Human RAD51 paralogue SWSAP1 fosters RAD51 filament by regulating the anti-recombinase FIGNL1 AAA+ ATPase

RAD51 assembly on single-stranded (ss)DNAs is a crucial step in the homology-dependent repair of DNA damage for genomic stability. The formation of the RAD51 filament is promoted by various RAD51-interacting proteins including RAD51 paralogues. However, the mechanisms underlying the differential control of RAD51-filament dynamics by these factors remain largely unknown. Here, we report a role for the human RAD51 paralogue, SWSAP1, as a novel regulator of RAD51 assembly. Swsap1-deficient cells show defects in DNA damage-induced RAD51 assembly during both mitosis and meiosis. Defective RAD51 assembly in SWSAP1-depleted cells is suppressed by the depletion of FIGNL1, which binds to RAD51 as well as SWSAP1. Purified FIGNL1 promotes the dissociation of RAD51 from ssDNAs. The dismantling activity of FIGNL1 does not require its ATPase but depends on RAD51-binding. Purified SWSAP1 inhibits the RAD51-dismantling activity of FIGNL1. Taken together, our data suggest that SWSAP1 protects RAD51 filaments by antagonizing the anti-recombinase, FIGNL1.


Introduction
Homologous recombination (HR) is essential to maintain genome stability by repairing DNA double-strand breaks (DSBs) and stalled DNA replication forks. In DSB repair, HR requires the formation of single-stranded DNA (ssDNA) through the processing of the DSB ends.
Subsequently, the ssDNA is used to find a homologous double-stranded DNA (dsDNA) and invade it to form a displacement loop called a D-loop. These homology search and strand invasion steps are catalysed by a RAD51/RecA family protein 1-3 . In eukaryotes, RAD51 binds to ssDNA in the presence of ATP to form a right-handed helical-filament on the DNA 4-7 . The RAD51 filament is a key protein machinery for the homology search and strand exchange. RAD51 filaments are highly dynamic. Defective assembly or improper assembly of RAD51 filaments leads to genomic instability. Therefore, RAD51 filament formation is tightly regulated both temporally and spatially by several positive and negative factors.
Once ssDNA is formed in vivo, the ssDNA binding protein, replication protein-A (RPA), tightly binds to the ssDNA, which inhibits the loading of RAD51. To promote RAD51 loading on RPA-coated ssDNA, RAD51 requires positive regulators called "RAD51 mediators" that act to load RAD51 on the RPA-bound DNA. RAD51 mediators, which bind directly to RAD51, include the breast cancer susceptibility gene, BRCA2, as well as RAD52 and RAD51 paralogues 8 . RAD51 paralogues share 20-30% amino acid sequence similarity with RAD51 9,10 .
RAD51 assembly and disassembly are controlled negatively and positively by various proteins. The negative regulators of RAD51 assembly include a RAD51 filament remodeller. In budding yeast, Srs2 DNA helicase has the ability to dismantle Rad51 filaments 24 . Similar to yeast Srs2, human BLM, RECQL5, FBH1, and PARI have DNA helicase activity and negatively regulate HR by disrupting the RAD51 filament 25-28 . These helicases are defined as "anti-recombinases". Although disruption of the RAD51 filament is necessary for genomic stability to prevent inappropriate HR, these anti-recombinases should be inert on active RAD51 filaments at a normal HR site. However, this highly dynamic control of the assembly/disassembly of RAD51 filament by multiple positive and negative regulators is poorly understood.
In this study, we report a novel function of the RAD51 paralogue, SWSAP1, in RAD51 assembly. Human SWSAP1 interacts with RAD51 through its conserved Phe-X-X-Ala (FxxA) BRC-like motif and this interaction is required for DNA damage-induced RAD51 focus formation. We also found that SWSAP1 interacts with FIGNL1 (fidgetin-like1), an AAA+ ATPase involved in HR 29 . We have shown that FIGNL1 depletion suppresses RAD51-assembly defects in SWSAP1-deficient cells, indicating that FIGNL1 facilitates RAD51 disassembly in the absence of SWSAP1. Notably, we found that purified FIGNL1 promotes the dissociation of RAD51 from ssDNA in an ATPase activity-independent manner. Purified SWSAP1 antagonizes the RAD51 filament dismantling activity of FIGNL1 in vitro. Based on our observations, we propose that SWSAP1 stabilizes the RAD51 filament by counteracting the FIGNL1 anti-recombinase.
Our findings reveal a novel stabilization mechanism of RAD51 filaments by a human RAD51 paralogue.

A conserved FxxA motif in SWSAP1 is crucial for RAD51 focus formation
To investigate the molecular mechanism by which SWSAP1 promotes RAD51 assembly, we examined the physical interaction between SWSAP1 and RAD51. FLAG-SWSAP1 was co-expressed with Myc-RAD51 in 293T cells and their association was confirmed by co-immunoprecipitation (co-IP) (Fig. 1a). To determine the region in SWSAP1 that binds to RAD51, we generated various truncated versions of SWSAP1 (Fig. 1b). The two middle regions (57-110 aa and 109-169 aa) of SWSAP1 are not important for the interaction with RAD51. On the other hand, a C-terminus (169-229 aa) deletion reduced the interaction and an N-terminus (1-57 aa) deletion almost abolished binding (Fig. 1d). We observed that the SWSAP1 N-terminus contains the FxxA motif, a highly conserved RAD51-binding motif 30 , which was originally identified in BRC repeats of the BRCA2 protein, referred as to BRC variant (BRCv) (Fig. 1c). We mutated phenylalanine 23 and alanine 26 on the FxxA motif to glutamate (F23E, A26E; SWSAP1-EE mutant). We also mutated a conserved 44-49 aa (PLQSMP) region downstream to the FxxA motif by substituting all 6 residues with alanine in the SWSAP1-A3 mutant (Fig. 1c). SWSAP1-EE showed reduced binding to RAD51 and the binding of the SWSAP1-A3 mutant to RAD51 in vivo was greatly abrogated relative to the wild-type SWSAP1 (Fig. 1e). These results indicate that the FxxA motif and its downstream 44-49 aa region on SWSAP1 are crucial for RAD51 interaction although we cannot exclude the possibility that these amino acid changes alter proper protein folding and activity.
To assess the effect of the SWSAP1-RAD51 interaction on RAD51 assembly, we monitored DNA damage-associated RAD51 focus formation in SWSAP1 mutants by immunofluorescence microscopy. Similar to previous observations 14 , SWSAP1 knockdown decreased the punctate staining of RAD51 foci, which was induced by the treatment of U2OS cells with camptothecin (CPT), a topoisomerase-I inhibitor (Fig. 1f). The expression of siRNA-resistant wild-type SWSAP1 suppressed the defect in RAD51 focus formation in the SWSAP1-depleted cells. However, this defect was not rescued in the SWSAP1-EE mutant (Fig. 1f). Hence, the interaction of SWSAP1 and RAD51 through the FxxA motif of SWSAP1 is required for DNA damage-induced RAD51 focus formation.

SWSAP1 interacts with FIGNL1
To gain insight into the function of SWSAP1 in HR, we looked for an interacting partner of SWSAP1 and found that SWSAP1 interacts with Fidgetin-like 1 (FIGNL1) protein. FIGNL1 is an AAA+ ATPase protein that is known to bind to RAD51 and has a role in HR 29 .
FLAG-SWSAP1 and Myc-FIGNL1 were co-expressed in 293T cells and subjected to Co-IP using an anti-Myc antibody. FLAG-SWSAP1 was recovered in the Myc-FIGNL1 immunoprecipitate from cells that expressed both proteins (Fig. 2a). This result is supported by the previous observation that human SWSAP1 peptides (C19orf39) were identified by mass spectroscopy as FIGNL1-binding proteins 29 . The interaction between SWSAP1 and FIGNL1 was also confirmed by using recombinant proteins. A His-tagged SWSAP1-SWS1 complex and GST-tagged FIGNL1∆N, in which N-terminal 284aa were deleted ( Supplementary Fig. 1a), were purified from bacteria and were used for a GST pull-down assay. The SWSAP1-SWS1 complex bound to GST-FIGNL1∆N fusion protein, but not GST alone (Fig. 2b). This result indicates that SWSAP1 directly interacts with FIGNL1 rather than a mediation by RAD51.
To identify a FIGNL1-binding site in SWSAP1, we used a series of deletion mutants of SWSAP1 (Fig. 1b, c). Co-IP revealed that the N-terminus (1-57 aa) and C-terminus (169-229 aa) of SWSAP1 are necessary for its interaction with FIGNL1 (Fig, 2c).
Since the N-terminus of SWSAP1 is also important for binding to RAD51, we examined the role of the FxxA motif of SWSAP1 in the interaction with FIGNL1. Co-IP assays revealed that SWSAP1-EE binds to FIGNL1 at a comparable level as the wild-type SWSAP1 (Fig.   2c), indicating that SWSAP1 uses a different interface for binding to FIGNL1 than for binding to RAD51. To narrow down the interaction region in SWSAP1, we generated two C-terminal truncation mutants, D4N and D4C, in which residues 169-193 and 194-229 were deleted, respectively (Fig. 2d). Reduced interaction of FIGNL1 with SWSAP1-D4C was observed, but not with the D4N mutant (Fig. 2e). Various substitutions of highly conserved amino acids in the C-terminal 35 aa region of SWSAP1 revealed that the single amino acid substitution of lysine 221 to arginine impaired the interaction of SWSAP1 with FIGNL1 ( Fig.   2f and Supplementary Figs. 2a-c).
In order to elucidate the role of the SWSAP1-FIGNL1 interaction in RAD51 assembly, we examined DNA damage-induced RAD51 focus formation. Reduced RAD51 focus formation in SWSAP1-depleted cells was suppressed by the expression of siRNA-resistant wild-type SWSAP1 but not by SWSAP1-K221R expression (Fig. 2g), suggesting that the SWSAP1-FIGNL1 interaction is critical for RAD51 assembly.

FIGNL1 depletion suppresses the RAD51 focus formation defect by SWSAP1 depletion
Next, we investigated the functional relationship between SWSAP1 and FIGNL1 in RAD51 assembly by depleting SWSAP1 and/or FIGNL1. Consistent with a previous study 29 , CPT-induced RAD51 foci appear normal in FIGNL1-depleted cells. Unexpectedly, the depletion of FIGNL1 suppressed defective RAD51 focus formation in SWSAP1-depleted cells. RAD51 focus formation in SWSAP1/FIGNL1-double depleted cells was similar to that in control and FIGNL1-depleted cells (Fig. 3a). We also verified that FIGNL1 depletion suppresses a defect conferred by the depletion of SWSAP1-binding partner, SWS1 14,15 .
Indeed, impaired RAD51 focus formation induced by SWS1 depletion was also suppressed by FIGNL1 knockdown (Supplementary Fig. 3). These data suggest that FIGNL1 could disrupt RAD51 foci, possibly as an anti-recombinase, in the absence of SWSAP1-SWS1.
In mammals, several anti-recombinases that inhibit HR have been described. These include BLM, PARI, and RTEL1 25,27 . We examined whether the depletion of other anti-recombinases suppresses the RAD51-assembly defect in SWSAP1-deficient cells.
When BLM, PARI, or RTEL1 was depleted along with SWSAP1, reduced RAD51 focus formation in SWSAP1-depleted cells was not rescued (Fig. 3a, b). These results suggest that the functional interaction of SWSAP1 is specific to FIGNL1 and not BLM, PARI, or RTEL1. SWSAP1 is one of the RAD51 paralogues required for efficient RAD51 assembly.
We extended the relationship of SWSAP1 and FIGNL1 to RAD51C, a RAD51 paralogue and a common component of the BCDX2 and CX3 complexes 13 . As previously reported, RAD51C deficiency leads to defective RAD51 focus formation 16,17 . DNA damage-induced RAD51 focus formation in RAD51C/FIGNL1-double depleted cells was almost similar to that in RAD51C-depleted cells, indicating that FIGNL1 depletion did not suppress the RAD51-assembly defect in RAD51C-deficient cells (Fig. 3c). These data suggest that FIGNL1 prevents RAD51 assembly in the absence of SWSAP1, but not in the absence of RAD51C.
Since FIGNL1 directly binds to the C-terminus of SWSAP1, we tested the role of this interaction of FIGNL1 in RAD51 assembly using the interaction-defective SWSAP1-K221R mutant. A marked increase in RAD51 assembly was observed by depleting FIGNL1 in the SWSAP1-K221R mutant, indicating that the SWSAP1-FIGNL1 interaction is required to inhibit the anti-recombinase activity of FIGNL1 (Fig. 2g).

Spontaneous chromosome fragmentations in SWSAP1-depleted cells are suppressed by FIGNL1 depletion
The restoration of RAD51 focus formation by FIGNL1 depletion in SWSAP1-depleted cells prompted us to test whether the restored RAD51 assembly possesses the ability to repair

FIGNL1 promotes RAD51 dissociation from DNAs
Suppression of the RAD51-focus formation defect in SWSAP1-deficient cells by FIGNL1 depletion raised the possibility that FIGNL1 prevents RAD51 filament formation in the absence of SWSAP1. To elucidate the role of FIGNL1 in the RAD51 filament formation, we examined the binding of RAD51 to an 83nt ssDNA immobilized to magnetic beads (Fig.4a).
We pre-incubated RAD51 with the ssDNA on beads in the presence of ATP and Mg 2+ . After the formation of the RAD51-ssDNA complex, the purified N-terminal-truncated human FIGNL1 protein (deletion of amino acids 1-284; FIGNL1ΔN) containing an ATPase domain with highly conserved Walker A/B motifs, RAD51-binding domain, and VPS4 domain 29 was added to the complex ( Supplementary Figs. 1a, b). After incubation, we separated beads containing ssDNA-bound RAD51 from the supernatant containing released RAD51 from the DNA. In the presence of ATP, RAD51 stably binds to the ssDNA without any dissociation in supernatant. When different concentrations of FIGNL1 were added to pre-assembled RAD51-ssDNA, RAD51 was recovered in the sup fractions in proportion to the FIGNL1 concentration, indicating the dissociation of RAD51 from ssDNA. At 0.5 µM of FIGNL1, almost 31.2% of the RAD51 was recovered, however, in the absence of FIGNL1 only 1.0% was recovered (Fig. 4b). Concomitantly, the amounts of RAD51 bound to ssDNA (ssDNA-bound) were reduced. These data indicate that FIGNL1 promotes the dissociation of RAD51 from ssDNA. Our in vivo and in vitro studies described here suggest that FIGNL1 is a new member of RAD51 filament remodelling proteins with anti-recombinase activity.
Ca 2+ ion is known to stabilize RAD51 filament 31,32 . We examined whether FIGNL1 is able to dissociate RAD51-ssDNA formed in the presence of ATP and Ca 2+ and found that FIGNL1 did not dissociate RAD51-ssDNA stabilized by Ca 2+ ion. This suggests that ATP hydrolysis by RAD51 might play a role in FIGNL1-mediated RAD51 remodelling ( Supplementary Fig. 5a). We also examined the effect of different lengths of ssDNA. In addition to the 83nt ssDNA, we formed RAD51 filament on 43nt and 153nt ssDNAs (same concentration in nucleotides) and examined the effect of FIGNL1 ( Supplementary Fig. 5b Fig. 1e), and promotes the dissociation of RAD51 from DNA with slight increase, indicating that the AAA+ ATPase activity of FIGNL1 is not necessary for RAD51-dismantling activity (Fig. 4c). Previously, it has been shown that FIGNL1 binds to RAD51 through the FIGNL1's RAD51 binding domain (FRBD) containing the FxxA BRCv motif. We mutated the FxxA motif (F295E, A298E; FIGNL1ΔN-EE) and examined purified FIGNL1-EE protein for RAD51 dismantling activity. Consistent with previous observations, the FIGNL1-EE mutation reduced binding to RAD51 (Supplementary Fig. 1e). FIGNL1-EE protein showed a 3.17-fold decrease in activity to dissociate RAD51 from ssDNA relative to wild-type protein (Fig. 4c). These results suggest that FIGNL1 disassembles RAD51 from DNA through its interaction with RAD51. We also measured ATP hydrolysis activity of wild-type FIGNL1ΔN, FIGNL1ΔN-KR and FIGNL1ΔN-EE proteins ( Supplementary Fig. 1g, h). Both wild-type FIGNL1ΔN and FIGNL1ΔN-EE proteins hydrolysed ATP. As expected, FIGNL1ΔN-KR proteins showed a large reduction of the ATPase activity.
To examine the effect of these mutations on in vivo FIGNL1 activity, we analysed the DNA damage-induced RAD51 focus formation in siRNA-resistant wild-type FIGNL1-, FIGNL1-EE-and FIGNL1-KR-expressing cells after double depletion of FIGNL1 and SWSAP1. The expression of wild-type FIGNL1 decreased RAD51 focus formation caused by the SWSAP1 depletion. However, FIGNL1-EE expressing cells did not show marked reduction of the RAD51-focus formation relative to wild-type FIGNL1-expressing cells, suggesting that FIGNL1-EE has a compromised ability to suppress RAD51 focus formation in vivo ( Supplementary Fig. 1i). Corresponding with results of in vitro RAD51 disassembly assay, RAD51 focus formation in FIGNL1-KR cells was slightly less than that in wild-type expressing cells, indicating the activity of FIGNL1-KR to suppress RAD51 focus formation is slightly higher than that of wild-type FIGNL1 (P-value =0.39; Supplementary Fig. 1i).
The above in vivo experiments showed that SWSAP1 protects RAD51 assembly from FIGNL1 (Fig. 3). To determine whether SWSAP1 protects RAD51 filaments from FIGNL1, we purified human SWSAP1 without its binding partner, SWS1, and examined the effect on RAD51 dismantling activity by FIGNL1. When SWSAP1 was added, most was subsequently recovered in the supernatant, indicating that the majority of SWSAP1 proteins do not bind to the ssDNA beads under these conditions. Thus, we pre-incubated SWSAP1 with FIGNL1 for 30 min and then added the mixture to RAD51-coated ssDNA beads (Fig. 4a). Only 21.2% of dissociated RAD51 was detected in the supernatant in the presence of SWSAP1, while 36.3% of RAD51 was found in the supernatant in the absence of SWSAP1 (Fig. 4d). This indicates that SWSAP1 inhibits the RAD51 filament dismantling activity of FIGNL1 when pre-incubated together.

Swsap1-deficient mice are viable and have a defect in spermatogenesis
To investigate the function of Swsap1 in vivo, we generated Swsap1 knockout mice by deleting exon 2, which encodes 82-278 aa of the SWSAP1 protein (Fig. 5a). Swsap1 -/mice were viable, showed normal growth and weight gain, and did not show any obvious developmental abnormalities ( Supplementary Fig. 4a, b). However, Swsap1 -/mice had ~1/3 smaller sizes of testis than did the wild-type and Swsap1 +/mice ( In human cells, SWSAP1 stabilizes RAD51 filaments (see above). To determine whether Swsap1 is required for RAD51 assembly during meiosis, leptotene spermatocytes, which were defined as a cell containing un-synapsed SYCP3 axes, were examined for RAD51 focus formation on nuclear spreads. A marked reduction in the number of RAD51 foci per cell was observed in Swsap1 -/spermatocytes relative to heterozygous littermates (47.1±3.3 per a spread versus 133.8±3.7, respectively) ( Fig. 5f). We also examined the focus formation of a meiosis-specific RecA homolog, DMC1, and found the reduced DMC1 focus formation (52.1±26.9 per a spread in Swsap1 -/versus 120±43.3 in its control; Fig. 5g).
These results are consistent with the recent observation 33 . We observed relatively normal γH2AX-staining in Swsap1 -/leptotene cells ( Supplementary Fig. 4e). These observations suggest that reduced RAD51 and DMC1 focus formation during meiosis is due to inefficient RAD51/DMC1 assembly in response to damage rather than defective DSB formation in the absence of SWSAP1.

Swsap1 -/cells are sensitive to DNA-damaging agents
Since Swsap1-deficient mice show a defect in RAD51 assembly during meiosis, we assessed whether Swsap1 mutants exhibit HR deficiency during the mitotic cell cycle.
Swsap1 -/immortalized fibroblasts were established and checked for DNA damage sensitivity. Swsap1 -/cells were highly sensitive to CPT, and also moderately sensitive to the DNA cross-linker, mitomycin C (MMC) (Figs. 5h, i). To assess the effect of Swsap1 deletion on RAD51 assembly in mitotic cells, we monitored RAD51 focus formation upon CPT-induced DNA damage. A reduction of approximately 25% in RAD51-foci positive cells was observed in Swsap1 -/fibroblasts relative to the wild-type control (53.4±4.3% in wild-type versus 27.7±5.8% in mutant), while γH2AX focus formation was not affected in Swsap1 -/cells (Figs. 5j, k). These results further supported the hypothesis that Swsap1 is required for DNA damage-induced RAD51 assembly during mitosis.

Discussion
HR is a highly regulated process for error-free repair of DNA damage as well as resolution of stalled DNA replication forks. HR comprises multiple pathways with different molecular mechanisms, including crossover/noncrossover and intersister/interhomolog. Inappropriate choice of the HR pathway results in genomic instability. A RecA homologue of eukaryotes, RAD51, plays a central role in homologous dsDNA searching using ssDNA and exchanging it by forming a nucleoprotein filament on ssDNA. RAD51 filament formation and RAD51 assembly/disassembly is thought to be highly dynamic and regulated by its own ATP hydrolysis cycle as well as by the interactions with various RAD51-partners. The ssDNA-binding protein RPA has an inhibitory effect on the binding of RAD51 to ssDNA.
BRCA2, whose mutations predispose to familial breast and ovarian cancers, has multiple activities including mediator activity to promote the RAD51-mediated strand exchange reaction. Human BRCA2 contains 8 repeats of BRC motif, each of which binds to RAD51 monomer and is critical for the RAD51 mediator activity 34-36 . RAD51 paralogues show amino-acid sequence similarity to RAD51. The similarity region forms a RAD51/RecA fold or core domain with or without nucleotide binding. Six RAD51 paralogues are known in humans: RAD51B, RAD51C, RAD51D, XRCC2, XRCC3 and SWSAP1. Knockdown experiments have demonstrated that these paralogues are required for DNA damage-induced RAD51 focus formation, indicating that all paralogues have the mediator activity of RAD51. However, the molecular mechanisms by which RAD51 paralogues collaborate to facilitate and/or stabilize RAD51 filament formation remain largely uncertain. In addition, the contribution of different RAD51 paralogues in the RAD51 assembly is unclear.
In this study, we found a novel activity of the RAD51 paralogue, SWSAP1, to assist in RAD51 assembly. SWSAP1 protects RAD51 filaments by inhibiting a member of a novel class of RAD51 anti-recombinase, FIGNL1 (Fig. 6).

SWSAP1 binds to 3'-end of RAD51 filament
To be polymerized on the DNAs, RAD51 possesses two different interfaces to the adjacent RAD51 monomer/protomer. One is a short β sheet located between the N-terminal domain (NTD) and ATPase core domain, referred to as "β0", and the other is β5, a β sheet on the ATPase core domain in which one RAD51 monomer forms stable binding with β0 in the adjacent RAD51 monomer at 3'-end with a polarity. Interestingly, β0 contains the conserved Phe-X-X-Ala (FxxA) sequence and is structurally similar to the BRC motif of BRCA2, which directly binds to β5 of RAD51 7,30 . In this study, we showed that the FxxA motif of SWSAP1 is critical for RAD51-binding as well as RAD51 focus formation. This suggests that SWSAP1 uses the FxxA motif for binding to the β5 of RAD51. In addition, we revealed that a second conserved motif, PLQSMP, located downstream of FxxA motif is also important for the interaction with RAD51 (Fig. 1e). Consistent with previous observations of a second interaction motif in BRC repeats 37 , downstream conserved amino acids could serve as interaction interface with a distinct pocket to RAD51.
If SWSAP1 binds to the RAD51 filament, it could be located in 3'-end of the filament which is constrained by the FxxA(β0)-β5 interaction (Fig. 6). A previous study has shown that the C. elegans RAD51 paralogue complex, RFS-1/RIP-1, binds to 5'-end of the RAD51 filament and stabilizes the complex by preventing the dissociation of RAD51 from the end 38 . We envision that the RFS-1/RIP-1 complex utilizes a β5-like motif for the binding to β0 of RAD51 on 5'-end. Based on structural analysis of budding yeast Psy3-Csm2, which contains a dimer of two structural variants of Rad51/RecA fold, we also proposed that Csm2 of the Psy3-Csm2 dimer binds to 5'-end of Rad51-filament for the stabilization of the filament 11 . Unlike the homologue of the budding yeast Csm2, it is likely that, human SWSAP1, and thus the SWSAP1-SWS1 complex, binds to 3'-end of RAD51-ssDNA filaments in order to prevent the dissociation of RAD51 from that end.

FIGNL1 is a novel anti-recombinase with RAD51 filament dismantling activity
Previous studies have identified FIGNL1 as a protein that binds to RAD51 and is involved in a step following RAD51 assembly in HR 29,39 . Here, we reveal a novel role of FIGNL1 as an anti-recombinase. We demonstrate that in vitro purified FIGNL1 promotes the dissociation of RAD51 from the RAD51-ssDNA complex. Although FIGNL1 belongs to AAA+ ATPase family, which has activity to remodel the conformation of different sets of proteins, we found that ATPase-deficient FIGNL1 has similar or slightly higher RAD51 dissociation activity than wild-type FIGNL1 (Fig. 4c, Supplementary Fig. 1i). These results indicate that ATPase activity of FIGNL1 is not critical for RAD51-dismantling activity. Given slightly higher activity of FIGNL1-KR than the wild-type, ATP hydrolysis of FIGNL1 may modulate its RAD51-dismantling activity. Based on the crystal structure analysis of human FIGNL1 dimer (PDB:3D8B), the interface between human FIGNL1 monomers sandwiches ATP/ADP molecule, implying ATP-binding stabilizes dimer formation of FIGNL1 and facilitates its RAD51 dissociation activity.
Our observations suggest a novel mechanism by which FIGNL1 dismantles RAD51 from RAD51 filaments. We found that the RAD51 interaction domain (FxxA motif) of FIGNL1 is critical for this activity. Since RAD51 dissociation by FIGNL1 was eliminated by preventing ATP hydrolysis of RAD51, the binding of FIGNL1 to RAD51 may stimulate ATP-bound RAD51 to hydrolyse ATP to ADP, converting RAD51 to a low-affinity DNA-binding status. Taken together, we conclude that FIGNL1 is a new type of anti-recombinase, which disassembles RAD51 filaments from ssDNA in an ATPase activity-independent manner. Alternatively, FIGNL1, which binds to RAD51 in solution, may affect the equilibrium between association and dissociation states of RAD51 to the DNA (Fig. 6).
Whereas FIGNL1 disassembles RAD51 filaments from ssDNA, its activity is relatively low compared to other anti-recombinases 24-28 . This low RAD51 filament disruption activity of FIGNL1 could be explained by the absence of its N-terminal regions (1-284aa). The N-terminal region of FIGNL1 may enhance this activity. Furthermore, SPIDR and FLIP, FIGNL1-binding proteins, interact with the N-terminus of FIGNL1 29, 41,42 and could support the full activity of FIGNL1.
Previously, several proteins that dismantle RAD51 filaments have been identified.
S. cerevisiae Srs2 DNA helicase is a primordial example of this class, which extends to BLM, RECQL5, PARI and FBH1 helicases. Importantly, all these factors are DNA helicases that have DNA binding activity and translocate on DNA molecules. It is shown that the ATPase activity of these helicases is critical for the dismantling activity of RAD51 filaments.
Given that the ATPase motif of FIGNL1 is dispensable for its RAD51 filament disruption activity and major fraction of FIGNL1 does not bind to DNA, FIGNL1 belongs to a novel class of the RAD51 anti-recombinase since it does not work as an enzyme but may act in a stoichiometric manner.
A previous study showed that FIGNL1 depletion causes a defect in HR using DR-GFP system without affecting RAD51 focus formation 29 , suggesting that FIGNL1 is involved in a later step of HR. We speculate that, similar to Srs2 {Elango, 2017 #65}, FIGNL1 could optimize RAD51 filament on the DNA, which is suitable for efficient homology search and strand exchange etc. In other words, FIGNL1 may positively control the recombination by disrupting improper RAD51 ensembles.

SWSAP1 antagonizes FIGNL1's RAD51 filament dismantling activity
We found that human FIGNL1 binds to human SWSAP1. In vitro, pre-incubation of SWSAP1 with FIGNL1 attenuates the RAD51 dismantling activity of FIGNL1, indicating that SWSAP1 antagonizes the anti-recombinase activity of FIGNL1. Consistent with this in vitro observation, the depletion of FIGNL1 in human cells suppresses a defect in the formation of DNA-damage-induced RAD51 foci in SWSAP1-depleted cells. These data simply show that SWSAP1 protects RAD51 foci from FIGNL1. We propose that SWSAP1 is a novel type of human RAD51 paralogue, which regulates RAD51 filament formation by antagonizing the anti-recombinase, FIGNL1. Previously, the budding yeast Rad51 paralogue complex, Rad55-57, was shown to protect yeast Rad51 filaments from the Srs2 anti-recombinase 43 .
Among the RAD51 paralogues, one class promotes RAD51 assembly by competing with the anti-recombinase. Interestingly, the protection function of SWSAP1 is very specific to FIGNL1 and is not observed against other anti-recombinases such as BLM, PARI, and RTEL1. Conversely, the other RAD51 paralogues such as RAD51C cannot antagonize the FIGNL1 activity. Previous observations that FIGNL1 does not bind to any RAD51 paralogues suggests that SWSAP1 functions in an HR pathway distinct from BCDX2 and CX3 complexes 29 . In the future, we may expand the list of interactions between RAD51 mediators and anti-recombinases.
The dismantling activity of FIGNL1 requires the RAD51-binding motif, FxxA on its own and the RAD51 mediator activity of SWSAP1 needs the FxxA motif by itself. Simply, SWSAP1 competes for the β5 motif of RAD51 with FIGNL1. If SWSAP1 has a higher affinity to RAD51 than FIGNL1, this scenario is possible. However, we show that the interaction between SWSAP1 and FIGNL1 is also important for the anti-FIGNL1 activity of SWSAP1. Therefore, the association of SWSAP1 with FIGNL1 inhibits the activity of FIGNL1 either in solution or on the end of the RAD51 filaments. Further study is required to elucidate the mechanisms for how SWSAP1 protects RAD51 filaments from FIGNL1, particularly considering the SWSAP1-binding partner, SWS1, as well as the FIGNL1-binding proteins, SPIDR and FLIP.

RAD51 filament formation by various sets of positive and negative regulators
Our results suggest that FIGNL1 inhibits inappropriate HR and SWSAP1 ensures HR when it is necessary. Alternatively, FIGNL1 may promote the fine-tuning of RAD51 assembly by working with SWSAP1. Our results, together with other studies, show that RAD51 filaments are highly dynamic and regulated by different combinations of positive and negative RAD51 factors, which may ensure the proper functional assembly on DNA substrates under different cellular conditions. Differential dynamics of RAD51 filaments may affect the fate of recombination intermediates into differential outcomes. In Arabidopsis, FIGNL1 has anti-crossover activity during meiosis, possibility by limiting RAD51 filament assembly 39, 44 .
In this case, FIGNL1 is not an anti-recombinase, but rather a pro-noncrossover recombinase. By balancing the activity of RAD51 mediators and remodellers under different cellular contexts, a proper recombination pathway may be chosen by controlling RAD51 filament dynamics.

Mouse Swsap1 regulates meiotic recombination during spermatogenesis
Unlike other RAD51 paralogue (Rad51b, Rad51c, Rad51d and Xrcc2) knockout mice that are embryonic lethal 8,19-23 , Swsap1 -/mice are viable, indicating that Swsap1 is dispensable for embryonic development. Mild reduction of CPT-induced RAD51 focus formation in Swsap1 -/cells relative to the control could explain the dispensability of Swsap1 for embryonic development. It suggests that, although Swsap1 is critical for the stabilization of RAD51 filament in mitosis, there may be other factors to compensate for the absence of SWSAP1 in RAD51 assembly (Fig. 5j). Whereas embryonic development is unaffected in Swsap1 -/mice, 99.0% abnormal seminiferous tubules and lack of developing ovarian follicles in Swsap1 -/- (Figs. 5d, e) imply that Swsap1 is essential for meiosis and probably meiotic recombination. Indeed, Swsap1 -/mice showed a decreased number of RAD51 and DMC1 foci (Figs. 5f, g). These observations raise the possibility that Swsap1 is a more specialized RAD51 paralogue for meiotic recombination than for mitotic HR. In addition, FIGNL1, a SWSAP1-binding protein, has been shown to limit crossover formation during meiosis in Arabidopsis and is highly expressed in mouse spermatocytes 39, 44 . The formation of crossovers, which are essential for the proper segregation of homologous chromosomes, is highly regulated during meiosis. The SWSAP1-FIGNL1 axis is critical to ensure meiotic HR and, especially, CO formation.

Immunoprecipitation (IP)
After washing with PBS, cells were resuspended in 500 µl of benzonase buffer (20 mM

Western blot
IP and WCE samples were separated by 12% SDS-polyacrylamide gels and transferred onto a PVDF membrane (Millipore). The blots were blocked with 5% milk in TBST for 30 min and incubated with primary antibody overnight and secondary antibody for 30 min.
Proteins were detected by alkaline phosphatase kits (Nacalai).

Clonogenic survival
SV40-immortalized mouse fibroblasts prepared as described below were seeded on 10cm-dishes. After 8 hours, cells were treated with indicated concentration of camptothecin (Wako) for 22h and mitomycin C (Nacalai), continuously. Ten days after treatment, cells were stained in 4% crystal violet (Sigma). The number of colonies was counted.

Immunofluorescence staining
Cells were cultured on coverslips in the presence or absence of 100 nM camptothecin for 22 h. Cells were permeabilized with CSK buffer (10 mM PIPES at pH 6.8, 100 mM NaCl, 300 mM Sucrose, 3 mM MgCl 2 , 1 mM EGTA, 0.5% Triton X-100, 1x protease inhibitor cocktail, 1x phosphatase inhibitor) for 5min on ice and fixed with 2% PFA (Sigma-Adrich) for 15min at room temperature. The coverslips were blocked in TBST containing 3% BSA.
Subsequently, coverslips were incubated with primary antibodies in TBST containing 3% BSA for overnight, washed 3 times with TBST and secondary antibodies for 1h. After washing with TBST, the coverslips were mounted with Vectashield media (Vector Laboratories).

Spermatocyte spreads
Testis was incubated in 2 ml testis isolation medium (104mN NaCl, 45mM KCl, 1.2mM MgSO 4 , 0.6mM KH 2 PO 4 , 0.1% glucose, 6mM sodium lactate, 1mM sodium pyruvate) containing 2 mg ml -1 collagenase (Worthington) at 32°C for 55 min 46 . Subsequently, testis was treated with 0.7 mg ml -1 trypsin (Sigma) and 4 µg/ml DNaseI (Roche) at 32°C for 15 min. The reaction was stopped by adding 20 mg ml 1 trypsin inhibitor in testis isolation medium (Sigma). The resultant suspension was filtered through a 70-µm cell strainer (Corning). After washing with testis isolation medium, cells were resuspended in 0.1 M sucrose and applied to a spot containing 1% PFA on glass slide. The slides were dried under moist condition for 2.5 h and dry condition for 1 h. After rinsing with DRIWEL (FujiFilm), slides were stored at -80°C. Bacterial pellets were resuspended in buffer-A and sonicated. Lysates were filtered and applied to GSTrap column on an ÄKTA Pure system. GST-SWSAP1 was eluted with glutathione containing buffer-A and treated with thrombin overnight to remove GST tag.

FIGNL1ΔN (N-terminal 284aa deletion) sequence was inserted into
After thrombin treatment, SWSAP1 protein was purified with HiTrap Q (GE healthcare) with a gradient of NaCl and concentrated by Amicon Ultra 10K centrifugal filter unit (Millipore).
Protein concentration was determined by Bradford method (BioRad) using BSA as a standard.

GST pulldown assay
GST pulldown was performed using

Metaphase spreads
Metaphase spread were performed as previously described 47   E. coli extracts expressing GST or GST-FIGNL1ΔN were incubated with glutathione beads. After wash, the beads were incubated with purified SWSAP1-SWS1 complex. After collection of the beads, samples were eluted with glutathione and subjected to western blotting. c, Co-immunoprecipitation analysis of SWSAP1 mutants. Myc-FIGNL1 and indicated FLAG-SWSAP1 truncation mutants were expressed in 293T cells and, after 72 h transfection, extracts were used for IP and analyzed by western blotting with indicated antibodies. d, Schematic representation of SWSAP1 C-terminal truncations and K221R mutants. e-f, Co-immunoprecipitation analysis of SWSAP1 C-terminus mutants with FIGNL1. FLAG-SWSAP1 C-terminal truncation and FLAG-SWSAP1-K221R mutants were co-expressed with Myc-FIGNL1 in 293T cells and, after 72 h, transfection, WCE was used for IP using anti-Myc to detect the binding to Myc-FIGNL1. g, Quantification of RAD51-positive cell in SWSAP1-K221R mutants. U2OS cells with siRNA transfection with or without expression of siRNA-resistant SWSAP1 or SWSAP1-K221R, the cells were treated with 100 nM of CPT for 22 h, RAD51-focus formation was analyzed as shown in Fig. 1f. Data are mean ± s.d.; Statistics and reproducibility, see accompanying Source data.    -/-+/-forward, β-geo-specific forward and reverse primers were used. Wild-type and mutant genes show 3kb and 0.75kb fragments, respectively. c, Swsap1 testis images. Representative images of Swsap1 +/+ and Swsap1 -/testis are shown. d, Testis sections of wild-type and Swsap1 mutant mice. Cross sections of fixed testis were stained with HE. Top: Representative images of Swsap1 +/+ and Swsap1 -/seminiferous tubules are shown. Bottom, quantification of atrophic tubules is shown. Ninety tubules in Swsap1 +/+ and 101 tubules in Swsap1 -/were examined. e, Cross sections of fixed ovary were stained with HE. f, RAD51 and SYCP3 immunofluorescence analysis of leptotene spermatocytes. Chromosome spreads were prepared and stained for RAD51 and SYCP3. Left, Representative images of Swsap1 +/and Swsap1 -/spermatocyte spreads are shown. Right, Quantification of RAD51 foci in leptotene spermatocytes. A number of RAD51 foci was counted per a nucleus. Data are mean ± s.d.; Statistics and reproducibility, see accompanying Source data. g, DMC1 and SYCP3 immunofluorescence analysis of leptotene spermatocytes. Chromosome spreads were prepared and stained for DMC1 and SYCP3. Left, Representative images of  RecA/RAD51 fold ATPase core (α/β) Fig. 6