Rtt105 stimulates Rad51-ssDNA assembly and orchestrates Rad51 and RPA actions to promote homologous recombination repair

Significance

Homologous recombination (HR) is a fundamental process of life and an essential mechanism for maintaining genome stability and avoiding cancer. Central to HR repair is the prompt assembly of Replication Protein A (RPA) on resected single-strand DNA (ssDNA), followed by the timely replacement of RPA by Rad51 recombinase to form Rad51 filament. How cells drive dynamic Rad51 assembly and coordinate these steps remains unclear. In this study, we identified Rtt105, a Ty1 transposon regulator, as a Rad51 regulator promoting dynamic Rad51-ssDNA assembly. Importantly, we showed that Rtt105 physically interacts with both Rad51 and RPA and plays a unique role in orchestrating their actions to ensure proper Rad51 nucleation and HR repair. Our work provides insights into the regulation and coordination of HR repair.

Abstract

Homologous recombination (HR) is essential for the maintenance of genome stability. During HR, Replication Protein A (RPA) rapidly coats the 3′-tailed single-strand DNA (ssDNA) generated by end resection. Then, the ssDNA-bound RPA must be timely replaced by Rad51 recombinase to form Rad51 nucleoprotein filaments that drive homology search and HR repair. How cells regulate Rad51 assembly dynamics and coordinate RPA and Rad51 actions to ensure proper HR remains poorly understood. Here, we identified that Rtt105, a Ty1 transposon regulator, acts to stimulate Rad51 assembly and orchestrate RPA and Rad51 actions during HR. We found that Rtt105 interacts with Rad51 in vitro and in vivo and restrains the adenosine 5' triphosphate (ATP) hydrolysis activity of Rad51. We showed that Rtt105 directly stimulates dynamic Rad51-ssDNA assembly, strand exchange, and D-loop formation in vitro. Notably, we found that Rtt105 physically regulates the binding of Rad51 and RPA to ssDNA via different motifs and that both regulations are necessary and epistatic in promoting Rad51 nucleation, strand exchange, and HR repair. Consequently, disrupting either of the interactions impaired HR and conferred DNA damage sensitivity, underscoring the importance of Rtt105 in orchestrating the actions of Rad51 and RPA. Our work reveals additional layers of mechanisms regulating Rad51 filament dynamics and the coordination of HR.

Homologous recombination (HR) is a fundamental process of life and a major driving force in evolution (1). It is essential for the protection and rescue of stalled or collapsed replication forks, the repair of chromosome breaks, and the exchange of genetic information during meiosis (2–6). Mice lacking key HR genes are embryonic lethal (7). Mutations in HR genes often lead to genomic instability and a predisposition to breast, ovarian, and prostate cancers, or other diseases (7, 8).

HR is initiated with the processing of the 5′-ends of DNA double-strand breaks (DSBs) by nucleases, yielding 3′-tailed single-strand DNA (ssDNA) (9–12). The exposed 3′-ssDNA tracts are promptly coated by the conserved ssDNA-binding protein complex Replication Protein A (RPA). RPA coated on ssDNA is then replaced by Rad51 recombinase with the aid of mediator proteins to form a helical nucleoprotein filament. The Rad51-ssDNA presynaptic filament serves as the key intermediate driving homology search and strand invasion (1–3, 5, 7, 11–13). Within the filament, DNA is extended 1.5-fold over the dsDNA contour length, with a stoichiometry of 3 nt per Rad51 monomer and six Rad51 monomers per helical turn (14, 15). A catalytically competent Rad51 filament initiates pairing and strand exchange with a homologous sequence with more than 8-nt contiguous homology tract. Expansion of the recognition forms a displacement loop (D-loop) (15, 16). DNA polymerases then extend the 3′-invading ends within the D-loop to synthesize new DNA. Finally, the intermediate is channeled into one of the alternative resolution pathways to complete repair (2–5, 12).

The concerted action of RPA and Rad51 is crucial for proper Rad51 filament formation, homology search, strand invasion, and subsequent repair. Yeast Rad52 and human BRCA2 act as the primary mediator proteins facilitating Rad51 nucleation on ssDNA coated with RPA that imposes an inhibitory effect on Rad51 loading (12, 14, 17–29). While the Rad51 paralogs, including Rad55-Rad57 and the Shu complex in yeast and RAD51B, RAD51C, RAD51D, XRCC2, and XRCC3 in humans, and the recently identified E3 ubiquitin ligase Bre1(RNF20 in human) implement subsidiary roles in enhancing Rad51 nucleation and filament formation (12–14, 27, 28, 30–33). How cells drive dynamic Rad51-ssDNA assembly and orchestrate Rad51 and RPA actions to ensure proper HR remains unclear.

Rtt105, a yeast Ty1 transposon regulator, has recently been recognized as an RPA chaperone playing important roles in preserving genome integrity (34–37). Specifically, Rtt105 interacts with RPA and promotes RPA nuclear import and binding to ssDNA on chromatin (34–36). As a result, Rtt105 is required to promote accurate DNA replication and HR repair, to stabilize replisome under replication stresses, and to facilitate heterochromatin and telomere maintenance (34–37). Unexpectedly, by mass spectrometry analysis, we identified that Rtt105 also functions as a Rad51 regulator. We found that Rtt105 physically interacts with Rad51 and directly stimulates dynamic Rad51 assembly, strand exchange, D-loop formation, and HR repair. Importantly, we found that the regulations of Rtt105 on both RPA and Rad51 are epistatic and necessary for proper Rad51 nucleation, strand exchange, and HR repair. Our study reveals a different layer of control on Rad51 filament dynamics and sheds light on the importance of Rtt105 in regulating concerted actions of RPA and Rad51 in the repair of DNA breaks.

Results

Rtt105 Interacts with Rad51 In Vitro and In Vivo.

To fully understand the role of Rtt105 in preserving genome integrity, we attempted to identify new potential Rtt105-interacting proteins. We incubated purified glutathione S-transferase (GST) or GST-tagged Rtt105 protein with cell lysates derived from yeast treated with methyl methane sulfonate (MMS), a DNA alkylating agent that can induce DSBs. Subsequently, we performed GST pull-down assays and mass spectrometry analysis. GST was used as a negative control. Notably, besides the known Rtt105-interacting partners, such as RPA and the importin Kap95 (35, 36, 38), several recombination proteins, including Rad51, Rad52, and Mms22, were also identified among the top hits (Fig. 1A). In this study, we focus on the interaction between Rtt105 and Rad51. We found that Rtt105 interacted with Rad51 in vivo, and this interaction was increased from the S through G2/M phases (Fig. 1B), correlating with the cell cycle-dependent fluctuation of Rtt105 protein level (35). Indeed, purified recombinant GST-Rad51, but not GST itself, interacted with His-Rtt105 in vitro, as reflected by GST pull-down assays (Fig. 1C). Interestingly, the addition of ssDNA, but not dsDNA, augmented the interaction in vitro (Fig. 1 D and E, lanes 4-6 and 10-12), suggesting that the interaction could occur in the chromatin context with exposed ssDNA.

Fig. 1.

Rtt105 interacts with Rad51 in vivo and in vitro. (A) Table showing the potential Rtt105-interacting proteins identified by mass spectrometry analysis. The unique peptide number for each protein is indicated. The score is an indicator of reliability. (B) Immunoprecipitation and Western blot analysis of the interaction between Rad51-3xHA and Rtt105-3xFLAG at indicated conditions. GAPDH serves as a loading control. Yeast cells were treated with α-factor (5 µg/mL), hydroxyurea (200 mM), or nocodazole (20 µg/mL) for 2 h to synchronize cells in the G1, S, and G2/M phases, respectively. (C) A GST pull-down assay showing the interaction between GST-Rad51 and His-Rtt105. GST was used as a negative control. The Bottom panel shows Coomassie blue staining of purified GST or GST-Rad51 used for the experiment. The Upper panel is the Western blot showing the pull-down products. (D and E) A GST pull-down assay and its quantification showing the effect of ssDNA or dsDNA on stimulating the interaction between Rad51 and Rtt105. Purified GST-Rtt105 and His-Rad51 were used for the experiment. (F) Scheme showing the full-length or truncated Rtt105 proteins used for pull-down assays. Dot lines represent deleted regions. The positions for these truncations are indicated. (G and H) GST pull-down assays showing the interactions between GST-Rad51 and His-tagged WT or mutant Rtt105 proteins. Cell lysates with His-tagged Rtt105 truncations (G) or purified His-tagged Rtt105 proteins (H) were used for the experiments. (I) Immunoprecipitation showing the interaction between Rad51-3xHA and Rtt105-3xFLAG in indicated strains. Values represent the mean ± SD of three independent experiments (n = 3). Statistical analysis was calculated with Student’s t test. **P < 0.01.

The Rtt105 Residues Y²⁰³ and P²⁰⁴ Are Critical to Mediate the Rtt105-Rad51 Interaction.

To delineate the Rtt105 motif required for the interaction with Rad51, we expressed a series of progressively truncated His-Rtt105 mutant proteins and tested their interactions with GST-Rad51 by pull-down assays (Fig. 1F). We noted that the truncated Rtt105 with residues up to 1 to 200 failed to interact with Rad51, while the one with residues 1 to 204 could efficiently bind Rad51 (Fig. 1G, lanes 6-10), suggesting that the residues between 201 and 204 (NSYP) are important for the interaction. Notably, simultaneous mutation of the residues tyrosine 203 and phenylalanine 204 to alanine (YP2A) was sufficient to completely abolish the Rtt105-Rad51 interaction in vitro and in vivo (Fig. 1 H and I). Thus, we identified Rad51 as a Rtt105-interacting protein and demonstrated that the residues Y²⁰³ and P²⁰⁴ of Rtt105 are crucial for mediating the interaction.

The Rtt105-Rad51 Interaction Promotes DNA Damage Response and Intra-/Interchromosomal HR Repair.

We then assessed the function of the Rtt105-Rad51 interaction in the DNA damage response and repair. We observed that specific disruption of the interaction (YP2A) conferred sensitivity to phleomycin (PHL) and MMS that can induce DSBs, but not to camptothecin (CPT) or hydroxyurea (HU) that gives rise to replication stresses (Fig. 2A). Consistently, the YP2A point mutant exhibited a defect in the recovery from short phleomycin exposure, as reflected by the formation and removal of Rad52-GFP foci, an indicator of DNA lesion (SI Appendix, Fig. S1 A and B). This result was confirmed by monitoring the integrity of chromosomes using pulse-field gel electrophoresis (SI Appendix, Fig. S1C). These results indicate an indispensable role of the Rtt105-Rad51 interaction in promoting DNA damage response and repair.

Fig. 2.

The interaction between Rtt105 and Rad51 promotes DNA damage response and HR repair. (A) Spotting assays showing the DNA damage sensitivity for indicated cells at indicated drug concentrations. (B) Scheme showing a haploid ectopic recombination system (MK203). (C) Survival rate of DSB repair by ectopic recombination for indicated strains. (D and E) PCR amplification and quantification showing DSB repair kinetics for indicated strains. (F) Schematic representation of a haploid yeast ectopic repair system (NA14, NA29). The DSB can be repaired by intrachromosomal (IGC) or interchromosomal ectopic recombination (EGC), or single-strand annealing(deletion). (G and H) Graphic representation of the survival rate (G) and proportion (H) of repair products for indicated cells in either NA14 or NA29 background. Values represent the mean ± SD of three independent experiments (n = 3). Statistical analysis was calculated with Student’s t test. *P < 0.05 and **P < 0.01. ns, no significance.

Next, we assessed the role of their interaction in HR repair using a haploid system (MK203) wherein a single HO-induced DSB is generated on chromosome V upon galactose induction, and the DSB is repaired by HR using the homologous template on chromosome II as donor (Fig. 2B) (39). Each strain bears two copies of the URA3 gene. One copy is located on chromosome V and contains an HO cut site, while the other is located on chromosome II and contains a BamHI and an EcoRI cut site (Fig. 2B) (39). This allowed us to follow the repair products by restriction digestion. Approximately 80% of wild-type (WT) cells completed the repair and survived, while only ~50% of YP2A cells and ~15% of rtt105Δ cells survived (Fig. 2C). Furthermore, the YP2A mutant cells completed the repair with slower kinetics, and the defect was more striking in rtt105Δ cells (Fig. 2 D and E). Thus, the Rtt105-Rad51 interaction promotes interchromosomal ectopic recombination repair.

To verify this result, we employed another haploid repair system(NA14) in which a single HO cut is induced at the URA3 locus inserted on chromosome V that can be repaired by intrachromosomal recombination or single-strand annealing (SSA) using the homologous template (3-kb away) on the same chromosome, or by interchromosomal recombination using the donor sequence on chromosome II as template (Fig. 2F) (40). The repair outcome can be distinguished by following the selection markers. ~80% of the WT cells completed the repair, while only ~50% of YP2A or rtt105Δ mutant cells survived (Fig. 2G). Notably, approximately 52% of WT cells were repaired by intra- or interchromosomal gene conversion, and 28% were repaired by SSA. In contrast, the repair by gene conversion was reduced to ~23% in both YP2A and rtt105Δ cells, while the repair by SSA remained unaffected (Fig. 2H). A similar result was obtained when the two URA3 sequences on chromosome V were separated by a 5.5-kb intervening sequence(NA29) (Fig. 2 G and H) (40). These results demonstrate that the Rtt105-Rad51 interaction is important for intra- and interchromosomal ectopic recombination in haploid cells. Interestingly, the rtt105Δ mutant displayed a lower survival rate in the MK203 than in the NA14 or NA29 strain, likely due to the lack of alternative repair pathways in MK203.

The Rtt105-Rad51 Interaction Affects Recombination Partner Choice.

Next, we analyzed the role of the interaction on HR repair in a diploid strain (MK235) wherein a single HO-induced DSB at the ura3-HOcs locus on chromosome V can be repaired by HR with either the allelic URA3 gene (chromosome V) or the ectopic ura3-HO-inc homologous sequence (chromosome II) as donor (SI Appendix, Fig. S2A) (39). The ectopic donor sequence contains a BamHI and an EcoRI cut site, which allowed us to trace the donors of repair products with restriction digestion. In both cases, the resulting cells cannot grow on plates containing the uracil analog 5-fluoroorotic acid (5-FOA), a poison killing uracil autotrophic cells. However, mitotic recombination and mutation in the heterozygotes can give rise to 5-FOA-resistant colonies, which indicates the repair by ectopic recombination (39).

Both the WT and the isogenic YP2A or rtt105Δ cells were fully proficient in repairing the HO-induced DSB and survived (SI Appendix, Fig. S2B). Notably, among the WT survivors, we detected a significant population of cells that were repaired by ectopic recombination using the homology sequence on chromosome II as a donor, as they were able to grow on 5-FOA selection plates (SI Appendix, Fig. S2C). This was validated by restriction digestion of the repair products (SI Appendix, Fig. S2D). Strikingly, both YP2A and rtt105Δ diploid strains were completely deficient in ectopic recombination in the presence of the allelic donor (SI Appendix, Fig. S2 C and D). This is likely because ectopic recombination has slower repair kinetics and a higher demand for dynamic Rad51 binding. Thus, the Rtt105-Rad51 interaction is critical for ectopic recombination in diploid cells and is likely required for sampling alternative recombination partners.

The Rtt105-Rad51 Interaction Stimulates Rad51 Loading and HR Repair Independently of Rad52 or Rad57.

To explore how the Rtt105-Rad51 interaction may impact HR repair, we first examined whether Rtt105 affects Rad51 nuclear import, as it does for RPA (35, 36). We transformed a high-copy plasmid expressing Rad51-GFP to WT or rtt105Δ cells and observed that Rad51 was properly localized to the nucleus in both populations of cells (SI Appendix, Fig. S3), indicating that Rtt105 is not required for Rad51 nuclear import.

Next, we tested Rad51 loading at DSBs in an unrepairable HO-cut system (Fig. 3A) (41, 42). We noted that Rad51 loading was significantly decreased in the YP2A point mutant, compared to that in the WT cells (Fig. 3A). This is not attributed to any alterations in Rad51 or Rtt105 protein levels, as they remained unaffected (SI Appendix, Fig. S4 A and B). Moreover, YP2A cells had a normal rate of DNA end resection (SI Appendix, Fig. S4 C and D). As a control, rtt105Δ cells exhibited a slight defect in resection at distal ends (4.8 kb). These results suggest that the interaction between Rtt105 and Rad51 facilitates Rad51 loading on ssDNA.

Fig. 3.

Rtt105 promotes dynamic Rad51-ssDNA assembly. (A) ChIP-qPCR analysis of Rad51-3xFLAG loading at 1, 5, or 10 kb away from the HO cut site. (B) Scheme showing the ssDNA pull-down assays. (C and D) An ssDNA pull-down assay and quantification showing the effect of Rtt105 or YP2A on His-Rad51 binding to ssDNA. (E) An electrophoretic mobility shift assay showing the effect of Rtt105 protein on Rad51 binding to M13 mp18 circular ssDNA. (F) An ssDNA pull-down assay followed by Western blot analysis indicating the presence of Rtt105 in His-Rad51-ssDNA complexes. (G) A single-molecule magnetic twister system for monitoring the kinetics of Rad51 assembling on ssDNA. (H) Single-molecule MT analysis showing the effect of Rtt105 or YP2A on Rad51 assembly with ssDNA. The Rad51-ssDNA filaments were assembled at 100 nM Rad51 with or without 20 nM Rtt105. (I) The effect of Rtt105 or YP2A on the ATP hydrolyzing activity of Rad51. Values represent the mean ± SD of three independent experiments (n = 3). Statistical analysis was calculated with Student’s t test. *P < 0.05 and **P < 0.01. ns, no significance.

Rad52 is the key mediator promoting Rad51 nucleation on RPA-coated ssDNA. The rad52-Y376A mutation abolishes the Rad52-Rad51 interaction and its mediator function (43). The Rad55-Rad57 complex acts in an epistatic manner to Rad52 (22–24, 31, 32). Both rad52-Y376A and rad57Δ single mutants exhibited an obvious defect in Rad51 loading and HR repair (SI Appendix, Fig. S5 A and B). However, additional mutations of YP2A in these cells further aggravated the defects. Consistently, combined mutation of YP2A in rad52Δ or rad57Δ cells exacerbated the DNA damage sensitivity (SI Appendix, Fig. S5C). These results imply that Rtt105 acts independently of Rad52 or Rad57 in promoting Rad51 loading and HR repair.

Rtt105 Directly Promotes Dynamic Rad51 Assembly on ssDNA.

To test whether Rtt105 directly regulates Rad51 filament formation, we incubated purified Rad51 with ssDNA (90 nt) immobilized to streptavidin-coated magnetic beads and carried out ssDNA pull-down assays, as described previously (Fig. 3B) (12). We noted that the addition of purified Rtt105, but not the rtt105-YP2A mutant protein (YP2A), stimulated Rad51 binding on ssDNA (Fig. 3 C and D). However, Rtt105 did not affect Rad51 binding to dsDNA (SI Appendix, Fig. S6 A–C). It is likely that Rtt105 binding causes conformational alterations of Rad51 that specifically favor the access of ssDNA. Consistently, the inclusion of Rtt105 stimulated the formation of Rad51-ssDNA complexes that exhibited retarded mobilities on gels, as reflected by the electrophoretic mobility shift assay (EMSA) (Fig. 3E, lanes 3-6). As previously noted, Rtt105 itself does not bind ssDNA (SI Appendix, Fig. S6D) (35, 36). However, it was detected in ssDNA pull-down products in the presence of Rad51 (Fig. 3F), suggesting that Rtt105 can at least transiently bind the Rad51-ssDNA filament.

We further employed single-molecule magnetic tweezers (MT) to monitor the impact of Rtt105 on the dynamics of Rad51-ssDNA assembly using a previously described method (12). We labeled 12.5-kb ssDNA molecules with digoxigenin and biotin at the 5′- and 3′-ends, respectively, and stretched the ssDNA molecules using MT at 8 pN, 22 °C, and pH 7.5 (Fig. 3G) (12). Assembly of Rad51-ssDNA filaments leads to an extension of the ssDNA that can be monitored using MT. Average values from multiple ssDNA molecules were calculated and plotted as a time-extension curve for each condition. The addition of Rad51 alone to ssDNA in the presence of 1 mM ATP led to a ~1.5-fold extension in ssDNA length (from ~3,000 nm to ~4,500 nm), consistent with the observation in Rad51 structural studies (15). Notably, the inclusion of Rtt105 in the reaction dramatically accelerated the assembly kinetics, shortened the time by threefold (from 4,500 s to 1,500 s), and resulted in further extension of ssDNA (Fig. 3H). In contrast, the inclusion of the YP2A mutant protein did not affect Rad51 assembly (Fig. 3H). As a control, both WT and mutant Rtt105 proteins themselves did not alter ssDNA length (SI Appendix, Fig. S7A). Thus, Rtt105 stimulates dynamic Rad51 assembly via direct physical interaction.

To test whether the stimulatory effect fits a distributive property within the filament, we performed stopped-flow fluorescence experiments following a previous method (44). Rad51 was premixed with different concentrations of Rtt105 before adding the 5′-Cy3-labeled ssDNA of varying lengths (45 nt, 60 nt, or 75 nt). The kinetics of Rad51-ssDNA assembly were monitored by observing the increase in Cy3 fluorescence. We performed the experiment as a function of increasing Rtt105 concentration and selected the Cy3 saturation point for each DNA substrate (SI Appendix, Fig. S7B). We then picked the minimal Rtt105 concentration where the Cy3 signal reaches the maximum for each DNA substrate and plotted them as a function of DNA length. Our results showed that the relationship between Rtt105 concentration and DNA length appeared to be linear (SI Appendix, Fig. S7C), suggesting that Rtt105 stimulates Rad51-ssDNA assembly in a distributive even manner.

A stable Rad51 binding to ssDNA requires the presence of ATP, while ATP hydrolysis by Rad51 leads to its disassociation from ssDNA (11, 45). Interestingly, when ATP was replaced with adenylyl imidodiphosphate (AMPPNP), a nonhydrolyzable ATP analog, in the single-molecule MT reactions, we could not detect significant stimulation of Rad51-ssDNA assembly by Rtt105 (SI Appendix, Fig. S7D), suggesting that the stimulation by Rtt105 involves Rad51 ATP hydrolysis activity. Notably, we found that the presence of Rtt105 but not the YP2A mutant protein strongly inhibited the ATP hydrolysis activity of Rad51 (Fig. 3I). Finally, we noted that the rad51-I345T point mutation, which causes a more stable Rad51-ssDNA complex (46), strikingly impaired ectopic recombination repair (SI Appendix, Fig. S7E). Interestingly, additional YP2A mutation in rad51-I345T cells significantly improved the HR survival rate (SI Appendix, Fig. S7E), further underscoring the role of Rtt105 in regulating dynamic Rad51-ssDNA assembly. Thus, we conclude that Rtt105 directly stimulates the dynamic assembly of Rad51 with ssDNA, which appears to act by inhibiting the ATP hydrolyzing activity of Rad51 and thereby stabilizing Rad51 on ssDNA.

Rtt105 Promotes Rad51 Nucleation, D-Loop Formation, and Strand Exchange.

We then tested whether Rtt105 could stimulate Rad51 nucleation on RPA-coated ssDNA using an ssDNA affinity pull-down assay (Fig. 4A), as described previously (12). In the absence of RPA, the WT Rtt105 but not the YP2A mutant protein stimulated Rad51 binding to ssDNA (Fig. 4 B and C, lanes 2-4). However, the addition of RPA markedly suppressed Rad51 loading on ssDNA (lanes 2, 5). Notably, this inhibition was alleviated by the inclusion of the WT Rtt105 but not the YP2A mutant protein (lanes 5-7), indicating that Rtt105 stimulates Rad51 nucleation on RPA-coated ssDNA.

Fig. 4.

Rtt105 promotes Rad51 nucleation, D-loop formation, and strand exchange. (A) Scheme illustrating the assay for Rad51 replacement of ssDNA-bound RPA. (B and C) An ssDNA pull-down assay and quantification showing the impact of Rtt105 or YP2A on His-Rad51 replacement of ssDNA-bound RPA. (D) Scheme showing the assay for D-loop formation. (E and F) A D-loop assay and quantification showing the effect of Rtt105 or YP2A on His-Rad51-mediated D-loop formation. (G) Scheme showing the assay for DNA strand exchange. (H and I) A strand exchange assay and quantification showing the effect of Rtt105 on His-Rad51-mediated DNA strand exchange reactions. (J) Invasion and DNA synthesis assay. Primers adjacent to homology on chromosomes V and II were used to amplify the polymerized invading intermediate. (K and L) PCR amplification and quantification showing the kinetics of DNA synthesis over time. Values represent the mean ± SD of three independent experiments (n = 3). Statistical analysis was calculated using Student’s t test. *P < 0.05 and **P < 0.01. ns, no significance.

During HR, the Rad51-ssDNA filaments execute homology search and strand invasion into a homologous duplex donor DNA, yielding the D-loop intermediate. We examined whether Rtt105 affects strand invasion and D-loop formation using a previously described assay (47, 48). In this reaction, a Cy5-labeled 90-mer ssDNA is paired with the supercoiled pRS306 duplex DNA by Rad51 in conjunction with RPA, yielding the D-loop product (Fig. 4 D and E, lane 4). We observed that the addition of Rtt105 increased the production of the D-loop in a dose-dependent manner (Fig. 4 E and F, lanes 4-6). In contrast, the inclusion of the YP2A mutant protein failed to do so (lanes 7-8). Similarly, the WT but not the YP2A mutant protein can weakly stimulate D-loop formation in the absence of RPA but with a much lower efficiency (SI Appendix, Fig. S8). Together, these results suggest that Rtt105 stimulates strand pairing or invasion during D-loop formation via associating with Rad51.

Next, we tested whether Rtt105 influences Rad51-mediated strand exchange using an in vitro assay as described previously (12, 49). We preincubated Rad51 with ssDNA (80 nt) before adding RPA and then initiated the reaction by adding a Cy3-labeled homologous duplex DNA (40 bps). The Rad51-mediated homologous pairing and strand exchange reactions yield a Cy3-labeled hybrid duplex DNA (Fig. 4 G and H, lane 4). We found that the inclusion of Rtt105 increased the exchange products in a dose-dependent manner (Fig. 4 H and I, lanes 4-7). Thus, Rtt105 stimulates Rad51-mediated strand exchange reaction.

Finally, we employed a PCR-based invasion-extension assay to follow strand invasion and DNA synthesis in vivo (50). In this system, a single HO cut is generated on chromosome V, which is repaired by HR with the donor sequence on chromosome II (50). Two primers, one located on chromosome V immediately adjacent to the homology region, while the other anneals to sequences on chromosome II, flanking the homology border, were used to amplify the invasion product (Fig. 4J). PCR products can only be detected when recombination occurs. The invasion intermediates on both arms occurred earlier and accumulated more in WT cells than in YP2A cells (Fig. 4 K and L). Together, these results indicate a direct role of Rtt105 in stimulating Rad51 nucleation, strand invasion, and exchange reactions in vivo and in vitro.

Rtt105 Regulates RPA and Rad51 with Different Motifs.

Previous studies showed that mutation of the residues E171 L172 on Rtt105 to alanine (EL2A) abolished the interaction between Rtt105 and RPA (35, 36). However, we noted that the Rtt105-EL2A mutation did not affect the Rtt105-Rad51 association (Fig. 1 H and I). Similarly, disruption of the Rtt105-Rad51 interaction by YP2A mutation did not impair the Rtt105-RPA interaction either (SI Appendix, Fig. S9 A and B). Thus, the interactions between Rtt105 and RPA or Rad51 are physically independent (Fig. 5A).

Fig. 5.

Rtt105 orchestrates RPA and Rad51 for effective Rad51 nucleation, strand exchange, and HR repair. (A) Predicted structural model of Rtt105 by AlphaFold 3. The residues for interacting with RPA(E¹⁷¹L¹⁷²) and Rad51(Y²⁰³P²⁰⁴) are indicated. (B and C) Single-molecule MT illustrating the impact of the WT or mutant Rtt105 protein (20 nM) on the assembly of RPA (20 nM, B) and Rad51(100 nM, C) on naked ssDNA, respectively. (D and E) Single-molecule MT showing the impact of the WT or mutant Rtt105 protein on Rad51 replacement of RPA coated on ssDNA. The RPA-ssDNA filaments were assembled at 100 nM RPA. After free RPA was rinsed away, 500 nM Rad51 with or without 20 nM Rtt105 was added to replace the ssDNA-bound RPA. (F and G) A strand exchange assay and quantification showing the effect of the WT or mutant Rtt105 on DNA strand exchange reactions. (H) The survival rate of DSB repair by ectopic recombination for indicated strains derived from MK203. (I and J) Graphs showing the survival rate (I) and proportion of repair products (J) for indicated cells in NA14 background. (K) DNA damage sensitivity test of the indicated cells at specified drug concentrations. Values represent the mean ± SD of three independent experiments (n = 3). Statistical analysis was calculated using Student’s t test. **P < 0.01. ns, no significance.

Next, we purified the WT or mutant Rtt105 proteins and tested their effects on RPA or Rad51 loading on ssDNA using single-molecule MT (SI Appendix, Fig. S10A). As expected, the addition of RPA itself led to a striking extension of ssDNA over time (Fig. 5B), indicative of RPA assembling on ssDNA (36). Notably, the inclusion of WT or the YP2A mutant Rtt105 remarkably accelerated RPA assembly, increasing the speed by ~sixfold (from 1,500 to 250 s) (Fig. 5B). The stimulatory effect of Rtt105 appeared to be dose-dependent (SI Appendix, Fig. S10 B–D). In contrast, inclusion of the EL2A or ELYP4A(4A) mutant protein completely failed to stimulate RPA assembly (Fig. 5B). On the other hand, the WT and the EL2A mutant Rtt105 possessed the abilities to enhance dynamic Rad51 assembly on ssDNA, yet the YP2A or 4A mutant protein failed to do so (Fig. 5C). The effects of these Rtt105 variants on RPA or Rad51 assembly were further confirmed by EMSAs (SI Appendix, Fig. S11 A–C). Thus, Rtt105 can promote the dynamic assembly of RPA or Rad51 on ssDNA separately via direct physical association.

Rtt105 Orchestrates RPA and Rad51 Actions for Effective Rad51 Nucleation, Strand Exchange, and HR Repair.

However, proper HR requires coordinated actions of both RPA and Rad51. Therefore, we monitored the effects of Rtt105 on Rad51 replacement of RPA coated on ssDNA using single-molecule MT (Fig. 5D) (12, 51, 52). The binding of Rad51 to ssDNA results in further extension of ssDNA (Fig. 5D). First, enough RPA (200 nM) was added to saturate the ssDNA (Fig. 5E and SI Appendix, Fig. S7F), and then the free RPA in the flow cell was rinsed. Subsequently, Rad51 (500 nM) was added with or without Rtt105 (20 nM) to the reaction. Rad51 alone could fully replace the ssDNA-bound RPA in ~70 min, while the inclusion of WT Rtt105 in the reaction dramatically stimulated the reaction, increasing the speed by ~fourfold (~17 min) (Fig. 5E). Surprisingly, none of the EL2A, YP2A, or 4A mutant Rtt105 protein could stimulate the replacement under this condition. However, when the ssDNA was first saturated with RPA with the inclusion of the WT Rtt105, then the addition of the WT or EL2A, but not the YP2A or 4A mutant protein, together with Rad51 could promote the replacement of RPA (SI Appendix, Fig. S12). The above results strongly suggest that proper Rad51 nucleation and filament elongation require coordinated regulations of both RPA and Rad51 by Rtt105.

Consistently, in the presence of RPA, only the WT Rtt105, but not the EL2A, YP2A, or 4A mutant protein, could stimulate the strand exchange reaction (Fig. 5 F and G, lanes 4-8 and SI Appendix, Fig. S13 A and B). However, when RPA was omitted in the reaction, the WT and the EL2A mutant Rtt105 protein exhibited a weak stimulation on the reaction, while the YP2A and 4A mutant protein failed to do so (SI Appendix, Fig. S13 C and D). Together, our results suggest that Rtt105 plays a role in orchestrating RPA and Rad51 actions on ssDNA to ensure proper Rad51 nucleation and strand exchange reactions.

In support of these in vitro results, we noted that the EL2A, YP2A, and 4A mutants all exhibited a similar extent of defect in ectopic recombination repair in the haploid cells MK203 (Fig. 5H). This conclusion was further validated with the other ectopic recombination system (NA14 or NA29) (Fig. 5 I and J and SI Appendix, Fig. S13 E and F). Similarly, all these mutants were also completely deficient in ectopic recombination in the diploid cells MK235 (SI Appendix, Fig. S2 C and D). Finally, these mutants exhibited a similar degree of DNA damage sensitivity (Fig. 5K). Together, these results strongly indicate that the effects of Rtt105 on RPA and Rad51 are epistatic during HR repair and that both regulations are necessary for efficient Rad51 nucleation, strand exchange, and HR repair. Although Rtt105 can regulate RPA and Rad51 separately in vitro (Fig. 5 B and C), our data indicate that Rtt105 plays an important role in orchestrating RPA and Rad51 actions to coordinate the successive reactions in HR repair (Fig. 5 D–K).

Rad51 K¹²⁸D¹³⁰K¹³¹ Residues Are Critical to Mediate the Interaction with Rtt105.

Next, we turned to determine the Rad51 motif required for associating with Rtt105. We expressed a series of truncated GST-tagged Rad51 variants and tested their interactions with His-Rtt105. We found that the Rad51 N-terminal fragment with residues 1 to 100 did not bind Rtt105, while the one with residues 1 to 200 or more was able to do so (Fig. 6A, lanes 1-5), implying that the segment between residues 100 and 200 is important for the interaction. Further mapping located the interacting motif to the residues between 120 and 140 (lanes 6-11). Finally, we found that simultaneous mutation of the seven residues S¹²⁵E¹²⁶K¹²⁸D¹³⁰K¹³¹N¹³⁴E¹³⁵, or minimally, the three residues K¹²⁸ D¹³⁰ K¹³¹ to alanine (7A or KDK3A) in Rad51 was sufficient to disrupt the interaction with Rtt105 (Fig. 6A, lanes 12-17). Indeed, the KDK3A mutation abolished the Rtt105-Rad51 interaction in cells (Fig. 6B). Thus, the conserved Rad51 K¹²⁸D¹³⁰K¹³¹ motif, which is embedded in an α-helix domain linking the N-terminal fragment and the downstream ATPase domain (53), is the key interface for mediating the interaction with Rtt105 (Fig. 6 C and D).

Fig. 6.

The Rad51 KDK motif is critical to mediate the interaction with Rtt105. (A) GST pull-down assays showing the interaction between His-Rtt105 and GST-tagged WT or mutant Rad51. GST was used as a control. The asterisks denote the full-length fragments. (B) Immunoprecipitation showing the interaction between Rtt105-3xFLAG and HA-tagged WT or KDK3A mutant Rad51 protein. (C) Schematic representation showing Rad51 domains and sequence conservation around the Rtt105-interacting motif. (D) The structural model of Rad51 showing the position of the KDK motif. The model was predicted using AlphaFold 3. (E and F) An ssDNA pull-down assay and quantification showing the effect of Rtt105 on the affinity of His-tagged WT or KDK3A mutant Rad51 to ssDNA. (G) ChIP-qPCR showing the recruitment of the WT or KDK3A mutant Rad51 protein at 1 kb away from the HO cut site. (H) ATP hydrolyzing activity of the His-tagged WT and KDK3A mutant Rad51. Values represent the mean ± SD of three independent experiments (n = 3). Statistical analysis was calculated with Student’s t test. *P < 0.05 and **p < 0.01. ns, no significance.

Mutation of the KDK Motif Impairs Rad51 ssDNA Binding Ability.

Notably, the KDK3A mutation attenuated Rad51 binding to ssDNA coupled to streptavidin-coated magnetic beads (Fig. 6 E and F, lanes 3-6). This result was confirmed by the EMSA (SI Appendix, Fig. S14A). Consistently, the loading of the KDK3A mutant Rad51 protein at DSBs was reduced (Fig. 6G). Interestingly, the KDK3A mutant protein exhibited an enhanced ATP hydrolysis activity compared to the WT Rad51(Fig. 6H). As a result, the KDK3A mutant exhibited a striking defect in response to the DNA damage (SI Appendix, Fig. S14B). However, additional mutation of YP2A in KDK3A cells did not further aggravate the sensitivity, suggesting that they are epistatic in response to the DNA damage. Therefore, the phenotypes seen in YP2A cells were resulted from deficient regulation of Rad51 by Rtt105. The defect was more severe in KDK3A than in YP2A cells, likely reflecting additional roles of the KDK motif in Rad51 functions. Thus, the Rad51 KDK motif appears to serve as an important element regulating Rad51 ATP hydrolysis and ssDNA binding capacities.

Discussion

Central to HR repair is the prompt assembly of RPA on resected ssDNA, followed by the timely replacement of RPA by Rad51 recombinase to form Rad51 filament. The concerted actions of RPA and Rad51 are crucial for proper Rad51 nucleation, homology search, strand invasion, and repair. Rtt105 was known as an RPA chaperone (35, 36). In this study, we identified Rtt105 as a Rad51 regulator that plays a unique role in promoting dynamic Rad51-ssDNA assembly and orchestrating Rad51 and RPA actions to ensure proper HR repair. We showed that Rtt105 implements several layers of roles in regulating HR repair: 1) facilitates Rad51 nucleation on RPA-coated ssDNA, 2) accelerates Rad51-ssDNA assembly dynamics, 3) suppresses Rad51 ATP hydrolyzing activity, 4) stimulates strand exchange, D-loop formation, and HR repair, and 5) orchestrates the actions of RPA and Rad51 during HR repair. Our results reveal additional mechanisms regulating Rad51 filament dynamics and coordination of HR.

Yeast Rad52 and the accessory Rad51 paralogs promote Rad51 filament formation and elongation (12, 27, 31, 32, 54–57). In mammals, BRCA2 and a group of Rad51 paralogs fulfill this function (11, 14, 28, 30, 55, 58, 59). Rad52 appears to facilitate Rad51 nucleation via two layers of mechanisms. First, Rad52 interacts with RPA and selectively modulates the dynamics of the DBD-D domain of Rfa1, preventing its full engagement with ssDNA and opening the 3′-end of the RPA-occluded sequence (60). Second, Rad52 orchestrates RPA ssDNA binding modes and RPA spacing to regulate ssDNA accessibility by Rad51 (61). In contrast, the Rad51 paralog Rad55-Rad57 or RFS-1/RIP-1 (in Caenorhabditis elegans) appear to act as a chaperone and transiently interact with Rad51-ssDNA complexes to promote their assembly and then disassociate from mature filaments (14, 28). Here, we found that Rtt105 functions independently of Rad52 or Rad55-Rad57. Unlike these mediators, Rtt105 lacks both ssDNA binding and ATP hydrolysis capacities, yet it can inhibit the ATP hydrolysis activity of Rad51. This could help explain how it stimulates Rad51 nucleation, filament growth, strand exchange, D-loop formation, and HR repair. However, how Rtt105 cooperates with Rad52 or other mediators and how it fits within the context of RPA-Rad52-Rad51 complexes await further investigation.

We found that Rtt105 interacts with Rad51 and RPA with distinct motifs and regulates their cooperative binding on ssDNA. We provided multiple evidence that the effects of Rtt105 on RPA and Rad51 are epistatic, and both regulations are required for proper Rad51 nucleation, strand exchange, and HR repair. Recent single-molecule studies have revealed that conformational changes of RPA-ssDNA complexes offer differential access to the ssDNA within the complex (34, 60–64). The ssDNA gaps between RPA molecules would facilitate Rad51 access (61). Whether Rtt105 regulates RPA spacing while simultaneously increasing Rad51 access remains unclear.

Interestingly, we identified the Rad51 KDK motif as a key element regulating Rad51 ATP hydrolysis and ssDNA binding capacities. Mutation of this motif leads to increased ATP hydrolysis activity, implying the existence of a conserved autoinhibition mechanism limiting Rad51 ATPase activity. It will be intriguing to identify the human counterpart of Rtt105 and characterize its role in HR repair and cancers. The advent of high-resolution single-molecule and cryo-EM techniques makes it possible to reveal the detailed mechanism of how Rtt105 and its human counterpart orchestrate RPA and Rad51 actions to couple HR repair.

Materials and Methods

Yeast Strains and Culture.

The yeast strains used in this study are listed in SI Appendix, Table S1. Construction of yeast knockout or point mutant strains was carried out as standard protocols. For the DNA damage sensitivity test, yeast cells were cultured in YEPD media overnight. Undiluted or 1/10 serially diluted cell cultures were spotted onto YPD plates containing the indicated concentrations of DNA damaging agents. The plates were then incubated at 30 °C for 3 d before capturing images. For DSB induction, yeast cells were cultured in YP-raffinose media at 30 °C to log phase prior to the addition of galactose (2%). All other methods are described in detail in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.