Skip to main content
NCBI home page
Elsevier Sponsored Documents logoLink to Elsevier Sponsored Documents
. 2017 Mar 21;18(12):2845–2853. doi: 10.1016/j.celrep.2017.02.068

Human RAD52 Captures and Holds DNA Strands, Increases DNA Flexibility, and Prevents Melting of Duplex DNA: Implications for DNA Recombination

Ineke Brouwer 1,3, Hongshan Zhang 2,3, Andrea Candelli 1, Davide Normanno 2, Erwin JG Peterman 1,∗, Gijs JL Wuite 1,∗∗, Mauro Modesti 2,4,∗∗∗
PMCID: PMC5379009  PMID: 28329678

Summary

Human RAD52 promotes annealing of complementary single-stranded DNA (ssDNA). In-depth knowledge of RAD52-DNA interaction is required to understand how its activity is integrated in DNA repair processes. Here, we visualize individual fluorescent RAD52 complexes interacting with single DNA molecules. The interaction with ssDNA is rapid, static, and tight, where ssDNA appears to wrap around RAD52 complexes that promote intra-molecular bridging. With double-stranded DNA (dsDNA), interaction is slower, weaker, and often diffusive. Interestingly, force spectroscopy experiments show that RAD52 alters the mechanics dsDNA by enhancing DNA flexibility and increasing DNA contour length, suggesting intercalation. RAD52 binding changes the nature of the overstretching transition of dsDNA and prevents DNA melting, which is advantageous for strand clamping during or after annealing. DNA-bound RAD52 is efficient at capturing ssDNA in trans. Together, these effects may help key steps in DNA repair, such as second-end capture during homologous recombination or strand annealing during RAD51-independent recombination reactions.

Graphical Abstract

graphic file with name fx1.jpg

Open in a new tab

Highlights

  • •

    RAD52 binds ssDNA rapidly and tightly using wrapping and bridging modes

  • •

    RAD52 binding to dsDNA is slower, weaker, and often diffusive

  • •

    RAD52 changes dsDNA mechanics and intercalates into the double helix

  • •

    RAD52 prevents DNA melting by clamping DNA strands

Brouwer et al. show that human RAD52 swiftly and tightly wraps ssDNA around itself. With dsDNA, interactions are weaker and diffusive but drastically change DNA mechanics, suggesting double helix intercalation. DNA-bound RAD52 efficiently captures ssDNA in trans. These features seem favorable for strand annealing, clamping, and second-end capture.

Introduction

Human RAD52 belongs to a ubiquitous class of proteins that helps to overcome the thermodynamic barrier required to anneal complementary DNA strands under biological conditions (Iyer et al., 2002, Sugiman-Marangos et al., 2016). In mammalian cells, RAD52 is important for repair of DNA double-strand breaks (DSBs) by the mutagenic RAD51-independent single-strand annealing pathway (SSA) (Bhargava et al., 2016, Morales et al., 2015, Stark et al., 2004). However, unlike its yeast ortholog, vertebrate RAD52 does not seem to be crucial for recombinational repair of DSBs via RAD51 (Rijkers et al., 1998, Yamaguchi-Iwai et al., 1998). The reason for this is that the RAD52 function to facilitate the loading of RAD51 on replication protein A (RPA)-coated single-stranded DNA (ssDNA) appears to have been taken over by breast cancer susceptibility protein 2 (BRCA2) (see commentary by Liu and Heyer, 2011 and references therein). However, the finding that RAD52 deficiencies are synthetically lethal with BRCA2 deficiencies suggests a functional redundancy between RAD52 and BRCA2 (Feng et al., 2011, Lok et al., 2013). These reports, together with the recent discovery that RAD52 is implicated in promoting DNA synthesis after replication stress (Bhowmick et al., 2016, Ciccia and Symington, 2016, Sotiriou et al., 2016) and in modulating antibody class-switch recombination (Zan et al., 2017), are fueling a regain of interest in studying the function of RAD52 for both fundamental and therapeutic purposes (Hanamshet et al., 2016).

RAD52 forms ring-shaped structures in vitro (Shinohara et al., 1998). In solution, full-length RAD52 forms stable heptameric rings with a large central channel, a structural organization reminiscent of hexameric DNA helicases (Stasiak et al., 2000). However, unlike hexameric DNA helicases, there is no evidence indicating that DNA passes through the channel of the RAD52 ring. In contrast, it has been proposed that ssDNA wraps around the outer surface of the RAD52 ring, interacting with an exposed positively charged groove (Singleton et al., 2002). Annealing of complementary ssDNA might then involve ssDNA wrapping and dynamic interactions between multiple RAD52 rings (Grimme et al., 2010, Rothenberg et al., 2008).

Here, we directly visualize and quantify the interaction of fluorescently labeled human RAD52 with individual ssDNA and double-stranded DNA (dsDNA) molecules using a single-molecule approach that combines optical trapping with microfluidics and fluorescence microscopy. We report intrinsic properties of RAD52-DNA interactions, including binding stoichiometry, diffusivity, and effect on DNA mechanics, and discuss the implications of our findings on the biological roles of RAD52.

Results and Discussion

Experimental Approach

After expression in bacteria, human RAD52 tagged at the N terminus with EGFP (GFP-RAD52) was purified (Figure S1A) and tested in strand-annealing kinetic assays (details of experimental procedures can be found in Supplemental Experimental Procedures). We found no appreciable differences in activity compared to the untagged variant (Figures S1B and S1C), in agreement with previous studies where GFP-RAD52 fully rescued the synthetic lethality of BRCA2 RAD52 double-deficient cells (Feng et al., 2011, Lok et al., 2013).

To study the interaction of GFP-RAD52 with individual DNA molecules, we used a combination of optical trapping, fluorescence microscopy, and microfluidics (Brouwer et al., 2016, Heller et al., 2014a, Heller et al., 2014b). Using two independent optical traps, individual DNA molecules could be manipulated while simultaneously detecting the tension on the DNA. dsDNA molecules were biotinylated on the 3′ ends of both top and bottom DNA strands (Candelli et al., 2013) and were tethered to optically trapped streptavidin-coated polystyrene microspheres (Figure S1D). Single ssDNA tethers (Figure S1E) were generated by biotinylation of both the 5′ and 3′ ends of the top strand of a dsDNA molecule and subsequent detachment of the bottom strand by force-induced melting (Candelli et al., 2013). After incubation of the dsDNA and ssDNA constructs in a GFP-RAD52 solution, the DNA tethers were brought into a buffer channel where DNA-bound proteins could be visualized in the absence of fluid flow and fluorescent proteins in solution (Figures S1F and S1G). To facilitate in situ generation of ssDNA templates and DNA-protein complexes, our approach included laminar flow microfluidics, which allows fast exchange between buffers containing microspheres, DNA, protein-free, and protein solutions (Figures S1H and S1I).

GFP-RAD52 Binding to ssDNA Is Avid and Shortens ssDNA Contour Length

The affinity of RAD52 for ssDNA has been reported to be much higher than for dsDNA (Van Dyck et al., 1998). To quantitatively assess the affinity for ssDNA, single ssDNA molecules were (unless otherwise indicated) incubated at a tension of 10 pN in a buffer containing 1 nM of GFP-RAD52; 20 mM Tris-HCl (pH 7.6); 100 mM KCl; and either 1 mM Ca2+, 1 mM Mg2+, or no divalent cations. Protein complexes formation on the DNA was assessed from the fluorescence intensity (under constant continuous excitation with 500 ms exposure time/frame) immediately after transfer of the construct to a protein-free environment. We observed discrete fluorescent patches along the ssDNA constructs, with each patch corresponding to an oligomeric DNA-bound GFP-RAD52 complex. In the presence of Ca2+, loading of GFP-RAD52 is remarkably fast: after 5 s incubation in 100 pM GFP-RAD52, significant amounts of fluorescent protein were bound to the ssDNA (Figure 1A, left panel). Under these conditions, the observed rate of patch formation was 35 ± 3 oligomers s−1 nM−1 (n = 43). In the presence of Mg2+, a much lower rate of (5 ± 1)·10−2 oligomers s−1 nM−1 (n = 11) was observed and loading of similar amounts of protein on ssDNA required a much longer incubation (100 s) at 10-fold higher GFP-RAD52 concentration (Figure 1A, middle panel). When no divalent cation was present, binding of GFP-RAD52 was even less efficient and the patch formation rate was (1.2 ± 0.5)·10−2 oligomers s−1 nM−1 (n = 25; Figure 1A, right panel). Interactions appeared independent of DNA sequence (Figure S2A); most patches detected on ssDNA appeared static (see analysis of protein diffusion below); and within the (limited) observation time of our experiments (about 2 min), we did not observe dissociation of the GFP-RAD52 patches from the DNA, which implies a dissociation rate smaller than 0.008 s−1. We thus show that GFP-RAD52 interacts efficiently and statically with ssDNA in a cation-dependent manner.

Figure 1.

Figure 1

Open in a new tab

Binding of GFP-RAD52 to ssDNA

(A) Fluorescence images (top panels) and kymographs (bottom panels) of GFP-RAD52 on ssDNA in the presence of the indicated divalent cations. ssDNA molecules were held at 10 pN tension and incubated in a buffer containing 100 pM GFP-RAD52 for 5 s in the presence of CaCl2 or 1 nM GFP-RAD52 for 100 s in the presence of MgCl2 or in the absence of divalent cation. The scale bars represent 2 μm and 5 s.

(B) Size distribution of GFP-RAD52 oligomers bound to ssDNA, measured in CaCl2 at 100 pM with an average of 11 ± 1 monomers (mean ± SEM; n = 238). Dashed blue lines indicate multiples of seven monomers. Similar distributions were obtained for all the conditions tested.

(C) Bar plot showing how the average patch size varies with varying divalent cations at 1 nM and 100 pM GFP-RAD52. At 100 pM, no binding was detected within our incubation times in the presence of Mg2+ or in the absence of divalent cation. The error bars represent statistical errors in the number of counts.

(D) Force-extension curves during successive extension and retraction of GFP-RAD52-ssDNA complexes formed by incubation of the ssDNA with 5 nM GFP-RAD52 in 30 mM KCl. Red trace shows the corresponding (calculated) contour length during the extension trace.

(E) (Top panel) Kymograph of the fluorescence signal corresponding to (D). (Bottom panel) Enlargement of the events indicated by (1), (2), and (3) shows clear ruptures of protein-protein bridges. The scale bars represent 5 μm and 10 s.

Next, we determined the stoichiometry of the DNA-bound complexes by quantifying the fluorescence intensity of the complexes and normalizing to that of an individual GFP (Figure S3). Stoichiometry distributions were typically very broad, ranging from one to several tens of GFP-RAD52 monomers per complex (Figure 1B). We found no evidence for the strict heptameric structure reported previously (blue dotted lines in Figure 1B; Stasiak et al., 2000). Heptamers, the dominant species in solution, may thus rearrange into different oligomeric complexes when the protein interacts with ssDNA. Nevertheless, further experiments are needed to directly observe this putative rearrangement and assess to what extent the deviation from the 7·n distribution expected for heptamers and multiples of heptamers is caused by quenching of the EGFP fluorescence by homo-FRET or because of a dark, non-fluorescent fraction in the GFP-RAD52 preparations. In the presence of Ca2+, the average patch size was significantly larger than in the presence of Mg2+ or without divalent cations (Figure 1C). Applying tension to the ssDNA substrate did not have a significant effect on the average number of patches (Figure S2B) and their size (Figure S2C). To further address how RAD52 interacts with ssDNA, we performed an experiment where a ssDNA construct was repeatedly incubated in the GFP-RAD52 channel, each incubation lasting 2 s. The position of the fluorescent patches was recorded and photobleached afterward. Subsequently, the recorded positions of the successive incubations were compared (Figure S2D; representative example out of eight experiments). The probability of detecting a fluorescent patch at the same position during successive incubations is in the order of 14% ± 4% (SEM), indicating that the rates of initial patch formation and of patch growth are of the same order, unlike RAD51, which exhibits rate-limiting nucleation and a fast polymerization rate (Candelli et al., 2014).

We also studied the impact of GFP-RAD52 binding on the mechanical properties of ssDNA. To this end, we incubated ssDNA constructs in a buffer containing GFP-RAD52 at very low tension, allowing different segments of the ssDNA to interact with each other. Next, the constructs were brought into a protein-free buffer channel, where force-extension and force-relaxation curves were measured. In these curves, two clear effects were observed. In the extension curves (Figure 1D), rupture events were observed, where a large, abrupt increase in DNA extension was observed without an increase in force (orange arrows). The average increase in contour length for a single rupture event was 0.35 ± 0.02 μm (mean ± SEM; Figure S2E). The relaxation curves, on the other hand, showed no such ruptures and appeared smooth, but the ssDNA was significantly shorter than naked ssDNA. This shortening persists up to forces above 80 pN (Figure 1D). The rupture events could be attributed to tension-induced rupture of protein-protein bridges, as can be observed in the corresponding fluorescence kymographs (Figure 1E) or to the rupture of short stretches of dsDNA that were formed through RAD52-mediated annealing of partially complementary DNA segments. In addition, the shortening of the GFP-RAD52-ssDNA construct with respect to naked ssDNA can be attributed either to very strong (and thus unbroken) protein-protein bridges or to ssDNA that is wrapped around the protein complexes. Hence, this behavior is consistent with the proposal that ssDNA wrapping and ring-ring contacts might be involved in RAD52-promoted strand annealing (Grimme et al., 2010, Rothenberg et al., 2008).

GFP-RAD52 Binding Increases Flexibility and Contour Length of dsDNA and Prevents Melting

Although biochemical studies have mostly focused on the binding of RAD52 to ssDNA, there is also evidence for RAD52 interacting with dsDNA (Hengel et al., 2016, Van Dyck et al., 1998), despite the biological relevance of such interaction remaining subject of debate. To detect dsDNA binding, we worked at 20 nM GFP-RAD52 concentration; as reference, the RAD52 concentration in yeast is ∼1 nM (Ghaemmaghami et al., 2003). Interestingly, RAD52 has a secondary DNA binding site that is important for dsDNA binding, likely regulated by phosphorylation, and required to introduce positive supercoiling in dsDNA upon RAD52 binding (Honda et al., 2011, Kagawa et al., 2008). Here, we used our single-molecule approach to directly observe the binding of GFP-RAD52 to dsDNA. As for the interaction with ssDNA, binding was dependent on the divalent cation present: GFP-RAD52 binds to dsDNA more readily in the presence of Ca2+ (average patch formation rate [26 ± 4]·10−3 oligomers s−1 nM−1 [n = 66]), than in the presence of Mg2+ ([3.9 ± 0.3]·10−3 oligomers s−1 nM−1 [n = 101]) or in the absence of divalent cations ([2.9 ± 0.5]·10−3 oligomers s−1 nM−1 [n = 38]; Figure 2A). From these numbers, it is clear that, in the presence of divalent ions, the affinity for dsDNA is one to three orders of magnitude lower compared to the affinity for ssDNA. Again, the interaction of GFP-RAD52 with dsDNA appeared not to depend on DNA sequence (Figure S2A), and dissociation was slower than photobleaching, as found for the interaction with ssDNA. Unlike GFP-RAD52 binding to ssDNA, which predominantly involved static complexes, binding to dsDNA involved both static and diffusive complexes (Figure 2A), depending on the cation and applied tension. Also, when measured in Ca2+ and at 50 pN, the average patch size on dsDNA did not depend on incubation time (Figure S4A), indicating that patch growth is not cooperative.

Figure 2.

Figure 2

Open in a new tab

Binding of GFP-RAD52 to dsDNA

(A) Fluorescence images (top panels) and kymographs (bottom panels) of GFP-RAD52 on dsDNA in the presence of the indicated divalent cations and dsDNA template tensions. The dsDNA molecule was incubated in a channel containing 20 nM GFP-RAD52. Incubation times were 20 s for CaCl2 at 5 pN, 30 s for CaCl2 at 50 pN, 50 s for MgCl2 at both forces and for no divalent cation at 50 pN, and 300 s for no divalent cation at 5 pN. As on ssDNA, most binding is observed in the presence of Ca2+, slightly less binding is observed in presence of Mg2+, and the lowest affinity is observed without divalent cation. GFP-RAD52 shows a dynamic behavior at low force, whereas it binds in a more static fashion at higher forces. The scale bars represent 2 μm and 5 s.

(B) Size distribution (n = 242) of GFP-RAD52 oligomers bound to dsDNA, average of 34 ± 3 monomers (mean ± SEM), measured at 20 nM GFP-RAD52, 50 pN tension, in the presence of CaCl2. Dashed blue lines indicate multiples of seven monomers. Similar distributions were obtained for all the conditions tested.

(C) Relation between average patch size and DNA tension for the cationic conditions studied. As dsDNA tension is increased, the average patch size decreases by 4-fold in the 5–50 pN range. In addition, cationic conditions slightly influence the patch size. The error bars represent SEM.

(D) Mechanical properties of GFP-RAD52-dsDNA complexes determined by force-relaxation experiments (red curve). Up to 30 pN, the curve is well described by the eWLC model (dark gray). Compared to bare dsDNA (blue and light gray), a significant decrease in persistence length and a slight increase in contour length are observed (see Table 1). The inset shows the fluorescence image of the dsDNA, covered by more than 4,500 GFP-RAD52 proteins, recorded before the stretching cycle. The scale bar represents 2 μm.

Stoichiometry distributions of complexes bound to dsDNA were even broader than for ssDNA (Figure 2B), and also for dsDNA, we did not discern a clear 7·n distribution reminiscent of heptameric complexes. Again, a dependence of the stoichiometry distributions on the divalent cation present was observed, as well as a significant dependence on the tension applied to the construct (Figure 2C). At lower DNA tensions, GFP-RAD52 oligomers were larger than at higher tensions. This observation is in contrast to the interaction with ssDNA, which did not show dramatic tension dependence. Most likely, this is due to structural changes of bare dsDNA upon application of tension, causing disruption of base pairing resulting in force-induced melting or other structural states that are more ssDNA-like where RAD52 would preferentially bind with a stoichiometry reminiscent of that observed for ssDNA (King et al., 2013). For ssDNA, such structural transitions do not occur. On both templates, the size of these complexes depends on the presence of Ca2+ and Mg2+. Whereas the difference in the presence or absence of divalent cation can be explained by the shielding effect of their positive charges on the negatively charged phosphodiester backbone of DNA, the difference between Ca2+ and Mg2+ cannot be directly explained. Nevertheless, it is interesting to note here that Ca2+ greatly favors association kinetics of GFP-RAD52 with DNA. In light of previous studies that showed an important role of Ca2+ in the control of homologous recombination in human by affecting the ATPase activity of RAD51 (Bugreev and Mazin, 2004, Mazina and Mazin, 2004), our findings suggest that Ca2+ could have a much wider impact on DNA recombination transactions in human cells not only by stimulating RAD51-mediated strand exchange but also RAD52-mediated strand annealing.

Next, we investigated the effect of GFP-RAD52 binding on the mechanical properties of dsDNA using dsDNA constructs with a high coverage of GFP-RAD52. We observe smooth force-extension and relaxation curves (Figures 2D, S4B, and S4C) that did not show evidence for protein-protein bridges or DNA wrapping, in contrast to our observation for ssDNA. Individual force-extension curves show substantial variations (Figure S4B), likely due to DNA-bound protein complexes sticking nonspecifically to the microspheres. Relaxation curves, however, were reproducible and smooth (Figure S4C). For forces below 30 pN, these curves can be accurately described by the extensible-worm-like-chain (eWLC) model (Broekmans et al., 2016; Figure 2D). From the fit parameters in the presence and absence of GFP-RAD52 (Table 1), we deduce that GFP-RAD52 binding results in a (9 ± 1)-fold decrease of the persistence length and a (1.21 ± 0.03)-fold increase in contour length, whereas the stretch modulus is not affected. These properties suggest a possible binding mechanism of RAD52 to dsDNA through intercalation and opening of the double helix, which might be crucial for ATP-independent homology recognition and strand exchange by RAD52 (Bi et al., 2004, Reddy et al., 1997).

Table 1.

GFP-RAD52-dsDNA Complexes as an Extensible Worm-like Chain

Conditions Stretching Direction Number of Experiments Lp (nm) Lc (μm) S (pN)
GFP-RAD52-dsDNA relaxation 8 5.5 ± 0.9 19.4 ± 0.4 (21 ± 7)·102
Naked dsDNA extension 20 45 ± 2 15.96 ± 0.02 (18 ± 1)·102
Naked dsDNA relaxation 18 47 ± 2 16.00 ± 0.02 (17 ± 1)·102
Open in a new tab

Lc, contour length; Lp, persistence length; S, stretch modulus. GFP-RAD52-dsDNA complexes and naked dsDNA were measured in a buffer containing CaCl2. Errors: SEM.

In the buffer conditions used, at forces above 30 pN, extension-relaxation cycles of bare dsDNA typically show a saw-tooth-like overstretching transition with a large hysteresis between the extension and relaxation curve (Figure S4D), signature of the force-induced melting of the DNA strands (Gross et al., 2011). For dsDNA coated with GFP-RAD52, the behavior is different: the curves remain smooth and the hysteresis between extension and relaxation curves is much smaller, indicating that force-induced melting of the DNA strands no longer occurs. The formation of ssDNA is thus prevented by GFP-RAD52, providing evidence for strand annealing and clamping activity for RAD52 reminiscent of the activity proposed for the bacteriophage λ Redβ ortholog, which is thought to clamp DNA strands together to secure homology recognition (Ander et al., 2015). RAD52 DNA-strand clamping might be an important property during second-end capture, for holding together annealed DNA repeats to allow processing of ssDNA flaps and DNA repair synthesis during the various types of homologous recombination after D-loop formation. Moreover, as was discovered in yeast, RAD52 could be part of a complex that tightly tethers the two ends of broken chromosomes, allowing them to withstand the pulling forces of the mitotic spindle (Lisby and Rothstein, 2004, Lobachev et al., 2004).

GFP-RAD52 Slides along dsDNA

When examining the binding of GFP-RAD52 to dsDNA (Figure 2A), we observed that, depending on the buffer and the tension applied on the construct, a fraction of the fluorescent patches moved along the DNA in a diffusive manner. To quantify this diffusive motion, we used custom-written tracking software to determine the trajectory of each individual fluorescent patch over time and used mean squared displacement (MSD) analysis to determine the one-dimensional diffusion coefficient of each patch (Figures S5A–S5D). Then, we applied a threshold of 583 nm2/s (the minimal detectable diffusion coefficient in our experimental conditions; see Supplemental Experimental Procedures) to determine whether the complex was static or diffusive. For each tension and buffer condition studied, we generated the distribution of above-threshold diffusion coefficients (Figure 3A) and used two parameters to quantify the diffusive behavior: the average diffusion coefficient (calculated based on only the diffusive particles) and the diffusive fraction (the fraction of particles that diffused). Diffusion was most prominent at forces below 15 pN, where almost 100% of the particles were mobile. Both the average diffusion coefficient (Figure 3B) and the diffusive fraction (Figure 3C) decreased with increasing tension. The transition between static and diffusive behavior is reversible: when the force is increased, particles switch from a diffusive to a static state, and when the force is decreased again, particles may switch back to the diffusive mode (Figure 3D). Yet, no dependence of the diffusion coefficient on the nature of the divalent cation was observed (Figure 3E), and the diffusive fraction also did not change accordingly (Figure S5E). Finally, our data show no strong correlation between the diffusion coefficient and the size of the fluorescent patch (Figure 3F), indicating that diffusion is not limited by the drag force acting on the protein complex. A quantitative analysis revealed that RAD52 complexes also diffuse on ssDNA, albeit with smaller diffusion coefficients and diffusive fractions than on dsDNA and in a manner independent of the tension applied on the ssDNA construct (Figures S5F and S5G). In total, 25% ± 3% of the complexes bound to ssDNA showed diffusive behavior and the average diffusion coefficient was roughly 50-fold smaller for particles diffusing on ssDNA than on dsDNA (Figure 3E). From these data, we conclude that GFP-RAD52 can interact with DNA in either static or diffusive binding modes. The diffusive binding mode is predominantly observed on dsDNA and is slowed down when tension is applied on the dsDNA, which favors immobilization of RAD52 complexes by intercalation in the double helix (Figure 3G, top panel). This diffusive binding mechanism suggests a role for RAD52 in a diffusive search mechanism for localizing DNA structural intermediates, such as ssDNA-dsDNA interfaces.

Figure 3.

Figure 3

Open in a new tab

GFP-RAD52 Can Diffuse along DNA

(A) Histogram (n = 77) of the diffusion coefficients of the diffusive GFP-RAD52 complexes along dsDNA, measured in 20 nM GFP-RAD52 at 5 pN and in the presence of CaCl2.

(B) Relation between average diffusion coefficient and dsDNA tension. A clear ∼8-fold decrease is observed as the tension on the dsDNA molecule increases. The error bars represent SEM.

(C) The fraction of complexes that are mobile decreases with increasing tension. At low force, virtually all complexes show diffusive behavior, whereas at high tension, only a small fraction of complexes is mobile. The error bars represent statistical errors in the number of counts.

(D) Kymograph recorded during successive extension-relaxation cycles of a GFP-RAD52-dsDNA complex showing a clear force dependence of the diffusion: at low force, most complexes diffuse. When the force is increased, complexes switch to a static binding mode. When the force is decreased, complexes may switch back to diffusive behavior. Intensity is scaled logarithmically. The scale bars represent 2 μm and 5 s.

(E) Bar plot of the average diffusion coefficient for different divalent cations, forces, and DNA substrates. Diffusion is fastest in the presence of Ca2+, slower in presence of Mg2+, and slowest in the absence of divalent cations. The error bars represent SEM.

(F) Relation between diffusion coefficient and patch size, measured on dsDNA with a tension of 5 pN in the presence of Ca2+. Grey dataset shows all individual data points, and red dataset shows the average diffusion coefficient of 20 consecutive data points of increasing complex size. The error bars represent SEM.

(G) Schematic summarizing the interaction of RAD52 with DNA. GFP-RAD52 interacts with dsDNA in a diffusive mode at low tension (1–5 pN; top). As the applied tension is increased, diffusion halts as GFP-RAD52 complexes intercalate the double helix. At high tension (>50 pN), GFP-RAD52 tightly clamps the DNA strands. The process is reversible, as GFP-RAD52 complexes resume diffusion when the applied tension is brought back to 5 pN. GFP-RAD52 binding to ssDNA is rapid, stable, and static, consisting of a combination of wrapping and bridging modes (bottom).

DNA-Bound GFP-RAD52 Captures ssDNA in trans

Given the involvement of RAD52 in second-end capture during homologous recombination (McIlwraith and West, 2008, Nimonkar et al., 2009, Shi et al., 2009), we explored its ability to capture DNA in trans. GFP-RAD52 was first bound to dsDNA or to ssDNA, and the constructs were subsequently exposed to a solution of 60-mer ssDNA oligonucleotides fluorescently end labeled with Atto647N. GFP-RAD52 bound to dsDNA exhibits a remarkably efficient ability to capture the ssDNA oligonucleotide in trans (Figure 4A). All out of 27 individual GFP-RAD52 fluorescent patches (green) observed on six independent dsDNA molecules held at 50 pN captured at least one ssDNA oligo (red). Under the same conditions, no Atto647N signal was detected on constructs that were not incubated with GFP-RAD52 (Figure 4B). Next, we performed similar experiments on ssDNA constructs held at 5 pN (Figure 4C). Again, analysis of six independent molecules pre-incubated with GFP-RAD52 showed detection of Atto647N signal co-localizing with the GFP-RAD52 patches. Control experiments with ssDNA constructs not pre-incubated with GFP-RAD52 revealed that the oligonucleotides bind to naked ssDNA to a much lesser extent than in the presence of GFP-RAD52 (Figure 4D). We conclude that DNA-bound GFP-RAD52 is efficient at capturing ssDNA from solution, reminiscent of its role in second-end capture.

Figure 4.

Figure 4

Open in a new tab

GFP-RAD52 Captures ssDNA in trans

(A) Two representative experiments showing ssDNA capture by GFP-RAD52 bound to dsDNA (n = 6). A dsDNA construct held at 50 pN was first incubated in a channel containing 50 nM GFP-RAD52 in the presence of Ca2+ for 30 s and then moved into a protein-free buffer channel to detect the binding positions of GFP-RAD52 (green). Next, the construct was incubated in a channel containing 10 nM Atto647N-labeled ssDNA Oligo for 30 s and subsequently moved to the observation channel to detect the binding positions of the oligo (red). Both signals were merged to detect where colocalization (yellow) has occurred.

(B) Two representative control experiments as in (A) but without GFP-RAD52.

(C) Two representative experiments showing ssDNA capture by GFP-RAD52 bound to ssDNA (n = 6). A ssDNA construct held at 5 pN was first incubated in a channel containing 5 nM GFP-RAD52 in the presence of Ca2+ for 30 s and then moved into a protein-free buffer channel to detect the binding positions of GFP-RAD52 (green). Next, the construct was incubated in a channel containing 10 nM Atto647N-labeled ssDNA Oligo for 30 s and subsequently moved to the observation channel to detect the binding of the oligo (red). Both signals were merged to detect where colocalization (yellow) has occurred.

(D) Two representative control experiments as in (C) but without GFP-RAD52. The scale bars represent 2 μm.

Conclusions

We have provided a quantitative assessment of the interaction of human RAD52 with DNA, suggesting properties important for its physiological roles as summarized in Figure 3G. While interacting tightly with ssDNA through a combination of wrapping and bridging, RAD52 complexes bound to dsDNA profoundly affect dsDNA mechanics and can diffuse in a tension-dependent way along dsDNA. The substantial decrease in persistence length and slight increase in contour length observed upon RAD52 binding indicate that RAD52 binding increases dsDNA flexibility probably by destabilizing and intercalating into duplex DNA. Our findings suggest that the way by which RAD52 promotes strand exchange in vitro is not by a strand-invasion mechanism like the RAD51 or RecA nucleoprotein filament but rather results from the ability of RAD52 to change dsDNA structure, intercalating in the helix to make the bases available for pairing. Consistent with our model, it was previously observed that increasing the fractional A·T content of DNA increases the yields of in vitro strand exchange reactions by RAD52, likely because it would be easier for RAD52 to intercalate in A·T regions (Bi et al., 2004, Kumar and Gupta, 2004). Also, the overstretching behavior of dsDNA is profoundly altered in the presence of RAD52, suggesting that RAD52 prevents force-induced melting and thus providing evidence for strand-clamping activity. The methodology and findings reported here can now be used in future experiments to extend this analysis by studying how RAD52 interacts with RPA, a pivotal ssDNA-binding protein. Indeed, RAD52 will need to deal with RPA-coated ssDNA in the physiological context, and its direct physical and functional interaction with RPA appears to be essential for homologous recombination in yeast and mammalian cells, especially when long ssDNA substrates need to be processed (Jackson et al., 2002, Park et al., 1996, Sugiyama et al., 1998). Further, the activities of BRCA2 could be similarly analyzed and directly compared to the ones of RAD52 to explore possible functional redundancies between the two proteins suggested by studies of the Ustilago maydis orthologs (Mazloum et al., 2007).

Experimental Procedures

Expanded methods and details about proteins and DNA substrates and annealing and ssDNA oligo capture assays are provided in Supplemental Experimental Procedures.

Single-Molecule Experiments

Single-molecule experiments have been executed using a custom-build instrument (Gross et al., 2010) integrating optical trapping, wide-field fluorescence microscopy, and microfluidics. Beads and DNA catching were done in PBS buffer (pH 7.3–7.5). DNA melting for generation of ssDNA templates was performed in 20 mM Tris-HCl (pH 7.6). Protein, imaging, and ssDNA oligo buffer consisted of 20 mM Tris-HCl (pH 7.6); 100 mM KCl; and either 1 mM MgCl2, 1 mM CaCl2, or no divalent cations.

Quantification of Fluorescence Intensity

The stoichiometry of the DNA-bound GFP-RAD52 complexes was inferred from the number of GFP molecules in each fluorescent patch, calculated dividing the initial fluorescence intensity of the patch by the average intensity of a single GFP. We used a step-fitting algorithm (Kerssemakers et al., 2006) to extract the intensity of single GFPs from photo-bleaching traces (Figure S3A) of individual patches. The average GFP intensity was obtained from a Lorentzian fit to the histogram of step intensities (Figure S3B).

Quantification of Protein Diffusion

Diffusion was analyzed tracking GFP-RAD52 complexes for a large number of frames (on average 29 ± 2 s) and quantified using 1D MSD analysis (Heller et al., 2014b). Diffusion coefficients were calculated by linear fit to the first five points of the MSD curves (Figure S5D).

Statistical Analysis

Data are presented as mean ± SEM, and histograms show statistical errors in the number of counts. Data and images have been analyzed using custom-written LabView routines.

Author Contributions

Conceptualization, M.M., I.B., A.C., and G.J.L.W.; Investigation, I.B., H.Z., and D.N.; Writing – Original Draft, I.B. and H.Z.; Writing – Review & Editing, I.B., H.Z., D.N., A.C., G.J.L.W., E.J.G.P., and M.M.; Funding Acquisition, G.J.L.W., E.J.G.P., and M.M.; Supervision, G.J.L.W., E.J.G.P., and M.M.

Acknowledgments

We thank Claire Wyman, Bertrand Llorente, and Murray Junop for critical reading of the manuscript. This work was supported by the French National Research Agency (project RADORDER ANR-10-BLAN-1521; to M.M.), the ARC Foundation for Cancer Research (to M.M.), the A∗MIDEX project (no. ANR-11-IDEX-0001-02) for the «Investissements d’Avenir» French Government program (to M.M.), funding from LASERLAB-EUROPE (grant agreement no. 284464; EC’s Seventh Framework Programme; to M.M.), a fellowship from the Collège of Aix-Marseille Université (to H.Z.), a VICI grant of the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (to G.J.L.W.), and a European Research Council starting grant (no. 260849-PhysGene; to G.J.L.W.).

Published: March 21, 2017

Footnotes

Supplemental Information includes Supplemental Experimental Procedures and five figures and can be found with this article online at http://dx.doi.org/10.1016/j.celrep.2017.02.068.

Contributor Information

Erwin J.G. Peterman, Email: e.j.g.peterman@vu.nl.

Gijs J.L. Wuite, Email: g.j.l.wuite@vu.nl.

Mauro Modesti, Email: mauro.modesti@inserm.fr.

Supplemental Information

Document S1. Supplemental Experimental Procedures and Figures S1–S5
mmc1.pdf (2.6MB, pdf)
Document S2. Article plus Supplemental Information
mmc2.pdf (5MB, pdf)

References

  1. Ander M., Subramaniam S., Fahmy K., Stewart A.F., Schäffer E. A single-strand annealing protein clamps DNA to detect and secure homology. PLoS Biol. 2015;13:e1002213. doi: 10.1371/journal.pbio.1002213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bhargava R., Onyango D.O., Stark J.M. Regulation of single-strand annealing and its role in genome maintenance. Trends Genet. 2016;32:566–575. doi: 10.1016/j.tig.2016.06.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bhowmick R., Minocherhomji S., Hickson I.D. RAD52 facilitates mitotic DNA synthesis following replication stress. Mol. Cell. 2016;64:1117–1126. doi: 10.1016/j.molcel.2016.10.037. [DOI] [PubMed] [Google Scholar]
  4. Bi B., Rybalchenko N., Golub E.I., Radding C.M. Human and yeast Rad52 proteins promote DNA strand exchange. Proc. Natl. Acad. Sci. USA. 2004;101:9568–9572. doi: 10.1073/pnas.0403205101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Broekmans O.D., King G.A., Stephens G.J., Wuite G.J.L. DNA twist stability changes with magnesium(2+) concentration. Phys. Rev. Lett. 2016;116:258102. doi: 10.1103/PhysRevLett.116.258102. [DOI] [PubMed] [Google Scholar]
  6. Brouwer I., Sitters G., Candelli A., Heerema S.J., Heller I., de Melo A.J., Zhang H., Normanno D., Modesti M., Peterman E.J. Sliding sleeves of XRCC4-XLF bridge DNA and connect fragments of broken DNA. Nature. 2016;535:566–569. doi: 10.1038/nature18643. [DOI] [PubMed] [Google Scholar]
  7. Bugreev D.V., Mazin A.V. Ca2+ activates human homologous recombination protein Rad51 by modulating its ATPase activity. Proc. Natl. Acad. Sci. USA. 2004;101:9988–9993. doi: 10.1073/pnas.0402105101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Candelli A., Hoekstra T.P., Farge G., Gross P., Peterman E.J.G., Wuite G.J.L. A toolbox for generating single-stranded DNA in optical tweezers experiments. Biopolymers. 2013;99:611–620. doi: 10.1002/bip.22225. [DOI] [PubMed] [Google Scholar]
  9. Candelli A., Holthausen J.T., Depken M., Brouwer I., Franker M.A.M., Marchetti M., Heller I., Bernard S., Garcin E.B., Modesti M. Visualization and quantification of nascent RAD51 filament formation at single-monomer resolution. Proc. Natl. Acad. Sci. USA. 2014;111:15090–15095. doi: 10.1073/pnas.1307824111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Ciccia A., Symington L.S. Stressing out about RAD52. Mol. Cell. 2016;64:1017–1019. doi: 10.1016/j.molcel.2016.11.036. [DOI] [PubMed] [Google Scholar]
  11. Feng Z., Scott S.P., Bussen W., Sharma G.G., Guo G., Pandita T.K., Powell S.N. Rad52 inactivation is synthetically lethal with BRCA2 deficiency. Proc. Natl. Acad. Sci. USA. 2011;108:686–691. doi: 10.1073/pnas.1010959107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Ghaemmaghami S., Huh W.-K., Bower K., Howson R.W., Belle A., Dephoure N., O’Shea E.K., Weissman J.S. Global analysis of protein expression in yeast. Nature. 2003;425:737–741. doi: 10.1038/nature02046. [DOI] [PubMed] [Google Scholar]
  13. Grimme J.M., Honda M., Wright R., Okuno Y., Rothenberg E., Mazin A.V., Ha T., Spies M. Human Rad52 binds and wraps single-stranded DNA and mediates annealing via two hRad52-ssDNA complexes. Nucleic Acids Res. 2010;38:2917–2930. doi: 10.1093/nar/gkp1249. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gross P., Farge G., Peterman E.J.G., Wuite G.J.L. Combining optical tweezers, single-molecule fluorescence microscopy, and microfluidics for studies of DNA-protein interactions. Methods Enzymol. 2010;475:427–453. doi: 10.1016/S0076-6879(10)75017-5. [DOI] [PubMed] [Google Scholar]
  15. Gross P., Laurens N., Oddershede L.B., Bockelmann U., Peterman E.J.G., Wuite G.J.L. Quantifying how DNA stretches, melts and changes twist under tension. Nat. Phys. 2011;7:731–736. [Google Scholar]
  16. Hanamshet K., Mazina O.M., Mazin A.V. Reappearance from obscurity: mammalian Rad52 in homologous recombination. Genes (Basel) 2016;7:63. doi: 10.3390/genes7090063. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Heller I., Hoekstra T.P., King G.A., Peterman E.J.G., Wuite G.J.L. Optical tweezers analysis of DNA-protein complexes. Chem. Rev. 2014;114:3087–3119. doi: 10.1021/cr4003006. [DOI] [PubMed] [Google Scholar]
  18. Heller I., Sitters G., Broekmans O.D., Biebricher A.S., Wuite G.J.L., Peterman E.J.G. Mobility analysis of super-resolved proteins on optically stretched DNA: comparing imaging techniques and parameters. ChemPhysChem. 2014;15:727–733. doi: 10.1002/cphc.201300813. [DOI] [PubMed] [Google Scholar]
  19. Hengel S.R., Malacaria E., Folly da Silva Constantino L., Bain F.E., Diaz A., Koch B.G., Yu L., Wu M., Pichierri P., Spies M.A., Spies M. Small-molecule inhibitors identify the RAD52-ssDNA interaction as critical for recovery from replication stress and for survival of BRCA2 deficient cells. eLife. 2016;5:e14740. doi: 10.7554/eLife.14740. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Honda M., Okuno Y., Yoo J., Ha T., Spies M. Tyrosine phosphorylation enhances RAD52-mediated annealing by modulating its DNA binding. EMBO J. 2011;30:3368–3382. doi: 10.1038/emboj.2011.238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Iyer L.M., Koonin E.V., Aravind L. Classification and evolutionary history of the single-strand annealing proteins, RecT, Redbeta, ERF and RAD52. BMC Genomics. 2002;3:8. doi: 10.1186/1471-2164-3-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jackson D., Dhar K., Wahl J.K., Wold M.S., Borgstahl G.E.O. Analysis of the human replication protein A:Rad52 complex: evidence for crosstalk between RPA32, RPA70, Rad52 and DNA. J. Mol. Biol. 2002;321:133–148. doi: 10.1016/s0022-2836(02)00541-7. [DOI] [PubMed] [Google Scholar]
  23. Kagawa W., Kagawa A., Saito K., Ikawa S., Shibata T., Kurumizaka H., Yokoyama S. Identification of a second DNA binding site in the human Rad52 protein. J. Biol. Chem. 2008;283:24264–24273. doi: 10.1074/jbc.M802204200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kerssemakers J.W.J., Munteanu E.L., Laan L., Noetzel T.L., Janson M.E., Dogterom M. Assembly dynamics of microtubules at molecular resolution. Nature. 2006;442:709–712. doi: 10.1038/nature04928. [DOI] [PubMed] [Google Scholar]
  25. King G.A., Gross P., Bockelmann U., Modesti M., Wuite G.J.L., Peterman E.J.G. Revealing the competition between peeled ssDNA, melting bubbles, and S-DNA during DNA overstretching using fluorescence microscopy. Proc. Natl. Acad. Sci. USA. 2013;110:3859–3864. doi: 10.1073/pnas.1213676110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kumar J.K., Gupta R.C. Strand exchange activity of human recombination protein Rad52. Proc. Natl. Acad. Sci. USA. 2004;101:9562–9567. doi: 10.1073/pnas.0403416101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lisby M., Rothstein R. DNA repair: keeping it together. Curr. Biol. 2004;14:R994–R996. doi: 10.1016/j.cub.2004.11.020. [DOI] [PubMed] [Google Scholar]
  28. Liu J., Heyer W.-D. Who’s who in human recombination: BRCA2 and RAD52. Proc. Natl. Acad. Sci. USA. 2011;108:441–442. doi: 10.1073/pnas.1016614108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lobachev K., Vitriol E., Stemple J., Resnick M.A., Bloom K. Chromosome fragmentation after induction of a double-strand break is an active process prevented by the RMX repair complex. Curr. Biol. 2004;14:2107–2112. doi: 10.1016/j.cub.2004.11.051. [DOI] [PubMed] [Google Scholar]
  30. Lok B.H., Carley A.C., Tchang B., Powell S.N. RAD52 inactivation is synthetically lethal with deficiencies in BRCA1 and PALB2 in addition to BRCA2 through RAD51-mediated homologous recombination. Oncogene. 2013;32:3552–3558. doi: 10.1038/onc.2012.391. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Mazina O.M., Mazin A.V. Human Rad54 protein stimulates DNA strand exchange activity of hRad51 protein in the presence of Ca2+ J. Biol. Chem. 2004;279:52042–52051. doi: 10.1074/jbc.M410244200. [DOI] [PubMed] [Google Scholar]
  32. Mazloum N., Zhou Q., Holloman W.K. DNA binding, annealing, and strand exchange activities of Brh2 protein from Ustilago maydis. Biochemistry. 2007;46:7163–7173. doi: 10.1021/bi700399m. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. McIlwraith M.J., West S.C. DNA repair synthesis facilitates RAD52-mediated second-end capture during DSB repair. Mol. Cell. 2008;29:510–516. doi: 10.1016/j.molcel.2007.11.037. [DOI] [PubMed] [Google Scholar]
  34. Morales M.E., White T.B., Streva V.A., DeFreece C.B., Hedges D.J., Deininger P.L. The contribution of alu elements to mutagenic DNA double-strand break repair. PLoS Genet. 2015;11:e1005016. doi: 10.1371/journal.pgen.1005016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Nimonkar A.V., Sica R.A., Kowalczykowski S.C. Rad52 promotes second-end DNA capture in double-stranded break repair to form complement-stabilized joint molecules. Proc. Natl. Acad. Sci. USA. 2009;106:3077–3082. doi: 10.1073/pnas.0813247106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Park M.S., Ludwig D.L., Stigger E., Lee S.H. Physical interaction between human RAD52 and RPA is required for homologous recombination in mammalian cells. J. Biol. Chem. 1996;271:18996–19000. doi: 10.1074/jbc.271.31.18996. [DOI] [PubMed] [Google Scholar]
  37. Reddy G., Golub E.I., Radding C.M. Human Rad52 protein promotes single-strand DNA annealing followed by branch migration. Mutat. Res. 1997;377:53–59. doi: 10.1016/s0027-5107(97)00057-2. [DOI] [PubMed] [Google Scholar]
  38. Rijkers T., Van Den Ouweland J., Morolli B., Rolink A.G., Baarends W.M., Van Sloun P.P., Lohman P.H., Pastink A. Targeted inactivation of mouse RAD52 reduces homologous recombination but not resistance to ionizing radiation. Mol. Cell. Biol. 1998;18:6423–6429. doi: 10.1128/mcb.18.11.6423. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Rothenberg E., Grimme J.M., Spies M., Ha T. Human Rad52-mediated homology search and annealing occurs by continuous interactions between overlapping nucleoprotein complexes. Proc. Natl. Acad. Sci. USA. 2008;105:20274–20279. doi: 10.1073/pnas.0810317106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Shi I., Hallwyl S.C.L., Seong C., Mortensen U., Rothstein R., Sung P. Role of the Rad52 amino-terminal DNA binding activity in DNA strand capture in homologous recombination. J. Biol. Chem. 2009;284:33275–33284. doi: 10.1074/jbc.M109.057752. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Shinohara A., Shinohara M., Ohta T., Matsuda S., Ogawa T. Rad52 forms ring structures and co-operates with RPA in single-strand DNA annealing. Genes Cells. 1998;3:145–156. doi: 10.1046/j.1365-2443.1998.00176.x. [DOI] [PubMed] [Google Scholar]
  42. Singleton M.R., Wentzell L.M., Liu Y., West S.C., Wigley D.B. Structure of the single-strand annealing domain of human RAD52 protein. Proc. Natl. Acad. Sci. USA. 2002;99:13492–13497. doi: 10.1073/pnas.212449899. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Sotiriou S.K., Kamileri I., Lugli N., Evangelou K., Da-Ré C., Huber F., Padayachy L., Tardy S., Nicati N.L., Barriot S. Mammalian RAD52 functions in break-induced replication repair of collapsed DNA replication forks. Mol. Cell. 2016;64:1127–1134. doi: 10.1016/j.molcel.2016.10.038. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Stark J.M., Pierce A.J., Oh J., Pastink A., Jasin M. Genetic steps of mammalian homologous repair with distinct mutagenic consequences. Mol. Cell. Biol. 2004;24:9305–9316. doi: 10.1128/MCB.24.21.9305-9316.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Stasiak A.Z., Larquet E., Stasiak A., Müller S., Engel A., Van Dyck E., West S.C., Egelman E.H. The human Rad52 protein exists as a heptameric ring. Curr. Biol. 2000;10:337–340. doi: 10.1016/s0960-9822(00)00385-7. [DOI] [PubMed] [Google Scholar]
  46. Sugiman-Marangos S.N., Weiss Y.M., Junop M.S. Mechanism for accurate, protein-assisted DNA annealing by Deinococcus radiodurans DdrB. Proc. Natl. Acad. Sci. USA. 2016;113:4308–4313. doi: 10.1073/pnas.1520847113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Sugiyama T., New J.H., Kowalczykowski S.C. DNA annealing by RAD52 protein is stimulated by specific interaction with the complex of replication protein A and single-stranded DNA. Proc. Natl. Acad. Sci. USA. 1998;95:6049–6054. doi: 10.1073/pnas.95.11.6049. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Van Dyck E., Hajibagheri N.M., Stasiak A., West S.C. Visualisation of human rad52 protein and its complexes with hRad51 and DNA. J. Mol. Biol. 1998;284:1027–1038. doi: 10.1006/jmbi.1998.2203. [DOI] [PubMed] [Google Scholar]
  49. Yamaguchi-Iwai Y., Sonoda E., Buerstedde J.M., Bezzubova O., Morrison C., Takata M., Shinohara A., Takeda S. Homologous recombination, but not DNA repair, is reduced in vertebrate cells deficient in RAD52. Mol. Cell. Biol. 1998;18:6430–6435. doi: 10.1128/mcb.18.11.6430. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Zan H., Tat C., Qiu Z., Taylor J.R., Guerrero J.A., Shen T., Casali P. Rad52 competes with Ku70/Ku86 for binding to S-region DSB ends to modulate antibody class-switch DNA recombination. Nat. Commun. 2017;8:14244. doi: 10.1038/ncomms14244. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Document S1. Supplemental Experimental Procedures and Figures S1–S5
mmc1.pdf (2.6MB, pdf)
Document S2. Article plus Supplemental Information
mmc2.pdf (5MB, pdf)

RESOURCES