BRCA2 stabilises RAD51 and DMC1 nucleoprotein filaments through a conserved interaction mode

BRCA2 is essential for DNA repair by homologous recombination in mitosis and meiosis. It interacts with recombinases RAD51 and DMC1 to facilitate the formation of nucleoprotein filaments on resected DNA ends that catalyse recombination-mediated repair. BRCA2’s BRC repeats bind and disrupt RAD51 and DMC1 filaments, whereas its PhePP motifs binds to recombinases in a manner that stabilises their nucleoprotein filaments. However, the mechanism of filament stabilisation has hitherto remained unknown. Here, we report the crystal structure of a BRCA2-DMC1 complex, revealing how PhePP motifs bind to recombinases. This novel mode of interaction is conserved for RAD51 and DMC1, which selectively bind to BRCA2’s two distinct PhePP motifs. In both cases, BRCA2 PhePP motifs enhance the stability of nucleoprotein filaments, protecting them from BRC-mediated disruption. Hence, we report the structural basis of how BRCA2’s PhePP motifs stabilise RAD51 and DMC1 nucleoprotein filaments for their essential roles in mitotic and meiotic recombination.


Introduc on
DNA double-strand break (DSB) repair by homologous recombina on is cri cal for genome integrity and fer lity (Page and Hawley, 2003, Wright et al., 2018, Sun et al., 2020).In soma c cells, DSBs arise due to exogenous damage and upon replica on fork collapse (Wright et al., 2018).These lesions can be repaired, and replica on restarted, through inter-sister recombina on, of which the machinery has an addi onal role in protec ng stalled replica on forks (Feng andJasin, 2017, Tye et al., 2021).In meiosis, a programme of DSB induc on triggers inter-homologue recombina on (Page and Hawley, 2003), enabling synapsis between homologues (Adams and Davies, 2023), and the forma on of crossovers that ensure correct chromosome segrega on and enhance gene c diversity (Hunter, 2015, Zickler andKleckner, 2015).Hence, defects in homologous recombina on are associated with chromosome instability, increased cancer risk and infer lity (Hall et al., 1990, Patel et al., 1998, Sharan et al., 2004).
The tumour suppressor BRCA2 performs a central role in the mechanics of recombina on by loading recombinases onto resected DNA ends to form nucleoprotein filaments that mediate strand invasion and homology search within the template DNA (Jensen et al., 2010, Liu et al., 2010, Thorslund et al., 2010).This involves the displacement of RPA from newly resected DNA ends (Shahid et al., 2014, Bell et al., 2023), and the remodelling of recombinases from their oligomeric assemblies into protein-DNA filaments that are ac ve in recombina on (Jensen et al., 2010, Liu et al., 2010, Thorslund et al., 2010).
Here, we report the crystal structure of BRCA2 Ex14 bound to DMC1, and that BRCA2 Ex27 binds to RAD51 through the same core interac on mode.Hence, we uncover the structural basis of how BRCA2 PhePP mo fs bind to RAD51 and DMC1 recombinases and stabilise their nucleoprotein filaments for mito c and meio c recombina on.

Crystal structure of a BRCA2-DMC1 complex
The mode of interac on between BRCA2's PhePP mo fs and recombinases has remained unknown since the discovery of this filament-stabilising binding site over two decades ago (Sharan et al., 1997, Esashi et al., 2005, Davies and Pellegrini, 2007, Esashi et al., 2007).As DMC1 exists in octameric rings, whereas RAD51 forms filaments, we reasoned that DMC1 may be more amenable for structure solu on of a BRCA2 PhePP-bound complex.Hence, we established a crystallisa on system for the structural core of human DMC1 in which its N-terminal domain was deleted (herein referred to as DMC1 N; amino-acids 83-340).We first validated this system by determining the na ve structure to a resolu on of 2.05 Å, using a previous DMC1 structure (PDB accession 4HYY: Du and Luo, 2013) as a template for molecular replacement (Table 1).We then soaked DMC1 N crystals with a pep de corresponding to a truncated form of BRCA2's Ex14 (herein referred to as Ex14-Tr; amino-acids 2401-2414).Whilst soaking substan ally reduced crystal quality, we obtained X-ray diffrac on data to anisotropic resolu on limits of 3.40-5.80Å, in which addi onal pep de density was clearly observed (Figure 1b, Table 1 and Supplementary Figure 1).The pep de density was located adjacent to a loop formed by DMC1 amino-acids 178-183, which is a crystal la ce contact.Hence, the reduc on in resolu on upon BRCA2 Ex14-binding is owing to the introduc on of a defect in the crystal la ce.
The electron density maps of the bound BRCA2 Ex14 pep de could not be interpreted unambiguously (Supplementary Figure 1).Nevertheless, we were able to build the BRCA2 Ex14 structure guided by an Alphafold2 model of the same BRCA2 Ex14 sequence bound to a DMC1 N dimer (Supplementary Figure 2a-e).This resulted in a refined model of the pep de at seven out of the eight DMC1 protomers, with more extensive structural informa on at instances away from rather than at crystal la ce contacts (Table 1 and Figure 1b).The observed BRCA2 Ex14 pep de contains the PhePP mo f (aminoacids 2406-2409), in which F2406 is buried in a hydrophobic pocket and the two consecu ve proline residues enable an unusual backbone conforma on (Figure 1c).The PhePP-binding site of DMC1 is formed from by two loops, corresponding to amino-acids 147-153 and 178-183.Importantly, the first loop corresponds to one of the most divergent regions between DMC1 and RAD51 (Supplementary Figure 3), which may explain the reported differences in Ex14-and Ex27-binding of RAD51 and DMC1 (Sharan et al., 1997, Esashi et al., 2005, Thorslund et al., 2007).The second loop incudes amino-acid residues V179 (DMC1) and L180 (RAD51) that substan ally contribute to the hydrophobic pocket of the differen al PhePP-binding site (Figure 1d).

BRCA2's PhePP mo fs mediate selec ve RAD51-and DMC1-binding
We next validated the BRCA2-DMC1 structure through interac on studies.Microscale thermophoresis (MST) confirmed that DMC1 N binds to Ex14 (amino-acids 2387-2420) with a K D of approximately 30 M, which is over threefold higher affinity than Ex27 (Figure 1e and Supplementary Figure 4).Further, the binding affinity was largely retained for the truncated Ex14-Tr pep de used in crystallographic studies (Figure 1e and Supplementary Figure 4).In contrast, the interac on was largely disrupted upon introduc on of alanine muta ons of the PhePP mo f (herein referred to as Ex14-AAA; amino-acids F2406A, P2408A, P2409A) (Figure 1e).These MST results agree with biochemical pull-down assays, in which DMC1 N was shown to bind to Ex14, Ex14-Tr and BRC4, but not Ex14-AAA or Ex27 (Figure 1f).
To confirm the interac on site on the DMC1 recombinase, we used a mutant in which the loops that form the PhePP-binding site of DMC1 were replaced by the sequences present in RAD51 (herein referred to as DMC1 loop mutant, in which loops 147-GAGGYPG-153 and 178-NVDHDA-183 were mutated to IDRGGGE and GLSGSD, respec vely).The DMC1 loop mutant abolished Ex14-binding, confirming that we had iden fied the correct PhePP-binding site (Figures 1e,g).However, it also failed to interact with Ex27, indica ng that addi onal contacts must contribute to Ex27-binding by RAD51 (Figure 1g).Finally, we used a monomeric DMC1 F85E mutant to demonstrate that Ex14 only binds to oligomeric DMC1, whereas BRC4 binds to both monomeric and oligomeric forms (Figure 1h).Hence, similar to the BRCA2 Ex27-RAD51 interac on (Esashi et al., 2007, Davies andPellegrini, 2007), we find that BRCA2 Ex14 selec vely binds to oligomeric DMC1 at a site formed by residues including V179.

BRCA2 Ex14 stabilises DMC1 nucleoprotein filaments
What is the func on of the BRCA2 Ex14-DMC1 interac on?We used electrophore c mobility shi assays (EMSAs) to assay the impact of BRCA2-binding on DMC1 nucleoprotein filaments.Firstly, using unfavourable nucleoprotein filament assembly condi ons (tris, acetate, EDTA; pH 7.5), we showed that BRCA2 Ex14 promotes the forma on of DMC1-ssDNA nucleoprotein filaments (Figure 2a).We then used favourable nucleoprotein filament assembly condi ons (triethanolamine + KCl; pH 7.5) to show that BRCA2 Ex14 binds to pre-formed DMC1-ssDNA filaments, inducing a substan al super-shi (Figure 2b).Finally, we demonstrated that binding of BRCA2 Ex14 protects DMC1-ssDNA nucleoprotein filaments from BRC4-mediated disrup on (Figure 2c).Importantly, promo on of assembly, stabilisa on and protec on of DMC1-ssDNA nucleoprotein filaments were abolished by the Ex14-AAA PhePP muta on (Figure 2a-c), indica ng that these func ons depend on the interac on observed in our BRCA2-DMC1 crystal structure.Hence, BRCA2 Ex14 binds to DMC1 via the PhePP mo f, promo ng the forma on of DMC1-ssDNA nucleoprotein filaments, and protec ng filaments from BRC4-mediated disrup on, in the same way as was previously observed for the BRCA2 Ex27-RAD51 interac on (Esashi et al., 2007, Davies andPellegrini, 2007).

BRCA2 Ex27 binds to RAD51 through a conserved interac on mode
We wondered whether BRCA2 PhePP mo fs bind to RAD51 and DMC1 recombinases through the same interac on mode.Firstly, we built an Alphafold2 model of BRCA2 Ex27 bound to a RAD51 dimer (Supplementary Figure 5a-e).We then built a model of BRCA2 Ex27 bound to a RAD51 filament by docking our BRCA2-RAD51 model onto a cryo-EM structure of the RAD51 filament (PDB accession 8BSC; Appleby et al., 2023a).We no ced that the N-terminal end of the modelled BRCA2 Ex27 pep de clashed with RAD51 self-associa on interac ons, so removed this and restricted the extent of the Ex27 pep de to amino-acids 3278-3309 (Figure 3a).The resultant model predicts that BRC2 Ex27 binds via precisely the same core PhePP interface as observed in our BRCA2-DMC1 crystal structure (Figure 3b,c).Further, the model predicts that the N-terminal end of BRCA2 Ex27 binds across RAD51's selfassocia on interface, explaining how it protects filaments from BRC4-mediated disrup on (Esashi et al., 2007, Davies andPellegrini, 2007).Hence, the N-terminal end of BRCA2 Ex14 likely binds across the DMC1 self-associa on interface in a similar manner to protect from BRC repeat disrup on.
We next validated our BRCA2-RAD51 model.Pull-down assays confirmed that RAD51 binds selec vely to BRCA2 Ex27 rather than Ex14, in a manner that is disrupted by muta on of the PhePP mo f (herein referred to as BRCA2 Ex27-AAA; amino-acids F3308A, P3310A, P3311A) (Figure 3d).Further, we found that protec on of RAD51-ssDNA nucleoprotein filaments from BRC4-mediated disrup on is abrogated by the BRCA2 Ex27-AAA mutant and by an L180E muta on that targets the divergent loop of RAD51's PhePP-binding site (Figure 3e).Hence, BRCA2 Ex27 confers protec on to RAD51-ssDNA nucleoprotein filaments by binding to RAD51's PhePP-binding site in the same manner as observed in our crystal structure of BRCA2-DMC1.Thus, we conclude that BRCA2's PhePP mo fs bind selec vely to RAD51 and DMC1 nucleoprotein filaments through a conserved interac on mode that stabilises filaments from BRC4-mediated disrup on.

Discussion
The interac on mode between BRCA2's PhePP mo fs and recombinases had remained unknown.
Here, we have presented a crystal structure of the BRCA2-DMC1 complex, and solu on data, in which we reveal a novel interac on mode, which is conserved between BRCA2's Ex14 and Ex27 PhePP mo fs and recombinases DMC1 and RAD51, respec vely.Further, similar PhePP mo fs have been reported in other recombinases-interac ng proteins, including RAD51AP1 and FIGNL1 (Dunlop et al., 2011, Yuan andChen, 2013), and are observable in the sequences of other proteins such as SYCP1 (Crichton et al., 2023) and RECQL5 (Islam et al., 2012).Hence, this may be a general mechanism whereby regulatory factors associate with RAD51 and DMC1 nucleoprotein filaments.Overall, our data support a dichotomy, in which BRCA2's BRC and PhePP mo fs have dis nct roles in remodelling recombinases and stabilising nucleoprotein filaments, respec vely, in both mito c and meio c recombina on.
How does BRCA2 PhePP-binding confer stability to RAD51 and DMC1 nucleoprotein filaments?The BRCA2-DMC1 crystal structure reported herein shows only the core PhePP interface, which is distant from the recombinase self-associa on interface.Nevertheless, the BRCA2 Ex27-RAD51 model suggests that the upstream sequence binds across the self-associa on interface, which would sterically hinder access to the self-associa on interface, thereby explaining its stabilisa on and protec on from BRCmediated disrup on (Esashi et al., 2007, Davies andPellegrini, 2007).In a parallel study, a cryo-EM structure of the BRCA2 Ex27-RAD51 complex was reported (Appleby et al., 2023b), which agrees with the model presented herein.Further, in the cryo-EM structure and in our model, the posi vely charged N-terminal end of Ex27 is located close to the filament axis, where bound DNA is located.Hence, we suggest that Ex27 further stabilises RAD51 nucleoprotein filaments by its N-terminus binding directly to DNA, in keeping with recent biochemical findings (Kwon et al., 2023).It is likely that BRCA2 Ex14 confers stability and protec on to DMC1 nucleoprotein filaments in a similar manner to Ex27, through its upstream sequence binding across the self-associa on interface.
There are some notable differences in the sequences of BRCA2 Ex14 and Ex27, which must affect their func on.Firstly, Ex14 show li le conserva on outside the PhePP mo f (Thorslund et al., 2007), sugges ng that whilst it must extend across the self-associa on interface to confer protec on, it forms fewer recombinase interac ons than Ex27.Secondly, Ex14 lacks the CDK phosphoryla on site that is present in Ex27 and is responsible for disrup ng binding at the G2-M transi on (Esashi et al., 2005).
Whilst BRCA2 PhePP interac ons have thus far been considered in isola on, they likely operate as part of a wider 'recombinosome' structure.BRCA2's PhePP mo fs are located on either side of its Cterminal DNA-binding domain (Yang et al., 2002).Hence, filament-stabilising interac ons by Ex14 and Ex27 may occur coopera vely with DNA-binding and nuclea on of nucleoprotein filament forma on by BRCA2, which is known to involve displacement of ssDNA-binding protein RPA (Shahid et al., 2014, Bell et al., 2023).Further, Ex14 is located adjacent to a binding site of meio c recombina on complex MEILB2-BRME1 (Zhang et al., 2019, Zhang et al., 2020), which is proposed to displace meiosis-specific ssDNA-binding complex MEIOB-SPATA22 and act as DNA clamp (Zhang et al., 2022, Gurusaran et al., 2023).Hence, BRCA2 may coordinate the forma on of a 'meio c recombinosome' in which the key func onali es of meio c recombina on are brought together to enable the loading and stabilisa on of DMC1 nucleoprotein filaments for inter-homologue recombina on.Our understanding of meio c recombina on would be truly transformed by structure elucida on of this mul -protein assembly.

Recombinant protein expression and purifica on
The sequences corresponding to human DMC1 (amino acids 1-340) and DMC1 N (83-340) were cloned into pHAT4 (Peranen et al., 1996) for expression as a TEV-cleavable His6-tag fusion in BL21 (DE3) cells (Novagen®) in 2xYT media, induced with 0.5 mM IPTG for 16 hours at 25°C.Cells were lysed by sonica on in 20 mM Tris pH 8.0, 500 mM KCl and fusion proteins were purified from clarified lysate through consecu ve Ni-NTA (Qiagen) and HiTrap Q HP (GE Healthcare) ion exchange chromatography.Protein samples were concentrated using 10,000/3,000 MWCO centrifugal units (Amicon) and were stored at -80C following flash-freezing in liquid nitrogen.Protein samples were analysed by SDS-PAGE with Coomassie staining, and concentra ons were determined by UV spectroscopy using a Cary 60 UV spectrophotometer (Agilent) with ex nc on coefficients and molecular weights calculated by ProtParam (h p://web.expasy.org/protparam/).
Structure solu on was achieved through molecular replacement using PHASER (McCoy et al., 2007), with a single chain from pdb accession 4HYY as a search model.Model building was performed through itera ve re-building by PHENIX Autobuild (Adams et al., 2010) and manual building in Coot (Emsley et al., 2010), with the addi on of PEG ligands.The structure was refined using PHENIX refine (Adams et al., 2010), using isotropic atomic displacement parameters with two TLS groups per chain.The structure was refined against data to 2.05 Å resolu on, to R and R free values of 0.1942 and 0.2191 respec vely, with 99.07% of residues within favoured regions of the Ramachandran plot (0 outliers), clashscore of 1.28and overall MolProbity score of 0.85 (Chen et al., 2010).

Crystallisa on and structure solu on of a BRCA2-DMC1 complex (8R2G)
DMC1 N protein crystals were obtained through vapour diffusion in hanging drops, by mixing 1 µl of protein at 20 mg/ml with 1 µl of crystallisa on solu on (50 mM HEPES-NaOH pH 7.4, 50 mM MgCl 2 , 500 mM NaCl, 8 % PEG 3350) and equilibra ng at 20°C for 2 weeks.Crystals were soaked overnight in crystallisa on solu on containing 3.5 mM BRCA2 Ex14-Tr pep de (RPTKVFVPPFKTKS; synthesised by Severn Biotech).Crystals were cryo-protected using 20 % glycerol and cryo-cooled in liquid nitrogen.
X-ray diffrac on data were collected at 0.9795 Å, 100 K, as 2000 consecu ve 0.10° frames of 0.050 s exposure on a Pilatus3 6M detector at beamline I04 of the Diamond Light Source synchrotron facility (Oxfordshire, UK).Data were processed using AutoPROC (Vonrhein et al., 2011), in which indexing, integra on, scaling and merging were performed by XDS (Kabsch, 2010) and Aimless (Evans, 2011), and anisotropic correc on with a local I/σ(I) cut-off of 1.2 was performed by STARANISO (Tickle et al., 2018).Crystals belong to tetragonal spacegroup P4 1 22 (cell dimensions a = 125.40Å, b = 125.40Å, c = 364.195Å, α = 90°, β = 90°, γ = 90°), with one DMC1 octamer in the asymmetric unit.Structure solu on was achieved through molecular replacement using Phaser (McCoy et al., 2007), through placement of four DMC1 dimers from its high-resolu on structure (pdb accession 6R3P).The intact molecular replacement solu on was refined using PHENIX refine (Adams et al., 2010) and alterna ve side-chain conforma ons were removed.BRCA2 pep des were built using Alphafold2 models of a BRCA2-DMC1 1:2 complex, which were docked onto the DMC1 protomers of the octameric ring using PyMOL.The completed BRCA2-DMC1 structure was refined using PHENIX refine (Adams et al., 2010), with isotropic atomic displacement parameters, using reference model restraints from the high-resolu on DMC1 structure (pdb accession 6R3P).The structure was refined against anisotropy-corrected data with resolu on limits between 3.4 Å and 5.8 Å, to R and R free values of 0.2635 and 0.3057 respec vely, with 98.80% of residues within the favoured regions of the Ramachandran plot (0 outliers), clashscore of 2.88 and overall MolProbity score of 1.08 (Chen et al., 2010).

Electrophore c mobility shi assays (EMSAs)
To form nucleoprotein filaments, DMC1 and RAD51 was incubated with fluorescent 100-base ssDNA (5'-FAM-AATTCTCATTTTACTTACCGGACGCTATTAGCAGTGGGTGAGCAAAAACAGGAAGGCAAAATGCCGCAAAAAAG GGAATAAGGGCGACACGGAAATGTTG-3') at the concentra ons indicated in the figures for 30 minutes at 4°C in 50 mM TEA (triethanolamine) pH 7.5, 1 mM ATP, 2 mM MgCl2, 0.1 mg/ml BSA, 2 mM DTT, 200 mM KCl.For assays requiring incomplete nucleoprotein filament assembly, the reac on buffer consisted of 50 mM TAE (tris, acetate, EDTA) pH 7.5, 2 mM ATP, 2 mM MgCl2, 2 mM DTT.In nucleoprotein forma on s mula on assays or supershi experiments, BRCA2Ex14 or BRCA2Ex14-AAA were added for a further 30 min at 4°C.For protec on experiments, the DMC1, RAD51, or RAD51 L180E-DNA complexes were incubated with BRCA2 Ex14, Ex14-AAA, Ex27, or Ex27-AAA for 30 min at 4°C prior to the addi on of BRC4 and further incuba on for 30 min at 4°C (concentra ons provided in figures).Glycerol was added at a final concentra on of 3%, and samples were analysed by electrophoresis on a 1% (w/v) agarose gel in 0.5x TAE pH 8.0 at 20-40 V for ~4 h at 4 °C.DNA was detected by FAM using a TyphoonTM FLA 9500 (GE Healthcare), with 473 nm laser at excita on wavelength 490 nm and emission wavelength 520 nm, using the LPB filter and a PMT voltage of 500 V.

Protein sequence and structure analysis
Mul ple sequence alignments were generated using Jalview (Waterhouse et al., 2009), and molecular structure images were generated using the PyMOL Molecular Graphics System, Version 2.0.4Schrödinger, LLC.Electrophore c mobility shi assay (EMSAs) analysing the ability of BRCA2 Ex14 to bind, promote and protect DMC1-ssDNA filaments.(a) EMSAs using TAE pH 7.5 condi ons in which (i) DMC1-ssDNA binding is not complete (boxed), demonstra ng that (ii) BRCA2 Ex14 but not (iii) BRCA2 Ex14-AAA mutant promotes the forma on of DMC1-ssDNA filaments, using the boxed condi ons from panel (i).

Table 1 Data collec on, phasing and refinement sta s cs
*Values in parentheses are for highest-resolution shell.