|Home | About | Journals | Submit | Contact Us | Français|
RNA can serve as a template for DNA double-strand break repair in yeast cells, and Rad52, a member of the homologous recombination pathway, emerged as an important player in this process. However, the exact mechanism of how Rad52 contributes to RNA-dependent DSB repair remained unknown. Here, we report an unanticipated activity of yeast and human Rad52: inverse strand exchange, in which Rad52 forms a complex with dsDNA and promotes strand exchange with homologous ssRNA or ssDNA. We show that in eukaryotes, inverse strand exchange between homologous dsDNA and RNA is a distinctive activity of Rad52; neither Rad51 recombinase nor the yeast Rad52 paralog Rad59 has this activity. In accord with our in vitro results, our experiments in budding yeast provide evidence that Rad52 inverse strand exchange plays an important role in RNA-templated DSB repair in vivo.
Mazina et al. report an unanticipated activity of yeast and human Rad52, in which Rad52 forms a complex with tailed or blunt-ended dsDNA and promotes strand exchange with homologous RNA or ssDNA. Genetic experiments show that this activity plays an important role in RNA-templated DNA repair in vivo.
Homologous recombination (HR) is a high-fidelity process that uses homologous DNA sequences as a template to repair damaged DNA (Kowalczykowski, 2000; Krogh and Symington, 2004; Moynahan and Jasin, 2010). In eukaryotes, HR is carried out by the Rad52 epistasis group of proteins (Krogh and Symington, 2004). In this group, Rad51 plays a key role by promoting a search for homologous double-stranded DNA (dsDNA) template and forming DNA joint molecules that provide both the template and the primer for DNA polymerase during repair of DNA double-strand breaks (DSBs) (Sung, 1994). However, we recently demonstrated that transcript RNA can serve as a template for DSB repair via HR in yeast cells either indirectly, if converted into cDNA, or directly (Keskin et al., 2014). Direct RNA-templated DSB repair is efficient in the absence of ribonuclease (RNase H) function, and in cis, that is when the RNA is used as template to repair a break occurring in its own DNA gene (Keskin et al., 2014). Currently, little is known about the enzymatic machinery that executes RNA-templated DSB repair. Our results from budding yeast implicated Rad52, but not Rad51, in this RNA-directed DSB repair mechanism (Keskin et al., 2014). The role of Rad52 in RNA-dependent DSB repair is also consistent with data from human cells, which show an RNA-dependent localization of Rad52 at sites of DSBs (Wei et al., 2015). However, the exact mechanism of how Rad52 contributes to RNA-dependent DSB repair remains to be elucidated.
It is known that recombinases of the Rad51/RecA family form a nucleoprotein filament on single-stranded DNA (ssDNA) and promote DNA strand exchange with homologous dsDNA. However, in addition to this canonical or “forward” reaction, E. coli RecA was shown to form a nucleoprotein filament on dsDNA. The filament can promote strand exchange with either homologous ssDNA or ssRNA (Kasahara et al., 2000; Zaitsev and Kowalczykowski, 2000). This unconventional pairing process was called the “inverse” DNA strand exchange reaction (Zaitsev and Kowalczykowski, 2000).
Previously, it was shown that Rad52, an important member of the HR pathway (Mortensen et al., 2009), promotes annealing either of two complementary ssDNA molecules (Grimme et al., 2010; Mortensen et al., 1996; Wu et al., 2006) or of ssDNA with complementary ssRNA (Keskin et al., 2014). Here, we tested whether Rad52 also carries “inverse strand exchange” activity between homologous dsDNA and ssRNA, which could also account for the role of Rad52 in RNA-dependent DNA repair identified in our genetic experiments. Our current results demonstrate that both human and yeast Rad52 (yRad52) efficiently promotes inverse strand exchange between dsDNA and homologous ssRNA or ssDNA. We show that in eukaryotes, inverse RNA strand exchange is a distinctive activity of Rad52; neither the major recombinase Rad51 nor the yRad52 paralog Rad59 carries this activity. Our experiments in Saccharomyces cerevisiae cells support the biological significance of inverse RNA strand exchange. Taken together, our biochemical and genetic data indicate that inverse RNA strand exchange promoted by Rad52 may play a central role in RNA-dependent DSB repair.
First, we tested whether human Rad52 (hRad52) can promote inverse DNA strand exchange between homologous dsDNA and ssDNA. hRad52 nucleoprotein complex was formed with 3′-tailed dsDNA (no. 1, 63-mer/no. 117, 94-mer) (Table S1), in which oligo no.1 was 32P-labeled, and then inverse DNA strand exchange was initiated by addition of homologous ssDNA (no. 2, 63-mer) (Figure 1A). We found that hRad52 promotes inverse strand exchange with remarkably high efficiency; the initial rate of inverse reaction was approximately 10-fold higher than that of forward DNA strand exchange promoted by hRad52 with the same DNA substrates (Figures 1B and 1C). The inverse reaction required both DNA sequence homology (Figure S1A) and hRad52 protein (Figure 1C). Forward and inverse DNA strand exchange reactions promoted by hRad52 show different requirements for Mg2+ concentrations. The inverse reaction occurs across a much broader (1–20 mM) range of Mg2+ concentrations (Figure S1B) than the forward reaction (0.1–1 mM) (Arai et al., 2011; Bi et al., 2004) (O.M.M., D.V. Bugreev, and A.V.M., unpublished data). hRad52 was significantly more efficient in promoting inverse DNA strand exchange than hRad51 under conditions that were optimal for both proteins (Figure 1D). The initial rate for hRad52-promoted reaction is about 6-fold higher than that for hRad51. The hRad52 reaction was efficient at an equimolar ssDNA:tailed dsDNA ratio and reached the maximal rate at a 3-fold excess of ssDNA (Figure S1C). In contrast, hRad51 (Figure S1D) or RecA (Zaitsev and Kowalczykowski, 2000) requires larger (10-fold) excess of ssDNA for efficient inverse DNA strand exchange. hRad52- and Rad51-promoted inverse reactions were only moderately sensitive to a 10-fold excess of non-homologous ssDNA inhibitor (Figure S1E), a hallmark of inverse DNA strand exchange, as opposed to forward DNA strand exchange that is highly sensitive to ssDNA inhibition (Zaitsev and Kowalczykowski, 2000).
We tested a possibility that the inverse DNA strand exchange may result from melting of dsDNA followed by annealing of the resultant ssDNA strand with homologous ssDNA. First, we found no evidence of dsDNA or tailed dsDNA melting when hRad52-DNA complexes were probed with P1 nuclease (Figure S1F). Then, this possibility was tested by a “cross-annealing” control when hRad52 was incubated with two DNA substrates, forked DNA (no. 71/no. 169) and tailed DNA (no. 1/no. 196), in which dsDNA regions were identical, but ssDNA parts were of different lengths and sequences (Figure S1G, top). If hRad52 protein melts these substrates, ssDNA strands produced from one DNA substrate could cross-anneal with the complementary strands of the other DNA substrate, producing an additional DNA product detectable by gel electrophoresis. However, no such product was produced during 1 hr of incubation (Figure S1G, lanes 2–8). We concluded that the inverse DNA strand exchange occurs via a distinctive mechanism that does not involve dsDNA melting and annealing. The cross-annealing experiment also shows that inverse strand exchange does not occur between two dsDNA substrates.
We found that yRad52 also promotes inverse DNA strand exchange, indicating an evolutionary conservation of this activity (Figure 1E). Taken together, our current results demonstrate that human and yeast Rad52 possess inverse DNA strand exchange activity. This activity appears to be distinct and stronger than the forward DNA strand exchange activity of these Rad52 orthologs.
Our recent data indicate that in yeast cells, RNA can serve as a template for DSB repair via HR and that Rad52 plays a significant role in this process. In a rad52 null mutant, the frequency of DSB repair by RNA was reduced by a factor of ten (Keskin et al., 2014). These data prompted us to test whether human and yeast Rad52 can carry out inverse strand exchange between tailed dsDNA that mimics processed DNA ends and homologous ssRNA. We found that hRad52 promotes inverse strand exchange between 3′-tailed dsDNA (no. 1, 63-mer/no. 117, 94-mer) and ssRNA (no. 2R, 63-mer) (Figure 2A). Under standard conditions (a 3-fold molar excess of ssRNA or ssDNA), the inverse reaction with ssRNA showed a 4- to 5-fold lower initial rate than with ssDNA, but the extents of the reactions were similar (53% and 69% for ssRNA and ssDNA, respectively) (Figure 2B). A 10-fold molar excess of ssRNA further stimulated inverse strand exchange (Figure S2A). The reaction required homologous RNA (Figure S2B). Consistent with its inverse RNA strand exchange activity, hRad52 binds both RNA and RNA-DNA hybrid, though with lower affinity than ssDNA or dsDNA (Figures S2C and S2D). We found that inverse RNA strand exchange activity is evolutionarily conserved, as yRad52 can also promote exchange between dsDNA and ssRNA, albeit with lower efficiency (Figure 2C).
We found that the hRad52 R55A mutant deficient in ssDNA binding (Kagawa et al., 2002; Lloyd et al., 2005) is much less efficient than wild-type (WT) Rad52 in inverse strand exchange with ssDNA and, especially, with ssRNA; the initial rate of the reaction was decreased approximately 6-fold and 20-fold with ssDNA and ssRNA substrates, respectively (Figures 2D and 2E). We also found that the hRad52 R55A shows decreased binding to RNA and almost no binding to RNA-DNA hybrid comparing to WT hRad52 (Figures S2C–S2F). These data indicate that the same ssDNA binding site of Rad52 is required for inverse strand exchange with ssDNA and RNA.
In Rad52, the N-terminal domain (NTD), spanning approximately half of the protein, is responsible for its ssDNA annealing, DNA strand exchange, and protein multimerization (Hanamshet et al., 2016; Mortensen et al., 2002; Mortensen et al., 2009; Seong et al., 2008). The Rad52 C-terminal domain carries the nuclear localization site and regions involved in interaction with Rad51 and RPA. We found that the hRad521–209 NTD is sufficient to promote inverse RNA strand exchange efficiently (Figures 3A and 3B).
We then examined the ability of hRad51 recombinase to promote inverse strand exchange between tailed dsDNA and ssRNA (Figure S3A). We found no significant activity under tested conditions either in the presence of a 7-fold or even 100-fold excess of ssRNA (Figure S3B). Under the same conditions, hRad51 was active in promoting inverse strand exchange with ssDNA (a 7-fold excess) (Figure S3B). Similarly, yeast Rad51 (yRad51) promotes inverse strand exchange with ssDNA, but is incapable of using ssRNA in this reaction (Figures S3C–S3E). We also tested inverse strand exchange activity of yeast Rad59 (yRad59), which shares homology with the Rad52 NTD (Feng et al., 2007). We found that yRad59 promotes inverse strand exchange with ssDNA, but not with RNA (Figures 3C and 3D), even though it promotes both ssDNA-ssDNA and ssDNA-RNA annealing (Figure 3E). Thus, Rad52 appears to be the only known eukaryotic HR protein with the capacity to promote inverse strand exchange between dsDNA and ssRNA.
In vivo, RPA, a ubiquitous ssDNA binding protein (Chen et al., 2016), plays an essential role in DSB repair and physically interacts with Rad52 (Park et al., 1996). Therefore, we tested the effect of RPA on inverse DNA strand exchange promoted by hRad52 between tailed dsDNA and homologous ssDNA or ssRNA (Figure 4A). We found that RPA inhibited the initial rate of inverse strand exchange with ssDNA by approximately 2-fold (Figure 4B). In contrast, under the same conditions RPA stimulated the rate of inverse strand exchange with ssRNA also by approximately 2-fold (Figure 4C). The stimulation appeared to be species specific; hRad52 was stimulated only by human RPA, whereas yeast RPA inhibited the reaction promoted by hRad52 (Figure 4D), but stimulated the yRad52-promoted reaction (Figure 4E). These data indicate that physical interaction of RPA with Rad52 is important for stimulation of inverse strand exchange between dsDNA and ssRNA, rather than by destabilization of DNA duplex. This conclusion was further strengthened by the observation that inverse RNA strand exchange promoted by the hRad521–209 NTD, which lacks the RPA binding region, was not stimulated by human RPA (Figure 4F). It is possible that RPA stimulates inverse RNA strand exchange by inducing a favorable conformation in Rad52. A partial inhibition of inverse DNA strand exchange by RPA could be due to formation of stable RPA-ssDNA complexes that hinder ssDNA binding to Rad52-dsDNA complexes.
Canonical DSB repair mechanisms by HR require extensive processing of DNA ends to generate long single-stranded 3′ ends. Here, we wanted to test the effect of dsDNA end resection on the hRad52-promoted inverse strand exchange with ssRNA or ssDNA. We found that hRad52 is capable of promoting the inverse reaction between blunt-end duplex DNA (no. 1, 63-mer/no. 2, 63-mer) and homologous ssRNA (no. 2R, 63-mer) (Figures 5A and 5B). The rate and the extent of the reaction were significantly reduced compared with the reaction utilizing a 31-nt tailed dsDNA (Figure 5B). However, addition of hRPA greatly stimulated reaction with blunt-ended DNA nearly to the level observed for tailed dsDNA (Figure 5C). Also, hRad52 promoted inverse strand exchange between blunt-end dsDNA and ssDNA (no. 2, 63-mer) (Figure 5D). However, no stimulation by RPA was observed for this reaction (unpublished data). Utilization of blunt-ended dsDNA by Rad52 in inverse strand exchange may have important biological implications, obviating the need for dsDNA end resection during DSB repair in vivo.
Next, using the specific features of Rad52-promoted inverse RNA strand exchange identified in this study, we wanted to test the relevance of this reaction to RNA-directed DSB repair in vivo. The system, which we developed in yeast to study DSB repair by RNA in cis, consists of a defective his3 gene expressed from the galactose inducible promoter, pGAL1, in its antisense orientation, and disrupted by an artificial intron (AI). This AI can only be spliced from the antisense transcript of his3 (Figure S4). The AI contains the site for the homothallic switching HO endonuclease. The expression of HO, also from a pGAL1 promoter, generates a DSB in his3 within the AI. Following induction of the his3 antisense RNA and the DSB by galactose, only repair of the DSB by the spliced his3 antisense RNA can restore the functional sequence of the HIS3 gene and produce histidine prototrophic (His+) cells.
To corroborate the importance of Rad52 function in RNA-dependent DSB repair, we overexpressed yRad52, or either the yeast or human Rad52 NTD, in strains carrying our cis system. As a reminder, Rad52 NTD retains catalytic activities of Rad52, including inverse DNA/RNA strand exchange (Figures 3A and 3B), but lacks the Rad51 and RPA binding domains (Hanamshet et al., 2016; Mortensen et al., 2002; Storici et al., 2006). We showed previously that in the absence of RNase H activity, DSB repair proceeds using RNA template directly, whereas in its presence it proceeds through a cDNA intermediate (Keskin et al., 2014). Therefore, we tested the effect of Rad52 or Rad52 NTD overexpression in WT yeast cells in a strain defective in RNase H activity, or in a strain that is both RNase H defective and also carries a null mutation in the SPT3 gene that activates reverse transcription in yeast and is thus required for cDNA formation (Table S2) (Keskin et al., 2014). In all these strains, we observed a significant increase in the frequency of DSB repair by cDNA and RNA upon overexpression of yRad52, or either the yeast or human Rad52 NTD (Table 1). Strains with deleted endogenous RAD52 gene showed the largest response; e.g., a 94-fold increase was observed when hRad521–209 NTD was expressed in rnh1 rnh201 rad52 cells (Table 1). In line with in vitro results (Figure 2E), the R55A mutation of hRad52, which strongly reduced the efficiency of inverse RNA strand exchange, as well as Rad52 binding to RNA or an RNA-DNA hybrid (Figure S2E), markedly decreased (a factor of 13) the frequency of His+ colonies in rad52 rnh1 rnh201 cells expressing hRad521–209 R55A (Table 1). These results corroborate the function of Rad52 in RNA-directed DSB repair. Importantly, the fact that overexpression of the hRad521–209 NTD stimulated DSB repair by RNA in all studied yeast strains including rnh1 rnh201 and rnh1 rnh201 spt3 cells suggests that hRad52 could catalyze DSB repair by RNA in human cells.
Previously, we found that deletion of the yRAD51 recombinase gene did not reduce the frequency of DSB repair by transcript RNA (Keskin et al., 2014). Instead, the frequency of DSB repair by RNA in rnh1 rnh201 spt3 rad51 cells was significantly elevated compared to that in rnh1 rnh201 spt3 cells. We proposed that suppression of DSB repair through recombination with sister chromatids resulted in channeling the broken DNA substrate into the RNA-dependent pathway (Keskin et al., 2014). Here, we examined whether the yRad59 protein, which shares homology with Rad52 NTD and has partial functional overlap with Rad52 (Davis and Symington, 2001, 2003), is required for DSB repair by RNA in cis. Remarkably, we found that the frequency of DNA repair by RNA is increased by a factor of 4 and 7 in rnh1 rnh201 and rnh1 rnh201 spt3 cells when the RAD59 gene is deleted, respectively (Table 1). This finding parallels the results in rad51 null cells (Keskin et al., 2014), suggesting that RNA-templated DSB repair does not require yRad59 or yRad51. On the contrary, in a control experiment using an ssDNA oligonucleotide as a template for DSB repair in his3 in the rad59 null strains, we found that the repair frequency by ssDNA was significantly reduced by a factor of 2.7 in rnh1 rnh201 spt3 rad59 cells, and a factor of 4 in rnh1 rnh201 rad59 cells compared to rnh1 rnh201 spt3 and rnh1 rnh201 cells, respectively (Tables S3 and S7), as previously shown in rad59 null mutant cells (Storici et al., 2006). These results suggest that yRad59 has an important role in DNA-templated, but not in RNA-templated, DSB repair. It is relevant to note that while both Rad52 (Keskin et al., 2014) and yRad59 (Figure 3E) can promote RNA-DNA annealing, only Rad52 has the inverse RNA strand exchange activity (Figure 2). No such activity was observed for Rad51 or Rad59 (Figures S3 and 3D). Thus, our findings in yeast and our biochemical data support inverse RNA strand exchange activity of both yeast and human Rad52 as a distinct activity in eukaryotes that can contribute to the mechanism of DSB repair directed by RNA.
While DNA end resection is an essential step in DSB repair via a single-strand annealing (SSA) mechanism (Heyer et al., 2010), we show that inverse strand exchange occurs between ssDNA or ssRNA and homologous duplex DNA that is either non- or minimally resected (Figure 5). To determine whether the process of resection of broken DNA ends is essential for DSB repair by transcript RNA via HR in yeast cells, we tested the effect of null mutations in SAE2 or EXO1 genes or the nuclease defect D16A of the MRE11 gene, which all code for major factors important for efficient DNA end resection (Cejka, 2015; Lewis et al., 2004; Mimitou and Symington, 2008; Zhu et al., 2008), in our cis system. In the absence of SAE2 or EXO1, or in the mre11 D16A mutant, the frequency of DSB repair by RNA (in the rnh1 rnh201 spt3 background) was either increased or not changed (Table 1), suggesting that efficient resection is not required or is even an obstacle for DSB repair by RNA in cis. In a control experiment testing DSB repair and using an ssDNA oligonucleotide in an SSA assay (Storici et al., 2006) in the same strains of the cis system, we found that in the sae2 null mutant the frequency of His+ colonies was significantly reduced, and in exo1 null or in the mre11 D16A mutant the frequency was also significantly reduced, although to a lesser extent than in sae2 null cells (Table S3). These results support an RNA-dependent mechanism of DSB repair mediated by Rad52 that catalyzes a reaction in which RNA invades a broken dsDNA that is minimally or not resected (Figure 6).
Our current in vitro and in vivo findings on the mechanism of RNA-templated DNA DSB repair bring a fresh perspective to the complex relationship between RNA and DNA in the context of genome stability (summarized in Plosky, 2016). Recent work revealed an important function of Rad52 in RNA-dependent DSB repair (Keskin et al., 2014). Here, we describe an unanticipated activity of Rad52, inverse strand exchange that may be responsible for this function. Our in vitro results demonstrate that (1) both yeast and human Rad52 promote inverse strand exchange much more efficiently than the forward reaction, in contrast to Rad51, which is more efficient in forward reaction; (2) Rad52 promotes inverse strand exchange much more efficiently than Rad51 or yRad59; (3) Rad52 is so far the only known eukaryotic HR protein that can utilize both ssDNA and ssRNA in inverse strand exchange; (4) the hRad521–209 NTD retains the inverse strand exchange activity; (5) the reaction with ssRNA is stimulated by RPA; and (6) Rad52 can use non-resected duplex DNA as a substrate in inverse strand exchange.
Our current data show that the inverse strand exchange activity is conserved between low and higher eukaryotes. Remarkably, the NTD of both yeast and human Rad52 that retains inverse strand exchange activity can carry out RNA-directed DNA repair in yeast cell, and particularly in the absence of the yeast endogenous RAD52 gene. Moreover, the R55A mutation that inactivates inverse RNA strand exchange activity disrupts this Rad52 function in yeast cells. yRad52 is also known to have a mediator function, so that it can load Rad51 recombinase on RPA-covered ssDNA (Hanamshet et al., 2016). In contrast to the inverse strand exchange activity, this function of Rad52 is apparently not conserved in mammals.
We found that the intact ssDNA binding site of Rad52 is important for inverse reaction, as the hRad52 R55A, defective in ssDNA binding (Kagawa et al., 2002), has significantly reduced inverse strand exchange activity with both ssDNA and ssRNA. These data indicate that ssRNA and ssDNA may bind to the same Rad52 site. Somewhat unexpectedly, the R55A mutation has a strong inhibitory effect on the binding of RNA-DNA hybrid. The role of this binding and the mechanism of inverse reaction remain to be investigated further. Our data argue against dsDNA melting by Rad52, followed by annealing of the resultant ssDNA strands with complementary ssDNA or ssRNA by Rad52 (Figures S1F and S1G). Previous work showed that Rad52 can change the topology of dsDNA, which is manifested by formation of supercoils in covalently closed circular DNA in the presence of hRad52 (Kagawa et al., 2008). We suggest that this change in dsDNA topology may be important for inverse strand exchange by facilitating pairing of dsDNA with homologous ssDNA or RNA.
The mechanism of stimulation of Rad52-promoted inverse RNA strand exchange by RPA is of significant interest. Our data indicate the importance of protein-protein interaction between Rad52 and RPA in this stimulation, as it occurs in a species-specific manner and is abolished when the hRad52 NTD lacking the region responsible for RPA interaction is used instead of the full-length protein. It is possible that Rad52 became activated in the complex with RPA via conformational changes. Our data also argue against the contribution of helix-destabilizing activity of RPA, which would require higher RPA concentrations than were used in our experiments. RPA is known to interact with RNA, although with a significantly lower affinity than with ssDNA (Kim et al., 1992); the significance of this interaction for stimulation of the inverse reaction remains to be investigated.
Since the activity of Rad52 in inverse DNA strand exchange appeared to be significantly higher than in the forward reaction, we cannot completely exclude the possibility that a fraction of the product of the forward reaction (Figure 1D) results, in fact, from the inverse reaction. This may happen if some amount of Rad52 is transferred from ssDNA to tailed DNA substrate and initiates inverse reaction with homologous ssDNA. We consider this possibility not very likely, though, because Rad52 has higher affinity for ssDNA than for tailed dsDNA, which limits this protein transfer significantly.
Our in vivo data in yeast cells corroborate these findings. While both RNA-DNA annealing and inverse RNA strand exchange activities of Rad52 can contribute to RNA-directed DSB repair, our genetic data support a mechanism of RNA-directed DSB repair driven by Rad52-mediated inverse strand exchange. An important watershed between RNA-DNA annealing and inverse RNA strand exchange mechanism lies on the structure of the DNA substrate. While annealing requires that both DNA and RNA be in a single-stranded form, in inverse strand exchange Rad52 may utilize unresected duplex DNA. In accord with the prerequisite of Rad52-promoted inverse strand exchange, RNA-templated DSB repair in yeast cells does not require the function of key resection proteins like Sae2, Exo1, and Mre11. In contrast, the frequency of DSB repair by an ssDNA oligo, which follows an SSA mechanism (Storici et al., 2006), was significantly reduced in strains carrying mutants of these proteins (Table S3). These results suggest that RNA-templated DSB repair does not depend on prompt and efficient end resection of broken dsDNA ends, which is consistent with the role of inverse RNA strand exchange in this process. In addition, yRad59 that has RNA-DNA annealing, but not inverse RNA strand exchange activity, is required for DSB repair by an ssDNA oligo, but cannot substitute for Rad52 in RNA-directed DSB repair in yeast cells.
Interestingly, Chakraborty et al. recently showed that non-homologous end joining proteins preferentially associate with transcribed sequences following DSB induction and facilitate an error-free mechanism of DSB repair in transcribed DNA in mammalian cells (Chakraborty et al., 2016), supporting an RNA-guided DSB repair mechanism occurring prior to extensive end resection at the DSB ends. Here, we propose that Rad52 inverse RNA strand exchange can contribute to RNA-directed DSB repair in conditions of limited end resection by generating a heteroduplex between RNA and homologous DNA at the site of DSBs, in which RNA serves as a bridging template guiding DSB repair with or without a short gap filling synthesis (Figure 6). This mechanism may be especially efficient for DSB repair with reduced end resection, which is encountered in cells that are in the G1 stage of the cell cycle (Symington, 2014).
It was demonstrated that while Rad52 inactivation alone does not show any significant deficiency in DSB repair in mammalian cells (Rijkers et al., 1998), it causes synthetic lethality in combination with mutations in several other HR proteins, including BRCA1 and BRCA2 (Lok et al., 2013), defects of which are associated with various types of cancer (Prakash et al., 2015). These data indicate an essential back-up function of Rad52, which may complement the BRCA-dependent HR mechanism in mammals. We suggest that the Rad52 inverse strand exchange activities described in the current study may contribute to this back-up function. Thus, our findings may also help to identify additional therapeutic targets for cancer.
|REAGENT or RESOURCE||SOURCE||IDENTIFIER|
|Chemicals, Peptides, and Recombinant Proteins|
|BSA, molecular biology grade||New England Biolabs||B9000S|
|Proteinase K, RCR grade||Roche||03115828001|
|T4 kinase||New England Biolabs||M02201S|
|Critical Commercial Assays|
|QuickChange II site-directed mutagenesis kit||Agilent Technologies||200524-5|
|Experimental Models: Organisms/Strains|
|See Table S2 for a list of yeast strains used in this study||Storici’s lab||Strain number|
|See Table S1 for a list of oligonucleotides used in this study||IDT||http://www.idtdna.com/|
|YEP-NAT, YEP-NAT-ScRAD52-327, YEP-NAT-hRAD52-209, and YEP-NAT-ScRAD52 plasmids||Storici et al. (2006) or this study||N/A|
|2-micron plasmid||(Bruschi and Ludwig, 1989)||N/A|
|Software and Algorithms|
|Prism 5||GraphPad Software||https://www.graphpad.com|
|Image Quant TL||GE healthcare||https://www.gelifesciences.com|
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dr. Alexander Mazin (ude.demlexerd@nizama).
Yeast strains were freshly thawed from frozen stocks and grown at 30°C using standard yeast genetics practices.
hRad52, hRad521-209 NTD, hRad51, and RPA proteins were purified as described (Henricksen et al., 1994; Kagawa et al., 2001; Sigurdsson et al., 2001; Singleton et al., 2002). The deoxyribonucleotides (Table S1) were purchased from IDT Inc. and further purified by electrophoresis in polyacrylamide gels containing 50% urea (Rossi et al., 2010). HPLC-purified oligoribonucleotides were purchased from IDT Inc. dsDNA substrates were prepared by annealing of equimolar (molecules) amounts of indicated complementary oligonucleotides (Rossi et al., 2010). When indicated, oligonucleotides were 5′ end labeled with 32P using T4 polynucleotide kinase (New England Biolabs). All DNA and RNA concentrations are expressed in moles of molecules.
Nucleoprotein complexes were assembled by incubating hRad52 (900 nM) with 32P-labeled dsDNA (no. 1/no. 2; 68.6 nM) or 3′-tailed DNA (no. 1/no. 117, or indicated otherwise; 68.6 nM) in buffer A containing 25 mM Tris-Acetate, pH 7.5, 100 μg/ml BSA, 2 mM magnesium acetate, and 2 mM DTT for 15 min at 37°C. The reactions were initiated by addition of ssDNA (no. 2; 205.8 nM) or ssRNA (no. 2R; 205.8 nM). Variations to these conditions are indicated in the legend for Figure 1A. Aliquots (10 μl) were withdrawn at indicated time points and DNA or RNA samples were deproteinized by incubation in 1% SDS, 1.6 mg/ml proteinase K, 6% glycerol and 0.01% bromophenol blue for 15 min at 37°C. Samples were analyzed by electrophoresis in 10% polyacrylamide gels (acrylamide:bis-acrylamide, 17:1) in 1x TBE buffer (89 mM Tris, 89 mM boric acid and 1 mM EDTA, pH 8.3); the gels were processed as described (Rossi et al., 2010) and the reaction yield was determined using a Storm 840 Phosphor Imager (GE Healthcare).
When human or yeast RPA (1 μM) were used, they were pre-incubated with ssDNA (no. 2; 411.6 nM) or ssRNA (no. 2R; 411.6 nM) in buffer A (40 μl of reaction mixture) for 15 min at 37°C. Separate reaction mixture (40 μl) containing hRad52 (1.8 μM) and labeled 3′-tailed DNA (137.2 nM) in buffer A was incubated for 15 min at 37°C. Inverse strand exchange reaction (80 μl) was initiated by addition of mixtures containing RPA and ssDNA or RNA to the hRad52 nucleoprotein complexes. The final concentration of RPA was 500 nM, which corresponding to a stoichiometry of 1 RPA trimer per 30 nt of ssRNA or ssDNA, including ssDNA tail of 3′-tailed DNA.
The conditions were the same as for hRad52, except of a 10-fold molar excess of ssRNA (no. 2R; 63-mer, 686 nM) or ssDNA (no. 2; 63-mer, 686 nM) were used and the concentration of hRad521-209 was 1 μM.
The conditions for yRad52 inverse DNA and RNA strand-exchange were the same as for hRad52, except of a 10-fold molar excess of ssRNA (no. 2R; 63-mer, 686 nM) was used. The concentration of yeast RPA in the reaction with ssRNA was 1.5 μM to maintain the stoichiometry of 1 yeast RPA trimer per 30 nt of ssRNA.
yRad59 (3.5 μM) was incubated with 32P-labeled 3′-tailed DNA (no. 1/no. 117; 68.6 nM) in buffer A for 15 min at 37°C. The reactions were initiated by addition of ssDNA (no. 2; 63-mer, 686 nM) or ssRNA (no. 2R; 63-mer 686 nM) and carried out for 1 hr. The reaction products were deproteinized and analyzed, as described for inverse DNA or RNA strand exchange promoted by hRad52.
To form nucleoprotein filament hRad51 (2.15 μM) was incubated with labeled 3′-tailed DNA (no. 1/no. 117; 68. 6 nM) in buffer containing 25 mM Tris-acetate, pH 7.5, 100 μg/ml BSA, 3 mM magnesium acetate, 2 mM ATP, and 2 mM DTT for 15 min at 37°C. Afterward, the concentration of magnesium acetate was increased to 10 mM. Inverse DNA strand exchange reactions were initiated by addition of 7-fold molar excess of a 63-mer ssDNA (no. 2; 480.2 nM) or ssRNA (no. 2R, 63-mer; 480.2 nM). The products were deproteinized and analyzed as described for inverse strand exchange promoted by hRad52.
hRad52 protein (900 nM) was incubated with 32P-labeled 63-mer ssDNA (no. 1; 68.6 nM), 63-mer dsDNA (no. 1/no. 2; 68.6 nM) or tailed dsDNA (no. 1/no. 117; 68.6 nM) in 9 μl of buffer A for 15 min at 37°C, followed by the addition of 0.5 units of P1 nuclease (USBiological Life Science) in 1 μl. Reactions were carried out for 15 min at 37°C, then quenched by addition of SDS to 1%, proteinase K to 1.6 mg/ml, glycerol to 6% and bromophenol blue to 0.01% followed by 15 min incubation at 37°C. The DNA products were analyzed by electrophoresis in 10% polyacrylamide gels (acrylamide:bis-acrylamide, 17:1) in 1x TBE buffer (89 mM Tris, 89 mM boric acid and 1 mM EDTA, pH 8.3).
All the strains used in this study are FRO-767 (Keskin et al., 2014) derivatives (Table S2). mre11-D16A strain was constructed using the delitto perfetto system for in vivo site-directed mutagenesis (Stuckey et al., 2011). Mre11.D16A. F and Mre11.D16A.R (Table S1) oligos were used to create mre11-D16A mutant. Plasmids YEP-NAT, YEP-NAT-ScRAD52-327 and YEP-NAT-hRAD52-209 are episomal vectors containing the URA3 and the nourseothricin (NAT) resistance marker genes, and the GAL1 promoter, and are described in Storici et al. (2006). YEP-NAT-ScRAD52 was constructed like YEP-NAT-ScRAD52-327 but using a PCR product with the full length of the yeast RAD52 gene. The sequence of the YEP-NAT-ScRAD52 vector was verified by sequencing. YEP-NAT is the control empty vector, YEP-NAT-ScRAD52-327 contains the first 327 codons of yeast RAD52 gene, lacking the C-terminal region for Rad51 binding, expressed under the GAL1 promoter. YEP-NAT-hRAD52-209 contains the first 209 residues from the cDNA of human Rad52 isoform # (NM_002879) expressed under the GAL1 promoter. YEP-NAT-ScRAD52 contains full length yeast RAD52 gene expressed under the GAL1 promoter. YEP-NAT-hRAD52-209-R55A vector was constructed from YEP-hRAD52-209 vector by using QuickChange II site-directed mutagenesis kit (Agilent Technologies Cat# 200524-5). hRAD52-R55A.F and hRAD52-R55A.R oligos (Table S1) were used to create R55A mutant. Plasmid transformation was done as described (Storici et al., 2006). Cells are made competent to uptake the plasmid DNA by treatment with lithium cation. Cells from each plasmid transformation were plated to Urā medium to select for transformants. YEp-NAT, YEP-NAT-ScRAD52, YEP-NAT-ScRAD52-327 and YEP-NAT-hRAD52-209 and YEP-NAT-hRAD52-209-R55A were transformed in strains CM-95, 96 (WT), CM-100, 101 (rnh1Δ rnh201Δ) and CM-107, 108 (rnh1Δ rnh201Δ spt3Δ) which were generated by introducing the yeast 2-micron plasmid. Yeast genetic and molecular biology methods like yeast growth, gene disruption, isolation of mutants, yeast marker selection, yeast genome engineering, yeast colony PCR and sequence analysis of yeast DNA were done as described (Keskin et al., 2014; Storici et al., 2007; Stuckey et al., 2011). All primers used for strain and plasmid constructions, PCR verifications and sequence analyses are available upon request. Samples for sequencing were submitted to Eurofins MWG Operon.
To determine the frequency of His+ colonies in the strains of the cis system following induction of DSB, we conducted a fluctuation experiment as previously described (Keskin et al., 2014). Briefly, yeast cells were inoculated in 50 mL lactic acid containing media (YPLac) and incubated in a shaker for 24 hr at 30°C. Cells were then counted and 107, or in some cases, 108 cells were plated on galactose containing medium (YPGal) to turn on transcription of the his3 antisense on chromosome III and expression of the homothallic switching endonuclease. In addition, 104 cells were plated on YPGal medium to determine cell survival on galactose. Cells were incubated for 48 hr at 30°C and then replica-plated on synthetic complete medium lacking histidine (SC-His−) and grown for 3 days at 30°C. The frequency of His+ colonies following DSB induction was calculated by dividing the number of His+ colonies obtained on SC-His− medium by the number of colonies obtained on YPGal medium. The survival was calculated by dividing the number of colonies obtained on YPGal medium by the number of cells plated on the same medium.
For experiments using plasmids YEP-NAT, YEP-NAT-ScRAD52, YEP-NAT-ScRAD52-327, YEP-NAT-hRAD52-209, and YEP-NAT-hRAD52-209-R55A 107 or 108 cells were plated on medium lacking uracil and containing galactose (Urā Gal) and 103 or 104 cells were plated on Urā Gal medium to determine the cell survival. After 48 hr of incubation at 30°C, cells were replica-plated on SC-His− medium.
For experiments without induction of the DSB, cells were grown on 50 mL YPLac overnight at 30°C shaker. Next day, cells were counted and 108 cells were plated on glucose containing medium (YPD) and incubated for 24 hr at 30°C. After incubation, cells were replica-plated on SC-His− medium. In addition, 103 cells were also plated on YPD for cell survival. Results obtained in glucose are shown in Table S4.
Transformation by oligonucleotide HIS3.F (Table S1) (1nmol) was performed as described (Storici et al., 2007). The HIS3.F DNA was heated to 100°C in a heat-block for 5 min and then moved to ice to eliminate secondary structures before adding it to the cells. Cells from each oligonucleotide transformation were plated to selective His− medium and to YPD medium to determine culture viability. Induction of the homothallic switching endonuclease DSB was done by incubating cells in 2% galactose medium for 3 hr.
For conducting statistical analysis, Prism 5 GraphPad software was used. In vitro experiments were repeated at least three times; standard deviations (SD) are presented on the graphs. Results of genetic experiments in yeast cells are expressed as median and 95% confidence interval is shown in parenthesis, or alternatively the range when number of repeated experiments was less than 6. The nonparametric two-tailed Mann-Whitney-U test (Sokal and Rohlf, 1981) was used to calculate differences between His+ frequencies and P values that are presented in Tables S5, S6, and S7.
Data have been deposited to Mendeley Data and are available at http://dx.doi.org/10.17632/rkfwy7fvxc.1.
We thank S. Kowalczykowski for discussion and providing yRad52, Rad59, and RPA; C. Meers for yeast strains CM-95, and CM-96 and comments on the manuscript; Y. Liu for construction of strains HK-977 and HK-978 and plasmid hRad521-209 R55A; T. Davis for critical review of the manuscript; and all members of the Mazin and Storici laboratories for assistance and feedback on this study. We acknowledge funding from the National Cancer Institute of the NIH (grant numbers CA188347 and P30CA056036 to A.V.M.), Drexel Coulter Program Award (to A.V.M.), the National Institute of General Medical Sciences (NIGMS) of the NIH (grant number GM115927 to F.S.), the National Science Foundation fund (grant number 1615335 to F.S.), and the Howard Hughes Medical Institute Faculty Scholar (grant number 55108574 to F.S.) for supporting this work.
SUPPLEMENTAL INFORMATIONSupplemental Information includes four figures and seven tables and can be found with this article online at http://dx.doi.org/10.1016/j.molcel.2017.05.019.
AUTHOR CONTRIBUTIONSO.M.M. conducted most of the in vitro experiments, K.H. conducted the in vitro experiments, and H.K. conducted all in vivo experiments and conducted all statistical analysis of in vivo data. A.V.M. together with O.M.M. and K.H., and F.S. with H.K., designed and analyzed in vitro and in vivo experiments. A.V.M., O.M.M., and F.S. wrote the manuscript with input and suggestions from all authors.