|Home | About | Journals | Submit | Contact Us | Français|
Displacement of single-stranded DNA (ssDNA)-binding protein (SSB) from ssDNA is necessary for filament formation of RecA on ssDNA to initiate homologous recombination. The interaction between RecO and SSB is considered to be important for SSB displacement; however, the interaction has not been characterized at the atomic level. In this study, to clarify the mechanism underlying SSB displacement from ssDNA upon RecO binding, we examined the interaction between Thermus thermophilus RecO and cognate SSB by NMR analysis. We found that SSB interacts with the C-terminal positively charged region of RecO. Based on this result, we constructed some RecO mutants. The R127A mutant had considerably decreased binding affinity for SSB and could not anneal SSB-coated ssDNAs. Further, the mutant in the RecOR complex prevented the recovery of ssDNA-dependent ATPase activity of RecA from inhibition by SSB. These results indicated that the region surrounding Arg-127 is the binding site of SSB. We also performed NMR analysis using the C-terminal peptide of SSB and found that the acidic region of SSB is involved in the interaction with RecO, as seen in other protein-SSB interactions. Taken together with the findings of previous studies, we propose a model for SSB displacement from ssDNA where the acidic C-terminal region of SSB weakens the ssDNA binding affinity of SSB when the dynamics of the C-terminal region are suppressed by interactions with other proteins, including RecO.
Recombinational DNA repair is critical for maintaining the genomic integrity and survival of all organisms, because it corrects serious DNA damage, such as a double-stranded DNA (dsDNA) break (DSB)2 and a single-stranded DNA (ssDNA) gap by using a homologous region of the genome as the template for repair (1). In Escherichia coli, there are two recombinational DNA repair pathways. The RecBCD pathway is the major recombinational DNA repair system and is generally responsible for DSB repair. The other DNA repair system, the RecF pathway, is thought to be involved in ssDNA gap repair (2,–4). However, in a recBC sbcB background, a DSB is also repaired by the RecF pathway (5). In addition, RecF pathway proteins are conserved in almost all bacteria, whereas those of the RecBCD pathway are not (6). These facts suggest that the RecF pathway is the fundamental mechanism for recombinational DNA repair in bacteria. Moreover, reconstitution of the initial step of DSB repair by RecF pathway proteins (RecA, RecF, RecJ, RecO, RecQ, RecR, and ssDNA-binding protein (SSB)) has recently been demonstrated (7). Interestingly, the proposed mechanism of the RecF pathway (DSB end processing, removal of SSB, and assistance with RecA nucleation on DNA) is quite similar to eukaryotic recombinational DNA repair (1, 8,–11). Therefore, the overall process for recombinational DNA repair may be conserved from prokaryotes to higher eukaryotes, suggesting that the investigation of RecF pathway proteins might provide general information regarding the repair process in all organisms.
On the basis of the results of studies on E. coli and other bacterial RecF pathway proteins, an overall mechanism for recombinational DNA repair has been proposed. The DSB end is processed by RecQ helicase and RecJ exonuclease, resulting in the formation of a 3′-overhang ssDNA region. The ssDNA is immediately coated by SSB. The concerted action of recombination mediators (RecF, RecO, and RecR) promotes the recruitment of RecA at the dsDNA-ssDNA junction site and formation of a RecA filament on the SSB-coated ssDNA. The active RecA nucleoprotein filament searches for a homologous sequence of the damaged site from among other regions and recovers the region lost by a DSB through homologous recombination (12,–14). In this process, the initial binding of a RecA protomer on ssDNA, which is the nucleation step of RecA, is important. However, the binding of SSB on ssDNA is a strong barrier to this step, because SSB strongly inhibits the ssDNA binding of RecA (15, 16). Therefore, SSB displacement from ssDNA by a recombination mediator is important for the progression of recombinational DNA repair. The mechanism of SSB displacement after RecA nucleation has recently been clarified. Once the RecA filament core is formed on an SSB-free ssDNA region, RecA can displace SSB by filament elongation, because SSB can diffuse along ssDNA. Furthermore, SSB assists RecA filament elongation, because its diffusion on ssDNA causes melting of the secondary structure of ssDNA (17). Thus, to assist RecA nucleation, the mediator should dissociate a small number of SSBs from ssDNA and produce a small SSB-free space on ssDNA.
RecO binds to both ssDNA and dsDNA and catalyzes ssDNA assimilation into homologous superhelical dsDNA in vitro (18). An important characteristic of RecO is its interaction with SSB (19). Because neither RecF nor RecR interacts with SSB, RecO is thought to play the central role in SSB displacement. Interestingly, RecO can anneal SSB-coated ssDNA (20); a recent study has suggested that this activity is involved in second end capture after strand invasion by RecA (21). The crystalline structure of Deinococcus radiodurans RecO (drRecO) indicates that RecO comprises an N-terminal oligonucleotide-binding (OB) fold and a C-terminal helical domain. The OB fold and positively charged cluster in the C-terminal domain are involved in DNA binding (22). RecO interacts with RecR and forms the RecOR complex (15), which is necessary for SSB displacement from and RecA loading on ssDNA (23). The crystalline structure of the drRecOR complex indicates that the OB fold of RecO is also the binding site of RecR (24).
Almost all bacterial SSBs possess a single OB fold and a flexible, highly acidic C-terminal tail region and form tetramers in solution (25). SSBs of Thermus or Deinococcus species are exceptions that have two OB folds in a single polypeptide chain and form stable dimers in solution (26, 27). However, the quaternary structures of Thermus or Deinococcus SSB dimers are similar to the structure of other SSB tetramers. The highly acidic region at the end of the C-terminal tail of SSB, which is also conserved in Thermus or Deinococcus SSBs, is involved in its interaction with several proteins that participate in DNA metabolism (28,–31). The activity of SSB-interacting proteins is reportedly stimulated by their interactions with ssDNA-bound SSB (32, 33). In the RecF pathway, RecQ and RecJ interact with the SSB C terminus, which stimulates their DSB end processing activity (34, 35). Recently, it has been reported that the SSB C terminus regulates the ssDNA binding activity of the protein itself (36) and that the ssDNA binding of SSB is inhibited by the presence of acidic C-terminal amino acids. This result suggests that the C terminus is positioned near the OB fold, which is a positively charged ssDNA-binding site, and acts as an inhibitor of ssDNA binding by SSB. However, when ssDNA approaches the OB fold, the acidic C terminus is displaced by ssDNA and released. As a result, the interaction of the SSB with other proteins is facilitated by the flexibility of the free C terminus. Thus, the functional modulation of SSB by its interaction with ssDNA is gradually becoming clear. However, it is still unclear how SSB dissociates from ssDNA after the recruitment of SSB-interacting proteins to ssDNA. In this study, we attempted to clarify the mechanism underlying SSB displacement from ssDNA upon RecO binding. We analyzed the interaction between Thermus thermophilus HB8 RecO (ttRecO) and SSB by using NMR analysis and mutant proteins. Here, we propose a general mechanism for SSB displacement from ssDNA upon SSB-protein interaction. Our model explains why SSB has a highly acidic region that is separate from its core structure and that counteracts its binding to ssDNA.
Unlabeled ttRecO protein was prepared as described previously (19). To obtain uniformly labeled 13C/15N- or 15N-ttRecO, bacteria expressing recombinant ttRecO were grown in M9 medium containing 15NH4Cl and/or 13C-labeled glucose as the sole nitrogen and carbon source, respectively. Expression and purification of ttRecO were performed as described previously (19). RecA (ttRecA), RecR (ttRecR), and SSB (ttSSB) were also prepared as described previously (19, 37). The synthesized ttSSB C-terminal peptide (17 amino acids, DIDEGLEDFPPEEELPF) was obtained from Invitrogen. Unless otherwise stated, the designations RecO, RecR, RecA, and SSB hereafter refer to T. thermophilus proteins.
To prepare RecO mutants R127E/A, H132E, Y174A, T186A, R188E, L189A, and H197E, point mutations were incorporated into the RecO expression plasmid by using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) and were confirmed by DNA sequencing. The RecO mutants were expressed and purified using the same methods used for wild-type RecO.
NMR samples of 15N- and 13C/15N-labeled RecO were prepared at a final concentration of 0.5 mm in 20 mm Tris-HCl (pH 7.5), 2 mm EDTA, 2 mm DTT, and 10 mm MgCl2. The final NMR samples contained 90% H2O, 10% D2O. Two-dimensional HSQC spectra were acquired from 15N- or 13C/15N-labeled RecO. All NMR experiments were performed at 313 K in a triple resonance cryoprobe fitted with a z axis pulsed field gradient coil by using a 600-MHz DRX or AVANCE spectrometer (Bruker BioSpin, Billerica, MA). The sequence-specific backbone 1HN, 13Cα, 13C′, and 15N and side chain 13Cβ resonance assignments of 13C/15N-labeled RecO were obtained from the CBCA(CO)NNH, CBCANNH, HNCO, and HN(CA)CO experiments (38, 39). Data were processed on a Linux PC by using the AZARA 2.7 software package (W. Boucher) (available on the World Wide Web). All spectra were analyzed on a Linux PC with a combination of customized macro programs in the OpenGL version of the ANSIG version 3.3 software (40).
Samples of 15N-labeled RecO for NMR analysis were prepared at a final concentration of 0.2 mm in 20 mm Tris-HCl (pH 7.5), 2 mm EDTA, 2 mm DTT, 10 mm MgCl2, and 250 mm KCl. Unlabeled proteins were directly added to the samples from a concentrated stock to prevent changes in concentration and pH. For titration experiments of SSB, 0.02, 0.04, 0.1, and 0.2 mm SSB were added to 15N-labeled RecO samples. The relative intensities of the HSQC signals in the absence (16 scans/free induction decay) or presence (64 scans/free induction decay) of SSB were plotted against the number of amino acids and mapped onto the ttRecO structure. To titrate SSB C-terminal peptide, 0.04, 0.1, 0.2, and 0.4 mm concentrations of the SSB C-terminal peptide were added to 15N-labeled RecO samples. The chemical shift change (Δave) was calculated and normalized by using the formula, ((Δ1HN)2 + (Δ15N)2)½, where Δ1HN and Δ15N are the chemical shift differences (Hz) along the 1H and 15N axes, respectively.
A structural model of RecO was generated with Modeler 8v2, a comparative homology modeling software (41). For homology modeling calculations, the drRecO structure was obtained from the Protein Data Bank (entry 1U5K) (42). The molecular diagrams were depicted by using PyMOL (Delano Scientific LLC) and MolFeat (FiatLux, Tokyo, Japan).
Analyses of protein-protein interactions by native PAGE were performed as described previously (19). SSB (10 μm) and wild-type or mutant RecO (10 or 50 μm) or cytochrome c (50 μm) were incubated in 10 μl of 50 mm Tris-HCl (pH 7.5) and 2 mm EDTA. After 10 min at 50 °C, the samples were analyzed by 10% PAGE under nondenaturing conditions.
The protein-protein interactions were observed with a Biacore 3000 surface plasmon resonance biosensor (GE Healthcare). Random amine coupling was used to immobilize SSB to the surface of a CM-5 sensor chip. Surface activation by Amine Coupling Kit (GE Healthcare) was followed by an injection of SSB (100 μg/ml), which was diluted in 10 mm NaOAc (pH 4.7) prior to injection. Unliganded sites on the chip were then blocked with ethanolamine. A 500-μl injection of SSB resulted in the immobilization of 2,800 response units (RU) of SSB. Subsequently, the interaction experiments were performed in HBS-EP buffer (10 mm HEPES (pH 7.4), 150 mm NaCl, 3 mm EDTA, 0.005% (v/v) Surfactant P20 (GE Healthcare) at a flow rate of 20 μl/min. Forty microliters of each of the different concentrations of wild-type or mutant RecO were injected, and the chip was treated with 40 μl of 1 m NaCl to regenerate the chip after each injection. The base-line response of the surface of the chip, which was treated with the same reagents but without an injection of SSB, was subtracted from the response obtained from the SSB surface. In RecO-RecR interaction, RecR was immobilized to the sensor chip as described for SSB. A 300-μl injection of RecR (100 μg/ml) resulted in the immobilization of 2,800 RU of RecR. Forty microliters of each of the different concentrations of RecO were injected, and the chip was treated with 40 μl of 1 m NaCl to regenerate the chip after each injection.
Various concentrations of wild-type or mutant RecO in 10 μl of 50 mm Tris-HCl (pH 7.5), 10 mm MgCl2, 50 mm KCl, and 1 mm DTT were incubated with 74-mer ssDNA (20 μm as nucleotide) at 50 °C for 5 min. The samples were then electrophoresed on a 2% agarose gel in TAE buffer (40 mm Tris acetate (pH 8.0), 1 mm EDTA). DNA was visualized by using GelStar (Lonza Group Ltd., Basel, Switzerland). The intensities of the bands corresponding to free ssDNA were analyzed with a CS Analyzer version 2.0 (ATTO, Tokyo, Japan). The amount of RecO-bound ssDNA was calculated by subtraction of the amount of free ssDNA from the total amount of ssDNA.
Circular dsDNA, pUC18, was digested by HincII and converted to ssDNA by heat denaturation. The ssDNA (10 μm) was incubated with SSB (1 μm) in 10 μl of 50 mm Tris-HCl (pH 7.5), 10 mm MgCl2, 1 mm DTT, and 50 mm KCl at 37 °C for 5 min, and wild-type or mutant RecO (1 μm) was added to the solution. To stop ssDNA annealing activity, at the indicated times, 1 μl of 5% SDS and 1 μl of 20 mg/ml proteinase K were added to the reaction mixtures, which were then incubated at 4 °C for 5 min. Next, the samples were incubated at 37 °C for 15 min to complete deproteinization and electrophoresed on a 1% agarose gel in TAE buffer. DNA was visualized by using GelStar (Lonza). The intensities of the bands corresponding to dsDNA were analyzed by using CS Analyzer version 2.0 (ATTO, Tokyo, Japan).
Measurement of the ssDNA-dependent ATP hydrolysis of RecA in the presence of SSB, RecO, and RecR was performed as described previously (19). The lag times were calculated by using Microsoft Excel, as previously described (43). To determine the relative activity, the value obtained during the stationary phase of the reaction (last 5 min) was used.
Our previous study indicated that ttRecO interacts with ttSSB (19). To determine the SSB-binding site of RecO, it is necessary to understand the structure of RecO. Therefore, we reconstructed the tertiary structure of ttRecO by homology modeling with the structure of drRecO (22) as the template (identity, 36%; homology, 54%). The structure of ttRecO generated had two domains: an N-terminal OB fold and a C-terminal helical domain (supplemental Fig. S1A). To map the SSB-binding site of RecO, we used NMR spectroscopic analysis. We measured two-dimensional HSQC spectra in which cross-peaks corresponding to about 70% of the non-proline residues of RecO were observed. Finally, the cross-peaks were completely assigned (supplemental Fig. S1B).
To analyze the SSB-binding site of RecO, we sequentially added 0.02, 0.04, 0.1, and 0.2 mm SSB to 0.2 mm RecO. As a result, decay in the signal intensity for all cross-peaks of RecO was observed, possibly reflecting the increased molecular weight of RecO following complex formation with SSB. To compensate for the decreased sensitivity of the RecO signals in the presence of 0.2 mm SSB, we repeated the scans four more times and compared them with those in the absence of SSB (Fig. 1A). Most cross-peaks were recovered, but some were still weak in the presence of SSB (Fig. 1A, red). Although we could not exclude other possibilities, we speculated that these weaknesses were caused by the interactions of SSB with RecO. Thus, the relative intensity of each cross-peak/residue was plotted as a histogram (supplemental Fig. S2A). We then color-coded the residues exhibiting a relative intensity of less than 10% as red and those exhibiting a relative intensity between 10 and 20% as orange and mapped them onto the model structure of RecO (Fig. 1B). The colored residues clustered primarily on the positively charged region of the C-terminal domain of RecO, suggesting that this region is responsible for SSB binding (Fig. 1, B and C).
The C-terminal acidic region of SSB reportedly participates in the ssDNA binding of the RecOR complex (43), which suggests that the SSB C terminus interacts with the RecOR complex. Considering that RecR does not interact with SSB, the SSB C terminus must interact with RecO of the RecOR complex. To confirm that the SSB C terminus binds to RecO, we designed a 17-amino acid SSB C-terminal peptide and performed titration experiments as for intact SSB. When 0.04, 0.1, 0.2, and 0.4 mm concentrations of the SSB C-terminal peptide were sequentially added to 0.2 mm RecO, we observed significant chemical shift changes in some RecO cross-peaks (Fig. 1D), indicating that the SSB C terminus binds directly to RecO.
To investigate the amino acid residues involved in the SSB C terminus binding, the result was expressed as a change in chemical shift/residue, in which the average caused by the addition of the SSB C terminus was 70 Hz (supplemental Fig. S2B). We color-coded the amino acid residues exhibiting a chemical shift change of greater than 600 Hz as red and those exhibiting a change between 400 and 600 Hz as orange and mapped them onto the model structure of RecO (Fig. 1E). The residues with chemical shifts that were affected by the addition of the SSB C-terminal peptide were found to be clustered mainly at the positively charged region of the C-terminal domain (Fig. 1, C and E). Thus, residues at the C-terminal positively charged region of RecO were affected by the addition of intact SSB and the SSB C-terminal peptide, suggesting that this region is the binding site of SSB and the interaction site of the SSB C terminus.
To verify that the C-terminal positively charged region of RecO is the binding site of SSB, we introduced several mutations into this region and analyzed their effect on the interaction between RecO and SSB. To construct RecO mutants, we focused on positively charged residues and hydrophobic residues in this region because electrostatic and hydrophobic interactions are important for protein-protein interactions. Thus, we constructed RecO mutants R127E, H132E, Y174A, T186A, R188E, L189A, and H197E. Interactions between SSB and these RecO mutants were analyzed by native PAGE. The negatively charged protein (SSB) entered the gel and formed a band (Fig. 2, A and B, lane 1). By contrast, the positively charged protein (RecO) did not enter the gel (Fig. 2C, lane 2). The intensity of the band corresponding to SSB decreased in the presence of wild-type RecO, indicating that wild-type RecO interacted with SSB (Fig. 2, A and B, lanes 2 and 3). By contrast, R127E had little effect upon the band intensity, indicating that the binding affinity of R127E for SSB was decreased (Fig. 2A, lanes 4 and 5). In addition, R188E and L189A exhibited weak interactions with SSB (Fig. 2B, lanes 4 and 5 and lanes 6 and 7, respectively). The affinity of the other mutants for SSB was similar to that of wild-type RecO. In the model structure of RecO, residues Arg-127, Arg-188, and Leu-189 were located close to each other (Fig. 1B), indicating that the region is the binding site of SSB. Because R127E seemed to be highly effective in decreasing the binding affinity for SSB and the arginine residue is well conserved in several species of RecO (supplemental Fig. S3), we performed additional analyses by using an Arg-127 mutation. To avoid the charge effect, we also prepared mutant R127A, which showed decreased binding affinity for SSB, as found for the R127E mutant (Fig. 2C).
To analyze the interaction of SSB with R127A quantitatively, we performed surface plasmon resonance (SPR) analysis. SSB was immobilized to the sensor chip by amine coupling, and then wild type and R127A were injected. When wild-type RecO was injected, an increase in the RU was observed (Fig. 2D). The dissociation constant of this interaction was 8.99 × 10−8 m. By contrast, when R127A was injected, normal sensorgrams were not obtained; in addition, the increase in the RU was significantly lower than that found with wild-type RecO (Fig. 2E). Thus, we could not determine the dissociation constant of this interaction. However, the result indicates that the binding affinity of R127A for SSB was much lower than that of wild-type RecO.
To confirm that other functions of RecO aside from its interaction with SSB were not disrupted by introduction of the R127A mutation, we analyzed the interaction of R127A with RecR by SPR. RecR was immobilized to the sensor chip in the same way as for SSB, and then RecO protein was injected. When wild-type RecO was injected, an increase of RU was observed (Fig. 3A). The dissociation constant for this interaction was 6.6 × 10−8 m. We also observed an increase of RU by injection of R127A (Fig. 3B). Interestingly, the increase of RU was saturated by injection of 1 μm R127A, although injection of the same concentration of wild-type RecO still caused an increase of RU. The dissociation constant for R127A-RecR was 4 × 10−8 m. These results indicate that R127A retains its binding affinity for RecR, which is slightly higher than that of wild-type RecO. Similar results were obtained by native PAGE analysis (supplemental Fig. S4A).
Next, we analyzed the ssDNA binding activity of R127A by the electrophoretic mobility shift assay (EMSA). As described previously (19), wild-type RecO bound to ssDNA and formed large protein-ssDNA complexes, which were trapped in the loading well (Fig. 3C, lanes 2–7, indicated with an arrow). When R127A was incubated with ssDNA, ssDNA was also trapped in the loading well (Fig. 3A, lanes 8–13). This result indicated that R127A bound to ssDNA and formed large protein-ssDNA complexes, as observed with wild-type RecO. The EMSAs were repeated four times and quantified by densitometric analysis (Fig. 3D). The results indicated that the ssDNA binding affinity of R127A was slightly lower than that of wild-type RecO. Another ssDNA binding assay using fluorophotometry and etheno-modified ssDNA also gave similar results (supplemental Fig. S4B). We previously demonstrated that ssDNA binding of RecO was affected by the addition of RecR (19). This modulation of RecO ssDNA binding was also observed with R127A (supplemental Fig. S4B, triangle).
Finally, comparison of the CD spectra of wild-type RecO with those of R127A showed that these two proteins have the same secondary structure and denaturing temperature (supplemental Fig. S4, C and D). These results indicate that R127A lost its ability to interact with SSB but folded properly and maintained the basic functions of RecO.
RecO can reportedly anneal SSB-coated ssDNA (20). We hypothesized that replacement of SSB on ssDNA by RecO is necessary for progression of this reaction. Therefore, the interaction of RecO with SSB is thought to be the initial step of ssDNA annealing. We speculated that the R127A mutation affects the activity of RecO. When SSB was added to heat-denatured ssDNA, spontaneous annealing of ssDNA was inhibited efficiently (Fig. 4A, lane 3). However, SSB-coated ssDNA was converted to dsDNA by the addition of wild-type RecO (Fig. 4A, lanes 4–8), and about 60% of the ssDNA was converted to dsDNA in 10 min (Fig. 4B, circle). By contrast, little annealing of SSB-coated ssDNA occurred when R127A was added to the reaction mixture (Fig. 4A, lanes 9–13), and about 20% of the ssDNA was converted to dsDNA (Fig. 4B, cross). This percentage was similar to that observed in the absence of wild-type RecO or R127A (Fig. 4B, square). These results indicated that the annealing activity of R127A occurred at only the background level and that R127A could not anneal the SSB-coated ssDNA. Because R127A could bind to ssDNA (Fig. 3, C and D), the abolition of the SSB-coated ssDNA annealing activity was most likely caused by the decreased binding affinity of R127A for SSB. This result also supported the idea that the Arg-127 residue was important for binding with SSB.
The R127A mutant abolished SSB-coated ssDNA annealing activity, which probably included the process of SSB displacement. Thus, we examined the effect of the R127A mutation on RecA filament formation on SSB-coated ssDNA. We previously demonstrated that inhibition of the ssDNA-dependent ATPase activity of RecA by SSB recovers efficiently when RecO and RecR are present at a molar ratio of 1:2 (19). Therefore, we used wild-type RecO or R127A and RecR at this molar ratio and referred to the complexes as RecOR and R127AR, respectively. At concentrations greater than 0.35 μm, both the RecOR and the R127AR complexes efficiently recovered the ATPase activity of RecA even in the presence of SSB (Fig. 5A). However, at a concentration of 0.1 μm, the R127AR complex was nearly inactive in this respect, and only the RecOR complex showed efficient recovery of the activity. Thus, the R127A mutation affected the recovery of the inhibition of RecA activity by SSB. Because the R127A mutant could interact with RecR (Fig. 3B and supplemental Fig. S4A), RecO-SSB interactions must be important for the recovery of the ATPase activity of RecA.
Because the initial ssDNA binding (nucleation) of RecA is the rate-limiting step, the rate of ATP hydrolysis does not reach steady state until a stable RecA filament has formed (44,–46). In the presence of SSB, the lag time increases because the ssDNA binding of SSB inhibits the nucleation step of RecA. Reportedly, the lag time in the presence of SSB is efficiently reduced in the presence of the RecOR complex (43). Thus, one of the roles of RecO and RecR is thought to be the displacement of SSB from ssDNA and assistance in the nucleation step of RecA. Therefore, we focused on the lag times of the ATPase activity of RecA and measured them as described by Hobbs et al. (43). In the absence of the RecOR complex, the rate of ATP hydrolysis did not reach steady state during the experiment (1 h) (data not shown). However, in the presence of the RecOR complex, the lag times were reduced (Fig. 5B, gray line); the average lag time in the presence of 0.2–1 and 0.1 μm concentrations of the RecOR complex was 7.5 min and about 11 min, respectively. These results suggested that the RecOR complex efficiently displaced SSB and assisted RecA nucleation. By contrast, in the presence of a 0.1 μm concentration of the R127AR complex, a long lag time was observed (28 min) (Fig. 5B, black line). When the concentration of this complex was increased, the lag times decreased; in the presence of 0.2 μm R127AR complex, the lag time was 15.5 min, and in the presence of 0.35–1 μm R127AR complex, the average lag time was 9.5 min. These results indicated that the lag times in the presence of the R127AR complex were longer than those in the presence of the RecOR complex at every concentration. They also indicated that the Arg-127 residue was important for the interaction with and displacement of SSB from ssDNA. It has been reported that the RecOR complex recovers little RecA activity in the presence of the C terminus-deleted mutant of SSB (43). Because the C-terminal region of SSB was involved in the interaction with RecO (Fig. 1, D and E), our results are consistent with those of the previous study. Taken together, the region surrounding Arg-127 of RecO interacted with the C-terminal region of SSB to assist in the displacement of SSB from ssDNA.
An important role of the RecOR complex is stabilization of the RecA nucleoprotein filament after SSB displacement (23). Our previous study indicated that in the absence of SSB, the RecOR complex has a positive effect on the RecA filament, although the ATPase activity of RecA is inhibited by RecO or RecR alone (19). Both RecO and R127A interacted with RecR (Fig. 3, A and B) and formed RecOR and R127AR complexes, respectively (supplemental Fig. S4A), but the results in Fig. 5A could have been a consequence of the disruption of RecOR function rather than the disruption of the RecO-SSB interaction. To exclude this possibility, we examined the effect of the R127A mutation on the function of the RecOR complex, namely the stabilization of the RecA nucleoprotein filament by the RecOR complex in the absence of SSB. We reasoned that if the function of the RecOR complex is disrupted by introduction of the R127A mutation, then the ATPase activity of RecA would be affected. When wild-type RecO alone was incubated with ssDNA before the addition of RecA, the ATPase activity of RecA decreased (Fig. 6A, gray line with crosses) to 20% of that observed in the absence of RecO (Fig. 6B). However, the ATPase activity of RecA was recovered by the addition of RecR (Fig. 6A, gray line with circles) and reached a level that was 110% of that observed in the absence of the RecOR complex (Fig. 6B). This result indicated that the RecOR complex stabilized the RecA nucleoprotein filament. When R127A was incubated with ssDNA prior to the addition of RecA, the ATPase activity of RecA was again inhibited (Fig. 6A, black line with crosses). However, the inhibitory effect was weaker than that of wild-type RecO because about 45% of the ATPase activity of RecA was still observed (Fig. 6B). The inhibitory effect was recovered by the addition of RecR (Fig. 6A, black line with circles). In addition, the ATPase activity of RecA was stimulated to 130% of that observed in the absence of the RecOR complex (Fig. 6B). These results indicated that the stabilization of the RecA nucleoprotein filament by the RecOR complex was not disrupted by the introduction of the R127A mutation. Therefore, the results in Fig. 5A must have been a consequence of the disruption of the SSB displacement activity of RecO rather than the disruption of RecOR function.
Interaction of RecO and SSB seems to be important for the function of RecO as a recombination mediator. In this study, we identified the SSB-binding site of RecO by biochemical and structural analyses and concluded that the SSB C terminus binds to the C-terminal positively charged region surrounding the Arg-127 residue of RecO. Several proteins participating in DNA metabolism interact with the SSB C terminus (33,–35, 47). Our result indicates that RecO is one of these proteins. The crystalline structure of the E. coli exonuclease I (ExoI)-SSB C terminus peptide complex provided information about the structural properties of the SSB C terminus binding region of ExoI (47). This report indicated that the SSB C terminus is coordinated by electrostatic and hydrophobic interactions because ExoI has a hydrophobic pocket surrounded by basic amino acid residues. Acidic amino acid residues of the SSB C terminus interact with basic residues, called the “basic ridge,” of ExoI. The C-terminal end of phenylalanine of the SSB C terminus is buried in the hydrophobic pocket of ExoI (Fig. 7, bottom). Shereda et al. (48) studied the interaction of RecQ with the SSB C terminus by using NMR analysis and stated that the SSB C terminus-binding site of RecQ has characteristics similar to those of ExoI. Interestingly, we found that the SSB C terminus-binding site of RecO also has characteristics similar to those of ExoI and RecQ, although there is no sequence similarity between RecO and these proteins. The region surrounding the Arg-127 residue is positively charged. In addition, Pro-138, Ala-142, and Leu-187 in this region formed a hydrophobic pocket (Fig. 7, left). This structural feature is more significant in drRecO. Arg-121 of drRecO corresponds to Arg-127 of ttRecO. Ala-118, Met-136, Val-198, and Phe-202 also form a hydrophobic pocket (Fig. 7, right). In Thermus or Deinococcus SSB, the last phenylalanine residue is conserved. Therefore, it is plausible that the phenylalanine residue fits into the hydrophobic pocket of RecO. Thus, our results along with previous findings demonstrate that SSB-binding sites of genomic maintenance proteins have conserved characteristics (48).
Although RecO variants R127A and R127E showed the strongest apparent defect in binding SSB, the chemical shift of Arg-127 was not changed significantly when the SSB C-terminal peptide was added. However, the intensity of the signal corresponding to Arg-127 decreased to some extent when full-length SSB was added. These results raise the possibility that SSB interacts with RecO not only through its C-terminal region but also through its core domain and that Arg-127 of RecO is involved in the interaction with the SSB core domain rather than with the SSB C-terminal region. In our model described below, RecO approaches the SSB core domain (OB-fold) after binding to the SSB C-terminal region (see Fig. 8, C and D). There is a possibility that Arg-127 is important for this process and for stabilization of the RecO-SSB complex.
When Arg-127 was substituted for alanine, the RecO mutant protein showed decreased binding affinity for SSB and failure to anneal SSB-coated ssDNA. Recent studies have suggested that annealing ssDNA coated with members of the SSB protein family is one of the important functions of a recombination mediator. In eukaryotes, this activity is involved in the single-strand annealing (SSA) pathway and second end capture (21, 49,–51). In bacteria, it has been reported that RecO can catalyze second end capture in vitro (21). Therefore, it is possible that the SSB-coated ssDNA annealing activity of RecO plays an important role in vivo, as shown by Rad52. We also demonstrated that the R127A mutant could not recover the inhibition of ATPase activity of RecA by SSB efficiently. A recent study indicated that the E. coli RecOR complex could not recover the ATPase activity of RecA in the presence of an SSB C-terminal deletion mutant (43). In the present study, we demonstrated a direct interaction between RecO and the SSB C terminus. These results indicate that the interaction of RecO with the C-terminal region of SSB is important for RecO function as a recombination mediator.
Our previous study indicated that complete displacement of SSB from the SSB-RecO-ssDNA complex is achieved by the interaction of RecR with the complex (19). Because the RecOR complex remains on ssDNA, one may speculate that the binding sites of SSB and RecR on RecO overlap and that SSB is dissociated by competitive binding of RecR to SSB. However, the crystalline structure of the drRecOR complex indicated that the OB fold of RecO is embedded in the central hole of RecR and that His-93 of drRecO is important for the formation of the RecOR complex (24). His-93 is located on the opposite side of Arg-121, which corresponds to Arg-127 of ttRecO. Actually, the R127A mutation did not disrupt the formation of the ttRecOR complex (Fig. 3B and supplemental Fig. S4A). In addition, the stabilizing effect of the RecOR complex on the RecA filament was not disrupted by this mutation (Fig. 6). These results indicate that complete displacement of SSB does not occur by simple competition between SSB and RecR.
RecA can nucleate and form a stable filament on ssDNA when the length of the ssDNA is longer than 30 nucleotides (52). Thus, when at least one SSB molecule dissociates from ssDNA, RecA can nucleate on the ssDNA, because the binding site size of SSB is larger than 30 nucleotides (53, 54). The role of the RecOR complex may be to dissociate SSB from ssDNA and create a small SSB-free space for nucleation of RecA. In that case, RecA would nucleate on the ssDNA and displace SSB by filament elongation because SSB can diffuse along ssDNA (17). Our result indicates that a relatively low amount of the RecOR complex can reverse the inhibitory effect of SSB on RecA activity (Fig. 5A). In addition, we previously demonstrated that the addition of a small amount of RecO causes a decrease in the fluorescence intensity of the SSB-etheno-modified ssDNA complex (19). These results suggest that RecO binding to the SSB-ssDNA complex causes the release of some ssDNA from the complex (53, 55). However, if the displacement of one SSB molecule is sufficient for RecA nucleation, substoichiometric concentrations of the RecOR complex should overcome the inhibitory effect of SSB. In practice, approximately a one-tenth concentration of the RecOR complex relative to SSB is required (12). The efficient formation of a RecA filament may require the displacement of SSB at not only one site but at several sites on the SSB-coated ssDNA.
The degree of inhibition of the ATPase activity of RecA by R127A was less than that caused by wild-type RecO (Fig. 6). Results of the ssDNA binding assay indicated that the ssDNA binding affinity of R127A was slightly lower than that of wild-type RecO (Fig. 3, C and D, and supplemental Fig. S4B). Because the inhibitory effect of RecO is caused by the strong ssDNA binding of RecO, the reduced ssDNA binding affinity of R127A allows for a larger increase in RecA activity. Our results also indicated that R127AR has a more positive effect on RecA activity than wild-type RecOR (Fig. 6). The binding affinity of R127A for RecR was slightly higher than that of the wild type (Fig. 3, A and B, and supplemental Fig. S4A), and the RecOR complex stabilizes the RecA filament and stimulates RecA activity. Therefore, the more stable R127AR complex may be involved in the stimulation of RecA activity. Alternatively, the RecOR complex associates with RecA after the formation of a RecA filament (23), suggesting that RecO or RecR interacts with RecA on DNA. If RecO interacts with RecA on DNA and stabilizes the RecA filament, the R127A mutation would have a positive effect on the stabilization. To elucidate these possibilities, we will examine the RecO-RecA interaction by NMR titration analysis.
The SSB C terminus reportedly interacts with several proteins and regulates their activities (32,–35). It is possible that some of the activity of RecO may be regulated by its interaction with the SSB C terminus. One possibility is that the ssDNA binding affinity of RecO is increased by the interaction between the SSB C terminus and RecO; in this case, SSB may be displaced from ssDNA by the competitive ssDNA binding of RecO. However, the SSB C-terminal peptide did not increase the ssDNA binding affinity of RecO (supplemental Fig. S5), suggesting that the interaction causes other effects for SSB displacement. The interaction may decrease the ssDNA binding affinity of SSB itself, which would facilitate SSB displacement. The SSB C terminus is disordered in the crystalline structure of the SSB-ssDNA complex, indicating that the region is highly flexible in the complex (56) and that the C terminus is not involved in ssDNA binding (Fig. 8A). However, it has recently been reported that the E. coli SSB C terminus plays an important role in the regulation of the ssDNA binding of SSB (36). This report suggested that in the absence of ssDNA, the acidic C terminus interacts with the positively charged OB fold, which acts as the ssDNA-binding site of the SSB core (Fig. 8A). Upon ssDNA binding, the C terminus is displaced from the OB fold and becomes more accessible to other proteins (Fig. 8B). Therefore, the C terminus has an intrinsic inhibitory effect on the ssDNA binding of SSB. The presence of the C terminus is disadvantageous for the ssDNA binding of SSB (36), which is its primary biological property. This raises the question of why SSB possesses such a highly acidic region that could interfere with ssDNA binding separating from its core domain. It is possible that the negative charge enables SSB to dissociate from ssDNA effectively after recruitment of other proteins to the ssDNA has occurred (Fig. 8, C–F). If SSB maintains strong ssDNA binding after protein recruitment, the ssDNA-bound SSB will interfere with ssDNA binding and functions of the recruited proteins. Therefore, it is plausible that SSB uses its highly flexible C-terminal region to weaken its ssDNA binding activity upon protein binding. The dynamics of the SSB C terminus must be restricted when the C terminus interacts with other proteins (Fig. 8C). If so, the proteins would interact with SSB by using other surfaces or interact with ssDNA through the OB fold of SSB. Thus, the SSB C terminus must be located close to the OB fold of SSB again to weaken the ssDNA binding of SSB (Fig. 8D). As a result, the acidic SSB C terminus bound to other proteins may compete with ssDNA for the OB fold of SSB and displace ssDNA from the OB fold. In this case, rebinding of ssDNA to the OB fold of SSB may be prevented by the interaction of displaced ssDNA with the OB fold and/or the ssDNA-binding sites of other proteins because proteins interacting with SSB also have an OB fold and/or ssDNA-binding site (Table 1). In the case of RecO, its N-terminal domain has an OB fold, whereas its C-terminal domain interacts with the SSB C terminus (Fig. 8, C–F). In addition, we have demonstrated that RecO binds to SSB-coated ssDNA and replaces SSB on ssDNA (19). Therefore, displaced ssDNA probably interacts with the N-terminal OB fold of RecO (Fig. 8E, circled). Alternatively, displaced ssDNA may interact with the C-terminal positively charged region of RecO because this region also has been reported to be involved in DNA binding (22). However, RecR, which is necessary for SSB displacement from the SSB-RecO-ssDNA complex, interacts with the OB fold of RecO (24), which affects the ssDNA binding of RecO (19) (see also supplemental Fig. S4B). Thus, although we cannot exclude other possibilities, it is plausible that 1) displaced ssDNA binds to the OB fold of RecO, 2) RecR binding to the OB fold displaces the ssDNA from the OB fold, and 3) a small SSB-free space on the ssDNA is created by the RecOR complex (Fig. 8F). If so, RecA would nucleate at the SSB-free space (Fig. 8F) and displace SSB by filament elongation (Fig. 8G), because SSB can diffuse along ssDNA. SSB then may assist RecA filament elongation, because its diffusion on ssDNA causes melting of the secondary structure of ssDNA (17).
We have also demonstrated that a single mutation in the C-terminal region of ttSSB (F255P) decreases its ssDNA binding affinity (57). In addition, the ssDNA binding affinity of a ttSSB C-terminal-truncated mutant (SSB(1–229)) was higher than that of intact SSB (data not shown). These results further indicate that the SSB C-terminal region regulates the ssDNA binding affinity of SSB. With regard to the F255P mutation, the residues around Phe-255, EDFPP, have a proline-proline sequence. Because the proline residue has a cyclic side chain and acts as a structural constraint, the proline-proline-proline sequence of the F255P mutant must affect the structure and dynamics of the SSB C terminus. As a result, the F255P mutation may situate the C-terminal region near an OB fold and mimic SSB in a SSB-protein complex (Fig. 8H).
Finally, several functions of replication protein A (RPA; eukaryotic SSB, a heterotrimeric protein) are reportedly regulated by its hyperphosphorylated RPA32 subunit, which binds to an OB fold of the RPA70 subunit (58). The acidic nature of hyperphosphorylated RPA32 influences ssDNA binding and protein-protein interactions of the OB fold of RPA70. Therefore, self-regulation by a separate highly acidic region, such as seen in SSB, may be a conserved feature of ssDNA-binding proteins from bacteria to eukaryotes.
We thank Professor Seiki Kuramitsu (Osaka University) for providing expression plasmids for the T. thermophilus RecF pathway proteins used in this study. We thank Prof. Yoshifumi Nishimura (Yokohama City University) for generous support in measuring CD using the Jasco J-720 spectropolarimeter. We thank Eriko Osada (Tokyo Metropolitan University) and Yohei Doi (Tokyo Metropolitan University) for assistance in NMR resonance assignment.
2The abbreviations used are: