Allosteric effects of E. coli SSB and RecR proteins on RecO protein binding to DNA

Abstract Escherichia coli single stranded (ss) DNA binding protein (SSB) plays essential roles in DNA maintenance. It binds ssDNA with high affinity through its N-terminal DNA binding core and recruits at least 17 different SSB interacting proteins (SIPs) that are involved in DNA replication, recombination, and repair via its nine amino acid acidic tip (SSB-Ct). E. coli RecO, a SIP, is an essential recombination mediator protein in the RecF pathway of DNA repair that binds ssDNA and forms a complex with E. coli RecR protein. Here, we report ssDNA binding studies of RecO and the effects of a 15 amino acid peptide containing the SSB-Ct monitored by light scattering, confocal microscope imaging, and analytical ultracentrifugation (AUC). We find that one RecO monomer can bind the oligodeoxythymidylate, (dT)15, while two RecO monomers can bind (dT)35 in the presence of the SSB-Ct peptide. When RecO is in molar excess over ssDNA, large RecO–ssDNA aggregates occur that form with higher propensity on ssDNA of increasing length. Binding of RecO to the SSB-Ct peptide inhibits RecO–ssDNA aggregation. RecOR complexes can bind ssDNA via RecO, but aggregation is suppressed even in the absence of the SSB-Ct peptide, demonstrating an allosteric effect of RecR on RecO binding to ssDNA. Under conditions where RecO binds ssDNA but does not form aggregates, SSB-Ct binding enhances the affinity of RecO for ssDNA. For RecOR complexes bound to ssDNA, we also observe a shift in RecOR complex equilibrium towards a RecR4O complex upon binding SSB-Ct. These results suggest a mechanism by which SSB recruits RecOR to facilitate loading of RecA onto ssDNA gaps.

Escherichia coli RecO and RecR are essential recombination mediator proteins (RMPs) in the RecF pathway that is primarily involved in repair of single stranded DNA gaps (58)(59)(60)(61)(62)(63) but also plays a secondary role in double strand breaks (64,65). RecO binds to both ss and dsDNA and facilitates the annealing of complementary DNA strands (31,66). A crystal structure of RecO shows the two Cterminal residues of SSB-Ct (Pro and Phe) bound in a hydrophobic pocket of the central alpha helical region, similar to ExoI and RecQ (36,49,67). E. coli RecR, exists in a pH-dependent dimer-tetramer equilibrium and can form two species of protein complexes with RecO--RecR 4 O and RecR 4 O 2 --depending on the molar ratio of the two proteins (68). The main role of RecO, together with RecR, is to displace SSB molecules that are tightly bound to ssDNA and load RecA protein filaments onto ssDNA to initiate homologous recombination (33,(69)(70)(71)(72)(73)(74)(75)(76). E. coli RecO is a SIP (32,55,57), and a proposed mechanism for the loading of RecA onto ssDNA by RecOR suggests that RecO is recruited by the SSB-Ct through a direct interaction (77). However, the details of the interactions between the components of the RecOR pathway, RecO, RecR, SSB and ss-DNA, and their stoichiometries are still unclear.
Binding of SIPs to the SSB-Ct was initially viewed only as a means to tether the SIP to SSB in order to facilitate its binding to DNA; however, it has been shown that SSB-Ct binding to at least some SIPs can exert an allosteric effect on SIP activities. SSB-Ct binding has a stimulatory effect on RecQ helicase activity (34,78). SSB-Ct peptide also stimulates ATP hydrolysis by E. coli RadD (54), a protein implicated in double strand (ds) break repair (79,80). Thus, the SSB-Ct may affect the properties of other SIPs (34,54,78). We have previously demonstrated an allosteric effect of an SSB-Ct peptide on the interaction of E. coli RecO with RecR (68). Although E. coli RecR does not interact with SSB or DNA (33,81), an SSB-Ct peptide allosterically stabilizes RecR 4 O complexes (68). Here we show that the SSB-Ct also affects the ssDNA binding activity of E. coli RecO and RecOR complexes.

Proteins, peptides and DNA
Escherichia coli RecO protein was overexpressed from plasmid pMCSG7 in E. coli strain BL21(DE3) pLysS (kindly provided by Dr Sergey Korolev, Saint Louis University) and purified using Ni-NTA affinity chromatography and a Hi-Trap Heparin HP affinity column (GE Healthcare, Chicago, IL, USA) after His-tag cleavage with TEV protease as described (36). The auto-inactivation-resistant S219V mutant of TEV protease with an N-terminal His-tag and Cterminal polyarginine tag (His-TEV(S219V)-Arg) was overexpressed from E. coli strain BL21(DE3) transformed with PRK793 and pRIL (Stratagene, San Diego, CA, USA) and purified as described (82). E. coli RecR protein was overexpressed from plasmid pMCSG7 in E. coli strain BL21 Rosetta 2(DE3)pLysS (kindly provided by Dr. Sergey Korolev) and purified using Ni-NTA affinity chromatography, followed by cleavage of His-tag with TEV protease as described (83). The concentrations of RecO and RecR in monomers were determined using extinction coefficients of ε 280 = 2.44 × 10 4 M −1 cm −1 and ε 280 = 5.96 × 10 3 M −1 cm −1 , respectively, as determined from their amino acid sequences by SEDNTERP (84).
SSB-Ct peptide, composed of the 15 C-terminal amino acids (PSNEPPMDFDDDIPF) of E. coli SSB, was purchased from WatsonBio (Houston, TX, USA). The SSB-Ct peptide concentration was determined using an extinction coefficient of ε 258 = 390 M −1 cm −1 .

Analytical ultracentrifugation (AUC)
Sedimentation experiments were performed with an Optima XL-A analytical ultracentrifuge and An50Ti or An60Ti rotors (Beckman Coulter, Fullerton, CA, USA) at 25 • C as described (18,68). Absorbance was monitored at 546 nm for the Cy3-labeled DNA (Figures 4-6, and S2a) and at 260 nm for unlabeled DNA ( Figure S2b). Absorbance was also monitored at 230 nm. All sedimentation experiments were performed at least twice. In fact, each experiment was repeated using two entirely different RecO and RecR protein preps, yielding identical results. Furthermore, all of the sedimentation results are fully consistent with our previous studies of RecO and RecR interactions that were performed under the same solution conditions in the absence of DNA (68).
The densities and viscosities of the buffers at 25 • C were determined using SEDNTERP (84). The partial specific volume,ῡ, of RecO and RecR were determined from independent sedimentation equilibrium experiments on each protein in buffer BTP (68). The values ofῡ determined in buffer BTP are 0.734 ml/g for RecO and 0.711 ml/g for RecR. These values differ from the ones calculated using SEDNTERP by 1.2% and 2.7% for RecO and RecR, respectively (0.743 ml/g for RecO and 0.731 ml/g for RecR). Thē υ of the SSB-Ct peptide was calculated using SEDNTERP, yielding 0.704 ml/g.ῡ of 0.56 ml/g was used for DNA (85). In experiments involving more than one species, the partial specific volumes of complexes were calculated assuming additivity using Equation (1), where n i = number of moles of species 'i', M i = molecular weight of species 'i', and υ i = partial specific volume of each species 'i'. . Sample (110 l) and buffer (120 l) were loaded into each sector of an Epon charcoal-filled six-channel centerpieces. Absorbance data were collected by scanning the sample cells at intervals of 0.003 cm in the step mode with 5 averages per step. Samples were sedimented to equilibrium at the indicated rotor speeds (ranging from 18 000 to 28 000 rpm) starting with the lowest speed. The resulting absorbance profiles, A r , were analyzed by NLLS fitting to Eq. (2) as implemented in Sedphat (87) to obtain molecular weights using 'Species Analysis with Mass Conservation Constraints' model: where r is the distance from the center of rotation, r 0 is an arbitrary reference radius, is angular velocity, T is absolute temperature, R is the gas constant, M i is the molecular weight of species 'i', υ i = partial specific volume of each species 'i', is the buffer density, A r 0 ,i is the absorbance of species 'i' at the reference position, and b r is a radialdependent baseline offset. All sedimentation equilibrium experiments in this study were described by a single exponential and globally fit to a one species model ( Figures 5  and 6C). When a two species model was attempted (nucleoprotein complex and unbound DNA), the fraction of the second species was less than 1%.

Confocal microscopy
Confocal fluorescence measurements were performed using a Picoquant MT200 instrument (Picoquant, Germany). The microscope (Olympus IX-73, Japan) was equipped with a piezo scanner and a high numerical aperture water immersion objective (60 × 1.2 UPlanSApo Superapochromat, Olympus, Japan). Fluorophores were excited using a 485 nm pulsed laser (LDH PC-485, Picoquant, Germany) with a repetition rate of 20 MHz. Excitation power was monitored before the objective with a laser photodiode and optimized to avoid photobleaching and saturation of detectors to maintain a constant power for each set of measurements. Emitted photons were collected through the objective, passed through a dichroic mirror (ZT488/594rpc-UF3, Chroma, Bellows Falls, VT, USA), and filtered by a 100 m pinhole (Thorlabs, Newton, NJ, USA). Photons were separated according to polarization using a polarizer beam splitter cube (Ealing, Scotts Valley, CA, USA) and further refined by a 642 ± 40 nm bandpass filter (E642/80m, Chroma, Bellows Falls, VT, USA) in front of the SPAD detectors (Excelitas, Waltham, MA, USA). Photons are counted and accumulated by a HydraHarp 400 TCSPC module (Picoquant, Germany) with 1 picosecond resolution (88). Measurements were performed in uncoated polymer coverslip cuvettes (30 l per well) (Ibidi, Germany), which significantly decrease the fraction of protein adhering to the surface compared to glass cuvettes. Measurements were performed at 23 ± 1 • C in a temperature-controlled room, as detected on the microscope stage.
Imaging was performed using both XY and Z monodirectional scanning with 1 ms collecting steps with 256 × 256 pixels resolution. Excitation power for image collection was either 1.0 or 11 W depending on sample concentration. Measurements were performed keeping a constant ratio between Cy3-labeled and unlabeled protein (labeled:unlabeled = 1:100) in buffer BTP with 0.002% Tween-20. Brightness thresholds were set at 50 and 1200 photons/pixel, which removed most of the background and prevented saturation in images. Images are colored in a hue scale running from blue at 100 photons/pixel to red at 800 photons/pixel.

Light scattering
Light scattering at 90 • was measured using a PTI QM-2000 fluorometer (Photon Technologies, Inc., Lawrenceville, NJ, USA) with excitation and emission wavelengths at 350 nm.
Nucleic Acids Research, 2023, Vol. 51, No. 5 2287 ssDNA (1.9 ml of 25 nM DNA molecules) was titrated with RecO (5 M stock) in a 3 ml quartz cuvette in buffer BTP. Samples were stirred throughout the experiments using magnetic stir bars. For experiments in the presence of SSB-Ct, SSB-Ct was pre-mixed with ssDNA at the start of the experiments. SSB-Ct (3.8 M) was in 6-fold molar excess of the final concentration of RecO in the cuvette at the end of titration. The stock solution of RecO was also pre-mixed with 3.8 M of SSB-Ct to keep [SSB-Ct] inside the cuvette constant throughout the titrations. Reference titrations were also performed in which protein titrant was added to a 1.9 ml of buffer that does not contain DNA both in the absence and presence of SSB-Ct. All sedimentation experiments were performed in duplicate, with each experiment using different RecO and RecR protein preps.
Light scattering intensities were normalized as in Eq. (3), where I i,norm is the normalized scattering intensity after 'i'th injection of titrant (RecO), I i is the scattering intensity after 'i'th injection of titrant, and I 0 is the initial scattering intensity before injection of any titrants.

Large RecO-ssDNA aggregates form in vitro when RecO is in excess over DNA
Previous studies have shown that E. coli RecO can bind both ss and dsDNA and anneal complementary ssDNA when complexed with SSB (31,36). One of these studies (36) was carried out in buffer containing high concentrations of arginine (50 mM NaGlu and 50 mM Arg-HCl, pH 8.0) that suppresses RecO aggregation. The experiments reported here were performed in a more conventional buffer (buffer BTP (pH 8.0, 50 mM NaCl)) that tends to promotes RecO aggregation as discussed below. We made several attempts to use fluorescence signals from either RecO or a labeled DNA to obtain quantitative information on the binding of RecO to ssDNA. These in-  18 , with RecO showed similar inconsistencies ( Figure 1C). These results suggest that large complexes form during the titrations and that the resulting light scattering interferes with the spectroscopic experiments.
To examine this further, we monitored light scattering during a titration of 0.2 M (εdA-dT) 71 with RecO under the same solution conditions (buffer BTP, pH 8.0, 50 mM NaCl, 25 • C) used to monitor εdA fluorescence as described in Methods. Figure 1A 15 with RecO ( Figure 2A, open circles). This indicates either the absence of (dT) 15 binding to RecO or formation of complexes that do not result in aggregation or phase separation. However, significant light scattering was observed for RecO titrations of (dT) 35 , (dT) 70 and (dT) 140 . Furthermore, the RecO to (dT) L ratio at which the onset of light scattering occurs increases with increasing DNA length at RecO to (dT) L molar ratios of 5.2 for (dT) 35 , 6 for (dT) 70 and 7.1 for (dT) 140 ( Figure 2B-D). This suggests that a critical binding density of RecO on the ssDNA is required to initiate light scattering. The maximum scattering intensities also increase with increasing DNA length ( Figure  2, empty circles). The observation of higher scattering intensities with longer (dT) L suggests that multiple RecO molecules binding to a longer DNA molecule facilitate formation of large RecO-ssDNA complexes, and that the size of the complexes increases with increasing DNA length.
Similar scattering experiments were performed with ds-DNA of 18 and 60 bp (ds18 and ds60, respectively). Figure  S2 shows increases in scattering intensities for both ds18 and ds60 with the onset of scattering occurring at higher RecO/DNA ratios for the longer DNA as in the case of ssDNA. However, a notable difference is that scattering is readily observed even for the short 18 bp DNA whereas ss-DNA of similar length, (dT) 15 , did not exhibit observable scattering. This indicates that RecO binding to dsDNA is more prone to aggregation compared to ssDNA of similar length.
We  70 , which were performed by mixing with 3'-Cy3-(dT) 68 . As shown in Figure 3A, we observe amorphous fluorescent structures under these conditions for all (dT) L . These structures did not merge or split over several minutes of imaging, suggesting that the structures are not dynamic, but are solid aggregates. These aggregates appeared larger and more elongated for the longer (dT) L , consistent with the higher light scattering intensity for these longer DNA lengths ( Figure 2). Interestingly, Figure 3Ai shows that aggregates form even for RecO binding to (dT) 15 , although significant light scattering was not observed at this RecO/(dT) 15 ratio ( Figure  2A). It is possible that the confocal imaging is more sensitive in detecting small aggregates than is light scattering.

SSB-ct peptide binding to RecO inhibits large RecO-ssDNA complex formation
Escherichia coli single stranded binding (SSB) protein interacts with RecO via the last nine amino acids of SSB's C-terminal intrinsically disordered tails (SSB-Ct) (26,31,33,36,57). We therefore examined whether the interaction of RecO with DNA is influenced by its binding to the SSB-Ct. For these studies we used a 15 amino acid peptide (PSNEPPMDFDDDIPF), that contains the last 15 amino acids of the SSB-Ct, including the region that binds RecO.
Our previous studies showed that SSB-Ct forms a 1:  Figure 2D) with the maximum scattering intensity ∼6-fold lower than in the absence of SSB-Ct. Experiments with RecO and dsDNA in the presence of SSB-Ct exhibited an increase in scattering intensity for both ds18 and ds60 in contrast to ssDNA ( Figure S2). In the presence of SSB-Ct, however, the onset of scattering occurred at higher RecO/DNA ratios than in the absence of SSB-Ct (0.5 and 7 for ds18, and 2.5 and 18 for ds60, respectively). The maximum scattering intensities were similar for ds18 in the absence and in the presence of SSB-Ct, but the maximum intensity was reduced ∼3-fold for ds60 in the presence of SSB-Ct.
We next used confocal fluorescence microscopy to examine the effect of the SSB-Ct peptide (24 M) on mixtures of RecO and(dT) L at a 20-fold molar excess of RecO (4 M) over (dT) L (200 nM DNA molecules, labeled:unlabeled = 1:100 molar ratio) as before. Images showed mostly black background indicating that the binding of SSB-Ct to RecO significantly reduced the formation of the aggregated RecO-DNA structures ( Figure 3B). No aggregates were observed for the RecO-(dT) 15 and (dT) 35 complexes, and only a few small fluorescent aggregates were observed for the RecO-(dT) 68 and (dT) 140 complexes, significantly reduced in size and number (Figure 3Biii, iv). This is consistent with the significantly reduced light scattering intensities observed in the presence of SSB-Ct (Figure 2, filled circles). Hence, the binding of SSB-Ct to RecO reduced its tendency to form aggregates with ssDNA. However, aggregates formed in the absence of SSB-Ct did not dissolve upon addition of SSB-Ct. It was unclear from these results whether the decreased aggregation was due to a lower binding affinity of SSB-Ct-bound RecO to (dT) L or to a difference in the properties of a SSB-Ct-RecO-(dT) L ternary complex. To clarify this, we performed sedimentation velocity experiments as described below.

SSB-ct peptide enhances RecO affinity for ssDNA
We used analytical ultracentrifugation to examine the binding of RecO to ssDNA labeled with a Cy3 probe on its 3' dicating increased RecO binding to (dT) L . Hence, SSB-Ct binding to RecO enhances the RecO-(dT) L binding affinity. We also note that the c(s) species distributions for Cy3-(dT) 68 and Cy3-(dT) 140 in Figure 4C are noticeably asymmetric indicating that multiple RecO-(dT) L complexes form when SSB-Ct-RecO binds to these longer ssDNA molecules.

One RecO molecule binds to (dT) 15 while two RecO molecules can bind to (dT) 35 in the presence of SSB-ct
The N-terminal domain of RecO contains the DNA binding domain (83), however, there is no information available on the occluded site size (90) or the ssDNA contact size for RecO binding to ssDNA. This information is important since these properties constrain the RecO binding stoichiometries for each (dT) L . In order to assess these stoichiometries, we performed sedimentation equilibrium experiments using a four-fold molar excess of RecO (2.24 M) over Cy3-(dT) 15 or Cy3-(dT) 35 (0.56 M), and a six-fold molar excess of SSB-Ct (13.4 M) over RecO at three rotor speeds (18 000, 23 000 and 28 000 rpm). The sedimentation equilibrium profiles showed only a single exponential for the RecO complexes with both Cy3-(dT) 15 and Cy3-(dT) 35 . This is consistent with the c(s) species distributions from sedimentation velocity that showed only a single species under these same conditions ( Figure 4). The sedimentation equilibrium data were therefore fit to a one species model with mass constraint (Eq. 2) to obtain molecular weights of 31.8 ± 1.2 kDa for the SSB-Ct-RecO-(dT) 15 complex (Figure 5A) and 70.1 ± 3.4 kDa for the SSB-Ct-RecO-(dT) 35 complex ( Figure 5B). These are consistent with the expected molecular weights of 33.8 kDa and 69.5 kDa, respectively, for a complex containing one SSB-Ct-RecO bound to (dT) 15 and two SSB-Ct-RecO bound to (dT) 35 .

RecR inhibits RecO-(dT) L aggregation
We  (2.24 M) in order to avoid free RecO protein in the mixture. Surprisingly, we did not observe aggregation under these conditions, even in the absence of SSB-Ct for any length of (dT) L ( Figure S3a). This contrasts with significant aggregation in the absence of RecR, particularly for longer DNA ( Figure 3A). In the presence of RecR, aggregates were also not observed in the presence of SSB-Ct ( Figure S3b). This demonstrates a sec- Ct (6.3% in the absence of SSB-Ct, 20.4% in the presence of SSB-Ct). The sedimentation coefficient of this peak (1.3 S) may represent either unbound (dT) 140 or a RecO-(dT) 140 complex. Since the presence of multiple species complicates the identification of the RecOR-ssDNA complexes for the longer (dT) L , we performed sedimentation equilibrium experiments on the mixture of RecO, RecR, SSB-Ct and Cy3-(dT) 15 , which displays a symmetric sedimentation velocity peak at 2.9 S, suggesting a homogeneous species ( Figure  6B). Furthermore, we note that the shape of the peak for the RecO-(dT) 15 species changes ( Figure 6A and B, blue) from a wider asymmetric distribution in the absence of SSB-Ct to a single symmetric peak in the presence of SSB-Ct.
Sedimentation equilibrium experiments were performed to estimate the MW and thus identify the composition of the RecOR complexes. A mixture of RecO (2.24 M) RecR (8.96 M) and Cy3-(dT) 15 (0.56 M) and SSB-Ct (13.4 M) was examined ( Figure 6C). Each sedimentation equilibrium profile can be described by a single exponential, consistent with the single symmetric c(s) peak observed for the SSB-Ct-RecOR-(dT) 15 complex by sedimentation velocity ( Figure 6B, blue). Therefore, the sedimentation equilibrium data were fit to a one species model with mass constraint (Eq. 2), yielding an estimated MW of 131.1 ± 9.2 kDa, consistent, within error, with a RecR 4 O-SSB-Ct complex bound to one (dT) 15 (121.2 kDa). Even the upper limit of the MW estimate is 10.3 kDa less than the predicted MW (155.6 kDa) for two (dT) 15 molecules bound to RecR 4 O 2 along with two SSB-Ct molecules. In contrast to the symmetric peak observed in the presence of SSB-Ct, the asymmetric peak for (dT) 15

Allosteric effects of SSB-ct and RecR on RecO-ssDNA aggregation
We showed previously that the SSB-Ct exerts an allosteric effect on RecOR complex formation (68). Two types of RecOR complexes can be formed: RecR 4 O and RecR 4 O 2 , and SSB-Ct binding to RecO preferentially stabilizes the RecR 4 O complex. In this study we report evidence for a second allosteric effect of the SSB-Ct on RecO binding to DNA. Aggregation of RecO-ssDNA complexes is inhibited when SSB-Ct is pre-bound to RecO. This is not due to the inability of SSB-Ct-bound RecO to interact with ssDNA, as we have shown that (dT) L from 15 to 140 nts interacts with SSB-Ct-bound RecO when RecO is in excess over ss-DNA. We also observe that RecO-ssDNA aggregation is completely inhibited in the presence of RecR even in the absence of SSB-Ct, demonstrating an allosteric effect of RecR on binding of RecO to ssDNA.
Using a pull-down assay with (dT) 45 and (dT) 70 , Ryzhikov et al. (36) showed that a complex of RecO-RecR-SSB-ssDNA can form with full-length SSB protein, as RecO and RecOR interact with both free and SSB-bound DNA (36). These interactions were examined in quite dif-ferent solution conditions (50 mM NaGlu and 50 mM Arg-HCl, pH 8.0) that inhibit RecO aggregation (36).
The potential biological significance of the RecO-ssDNA aggregates is not clear. Although prior binding to SSB-Ct inhibits the aggregation process, the irreversible nature of the RecO-ssDNA aggregates, even upon addition of SSB-Ct, suggests that the aggregates may not interact with SSB-Ct. As the RecOR pathway for loading RecA requires a direct interaction between RecO and SSB (77), it is likely that RecOR and DNA interact under conditions where aggregates do not form. In fact, it has been suggested that RecO first interacts with SSB-Ct (36,68), which promotes RecO binding to DNA while remaining bound to SSB-Ct (Figure 7). In this sequence of events, it is unlikely that aggregates would form. Furthermore, subsequent formation of the RecOR complex also inhibits aggregation. While the RecO concentration has not been determined accurately, the relative abundance of RecO, RecR and SSB can be inferred from the reported rates of protein synthesis in E. coli. Li, et al. (92), reported that the rate of SSB synthesis is much faster than that of RecR, and RecR synthesis is faster than RecO synthesis in MOPS complete media growth conditions as follows: SSB (14444 molecules per generation) >> RecR (1342 molecules per generation) > RecO (85 molecules per generation) (92). At such ratios where RecR is in large excess over RecO, the RecR 4 O complex should be populated rather than RecR 4 O 2 complexes (68). Together with the observation that the formation of RecO-DNA aggregates are inhibited by RecR( Figure S3), we suggest that the functional state of the quaternary SSB-RecO-RecR-DNA complex to be soluble without any RecO-DNA aggregates.
Harami et al. (93) have reported condensate formation of SSB and that RecQ, another SIP, can bind to SSB within these condensates that may function to store SSB that can be released rapidly upon DNA damage or stress. SSB condensate formation is promoted by potassium glutamate, the major monovalent E. coli salt (21). A translesion synthesis polymerase Pol IV has also been proposed to function by forming a pool of SSB and Pol IV at the site of DNA replication stress. It is possible that RecO or a RecOR complex can also form a condensate together with SSB and other SIPs to play a similar role as RecQ and Pol IV.
Under certain solution conditions requiring the presence of acetate or glutamate salts, SSB protein can promote the condensation or collapse of polymeric ssDNA beyond the compaction that occurs due to wrapping of ssDNA around the SSB tetramer in the (SSB) 65 complex, indicating long-range, non-nearest neighbor intramolecular interactions (18)(19)(20)(21). The binding of RecO results in a further condensation of the ssDNA-SSB nucleoprotein complex, possibly by inducing a change in the binding mode of SSB (20). Such a change in the binding mode of SSB due to SSB-RecO interaction was also suggested by Ryzhikov et al. (36), which would result in a release of ssDNA, rendering the nucleotides available for RecO to bind and bridge distant sites on the DNA. Furthermore, it has been suggested that RecO(R) can interact with long ssDNA-SSB filaments in trans to facilitate annealing of complementary strands by RecO (20). Our observation of reduced aggregation of RecO-ssDNA in the presence of SSB-Ct peptides, without the DNA binding domains of SSB, suggests that there is also a change in the properties of RecO-ssDNA complexes, such that the complex remains soluble during annealing even at or beyond the critical binding density that promotes aggregation. This additional condensation of SSB-ssDNA complex in the presence of RecO and RecR brings remote regions of ssDNA together, which may facilitate a homology search by RecA (94).

SSB-ct affects both RecO and RecOR binding to ssDNA
In addition to the allosteric effect of SSB-Ct that inhibits RecO-ssDNA aggregation, we also observed that SSB-Ct affects binding of both RecO and RecOR to ssDNA. Since the SSB-Ct interacts only with RecO in a hydrophobic pocket remote from the N-terminal DNA binding domain and not with RecR or ssDNA (33,81,91), we suggest that the effects of SSB-Ct on the DNA binding activity of RecO are allosteric.
Ryzhikov et al. (36) have shown that ssDNA bound to SSB C8, a construct which lacks the C-terminal 8 amino acids of the acidic tip that binds RecO, does not bind to RecOR as well as ssDNA-bound to wild-type SSB, suggesting that the SSB-Ct facilitates recruitment of RecOR to ss-DNA (36). Our observation of enhanced binding of RecO-SSB-Ct to ssDNA would ensure that RecO remains bound to ssDNA until a RecOR complex is formed. Since RecR exists in a dimer-tetramer equilibrium and that RecO facilitates RecR tetramer formation (68), it is possible that the RecR tetramer is loaded by the SSB-RecO complex as a ring around ssDNA. Furthermore, the binding of the SSB-Ct to RecOR favors the RecR 4 O species, rather than RecR 4 O 2 (68). Based on this, we hypothesize that the SSB-RecR 4 O-ssDNA is the functional complex involved in RecA loading.
Formation of the SSB-RecR 4 O-ssDNA complex may induce a change in ssDNA binding mode of SSB (20,36). Other SSB interacting proteins (SIPs), such as E. coli RecQ, PriA and PriC, have been shown to influence the SSB-ssDNA binding mode, favoring the (SSB) 35 mode (34,56,95) that occludes less ssDNA, thus making more ssDNA available for SIP binding. RecOR would then bind to the free ssDNA where RecR can function to stimulate RecA loading (33,77,81). Furthermore, SSB molecules that are tightly bound to DNA also must be displaced in order to load RecA onto ssDNA (33,(69)(70)(71)(72)(73), as shown also in D. radiodurans and B. subtilis (75,76). A change in SSB binding mode to a less compact (SSB) 35 would make more ssDNA available and facilitate RecA loading.

DATA AVAILABILITY
The data underlying this article will be shared on reasonable request to the corresponding author.