction of RAD51 or yRad51 with BRCA1–BARD1. e, Far western analysis for interaction of BRCA1 and BARD1 with RAD51. B1–B1, BRCA1–BARD1. BSA and RAD54 were the negative and positive controls, respectively.

PowerPoint slide

Source data

Full size image

BRCA1 is known to bind DNA34,35. Using the southwestern assay, we found that both BRCA1 and BARD1 bind the D-loop, with BARD1 showing an apparently higher affinity for the substrate (Fig. 1c). Consistent with this, BRCA1–BARD11–142 (full-length BRCA1 in complex with only the RING domain of BARD1) exhibited a lower affinity for various DNA substrates (Extended Data Fig. 1i, j). Together, our results show that both BRCA1 and BARD1 contribute to the DNA-binding capability of the BRCA1–BARD1 complex. The DNA-binding domain of BRCA1 was previously found to reside within the protein’s middle region34,35 and our mapping effort has led to the isolation of the BARD1 DNA-binding domain (Extended Data Fig. 2c–e). Notably, the BARD1 domain exhibited similar DNA-binding properties to the complex (Extended Data Fig. 2f–i). Thus, BARD1 is a structure-specific DNA-binding protein with the highest affinity for the D-loop and DNA bubble.

RAD51 interaction with BRCA1–BARD1

BRCA1 has been shown to co-immunoprecipitate with RAD51 from cell extracts17, but it has remained unclear whether it associates with RAD51 directly. By affinity pull-down, we found that BRCA1–BARD1 interacts with human RAD51 but has little or no affinity for yeast Rad51 (yRad51) (Fig. 1d) or Escherichia coli RecA (Extended Data Fig. 3a). We also determined that four or five RAD51 molecules are bound by BRCA1–BARD1 (Extended Data Fig. 3b, c). Notably, formation of the BRCA1–BARD1–RAD51 complex was not affected by benzonase or ethidium bromide (Fig. 1d and Extended Data Fig. 3d), indicating that the association is not bridged by nucleic acid. Surprisingly, both BRCA1 and BARD1 retained RAD51 in the far western assay, with BARD1 showing a more robust signal (Fig. 1e), whereas the homologous recombination factors RAD51D–XRCC2 and DSS1 did not bind RAD51 under the same conditions (Extended Data Fig. 3e). These results helped to establish that BRCA1–BARD1 associates with RAD51 in a species-specific manner, and that both proteins in the complex participate in this interaction.

BRCA1–BARD1 enhances homologous DNA pairing

Given that BRCA1–BARD1 binds DNA and interacts with RAD51 (Fig. 1), we hypothesized that it would enhance either the assembly of the presynaptic filament or the potential of the presynaptic filament to mediate DNA strand invasion, or both. We used a DNA strand exchange assay36,37 (Extended Data Fig. 4a) to test whether BRCA1–BARD1 could facilitate RAD51 presynaptic filament assembly. Although the BRCA2–DSS1 complex promotes RAD51 presynaptic filament assembly on RPA-coated ssDNA36,38, BRCA1–BARD1 does not (Extended Data Fig. 4b, c). Moreover, unlike BRCA2–DSS137,38, BRCA1–BARD1 cannot target RAD51 to ssDNA when dsDNA is present (Extended Data Fig. 4d–f).

Next, we conducted a D-loop assay to test whether BRCA1–BARD1 could promote DNA strand invasion (Fig. 2a). Notably, BRCA1–BARD1, in amounts substoichiometric to RAD51, strongly enhanced the reaction, regardless of whether ATP (Extended Data Fig. 4g–i) or the non-hydrolysable analogue AMP-PNP (Fig. 2a–c) was used as the nucleotide cofactor, while BRCA2–DSS1 did not stimulate strand invasion (Fig. 2a–c). By contrast, BRCA1–BARD1 did not enhance the activity of yRad51 (Extended Data Fig. 4j, k). BRCA2–DSS1 did not stimulate D-loop formation by BRCA1–BARD1–RAD51 when the ssDNA substrate was pre-incubated with RAD51 or when ssDNA and plasmid DNA were premixed (Extended Data Fig. 5a, b). However, with RPA-coated ssDNA, D-loop formation became more robust in the presence of BRCA1–BARD1 and BRCA2–DSS1 than with either complex alone (Extended Data Fig. 5c, d). Together, these results reveal an unexpected role of BRCA1–BARD1 in promoting DNA joint formation that is catalysed by RAD51 (Fig. 2d).

Figure 2: Enhancement of RAD51-mediated D-loop formation by BRCA1–BARD1.
figure 2

a, Assay schematic of D-loop reactions. b, D-loop reactions performed with the BRCA1–BARD1 or BRCA2–DSS1 complex. c, Quantification of the reactions in b. Data are means ± s.d., n = 3. d, Schematic depicting the role of BRCA1–BARD1 in DNA strand invasion during homologous recombination.

PowerPoint slide

Full size image

In homologous DNA pairing, the presynaptic filament captures the duplex partner and then assembles the synaptic complex in which the recombining DNA molecules are aligned in homologous registry and base switching has occurred32. By monitoring the protection of dsDNA against restriction enzyme digestion (Fig. 3a), we found that BRCA1–BARD1 stimulates synaptic complex formation (Fig. 3b, c). Next, we used our DNA curtain assay39,40 to examine the pairing of homologous DNA sequences in real time (Fig. 3d). As previously reported39,40, the RAD51 presynaptic filament was able to engage a 70-base pair dsDNA fragment harbouring 9-nucleotide homology (Fig. 3e). Importantly, the results showed enhancement of DNA engagement by BRCA1–BARD1 (Fig. 3e). However, we found no evidence that BRCA1–BARD1 has any effect on the binding site distributions, the pairwise distance distributions, or the resident time (koff) of the bound dsDNA (Fig. 3f, g and Extended Data Fig. 5e). Since BRCA1–BARD1 does not affect the koff of the aligned dsDNA, we speculate that it acts by increasing the kon of dsDNA engagement. We note that BRCA1–BARD1 mutants impaired for BARD1–RAD51 interaction or lacking the RAD51-interaction domain of BRCA1 cannot promote pairing with the duplex target (see below). We also verified that BRCA1–BARD1 does not affect the ability of presynaptic filaments harbouring yRad51 to engage dsDNA (Fig. 3 and Extended Data Fig. 5f).

Figure 3: Promotion of synaptic complex formation by BRCA1–BARD1.
figure 3

a, Schematic of the synaptic complex assay. b, Synaptic complex formation by the RAD51–ssDNA filament and BRCA1–BARD1. c, Quantification of synaptic complex formation. Data are means ± s.d., n = 3 (for lanes 6 and 7) or 6 (for lanes 2–5). d, Schematic of DNA curtain assay39,40. e, Number of dsDNA oligonucleotides bound by each RAD51–ssDNA or yRad51–ssDNA filament as a function of BRCA1–BARD1 concentration. Data are means ± 95% confidence intervals, n = 49 (RAD51), 50 (RAD51 + 10 nM BRCA1–BARD1), 38 (RAD51 + 50 nM BRCA1–BARD1), 54 (RAD51 + 100 nM BRCA1–BARD1), 51 (yRad51) or 53 (yRad51 + 100 nM BRCA1–BARD1). f, Binding distribution for Atto565–dsDNA with or without BRCA1–BARD1. g, Semi-log survival plot of the synaptic complex with and without 100 nM BRCA1–BARD1. *P < 0.05; **P < 0.01. NS, non-significant. In f and g, data are means ± errors (determined by bootstrapping). The multiGaussian in f and the lines in g were fitted with least squares analysis.

PowerPoint slide

Source data

Full size image

Functional relevance of BARD1–RAD51 interaction

We sought to isolate RAD51-binding defective mutants of BRCA1–BARD1 for biochemical and genetic testing. First, we co-expressed RAD51 with various BRCA1 fragments in insect cells and conducted co-immunoprecipitation. Consistent with a previous study17, BRCA11–1527 could interact with RAD51, whereas BRCA11–1000 and BRCA11–500 were impaired in this regard (Extended Data Fig. 6a–c). Notably, BRCA1–BARD1 could co-precipitate much more RAD51 than BRCA1 alone. This result, together with the far western data (Fig. 1e), indicates that BARD1 harbours a major RAD51-interaction domain (Fig. 4a). Deletion analysis showed that the region between residues 123 and 162 of BARD1 is indispensable for RAD51 interaction (Fig. 4a, b and Extended Data Fig. 6d–g). Moreover, a GST-tagged BARD1 fragment harbouring these residues could efficiently associate with RAD51 (Extended Data Fig. 6h), indicating that it encompasses the RAD51 interaction domain. We also discovered that the core domain of RAD51 (referred to as T3), which has been implicated in BRCA2 binding via the BRC4 repeat of BRCA241, can interact with BRCA1–BARD1 (Extended Data Fig. 3f), but not with BRCA1–BARD11–142 or BRCA11–500–BARD1 (Extended Data Fig. 3g). Interestingly, BRCA1–BARD1 could compete with BRCA2–DSS1 for RAD51 association (Extended Data Fig. 3h).

Figure 4: Relevance of the BARD1–RAD51 complex in DNA strand invasion.
figure 4

a, Domains in BRCA1–BARD1. b, Alignment of the RAD51 interaction domain in BARD1 orthologues. The highlighted residues (green) were changed to AAE or N (red). The asterisks denote BARD1 mutations found in human cancers (cBioPortal for Cancer Genomic). c, Testing of RAD51 interaction with wild-type or mutant BRCA1–BARD1. d, Examination of BRCA1–BARD1 mutants in the D-loop reaction. e, Quantification of data from experiment in d. Data are means ± s.d., n = 3 (BARD1Δ123–162), 4 (BARD1WT and BARD1AAE) or 5 (BARD1K140N). P values were calculated using two-way ANOVA and multiple comparisons were corrected by the Bonferroni method. **P < 0.01.

PowerPoint slide

Full size image

We expressed and purified the mutant BRCA1–BARD1∆123–162 complex, in which the RAD51-interaction domain of BARD1 has been deleted. BRCA1–BARD1∆123–162 retained normal DNA-binding activity (Extended Data Fig. 7a, b) but was defective in RAD51 interaction (Fig. 4c) and, accordingly, failed to enhance D-loop formation (Fig. 4d, e) or synaptic complex assembly (Extended Data Fig. 7c, d). Sequence alignment of the RAD51-interaction domain in BARD1 orthologues revealed a number of conserved amino acid residues (including the FXDA motif; Fig. 4b). On the basis of this information, we generated a compound mutant that changes the conserved residues F133 and D135 to alanine and A136 to glutamic acid (the AAE mutant); F133 was included because other RAD51-interaction motifs41,42,43, such as BRC4 in BRCA241, also harbour a functionally indispensable phenylalanine residue. We expressed and purified the mutant BRCA1–BARD1AAE complex. Biochemical testing revealed that, even though the mutant complex binds DNA normally (Extended Data Fig. 7a, b), it is impaired not only for RAD51 association (Fig. 4c), but also for the ability to stimulate D-loop formation and synaptic complex assembly (Fig. 4d, e and Extended Data Figs 5g, 7c, d). Together, these results provided evidence that the BRCA1–BARD1–RAD51 complex is indispensable for the enhancement of RAD51-mediated DNA strand invasion.

Cancer-associated mutations have been identified within the RAD51-interaction domain of BARD1. One such mutation (K140N), found in two patients with colorectal adenocarcinoma or uterine corpus endometrial carcinoma, alters the conserved residue K140 (cBioPortal for Cancer Genomic)44,45 next to the FXDA motif (Fig. 4b). To determine the relevance of this mutation, we expressed and purified the BRCA1–BARD1K140N mutant complex for testing. Although the mutation had no impact on DNA binding (Extended Data Fig. 7a, b), it attenuated the affinity of BRCA1–BARD1 for RAD51 (Fig. 4c) and also compromised the ability of the tumour suppressor complex to enhance D-loop formation and synaptic complex assembly (Fig. 4d, e and Extended Data Fig. 7c, d).

Cellular role of the BRCA1–BARD1–RAD51 complex

We conducted cell-based studies to investigate the association between BRCA1–BARD1 and RAD51 and to ascertain the importance of the BRCA1–BARD1–RAD51 complex. The amount of RAD51 that co-immunoprecipitated with wild-type siRNA-resistant BARD1 (BARD1WTres) was increased by treatment of cells with mitomycin C (MMC) (Fig. 5a), and the BARD1AAEres mutation impaired the DNA damage-induced association with RAD51 (Fig. 5a). Cellular fractionation confirmed that the nuclear localization of BRCA1 and BARD1 is not affected by the BARD1AAEres mutation (Extended Data Fig. 8a).

Figure 5: Biological relevance of the BARD1–RAD51 complex.
figure 5

a, Immunoprecipitation to test BARD1WT and BARD1AAE for RAD51 association after treatment with MMC. Asterisk denotes a non-specific band. b, Schematic of the DR-GFP reporter assay (top). Results obtained with cells expressing BARD1WTres or BARD1AAEres upon treatment with BARD1 siRNA or control siRNA (siCtrl) (bottom). Data are means ± s.d., n = 3. c, Schematic of the CRISPR–Cas9 gene targeting assay (top). Results obtained with cells expressing BARD1WTres or BARD1AAEres upon treatment with BARD1 siRNA or siCtrl. Data are means ± s.d., n = 3. d, Clonogenic survival of cells expressing BARD1WTres or BARD1AAEres after treatment with olaparib or MMC. Data are means ± s.d., n = 3. EV, empty vector. P values were calculated using two-way ANOVA and multiple comparisons were corrected by the Bonferroni method. **P < 0.01.

PowerPoint slide

Source data

Full size image

Next, we used the direct repeat-green fluorescent protein (DR-GFP) reporter46,47, which measures DSB-induced homologous recombination, and a CRISPR–Cas9-stimulated gene-targeting assay48,49 to investigate whether the BARD1AAEres mutation affects homologous recombination. As expected, knockdown of endogenous BRCA1 or BARD1 using small inhibitory RNA (siRNA) impaired homologous recombination in both systems (Extended Data Fig. 8b–d). Notably, although the ectopic expression of BARD1WTres in BARD1-deficient cells fully restored homologous recombination, ectopic expression of BARD1AAEres resulted in only partial complementation (Fig. 5b, c and Extended Data Fig. 9a, b). Moreover, in clonogenic cell survival assays, BARD1-deficient cells expressing BARD1AAEres were markedly more sensitive than cells expressing BARD1WTres to MMC and to the poly-ADP ribose polymerase (PARP) inhibitor olaparib (Fig. 5d and Extended Data Fig. 9c).

We also investigated whether the BARD1AAEres mutation would affect the DNA damage-induced assembly of RAD51 nuclear foci. As expected, knockdown of endogenous BRCA1 diminished RAD51 focus formation, either spontaneously or after γ-ray exposure (Extended Data Fig. 8e–g). However, treatment with siRNA targeting BARD1 impaired RAD51 focus formation to a lesser extent (Extended Data Fig. 8g). In cells depleted of endogenous BARD1 and expressing BARD1WTres or BARD1AAEres, RAD51 focus formation occurred similarly, both spontaneously and after γ-irradiation (Extended Data Fig. 9d, e). However, as indicated by S4/S8 phosphorylation of RPA32, BARD1AAEres cells retained a much higher level of DNA damage 72 h after release from MMC treatment (Extended Data Fig. 9f). These results showed that homologous recombination-mediated repair is deficient in BARD1AAEres-expressing cells despite the fact that RAD51 focus formation is not affected. Even though depletion of 53BP1, an inhibitor of DNA end resection50, partially overcame the homologous recombination defect associated with BRCA1 deficiency (Extended Data Fig. 8h, i), it did not suppress the homologous recombination defect in BARD1-deficient cells (Extended Data Fig. 8j, k). In cells lacking both BARD1 and 53BP1, BARD1AAEres was less able than BARD1WTres to reverse the homologous recombination deficiency (Extended Data Fig. 8l, m). Together, our results help to establish the biological importance of the BRCA1–BARD1–RAD51 complex in DNA damage repair via homologous recombination and provide cellular evidence for a role of BRCA1–BARD1 in the DNA strand invasion step of homologous recombination.

Role of BRCA1 in RAD51-mediated DNA pairing

To investigate the role of BRCA1 in RAD51-mediated reactions, we expressed and purified BRCA11–500–BARD1, which lacks the RAD51 interaction and DNA binding domains of BRCA117,34, and also BRCA1∆758–1064–BARD1, in which the RAD51-interaction domain of BRCA1 has been deleted. These mutant complexes appeared to be proficient in DNA binding (Extended Data Fig. 10a–e) but were weakened for RAD51 interaction (Extended Data Fig. 10f–h). Importantly, neither mutant complex could strongly enhance RAD51-mediated D-loop formation (Extended Data Fig. 10i, j) or synaptic complex assembly (Extended Data Figs 5g and 7c–f). Thus, BRCA1 is also indispensable for the functional integrity of BRCA1–BARD1 as a co-factor of RAD51.

Discussion

Our study has revealed that BRCA1–BARD1 enhances DNA invasion in homologous recombination by interacting directly with RAD51 (Fig. 6). Mechanistically, BRCA1–BARD1 functions with the RAD51 presynaptic filament in the assembly of the synaptic complex, a critical precursor to D-loop formation (Fig. 6). Both BRCA1 and BARD1 are indispensable for this attribute. It is likely that physical association with RAD51 facilitates dsDNA engagement by the presynaptic filament and that the specific recognition of unwound DNA by BRCA1–BARD1 enhances the formation of the nascent DNA joint in the D-loop reaction. As we have not observed significant stimulation by BRCA1–BARD1 of DNA strand exchange between an oligonucleotide and a short linear duplex, it remains possible that the complex facilitates DNA homology search within a long duplex DNA target. A model incorporating this new function and the known roles of BRCA1–BARD1 in DNA end resection and RAD51 presynaptic filament assembly is shown in Fig. 6.

Figure 6: Model of BRCA1–BARD1 functions.
figure 6

Aside from a co-operative role with PALB2–BRCA2 in RAD51 presynaptic filament assembly (green arrows), our work has revealed a function of BRCA1–BARD1 in the promotion of homologous DNA pairing (red arrows). Previous studies have provided evidence that BRCA1–BARD1 antagonizes 53BP1 in DNA end resection (green blocks) and promotes the activity of the resection nuclease MRN/CtIP (green arrows), and for a role of the complex in cell cycle checkpoint regulation (green arrows).

PowerPoint slide

Full size image

The findings from our study open up a new avenue towards understanding how mutations in BRCA1–BARD1 affect its DNA damage repair and tumour suppression functions. Indeed, we have provided evidence that the cancer-associated mutation K140N in BARD1 compromises the physical and functional interactions of BRCA1–BARD1 with RAD51. We note that the region of BRCA1 (amino acid residues 758–1,064) that harbours the RAD51-interaction domain17 is frequently mutated in cancer (cBioPortal for Cancer Genomic)44,45 and that deletion of this domain abolishes the activity of BRCA1–BARD1 in RAD51-mediated DNA strand invasion. The biochemical systems established in our work should be valuable for determining the impact of pathogenic mutations on BRCA1–BARD1 functions. Moreover, our findings may guide the development of targeted therapies for breast, ovarian, and other cancers.

Methods

Construction of plasmids

A His6 affinity tag was fused to BARD1 in pFastbac-BARD1 (from J. Parvin) using the QuikChange mutagenesis kit (Stratagene). The mammalian pS-Flag-SBP-BARD1res expression vector was modified by removing the GFP coding sequence from the pS-Flag-SBP-BARD1 vector (from X. Yu) and introducing silent mutations into the siRNA target regions of BARD1 using oligos 1 (5′-GATGATAATATGGCCACAACCAGCGGCCGCGACTCTAGATC-3′) and 2 (5′-GATCTAGAGTCGCGGCCGCTGGTTGTGGCCATATTATCATC-3′) and oligos 3 (5′-GAAAGTCAGATATGTTGTGAGCAAGGCAAGTGTCCAGACCCAGCCTGCAATAAAAA-3′) and 4 (5′-TTTTTATTGCAGGCTGGGTCTGGACACTTGCCTTGCTCACAACATATCTGACTTTC-3′), respectively. QuikChange site-directed mutagenesis was used to construct the mutant forms of BARD1: BARD11–122, BARD11–162, BARD11–261, BARD1∆123–162, BARD1∆123–261, BARD1∆163–261, BARD1AAE and BARD1K140N (the sequences of the primers used are available upon request). BARD1123–162 was introduced into pDEST15 for expression of the GST-tagged form of this BARD1 fragment in E. coli. BARD1124–270 was cloned into pE-SUMO vector (LifeSensors Inc.) for expression of the SUMO-tagged form of this BARD1 domain in E. coli.

Protein purification: purification of BRCA1–BARD1 from insect cells

pFastbac-Flag-BRCA1 (from J. Parvin) and pFastbac-His-BARD1 were introduced into E. coli strain DH10Bac for bacmid generation. The bacmids were used to transfect SF9 insect cells to generate recombinant baculoviruses. After amplification in SF9 cells, the viruses were used to infect Hi5 insect cells for expression of BRCA1 and BARD1 (10 ml BRCA1 and 10 ml BARD1 P3 viruses for 600 ml culture). After a 44-h incubation at 27 °C, cells were harvested by centrifugation, frozen in liquid nitrogen, and stored at −80 °C. All purification steps were carried out at 0–4 °C. To prepare extract, the frozen cell pellet (8 g, from 600 ml culture) was thawed and suspended in 40 ml cell breakage buffer A (50 mM Tris-HCl, pH 7.5, 500 mM KCl, 1 mM 2-mercaptoethanol, 0.5% NP-40, 5 mM MgCl2, 2 mM ATP and the following protease inhibitors: aprotinin, chymostatin, leupeptin, and pepstatin A at 3 μg ml−1 each, and 1 mM PMSF) for cell lysis using a Dounce homogenizer type B pestle (30 strokes). The lysate was cleared by centrifugation at 10,000g for 15 min, and the supernatant was incubated with 3 ml anti-Flag M2 affinity resin (Sigma) for 2 h. The resin was transferred to a column (1.5 × 15 cm), washed with 50 ml lysis buffer and then with 50 ml buffer B (25 mM Tris-HCl, pH 7.5, 300 mM KCl, 10% glycerol, 0.5 mM EDTA, 0.01% Igepal CA-630, 1 mM 2-mercaptoethanol, 5 mM MgCl2 and 2 mM ATP), before the bound proteins were eluted four times with 2 ml buffer B containing the single Flag peptide (200 μg ml−1). The eluates were combined and mixed with 32 ml buffer C (25 mM Tris-HCl, pH 7.5, 10% glycerol, 0.5 mM EDTA, 0.01% Igepal CA-630, 1 mM 2-mercaptoethanol) before being further fractionated in a 1 ml HiTrap SP Sepharose HP column (GE Healthcare) using a 12 ml gradient of 75–500 mM KCl in buffer C. The pooled BRCA1–BARD1 fractions (250–350 mM KCl) were further fractionated in a gel filtration column of Superose 6 10/300 GL (GE Healthcare), which was developed with 24 ml buffer C containing 300 mM KCl. The peak fractions were pooled, divided into 10-μl portions, frozen in liquid nitrogen, and stored at −80 °C. The mutant forms of BRCA1–BARD1 were expressed and purified using the same procedures. The yield of highly purified BRCA1–BARD1 from 600 ml insect cell culture ranged from 150 to 300 μg with a final concentration of 300–500 μg ml−1.

Protein purification: purification of BARD1123–162 and BARD1124–270 from E. coli

The GST–BARD1123–162 expression plasmid pDEST15–BARD1123–162 or the BARD1124–270 expression plasmid pET-SUMO-BARD1124–270 was introduced into E. coli Rosetta (DE3) cells. An overnight culture derived from a single colony in 50 ml LB medium grown at 37 °C was used to inoculate 2 l fresh LB medium. IPTG was added to 0.4 mM when the cell density had reached OD600 = 0.8, and cells were harvested after a 16-h incubation at 16 °C. All the subsequent steps were carried out at 0–4 °C. The cell pellet (8 g) was suspended in 50 ml buffer D (20 mM KH2PO4, pH 7.5, 10% glycerol, 0.5 mM EDTA, 0.01% Igepal CA-630, 1 mM 2-mercaptoethanol and 300 mM KCl) containing the protease inhibitors (aprotinin, chymostatin, leupeptin, and pepstatin A at 3 μg ml−1 each, and 1 mM PMSF) and cell lysate was prepared by sonication. After centrifugation (100,000g for 90 min), the clarified lysate was incubated with 2 ml Glutathione Sepharose 4 Fast Flow resin (GE Healthcare; for GST-BARD1123–162) or Ni-NTA resin (GE healthcare; for BARD1124–270) for 2 h. The affinity resin was transferred to a glass column (1.5 × 15 cm) and washed with 20 ml buffer D before being eluted three times with 3 ml of 20 mM glutathione or 150 mM imidazole in buffer D. For BARD1124–270, the His6-SUMO tag was cleaved by the Ulp1 protease by overnight incubation at 4 °C. The eluates were pooled and concentrated in a Centricon-10K concentrator (Amicon) to 0.5 ml before being further fractionated in a Superdex 200 10/300 GL column (GE Healthcare) with 24 ml of buffer C containing 300 mM KCl. The peak fractions were pooled, concentrated to ~100 μl as above, divided into 5 μl portions, frozen in liquid nitrogen, and stored at −80 °C.

Protein purification: other recombination proteins

BRCA2–DSS1, RAD51, RPA and yeast Rad51 were purified to near homogeneity using our previously described procedures36,51,52.

DNA substrates and DNA binding assay

D-loop, DNA bubble, replication fork and double-stranded DNA were assembled from oligonucleotides 5/6/7, oligonucleotides 5/6, oligonucleotides 8/9/10/11 and oligonucleotides 12/13, respectively; the asterisk identifies the oligonucleotide that was 32P-labelled at its 5′ end in each substrate. The single-stranded DNA substrate was 5′ 32P-labelled oligonucleotide 12. These DNA substrates (10 nM each) were incubated with wild-type or the specified mutant form of BRCA1–BARD1 at 37 °C in 10 μl buffer E (25 mM Tris-HCl, pH 7.5, 90 mM KCl, 1 mM DTT, and 100 μg ml−1 bovine serum albumin (BSA)) for 10 min. After the addition of loading buffer (50% glycerol, 20 mM Tris-HCl, pH 7.4, 0.5 mM EDTA, 0.05% Orange G), the reaction mixtures were resolved by 6% native polyacrylamide gel electrophoresis in TAE buffer (30 mM Tris-acetate, pH 7.4 and 0.5 mM EDTA) at 4 °C. The gels were dried, and DNA species were visualized by autoradiography and quantified using the Personal Molecular Imager and Quantity One software (Bio-Rad). Since the nucleoprotein complexes formed by BRCA–1-BARD1 do not always migrate as well-defined species, we quantified DNA binding by measuring the disappearance of the DNA substrate.

Oligo 5: CATTGCATATTTAAAACATGTTGGAAGGCTCGATGCATGCTGATAGCCTACTAGTGCTGCTGGCTTTCAAATGACCTCTTATCAAGTGAC

Oligo 6: GTCACTTGATAAGAGGTCATTTGAATTCATGGCTTAGAGCTTAATTGCTGAATCTGGTGCTGGGATCCAACATGTTTTAAATATGCAATG

Oligo 7: CTGCTACGATGCTAGTCGTAGCTCGGCAGTCGTAGCAGGTTCCCAGCACCAGATTCAGCAATTAAGCTCTAAGCCATGAA

Oligo 8: GACGCTGCCGAATTCTACCAGTGCCTTGCTAGGACATCTTTGCCCACCTGCAGGTTCACCC

Oligo 9: GGACATCTTTGCCCACCTGCAGGTTCACCC

Oligo 10: TGGGTGAACCTGCAGGTGGGCAAAGATGTCC

Oligo 11: GGGTGAACCTGCAGGTGGGCAAAGATGTCCCAGCAAGGCACTGGTAGAATTCGGCAGCGTC

Oligo 12: TTATATCCTTTACTTTGAATTCTATGTTTAACCTTTTACTTATTTTGTATTAGCCGGATCCTTATTTCAATTATGTTCAT

Oligo 13: ATGAACATAATTGAAATAAGGATCCGGCTAATACAAAATAAGTAAAAGGTTAAACATAGAATTCAAAGTAAAGGATATAA

Affinity pull-down

RAD51, yRad51 or RecA (5 μM) was incubated with 0.5 μM Flag–BRCA1–BARD1 or 3 μM GST–BARD1123–162 at 4 °C for 30 min in 30 μl buffer F (25 mM Tris-HCl pH 7.5, 10% glycerol, 0.5 mM EDTA, 0.05% Igepal CA-630, 1 mM 2-mercaptoethanol, 150 mM KCl). Then the reaction mixture was mixed with 12 μl anti-Flag M2 affinity resin or Glutathione Sepharose 4 Fast Flow resin at 4 °C for 30 min to capture protein complexes through the Flag tag on BRCA1 or the GST tag on BARD1, respectively. After the resin was washed three times with 200 μl buffer F, bound proteins were eluted with 20 μl 2% SDS at 37 °C for 5 min. The supernatant (S), last wash (W) and SDS eluate (E), 8 μl each, were analysed by SDS–PAGE and Coomassie blue staining.

Southwestern analysis

BRCA1–BARD1 was resolved in a 7.5% SDS–PAGE gel and transferred onto a nitrocellulose membrane (Bio-Rad) at 4 °C in transfer buffer (25 mM Tris-HCl, 192 mM glycine, pH 8.3, 20% methanol). After being soaked in buffer G (25 mM Tris-HCl, pH 7.5, 2 mg ml−1 BSA, 1 mM DTT, 0.1% Triton X-100, 10% glycerol and 100 mM KCl) at 4 °C for 20 h, the membrane was rinsed twice with buffer H (25 mM Tris-HCl, pH 7.5, 200 μg ml−1 BSA, 90 mM KCl, 4 mM MgCl2 and 1 mM DTT) and then incubated in 10 ml buffer H containing 32P-labelled D-loop DNA (2 nM) at 25 °C for 1 h. The membrane was washed four times with 10 ml buffer H before analysis by phosphorimaging.

Far western analysis

After SDS–PAGE, BRCA1 and BARD1 were transferred onto a nitrocellulose membrane, as described for the southwestern analysis. The membrane was soaked in buffer I (10 mM KH2PO4 at pH 7.4, 150 mM KCl, 15 mg ml−1 BSA, 2 mM 2-mercaptoethanol, 0.05% Tween 20) at 25 °C for 2 h and then incubated with 5 μg ml−1 RAD51 in buffer I at 25 °C for 2 h. Then, the membrane was washed with 10 ml buffer I three times, incubated with anti-RAD51–HRP antibodies (Abcam, ab195548) for 1 h in buffer I, washed again with 10 ml buffer I three times, and developed with the Super Signal Substrate Kit (Pierce).

Homologous DNA pairing assay

The homologous DNA pairing assay was conducted as described36,37. The reaction was assembled in buffer J (25 mM Tris-HCl, pH 7.5, 60 mM KCl, 1 mM DTT and 100 μg ml−1 BSA) containing 1 mM ATP and 2 mM MgCl2 in a final volume of 12.5 μl. For mediator activity, the 150-mer oligonucleotide 14 (6 μM nucleotides) was first incubated with RPA (600 nM) at 37 °C for 5 min, and then RAD51 (2 μM) with or without the indicated concentration of BRCA1–BARD1 or BRCA2–DSS1 was incorporated into the reaction. Following a 5-min incubation at 37 °C, 32P-labelled homologous dsDNA (40 bp; oligos 15 (5′-TAATACAAAATAAGTAAATGAATAAACAGAGAAAATAAAG-3′) and 16 (5′-CTTTATTTTCTCTGTTTATTCATTTACTTATTTTGTATTA-3′); 1.6 μM base pairs) and 4 mM spermidine hydrochloride were added. For testing of ssDNA targeting activity, RAD51 was incubated with the mixture of ssDNA, 32P-labelled dsDNA and spermidine hydrochloride with and without BRCA1–BARD1 or BRCA2–DSS1 for 30 min. The reaction was terminated by adding an equal volume of 1% SDS containing 1 mg ml−1 proteinase K. Following a 5-min incubation at 37 °C °C, the deproteinized reaction mixtures were resolved in an 8% non-denaturing polyacrylamide gel in TAE buffer. The gel was dried onto 3MM CHR cellulose chromatography papers (GE Healthcare), and DNA species were visualized by autoradiography and quantified, as above.

D-loop assay

The D-loop assay was conducted as described42,53. In brief, the 32P-labelled 90-mer oligonucleotide 17 (AAATCAATCTAAAGTATAT ATGAGTAAACTTGGTCTGACAGTTACCAATGCTTAATCAGTGAGGCACCTATCTCAGCGATCTGTCTATTT; 2.4 μM nucleotides) was incubated with RAD51 (1 μM) at 37 °C for 5 min in buffer J containing 1 mM MgCl2 and 2 mM ATP or AMP-PNP. Following the incorporation of the indicated concentration of BRCA1–BARD1 and a 5-min incubation at 37 °C, the D-loop reaction was initiated by adding pBluescript SK replicative form I DNA (37 μM base pairs) and was incubated at 37 °C for 7 min. The molar ratio of the 90-mer to pBluescript plasmid in the reactions was 2.1 to 1. The reaction was terminated by adding an equal volume of 1% SDS containing 1 mg ml−1 proteinase K and a 5-min incubation at 37 °C. The deproteinized reaction mixtures were resolved by electrophoresis in a 1% agarose gel, which was dried onto Hybond-N membrane (GE Healthcare). Phosphorimaging analysis was used to visualize and quantify the radiolabelled DNA species.

Synaptic complex assay

The synaptic complex assay was conducted at 37 °C as described5,6. In brief, RAD51 (4 μM) was incubated with the 60-mer oligonucleotide 18 (12 μM nucleotide; 5′-AATGTTGAATACTCATAC TCTTCCTTTTTCAATATTATTGAAGCATTTATCAGGGTTATT-3′), which is homologous to the SspI restriction site in the target pUC19 dsDNA, in 8 μl buffer K (35 mM Tris-HCl, pH 7.5, 50 mM KCl, 2 mM ATP, 2 mM MgCl2, 100 μg ml−1 BSA, and 1 mM DTT) for 5 min. After adding the indicated amounts of BRCA1–BARD1 in 1 μl volume, the reaction mixture was incubated for 5 min. Then, linear pUC19 plasmid DNA (85 μM nucleotides) was added in 1 μl, followed by a 5-min incubation and treatment with 2.5 units of SspI for 10 min. The reaction mixtures were resolved by agarose gel electrophoresis in TAE buffer, and DNA species were stained with ethidium bromide. The heterologous oligonucleotide 19 (5′-CAGAATCAGGGGATAACGCAGGAAAGA ACATGTGAGCAAAAGGCCAGCAAAAGGCCAGGA-3′) was used as a control.

DNA curtain imaging analysis

RAD51 filaments were assembled on ssDNA curtains, and the dsDNA-binding properties of the resulting RAD51–ssDNA filaments were measured as described39,40. To determine the number of dsDNA-binding events, BRCA1–BARD1 was diluted from a 1.5 μM stock with buffer K (30 mM Tris–acetate pH 7.5, 1 mM MgCl2, 5 mM CaCl2, 100 mM KCl, 2 mM ATP, 1 mM DTT, and 0.2 mg ml−1 BSA) to concentrations ranging from 10 to 100 nM, introduced into the flow cell chamber, and incubated for 10 min at 37 °C. After washing with 1 ml buffer K (at 1 ml min−1), 2.16 nM Atto565-labelled dsDNA (70 bp) with 9 nt of homology (oligos 20 (5′- Atto565CCGGAGGCCTTAGGCCTTAGGCCTT AGGCCTTCAGCTGTTAGCCTTAGCTAGCTAGCTAGCTAGCTAGCT-3′; the underlined sequence is homologous to the ssDNA substrate) and 21 (5′-AGCTAGCTAGCTAGCTAGCTAGCTAAGGCTAACAGCTGAAGGCCTAAGGCCTAAGGCCTAAGGCCTCCGG-3′) to the RAD51–ssDNA filaments was introduced into the chamber, followed by a 10-min incubation at 37 °C. Then, the cell was washed with 0.5 ml buffer K (at 1 ml min−1) and three images were taken. The length of each of the RAD51–ssDNA filaments and the number of labelled dsDNA molecules bound were recorded and normalized to a length of 50 pixels (~40 kb). The weighted average and standard deviation based on the length of each filament were calculated. Confidence intervals of 95% are represented as error bars. For survival probabilities, experiments were conducted without or with 100 nM BRCA1–BARD1 and 100 ms exposures were recorded every 30 s over 90 min. Dwell times of ~180 molecules were determined for each experiment by kymograph and survival probabilities were plotted on a semi-log plot. Error bars represent 70% confidence as measured by bootstrap analysis, a close approximation of one standard deviation from the mean.

Cell culture and transfection

U2OS and HeLa cells from ATCC were grown in Dulbecco’s modified Eagles medium (DMEM) supplemented with 10% fetal bovine serum (Sigma), 100 μg ml−1 streptomycin, and 100 U ml−1 penicillin (Sigma). The cells were tested for mycoplasma contamination by Bionique testing labs (http://www.bionique.com/). Control siRNA (UAGCCGGUAGACUUAGGUCUG), BARD1 siRNA (AAGAGUAAAGCUUCAGUGCAA), BRCA1 siRNA (AAGCUCCUCUCACUCUUCAGU) and BRCA2 siRNA (UUGGAGGAAUAUCGUA GGUAA) oligonucleotides were purchased from Qiagen. TP53BP1 siRNA (s14313) was purchased from Ambion-Thermo Fisher Scientific. Transfection of siRNA, pS-Flag-SBP-BARD1res and pCMV-I-SceI-3×NLS was carried out using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. To generate stable HeLa and U2OS cell lines expressing Flag–SBP–BARD1 or its mutants, cells were transfected with their respective plasmids (pS-Flag–SBP–BARD1WTres, pS-Flag–SBP–BARD1AAEres, and pS-Flag–SBP–BARD1K140Nres) and individual clones were selected with 800 μg ml−1 G418.

Co-immunoprecipitation analysis

HeLa cells grown on 15-cm cell culture dishes were treated with or without 1 μM MMC overnight before collection. Following a wash with PBS, cells were scraped off and transferred to Eppendorf tubes. Whole cell lysate was prepared by adding 1 ml lysis buffer (300 mM NaCl, 1.0% Triton X-100, 5 mM EDTA, 2 mM NaVO4, 2 mM Na4O7P2, 0.02% NaN3, and 50 mM Tris-HCl, pH7.4) with protease inhibitors (Roche Complete Protease Inhibitor Cocktail Tablet) to cell pellets. Following a 12-s sonication, the cell extract was cleared by centrifugation at 18,400g for 15 min at 4 °C. The supernatant fraction (2 mg protein in total) was incubated with DNase I (20 U) for 15 min at room temperature and 15 min at 37 °C. Then, 50 μl anti-Flag resin (Sigma) or anti-mouse IgG resin (Santa Cruz) was added, followed by a 12 h incubation at 4 °C overnight. After the resin was washed four times with lysis buffer, bound proteins were eluted with 100 μl SDS gel loading buffer (50 mM Tris-HCl pH 6.8, 2% SDS, 0.1% bromophenol blue, 10% glycerol, 10% 2-mercaptoethanol) and 100 μl of the eluates were subjected to western blot analysis with anti-Flag and anti-RAD51 antibodies.

Immunoblot analysis

Protein was extracted from cells collected two days after transfection with the indicated siRNAs using NETN buffer (20 mM Tris-HCl pH 8, 420 mM NaCl, 1 mM EDTA, 0.5% Igepal CA-630, 1 mM DTT, and Roche Protease Inhibitor Cocktail) and 8 freeze–thaw cycles. Blots (20–50 μg total protein) were probed with the following antibodies: BARD1 (Bethyl, A300-263A; Santa Cruz Biotech, Sc11438), BRCA1 (Abcam, ab16780), 53BP1 (Abcam, ab36823), Flag M2-HRP (Sigma, A8592), Phospho RPA32 S4/S8 (Bethyl, A300-245A), BRCA2 (EMD Millipore, OP95-100UG), RAD51 (Santa Cruz Biotech, sc-8349), Actin (Abcam, ab3280), Tubulin (Santa Cruz Biotech, sc-53030), HA.11 (16B12) (Covance, MMS-101P), or GST-HRP (NEB, E2624S) according to the instructions provided by the manufacturers. If needed, the blots were incubated with HRP-conjugated secondary antibodies (Pierce 31450 for rabbit anti-mouse IgG-HRP; Sigma A6154 for goat anti-rabbit IgG-HRP; Santa Cruz Biotech Sc-2032 for goat anti-rat IgG-HRP) before visualization of protein signals using the ECL kit (Thermo Scientific Pierce).

DR-GFP reporter assay

The DR-U2OS cell line containing a single integrated copy of the DR-GFP reporter was used46,47. Exponentially growing cells were seeded in 6-well plates at 2 × 105 cells per well before transfection with 2 μl siRNA (20 μM) and 5 μl Lipofectamine 2000 (Invitrogen). One day after siRNA transfection, cells were transfected with 2 μg I-SceI expression vector (pCBASce) and 5 μl Lipofectamine 2000. Homologous recombination proficiency was determined by counting the fraction of GFP-positive cells using a BD FACScalibur S 72 h after I-SceI transfection. The results were derived from between three and five transfections of at least three independent experiments.

CRISPR–Cas9-induced gene targeting assay

The assay was conducted as described48. U2OS cells were seeded in 6-well plates at 2 × 105 cells per well before transfection with 2 μl siRNA (20 μM) and 5 μl Lipofectamine 2000 (Invitrogen). One day after siRNA transfection, cells were co-transfected with 1.6 μg sgRNA plasmid pX330-LMNA1 (from G. Dellaire) and 0.4 μg donor template pCR2.1-CloverLamin (from G. Dellaire) and 5 μl Lipofectamine 2000 (Invitrogen). Gene targeting efficiency was determined by counting the per cent of Clover-positive cells using a BD FACScalibur S 72 h after plasmid co-transfection. The results were derived from between three and five transfections of at least three independent experiments.

Immunofluorescence microscopy and image analysis

HeLa cells in exponential growth were transfected on two consecutive days in Opti-MEM medium using RNAiMAX (Invitrogen) with 20 nM BARD1 or control siRNA, as recommended by the manufacturer. Exposure to γ-rays was performed using a 137Cs γ-irradiator (J.L. Shepherd, model 81-14) and a dose rate of 1.05 Gy min−1. Immunohistochemistry was performed as previously described54, except that cells were fixed in 4% paraformaldehyde at room temperature for 10 min and permeabilized in 0.4% Triton X-100 in PBS for 5 min. Rabbit anti-RAD51 (H-92; Santa Cruz Biotechnology; 1:2,000) and goat anti-rabbit AlexaFluor-488 secondary antibodies (Invitrogen; 1:750) were used. For image capture of RAD51 foci, Z-stack section images consisting of 20 stacks (0.2-μm intervals) from 100 to 150 nuclei per sample were taken using a 63× oil objective and a Zeiss Axio-Imager.Z2 microscope equipped with Zen Blue software (Carl Zeiss Microscopy). For computational analyses of foci, Z-stacks were collapsed down to the maximum intensity projections, and a combination of ImageJ (http://rsb.info.nih.gov/ij/) and Cell Profiler (http://www.cellprofiler.org/) software programs was used with the following custom program settings for image processing: minimum object size = 3; maximum object size = 35; despeckle ratio = 0.3; rolling ball size = 5. A custom-built pipeline for automated cell (80–300 pixel units) and foci counting with settings for shape (0.5) and dimensions (5 pixels diameter) was employed. The threshold for foci detection was determined based on sham-irradiated samples, and nuclei with >5 foci per nucleus were counted positive. Group allocation and outcome assessment were done in a fully blinded manner.

Clonogenic survival assay

HeLa cells were transiently transfected with siRNA as described above. After 48 h, cells were seeded into 6-well plates at 50–32,000 cells per well, and treated with 0, 5, 10 and 20 nM MMC (Sigma) or 0, 0.5, 1 and 2 μM Olaparib (Selleckchem) in regular growth medium for 14 days. Cells were fixed with 10% methanol and 10% acetic acid, and stained with 1% crystal violet in methanol before colonies were counted. Clonogenic survival was determined for a given concentration of cells that were plated by dividing the number of colonies on each treated plate by the number of colonies on the untreated plate, taking the plating efficiency of untreated cells into account.

Preparation of cytoplasmic and nuclear extracts

The Dignam method for the preparation of cytoplasmic and nuclear extracts was followed55. In brief, 109 cells were washed with PBS and Dignam buffer A (10 mM Hepes pH 7.9, 1.5 mM MgCl2, 10 mM KCl, 0.5 mM DTT and 0.5 mM PMSF), collected by centrifugation, and lysed in two packed cell volumes of Dignam buffer A using a Dounce homogenizer (50 strokes) with the type A pestle. After centrifugation, the supernatant containing cytoplasmic proteins was saved for analysis. The pelleted nuclei were resuspended and lysed in 3 ml Dignam buffer C (20 mM Hepes pH 7.9, 1.5 mM MgCl2, 420 mM NaCl, 0.2 mM EDTA, 0.5 mM DTT, 25% glycerol, and 0.5 mM PMSF) using a Dounce homogenizer (80 strokes) with the type B pestle. Debris was removed by centrifugation to yield the nuclear extract fraction. The cytoplasmic and nuclear fractions, 20 μg each, were analysed by immunoblotting for their content of BRCA1, Flag–SBP–BARD1, tubulin and histone H3.

Statistics and reproducibility

The statistical analysis was performed using Prism 7 (GraphPad Software, Inc.; http://www.graphpad.com/quickcalcs/ttest1.cfm) on the data from at least three independent experiments, as specified. Unless stated otherwise, statistical significance was assessed by two-tailed unpaired Student’s t-test. *P ≤ 0.05 and **P ≤ 0.01 were considered significant.

Data availability

The data that support the findings of this study are available from the corresponding authors upon reasonable request.