Structural and Functional Studies of H. seropedicae RecA Protein – Insights into the Polymerization of RecA Protein as Nucleoprotein Filament

The bacterial RecA protein plays a role in the complex system of DNA damage repair. Here, we report the functional and structural characterization of the Herbaspirillum seropedicae RecA protein (HsRecA). HsRecA protein is more efficient at displacing SSB protein from ssDNA than Escherichia coli RecA protein. HsRecA also promotes DNA strand exchange more efficiently. The three dimensional structure of HsRecA-ADP/ATP complex has been solved to 1.7 Å resolution. HsRecA protein contains a small N-terminal domain, a central core ATPase domain and a large C-terminal domain, that are similar to homologous bacterial RecA proteins. Comparative structural analysis showed that the N-terminal polymerization motif of archaeal and eukaryotic RecA family proteins are also present in bacterial RecAs. Reconstruction of electrostatic potential from the hexameric structure of HsRecA-ADP/ATP revealed a high positive charge along the inner side, where ssDNA is bound inside the filament. The properties of this surface may explain the greater capacity of HsRecA protein to bind ssDNA, forming a contiguous nucleoprotein filament, displace SSB and promote DNA exchange relative to EcRecA. Our functional and structural analyses provide insight into the molecular mechanisms of polymerization of bacterial RecA as a helical nucleoprotein filament.


Introduction
The bacterial RecA protein plays a role in the complex system of DNA damage repair. RecA protein catalyzes strand exchange reaction between single-strand DNA (ssDNA) and homologous double-strand DNA (dsDNA) molecules, and also induces the expression of DNA repair proteins in response to DNA damage through a regulatory network known as the SOS response [1][2][3][4][5][6][7][8].
The Escherichia coli RecA protein (EcRecA) monomers bind onto DNA producing a righthanded helical nucleoprotein filament [9,10]. Recently the EcRecA crystal structures of RecA-ssDNA and RecA-dsDNA complexes have been reported [11], confirming that each RecA monomer interacts with three nucleotides of ssDNA or three base pairs of dsDNA. In the ssDNA complex structure the ATP-binding site is located at the subunit-subunit interface. The EcRecA-ssDNA nucleoprotein filament represents the structural intermediate responsible for homology pairing to a donor dsDNA, and the RecA-dsDNA structure is an end product after the strand exchange reaction [11].
The bacterial RecA monomer has three major domains: a small N-Terminal (NTD), a core ATPase domain and a large C-terminal domain (CTD). Members of RecA family, including the RadA protein of Archaea and the Rad51 and Dmc1 proteins of eukaryotes, share a core ATPase domain that contains the nucleotide binding site, the conserved Walker A and B motifs, and the putative DNA binding site(s), designed loops L 1 and L 2 [12][13][14][15][16][17][18].
The NTD of RecA family also shares an important domain responsible for self-polymerization of monomers in the presence or absence of DNA. The NTDs of the human Rad51 and the archaeal RadA protein have been both implicated in dsDNA binding [12,19], while RecA protein has an extra CTD for dsDNA binding [20]. The NTD of EcRecA protein structure contains 33 amino acids and assumes a conformation including an α-helix motif and a coil region. The coil region has a short β-loop polymerization motif, which interacts with a β-strand from the core ATPase domain, and is responsible for polymerization of the helical nucleoprotein filament in the compressed and extended RecA nucleoprotein conformations. Similar to the βloop of RecA protein, RadA/Rad51/Dmc1 proteins have a β-polymerization motif in the NTD which together with a subunit rotation motif is responsible for structural transitions from an inactive hexameric ring to right and left-handed filaments [12,13,21,22].
The RecA protein can bind both ssDNA and dsDNA, but the nucleation occurs much more rapidly on ssDNA [8]. RecA protein nucleates onto DNA as oligomers of 4-6 subunits, and then the polymerization rapidly extends primarily in the 5 0 to 3 0 DNA direction at rates greater than 1000 monomers/min at 37°C [23][24][25]. Two conformations of helical RecA filaments have been structurally characterized. The first is an inactive and compressed form, formed by RecA alone or bound to ADP, either in the absence of DNA or bound to ssDNA or dsDNA. The second one is an active and extended filament, formed in the presence of DNA and ATP-γ-S or ATP [24,[26][27][28].
In the present study, we report a functional and structural characterization of a RecA protein from Herbaspirillum seropedicae SmR1 (HsRecA), an endophytic bacterium capable of fixing nitrogen under ammonium and oxygen limiting conditions. Analysis of the H. seropedicae SmR1 genome sequence indicates the absence of recF, dinI, dinB, recBCD, sbcA and sbcCD genes and the presence of a truncated umuC gene [29,30], suggesting that RecA plays an important and novel role in DNA recombination and repair systems in this bacteria. Using ATPase, strand exchange assays and electron microscopy images, we compared wild type HsRecA and EcRecA protein activities, focusing on HsRecA protein binding onto ssDNA. We also determined the three-dimensional structure of HsRecA protein and explored the molecular mechanisms of polymerization of bacterial RecA as a helical nucleoprotein filament.

Material and Methods Reagents
Restriction enzymes were purchased from New England Biolabs. Glycerol, Tris buffer were purchased from Fisher. All other reagents were obtained from Sigma unless otherwise described.

Plasmid Construction, Cloning and Overexpression Host
The fragment NdeI/BamHI containing the wild type recA gene of H. seropedicae from the plasmid pAETWT-HMK [29] was cloned into NdeI-BamHI digested vector pET21a (Novagen). The integrity of the cloned fragment was confirmed by sequencing. E. coli STL2669, a nuclease-deficient strain, was used as host to overexpress the H. seropedicae RecA protein after 0.42 mM isopropyl-β-D-thiogalactopyranoside (IPTG) addition and 3-4 h incubation.

Protein Purification
The native wild type EcRecA protein was purified using previously described protocols [31,32] and the native wild type HsRecA was purified as follows. All purification steps were carried out at 4°C. Cell paste (approximately 13 g) containing RecA protein was flash-frozen with liquid N 2 , then thawed overnight on ice in a lysis solution of 250 mM Tris-HCl (80% cation, pH 7.8) and 25% (w/v) sucrose, adjusting cell to 20% (w/v) ratio. The cells were frozen 2 times in N 2 , following addition of lysozyme solution (2.5 mg/mL final concentration lysozyme in 250 mM Tris-HCl (80% cation, pH 7.8), and addition of 0.02 mL of 500 mM EDTA per mL of final lysis solution. The lysate was sonicated for 20 min, using 30s on / 30s off cycle, 60% output, and then centrifuged per 1h30min to remove cells debris. DNA-binding proteins and HsRecA protein was precipitated from the lysate supernatant with addition of 0.111 mL of 5% (w/v) polyethyleneimine per mL of lysate and incubated for 1 h. The pellet was washed with R-buffer (20 mM Tris-HCl (80% cation, pH 7.8), 10% glycerol, 0.1 mM EDTA and 1 mM dithiothreitol) and then HsRecA protein was extracted from the pellet by addition of R-buffer + 300 mM ammonium sulfate two times.
The BioRex-70 (BioRad) (pH adjusted to 7.5 using 12 N HCl) was equilibrated with Rbuffer + 300 mM ammonium sulfate and used to remove the excess of polyethyleneimine. The extracted HsRecA protein was then precipitated with 0.28 g/mL ammonium sulfate and centrifuged. The resulting pellet was washed twice with R-buffer + 0.28 g/mL ammonium sulfate. The pellet was resuspended in R-buffer + 50 mM KCl and loaded onto DEAE Sepharose fast flow column (GE Healthcare). HsRecA protein was eluted from the DEAE Sepharose column with a 7 column volume of a linear KCl gradient (0.05-1 M) in R-buffer. The eluted HsRecA protein was dialyzed into P-buffer (20 mM potassium phosphate, 10% glycerol, 0.1 mM EDTA and 1 mM dithiothreitol) and applied to ceramic hydroxyapatite column HAP (BIO-RAD). The HsRecA protein was eluted with a 10 column volumes of P-buffer containing a linear gradient of potassium phosphate (0.02-1 M). The most pure fractions containing HsRecA protein were pooled and dialyzed into R-buffer + 50 mM KCl and, then applied to Source 15-S column (GE Healthcare). HsRecA protein was not retained in the column and it was collected in the flow-through fraction. The pooled collected was applied directly into the Source 15-Q column (GE Healthcare) and submitted to a 10 column volume of a linear KCl gradient (0.05-1 M) in R-buffer. The fractions containing HsRecA protein were pooled, adjusted to 1 M ammonium sulfate and then applied to Butyl Sepharose column (GE Healthcare) which was submitted to a 10 column volume of a linear ammonium sulfate gradient (1-0 M) in R-buffer. The purified protein was free of detectable nuclease activity and showed 95% purity in SDS-PAGE. HsRecA and EcRecA concentration were determined using absorbance at 280 nm and the extinction coefficient 1.60×10 4 and 2.23×10 4 M −1 cm −1 , respectively.
E. coli single-stranded DNA binding protein (SSB) was purified as described previously [33] and its concentration was determined using an extinction coefficient of 2.83× 10 4 M −1 cm −1 at 280 nm.

ATPase Assay
ATP hydrolysis was measured by a coupled spectrophotometric assay as previously described [34,35] at 37°C. The assays were carried out in a Varian Cary 300 dual beam spectrophotometer equipped with a temperature controller and a 12-position cell changer. Regeneration system of ATP (10 U of pyruvate kinase mL −1 and 3 mM phosphoenolpyruvate) from ADP and phosphoenolpyruvate is coupled to the conversion of NADH to NAD + (2 mM NADH and 10 U of lactate dehydrogenase mL −1 ), which can be monitored by a decrease in absorbance at 380 nm. Although the maximum absorbance for NADH occurs at 340 nm, absorbances were measured at 380 nm to remain within the linear absorbance range of the spectrophotometer for the extended length of time required in these experiments. The amount of ATP hydrolyzed was calculated using the extinction coefficient of NADH, 1.21 mM -1 cm -1 . The standard reaction condition was RecA-buffer (25 mM Tris-OAc (80% cation, pH 7.5), 1 mM dithiothreitol, 5% (w/v) glycerol, 3 mM potassium glutamate, and 10 mM magnesium acetate). Concentrations of HsRecA, EcRecA, DNA, SSB, and ATP are indicated in figure legends. E. coli SSB was used to stimulates the ATPase activity of both HsRecA and EcRecA proteins when due.

DNA Three Strand Exchange Reactions
Three strand exchange reactions were carried out as described previously [36,37] in a RecAbuffer at 37°C. An ATP regeneration system (10 U of pyruvate kinase mL −1 and 2.5 mM phosphoenolpyruvate) was also included. The EcRecA or HsRecA protein were pre-incubated with circular ssDNA M13mp18 (7249 mer) for 20 min, SSB protein and ATP were then added, followed by another 10 min of incubation. The reaction was initiated by the addition of M13mp18 linear dsDNA and incubated for 60 min. To stop the reaction, 5 μL of a stop and load reaction buffer (15% Ficoll, 0.25% bromophenol blue, 0.25% xylene cyanol, 25 mM EDTA, and 10% SDS) was added into each 10 μL aliquot collected. Samples were subjected to electrophoresis in 0.8% agarose gels with TAE buffer. Concentrations of HsRecA, EcRecA, DNA, SSB, and ATP are indicated in figure legends.

Electron Microscopy
A modified alcian method was used to visualize RecA filaments. Activated grids were prepared as described previously [20]. Samples for electron microscopy analysis were prepared as follows. All incubations were carried out at 37°C. EcRecA or HsRecA (6.7 μM) was preincubated with 20 μM M13mp18 circular ssDNA in a RecA-buffer containing ATP regeneration system (10 U mL -1 creatine phosphokinase-PK and 12 mM phosphocreatine-PC) for 20 min. Then, 3 mM ATP with or without 2 μM SSB protein were added, and the reaction was incubated for another 10 min. To stabilize the filaments, 3 mM ATPγS was added and further incubated for 3 min.
The reaction mixture was diluted to a final DNA concentration of 0.4 ng/μL with diluting buffer (200 mM ammonium acetate, 10 mM Hepes and 10% glycerol (pH adjusted to 7.5)) and adsorbed to an activated alcian grid for 3 min. The grid was then touched to a drop of the diluting buffer followed by floating on a drop of the same buffer for 1 min. The sample was then stained by touching to a drop of 5% uranyl acetate followed by floating on a fresh drop of the same solution for 30 s. Finally, the grid was washed by touching to a drop of double distilled water followed by immersion in two 10 mL beakers of double distilled water. After the sample was dried, it was rotary-shadowed with platinum. This protocol is designed for visualization of complete reaction mixtures, and no attempt was made to remove unreacted material. Although this approach should yield results that give a true insight into reaction components, it does lead to samples with a high background of unreacted proteins. Imaging and photography were carried out with a TECNAI G2 12 Twin Electron Microscope (FEI Co.) equipped with a GATAN 890 CCD camera. Digital images of the nucleoprotein filaments were taken at X 15000 Magnification. Ten filaments each from EcRecA and HsRecA were measured three times using Metamorph analysis software and the average length was calculated in nm. The 0.5 μm scale bar was used as a standard to calculate the number of pixels per μm.

Crystallization, Data Collection and Structure Refinement
All crystallization experiments were performed using the hanging-drop vapor diffusion method in 24-well plates at 20°C. All conditions were tested with the apo HsRecA form and with the ligands ATPγS and ADP. The initial crystallization screening was carried out using a reservoir consisting of 1 mL of each composition from the Cryo Suit (QIAGEN), PEG Ion, Natrix, Crystal Screen 1 and 2 and Crystal Screen Lite (Hampton Research) crystallization kits. Each drop contained 1 μL reservoir solution and an equal volume of the HsRecA protein solution (4 mg/mL) plus ligands when used. Clusters of multiple needle crystals were observed after 4 days under condition 0.2 M CaCl 2 , 20% w/v Polyethylene glycol 3,350 (PEG 3,350), pH 6.8. Further optimization of the condition was performed, initially varying the CaCl 2 and PEG concentrations, and then screening for additives by using Additive Screens (Hampton Research). The best single crystals were finally obtained with 0.25 M CaCl 2 and 14% w/v PEG 3,350 with further addition of Polypropylene glycol P 400 (PPG 400) onto the drop to give a final concentration of 5-8% w/v. Crystals from 0.1 mM HsRecA, 10 mM MgCl 2, 2 mM ADP solution reached dimensions of 0.8 mm × 0.2 mm × 0.2 mm after 4 days.
Diffraction data were collected using a Bruker diffractometer equipped with a rotatinganode X-ray generator operated at 40 kV and 40 mA and a Smart-6000 detector. The crystals were flash-cooled in liquid nitrogen at 100 K using the reservoir solution supplemented with 25% ethylene glycol as cryoprotectant. The crystals diffracted to a maximum resolution of 1.7 Å. A total of 1056 frames of data were collected with an oscillation angle of 0.5°, an exposure time of 120 s per frame and a crystal-to-detector distance of 60 mm.
Diffraction data were processed with the proteum2 software (Bruker AXS (2010) PRO-TEUM2, Version 2010.5, Bruker AXS Inc., Madison, Wisconsin, USA). Initial phasing was obtained by molecular replacement with the program Phaser [38]. Two domains of E. coli RecA (PDB entry 1xmv, residues 3 to 282, NTD and central ATPase domain, and residues 283 to 328, C-terminal) were used as separate search models. The initial structure was built with Phenix Autobuild [39] and it was completed by repeating cycles of manual model building with COOT [40] and refinement with Phenix.refine [41]. Translation, libration, and screw (TLS) groups were determined with TLSMD [42] and were used during refinement with Phenix.refine.

ATPase Assay and Strand Exchange Activity
To investigate how HsRecA dynamically interacts with ssDNA the DNA-dependent ATPase activity of HsRecA was measured in a coupled spectrophotometric assay using circular ssDNA M13mp18 (cssDNA), ATP, an ATP regeneration system in the presence or absence of E. coli single-stranded binding protein (SSB). All ATPase assays were performed alongside EcRecA wild type in order to compare the two proteins' activities.
A maximum steady state ATPase reaction is typically associated with the formation of complete and contiguous RecA filaments on the circular ssDNA [35]. The EcRecA and HsRecA apparent k cat was determined after the steady-state rate of ATP hydrolysis was achieved, assuming one monomer of RecA bound for each three nucleotides of ssDNA. The results exhibited an apparent k cat of 25.63 ± 0.10 and 28.91 ± 0.11 min -1 , to EcRecA and HsRecA, respectively ( Fig  1A-Reaction 1). The literature reports a k cat of about 30 min -1 when EcRecA is bound to ssDNA under similar conditions [35,43,44]. In multiple trials, the ssDNA-dependent ATPase activity of the HsRecA protein was consistently about 10% higher than that exhibited by EcRecA protein.
The SSB protein is typically used in in vitro assays to remove the ssDNA secondary structures and to facilitate formation of contiguous RecA nucleoprotein filaments when it is added after DNA-RecA pre-incubation. However, when SSB is added prior to RecA, it strongly inhibits RecA filament nucleation and a long lag is seen before ATP hydrolysis is observed [45][46][47]. We utilized the EcSSB for all experiments, in part because the HsSSB was not available. Studies of RecA proteins from bacterial species ranging from Neisseria gonorrhoeae (Ng) to Streptococcus pneumoniae (Sp) to Deinococcus radiodurans (Dr) have failed to find any difference in SSB effects when EcSSB replaced cognate SSB proteins in standard RecA assays [48][49][50]. We note that the HsSSB protein is more closely related to the EcSSB (57% identity/69% similarity) than is the SSB from any of the three species mentioned above (50%/63%, 31%/50%, and 38%/49% for NgSSB, SpSSB, and DrSSB, respectively).
To evaluate the effects of SSB on the ATPase activity of HsRecA, we omitted SSB from the reaction described above (Fig 1A-Reaction 2). In this reaction HsRecA showed higher ATPase activity than EcRecA, with an apparent k cat of 16.49 ± 0.11 and 8.89 ± 0.01 min -1 , respectively. However, neither protein reached the same ATPase activity registered when SSB was present during the 70 minutes evaluated. In addition, a linear ATPase profile throughout the reaction indicated that there was no significant increase of RecA bound to the cssDNA over the course of the experiment.
We also investigated the influence of the moment that SSB is added in HsRecA activity. We added SSB, cssDNA and ATP 10 min before RecA addition, to allow SSB coating the cssDNA ( Fig 1B). As already reported in the literature [45][46][47], EcRecA exhibited a low initial ATPase activity, followed by a very gradual increase, indicating a slow displacement of SSB. However, HsRecA protein displaced the SSB bound to the cssDNA much more effectively. The lag in HsRecA protein mediated ATP hydrolysis was short, and a high rate suggesting formation of contiguous HsRecA filaments was achieved within 10-15 min. HsRecA showed an apparent k cat of 24.66 ± 1.45 min -1 , almost as great as the k cat determined in the previous assay when SSB was added after HsRecA-DNA pre-incubation. In comparison, EcRecA could not reach the same ATPase activity during the 80 min evaluated.
We also examined the ability of the EcRecA and HsRecA proteins to catalyze strand exchange between cssDNA and linear dsDNA (ldsDNA) and form nicked circular dsDNA (NC product). Strand exchange promoted by HsRecA was significantly more efficient than the same reaction promoted by EcRecA. The reaction rates appeared to be similar for the reactions promoted by both proteins, with NC product detectable after 5 min incubation, as well as intermediates (Fig 2A). The reactions reached an apparent endpoint after 10 min. However, the extent of reaction was greater for the HsRecA protein at all time points (Fig 2B).

Visualization of HsRecA Filaments by EM
We used electron microscopy to evaluate the characteristics of HsRecA and EcRecA protein filaments in the absence of SSB (Fig 3A and 3B) and when it was added after RecA-cssDNA preincubation (Fig 3C and 3D).
In the absence of the EcSSB protein, EcRecA formed small filamented circles, with discontinuous regions being common. On the other hand, HsRecA protein formed more open filaments with fewer discontinuities. The addition of SSB eliminated detectable differences between the EcRecA and HsRecA filaments, forming large and completely filamented circles on the circular ssDNA substrate. Representative filaments are shown in Fig 3, panels A-D. The longer filaments formed by HsRecA in the absence of SSB indicate an enhanced capacity to extend filaments into regions of secondary structure. Note that in these images, only the RecA filaments are generally visible. The ssDNA is not seen unless it is bound with SSB (giving it a distinctive beaded appearance).
The general impressions gained from the initial survey of the grids was confirmed by direct filament length measurements. The length of ten filaments each from EcRecA and HsRecA-cssDNA filaments were measured in the presence and absence of SSB. In the presence of SSB, HsRecA and EcRecA exhibited filament lengths that were indistinguishable. There was no statistical significant difference (Tukey p<0.05) between HsRecA and EcRecA filaments lengths in the presence of SSB. On the other hand, in the absence of SSB, the HsRecA filaments were significantly longer than those formed by the EcRecA (Fig 3E).

X-Ray Crystal Structure of HsRecA Protein
We solved the X-ray crystal structure of H. seropedicae RecA to 1.7 Å resolution (Fig 4). The structure was refined with good bond geometry and crystallographic quality statistics, with no residues in disallowed regions of the Ramachandran space ( Table 1). The crystals contained a single HsRecA monomer per asymmetric unit.
Electron density maps derived from molecular replacement phasing permitted assembly of a model of the 351-residue HsRecA protein, with the exception of loops for which density was  . The core ATPase domain contains one Ca 2+ ion (magenta sphere), coordinated by Asn119 and Asp120 and the ATPase activity site is partially occupied by ATP and ADP. Figure was prepared using the STRIDE program for secondary structure assignment [55], and visualized using PyMOL [56].  [10,11,27,[51][52][53], the tertiary structure of HsRecA can be subdivided into three domains: a small N-terminal domain (NTD), a central core ATPase domain, and a large C-terminal domain (CTD) (Fig 4). The NTD (residues 6-40) consists of a α-helix and a random coil motif and the CTD (residues 276-343) consists of 3 α-helices and one antiparallel β-sheet motif. The core domain (residues 44-275) consists of a conserved α/β ATPase motif, which in its turn contains the Walker A (residues 72-81) and B (residues 147- 157) motifs, and the disordered DNA binding loops L 1 (residues 163-176) and L 2 (residues 202-219) for which electron density was not observed in our structure. A Ca 2+ ion is bound by Asn119 and Asp120 in the ATPase core. Proper Ca 2+ coordination in the structure was confirmed by the CheckMyMetal web server [54]. The amino acid sequence alignment from HsRecA and EcRecA, shows that these two proteins shares 68.12% identity and 80.6% similarity. The overall structure of HsRecA is similar to that of EcRecA (PDB entry, 1XMV) with root-mean square deviation (RMSD) of 1.324 Å for Cα, 1.335 Å for backbone and 1.486 Å for all atoms. The structural comparison was performed using LSQKAB from the CCP4 program suite [57].

ATPase Active Site
The co-crystallized adenosine diphosphate (ADP) interacts directly with Ser76, Ser77, Thr80, Thr81, Asp107, Tyr110, and Gly272 in the HsRecA structure. Despite the fact that Mg 2+ ion was present in the crystallization solution there was no electron density for this ion around the nucleotide ADP to suggest proper coordination. Mg 2+ ion was also not observed in Deinococcus radiodurans and Mycobacterium tuberculosis RecA crystal structures [53,58]. We observed an extra electron density beyond the β-phosphate of ADP which could not be properly modeled with waters, SO 4 3or PO 4 3ions. The refinement of PO 4 3and ADP as the only nucleotide showed electron density that suggests them to be covalently linked, but there also remained residual negative electron density in the Fourier difference map over PO 4 -3 at 100% occupancy ( Fig 5A and 5B). Therefore, we considered that this site is partially occupied by ATP, which might have come from the cells. The refinement with ADP and ATP sharing the position improved the adjustment to the electron density and resulted in occupancies of 61% and 39%, respectively. The amide nitrogen of Gln201 is at 2.88 Å from the γ-phosphate. This residue precedes the DNA binding Loop L 2 and plays a role in the activation of the RecA function by induction of an allosteric effect of ATP [59]. Yet in the DrRecA-ATPγS complex this distance is 3.0 Å from the γ-thiophosphate. Both glutamines present the same conformation, different from the EcRecA structure (pdb entry 1xmv) which is bound to ADP.

Assembly of HsRecA Protein in the Crystal Structure
Bacterial RecA and other members of the RecA family (RadA, Dmc1 and Rad51) were aligned using their primary and tertiary structures to identify common regions (Fig 6). The monomer assembly regions were first identified in the RadA/Rad51/Dcm1 proteins [12,13,21,22,60]. They encompass residues in the NTD (Fig 6, N-PM region) and in the core ATPase domain (Fig 6, Core-PM region). The crystal structures of several RecA family members have revealed that the N-PM region is a polymerization motif, which is a short β-strand that just precedes the ATPase core and has conserved hydrophobic residues. Yet, using the alignment and the protein structures from the bacteria E. coli (pdb entry 3CMU, 1XMS, 1U98 and 1XMV), D. radiodurans (pdb entry 1XP8), M. smegmatis (pdb entry 1UBE, 1UBF and 1UBG), M. tuberculosis (pdb entry 1MO3, 1MO5, 1MO6 and 1G18) and ours, we noticed that this polymerization motif (N-PM) consists of a conserved sequence (Ser, Val/Ile, Met, Arg/Lys, Leu, Gly; residues 25-30 in the EcRecA sequence after elimination of the N-terminal Met, and residues 30-35 in the HsRecA sequence). This sequence interacts with the core ATPase domain (Core-PM) of the adjacent subunit at another conserved sequence (Asp, Asn, Leu, Leu, Val/Cys, Ser; residues 113-118 and 118-123 in the EcRecA and HsRecA sequences, respectively). In the N-PM, the conserved hydrophobic residue phenylalanine found in RadA, Rad51 and Dmc1 proteins is replaced by a different hydrophobic residue, valine or isoleucine, in the RecA proteins. However, none of these substitutions affects a hydrogen bond with the Core-PM residues. The conserved methionine in the N-PM may play a role in binding/recognition of the hydrophobic β-strand polymerization motif in the core ATPase domain, promoting thus the polymerization of RecA filaments on the DNA substrate. The Table 2 shows the hydrogen bonds formed between the N-PM and the Core-PM from RecA proteins.
In all inactive RecA protein structures, the short β-strand motif becomes a β-loop motif conformation in the NTD (Fig 7A). This motif is observed in the HsRecA structure as well, although in the EcRecA-ssDNA presynaptic filament this stands as a β-strand motif, as found in the RecA family proteins (Fig 7B) and it is an antiparallel β-strand with respect to the second motif in the EcRecA-ssDNA pre-synaptic crystal structure filament. The Ca 2+ ion in the HsRecA Core-PM is coordinated by Asn119 and Asp120 and the Mg 2+ ion in the EcRecA N-PM is coordinated by Glu18, Lys23 and Ile26 [27]. Both divalent ions may play a role in the formation and stability of RecA protein as a helical nucleoprotein filament. Previous studies showed that addition of excess Mg 2+ (relative to the available ATP) helps to produce the active and extended conformation of the filament, required for DNA pairing and the strand exchange reaction in vitro [66]. HsRecA E36 is at the subunit interface border and might interact with excess ions along with other residues nearby, possibly increase the interaction within the subunit-subunit surface.

Electrostatic Potential Surface
We also reconstructed the HsRecA protein on the basis of its electrostatic potentials in order to understand the structural basis of their functional activities and subunit-subunit interactions. The electrostatic potential distribution on the solvent-accessible surface of the HsRecA protein for two subsequent helical hexamers is shown in Fig 8. Due to the CTD portion of each monomer, there is a relatively large negatively charged region on the outer surface in the HsRecA filament structure. As has been observed for many RecA proteins, HsRecA protein has increased positive charge along the inner side of the filament, near the DNA binding sites (Fig 8B). There are subtle differences in the charge density and distribution relative to EcRecA protein,. These have not been analyzed in detail, but may eventually help explain the DNA binding properties of HsRecA protein. Important domains that affect filament polymerization stand orderly along the filament axis, such that there is a charge polarity change from negative to positive that binds the DNA from the 3 0 to 5 0 direction (Fig 8B and 8C). Therefore, we speculate that electrostatic interactions would facilitate proper filament assembly, after monomers had been stabilized by hydrophobic interactions between the N-PM of one monomer and the Core-PM of the subsequent one.

Helical Pitch
The crystallography 6 1 symmetry leads to the formation of a helical filament that has a pitch of 91.3 Å (Fig 9), a value close to the pitch range 90-100 Å determined by electron microscopy corresponding amino acid sequences: N-PM and core-PM, N-terminal and core polymerization motif, respectively, the ATP binding Walker A and B motifs, the putative DNA binding sites Loop L 1 and L 2 . The secondary structure of HsRecA and HsDmc1 are indicated at the top and bottom, respectively. The 3D structural and amino acid alignment was performed using the MultiProt and T-Coffee programs [61,62]. The results are visualized using ESPript-http://espript.ibcp.fr [63].
doi:10.1371/journal.pone.0159871.g006 Table 2. Summary of hydrogen bonds formed between the N-PM and the Core-PM in RecA protein structures. The residues and their atoms involved in the interaction are indicated, as well as the atomic distances. Interactions were determined using the CCP4 application (Protein Interfaces, Surfaces and Assemblies-PISA) [ for active filaments of EcRecA formed in the presence of DNA, ATP-γ-S or ATP [70]. Inactive and compressed filaments characterized to date have a helical pitch of 65-85 Å and are formed by RecA alone or bound to ADP, either in the absence of DNA or bound to ssDNA or dsDNA [10,26,51,53,58,71]. Despite the fact that HsRecA-ADP/ATP protein was crystallized with ADP and ATP, our structure presented a helical pitch characteristic of an active RecA filament form. However, as noted above, the structural packing is that of the inactive state.

Discussion
In the present work, we report a functional and structural characterization of the RecA protein from H. seropedicae, an important bacterium to environmental crops in agricultural systems. The work has several conclusions. The HsRecA protein forms filaments on ssDNA, and promotes ATP hydrolysis and DNA strand exchange as do other bacterial RecA proteins. The HsRecA distinguishes itself with a somewhat more efficient DNA strand exchange reaction, and a moderately increased rate of ATP hydrolysis relative to that promoted by EcRecA protein. The HsRecA also displaces pre-bound SSB protein much faster than EcRecA. The structure of HsRecA protein, also presented here, is consistent with the RecA structures found for RecA proteins from other bacteria [10,11,27,[51][52][53]. However, in the absence of DNA, the HsRecA crystal structure showed a helical pitch closer to the extended form of a RecA filament, although the structural packing remained that of the inactive state. Structural comparisons offer an opportunity to gain new structure-function insights. The amino acid sequence alignment from H. seropedicae SSB and E. coli SSB, shows that these two proteins are 57.1% identical and 69.0% similar. Our results showed that the E. coli SSB protein stimulates the formation of contiguous HsRecA nucleoprotein filaments (Fig 1A-Reaction 1), as previously seen for E. coli, Pseudomonas aeruginosa and Neisseria gonorrhoeae RecA nucleoproteins [45,72,48,73]. However, the effect of SSB is different in D. radiodurans: either E. coli or D. radiodurans SSB protein suppresses DrRecA ATPase activity in vitro [74].  2+ and Ca 2+ ions, and the ssDNA bound in the EcRecA structure was removed for clarity. Figure was prepared using the STRIDE program for secondary structure assignment [55]. The superposition was generated using the MultiProt program [61] and visualized using PyMOL [56]. HsRecA reached an apparent k cat of 28.91 ± 0.11 min -1 . When compared under the same conditions, these rates were about 10% higher than those observed when EcRecA was bound to ssDNA.
The HsRecA protein has a higher capacity to bind to regions of secondary structure in ssDNA. Omission of the SSB from the ATPase assay decreases EcRecA ATPase activity about 65% (k cat of 8.89 ± 0.01 min -1 ), while for HsRecA protein this reduction was about 43% (k cat of 16.49 ± 0.11) (Fig 1A-Reaction 2). Based on the fact the ATPase activity is correlated to the amount of RecA bound to DNA, we suggest that HsRecA can better bind to the secondary structure of ssDNA in absence of SSB than the other proteins previously reported. Electron The surface potential representation has charge levels from -3kT/e (red) to +3kT/e (blue). The electrostatic potential distribution was generated using the APBS program, side chain atom not ordered in the crystal were added using the PDB2PQR program and protonation states at pH 7.5 were assigned with the PROPKA program [67][68][69], and visualized using PyMOL [56]. microscopy experiments performed to visualize EcRecA and HsRecA interaction with cssDNA and ATPγS also demonstrated a greater capacity to bind secondary structure for the HsRecA protein. When the SSB protein was absent, HsRecA filament lengths were significantly greater on M13mp18 ssDNA than those observed for EcRecA. When SSB protein was added after RecA and DNA pre-incubation, the filament lengths and characteristics were indistinguishable.
HsRecA protein displaces pre-bound SSB much more quickly than does EcRecA, and reaches a steady state with an apparent k cat of 24.66 ± 1.45 min -1 within 10-15 min (Fig 1B). SSB protein is important in recombination processes because it stimulates the first phase, presynaptic polymerization of RecA protein on ssDNA, and the last phase, strand exchange [75][76][77]. Previous studies showed that some mutations in EcRecA (recA730 (E38K) [78], recA803 (V37M) [79], recA2020 (T121I) [80] and recA441 (E38K/I298V) [78,81,82]) can produce a similar capacity for bypassing the SSB block to nucleation. In HsRecA protein, these residues are conserved, except V37 and E38, which correspond to I44 and Q45 in HsRecA, respectively (Fig 6). Like EcRecA, the HsRecA has a C-terminus that is predominantly negatively charged (four Glu residues and one Lys among the 16 C-terminal amino acid residues. The concentration of negative charge is greater in the EcRecA C-terminus (7 of the last 17 amino acid residues, with no positive charges). Elimination of the EcRecA C-terminus also results in faster nucleation on SSB-coated ssDNA [20]. We have postulated that an interaction between the RecA C-terminus (unstructured in most RecA protein structures, including this one) and the surface at E38 could mask a RecA surface required for rapid SSB displacement. We postulate that the decrease in negative charge in the HsRecA, coupled with the amino acid residue substitutions at positions 44 and 45, may decrease the masking effect of the C-terminus and permit the observed rapid SSB displacement.
The monomeric structure of native HsRecA protein exhibits an architecture similar to that of bacterial RecA proteins previously crystallized. HsRecA protein has a small NTD, a core ATPase domain and a large CTD, with the same secondary structure elements of bacterial RecA proteins [9,51,53,58]. However, the crystallography 6 1 symmetry leads to the formation of a helical filament that has a pitch of 91.3 Å (Fig 9), a value within the pitch range of 90-100 Å determined by electron microscopy for active filaments of EcRecA formed in the presence of DNA, ATP-γ-S or ATP [70]. In contrast, the inactive and compressed filament has a helical pitch of 65-85 Å and is formed by RecA alone or bound to ADP in the absence of DNA [10,26,51,53,58,71]. Despite the fact that HsRecA-ADP/ATP protein formed an inactive filament in the presence of ADP and absence of DNA, our structure presented a helical pitch of an extended RecA filament form. Although previous studies using single particle analysis revealed that there is a considerable overlap in pitch between the active and inactive states [83], high salt concentrations induce RecA ATP hydrolysis and also facilitate the crystallization of the extended filament in the absence of ssDNA [84][85][86]. Curiously, polyethylene glycol (PEG) of various sizes also increased M. voltae RadA ATPase activity in the absence of DNA [87]. We suggest that the high concentrations of salt and PEG used in the crystallization solution of HsRecA (0.25 M CaCl 2, 14% w/v PEG 3,350 and 8% w/v PPG 400) could explain the crystallization of the HsRecA extended filament form even in the absence of DNA.
The mechanism of polymerization of RadA/Rad51 proteins as helical filaments have already been reported [17]. However, analyzing the primary and tertiary structures from bacterial RecA protein we found some differences. A conserved residue sequence in the NTD plays a role in the Each monomer is colored differently. The hexameric structure has 6 1 -fold symmetry, the helical filament has a pitch of 91.3 Å. Figures were prepared using the STRIDE program for secondary structure assignment [55], and visualized using PyMOL [56].
doi:10.1371/journal.pone.0159871.g009 polymerization: hydrophobic residues from two motifs (N-PM and Core-PM) stabilize the subunit-subunit interaction in inactive structures. The N-PM assumes a β-loop conformation, while the Core-PM assumes a β-strand conformation. Sequence alignment revealed the presence of a conserved methionine in the bacterial N-PM RecAs, which we speculate may play a role in binding/recognition of the hydrophobic β-strand polymerization motif in the core ATPase domain. We also noticed that the transition of inactive and compressed to active and extended RecA structure follows the transition from a β-loop to a β-strand motif conformation in N-PM. The presence of the divalent ions Mg 2+ and Ca 2+ as observed in the N-PM and Core-PM, respectively, in EcRecA (pdb entry 1XMV) and HsRecA structures may play a role in the structural transition from compressed to extended conformation. This is similar to structural transitions in yeast Dmc1 protein, in which Ca 2+ ion promotes the formation of an extended helical filament onto ssDNA [88].
The polymerization and dissociation of RecA protein from DNA are regulated by an array of proteins [6]. HsRecA was efficient at displacing SSB from ssDNA. In its absence, HsRecA was still able to bind more of the M13mp18 ssDNA circle, with its extensive secondary structure, than was observed with EcRecA when ATP or ATPγS is present. Most bacterial RecAs are dependent on RecO and RecR to displace SSB from DNA [89,90], but HsRecA protein appears to require less assistance. The H. seropedicae SSB, RecO and RecR proteins may play a somewhat different role in the recombination processes of this bacterium. Further experiments need to be performed to investigate the role of H. seropedicae SSB, RecO and RecR proteins in the mechanism of polymerization and stability of HsRecA filaments.
The HsRecA protein has a greater capacity to bind ssDNA in the presence of SSB, and this may reflect the plant-bacteria interaction and the apparent absence of some auxiliary proteins normally associated with RecA. RecA protein has recently been implicated in bacterial swarming, a flagellar-driven highly coordinated translocation of a bacterial colony across a moist surface [91,92]. This motility has been linked to a chemotaxis signaling pathway which includes methylaccepting chemotaxis proteins (MCPs), CheW adaptor proteins and CheA kinase [93]. Irazoki et al. (2016) [94] showed that activation of the SOS response by the presence of a DNA-damaging compound increases the RecA concentration, thereby disturbing the equilibrium between RecA and CheW and resulting in the cessation of swarming. When the DNA-damaging source decreases or disappears, the repair of the DNA damage seems to restore colony swarming ability [94]. Based on the fact that H. seropedicae is subjected to many environmental factors that cause DNA-damage, such as solar ultraviolet radiation, reactive oxygen species, pH, among others, the enhanced capacity of HsRecA to bind DNA may facililtate more rapid repair and reestablishment of swarming, providing the bacteria a competitive advantage in colonization of plant root surfaces. Many genes related to chemotaxis were found in H. seropedicae SMR1 genome but further experiments are needed to evaluate the role of RecA to swarming in this bacterium.