Chimeras of Escherichia coli and Mycobacterium tuberculosis Single-Stranded DNA Binding Proteins: Characterization and Function in Escherichia coli

Single stranded DNA binding proteins (SSBs) are vital for the survival of organisms. Studies on SSBs from the prototype, Escherichia coli (EcoSSB) and, an important human pathogen, Mycobacterium tuberculosis (MtuSSB) had shown that despite significant variations in their quaternary structures, the DNA binding and oligomerization properties of the two are similar. Here, we used the X-ray crystal structure data of the two SSBs to design a series of chimeric proteins (mβ1, mβ1′β2, mβ1–β5, mβ1–β6 and mβ4–β5) by transplanting β1, β1′β2, β1–β5, β1–β6 and β4–β5 regions, respectively of the N-terminal (DNA binding) domain of MtuSSB for the corresponding sequences in EcoSSB. In addition, mβ1′β2ESWR SSB was generated by mutating the MtuSSB specific ‘PRIY’ sequence in the β2 strand of mβ1′β2 SSB to EcoSSB specific ‘ESWR’ sequence. Biochemical characterization revealed that except for mβ1 SSB, all chimeras and a control construct lacking the C-terminal domain (ΔC SSB) bound DNA in modes corresponding to limited and unlimited modes of binding. However, the DNA on MtuSSB may follow a different path than the EcoSSB. Structural probing by protease digestion revealed that unlike other SSBs used, mβ1 SSB was also hypersensitive to chymotrypsin treatment. Further, to check for their biological activities, we developed a sensitive assay, and observed that mβ1–β6, MtuSSB, mβ1′β2 and mβ1–β5 SSBs complemented E. coli Δssb in a dose dependent manner. Complementation by the mβ1–β5 SSB was poor. In contrast, mβ1′β2ESWR SSB complemented E. coli as well as EcoSSB. The inefficiently functioning SSBs resulted in an elongated cell/filamentation phenotype of E. coli. Taken together, our observations suggest that specific interactions within the DNA binding domain of the homotetrameric SSBs are crucial for their biological function.

SSB from Escherichia coli (EcoSSB) has been an archetype to understand the biochemical, biophysical and the structural properties of the related SSBs [6]. EcoSSB consists of an Nterminal domain (,115 amino acids) rich in bsheets and a Cterminal domain without a defined tertiary structure [17,18]. The C-terminal domain can be divided into a spacer region rich in glycine and proline residues, and a highly conserved region consisting of negatively charged residues (acidic tail). EcoSSB functions as a homo-tetramer consisting of four OB folds and interacts with ssDNA in different binding modes. In low salt (,20 mM NaCl) and high protein to DNA ratios, only two of the four subunits bind to ,35 nucleotides in an unlimited cooperative manner to long ssDNA, known as SSB 35 mode [19][20][21][22]. While in high salt (.0.2 M NaCl) and low protein to DNA ratios, all four subunits bind to ,65 nucleotides in a limited cooperative manner to polynucleotides known as SSB 65 mode [19][20][21][22]. The dynamic transition between these binding modes may be relevant for the in vivo function of SSBs [6,23].
Unlike most other bacterial SSBs, SSBs from Deinococcus/ Thermus group have been characterized to form homodimers [24][25][26][27][28]. However, in these SSBs, each monomer contains two OB folds. Studies with Deinococcus radiodurans SSB (DraSSB) show that the mechanism of DNA wrapping onto it is not identical to that of EcoSSB [27]. However, the DNA binding affinity, rate constant and association mechanisms of DraSSB are similar to those of EcoSSB. Interestingly, DraSSB complements E. coli for the essential function of SSB [26]. SSB from Helicobacter pylori which is closer to EcoSSB for its various properties is also known to function in E. coli [29,30]. SSB from M. tuberculosis (MtuSSB) shares ,30% identity and ,39% similarity with EcoSSB in its primary sequence. Although the dynamics and the mode of DNA binding to MtuSSB have not been studied in detail, the initial biochemical characterization has shown that like EcoSSB, MtuSSB is a homotetramer and binds to ssDNA in two modes similar to EcoSSB [31]. The threedimensional structure of SSB from M. tuberculosis suggested significant variability in its quaternary structure. The MtuSSB has unique dimeric interface facilitated by the clamp structures formed by b6 strands of the interacting subunits [32]. Such structural differences were also observed in SSBs of other mycobacteria (M. smegmatis and M. leprae) [33,34].
To further our understanding of the structure-function relationship of eubacterial SSBs, in this study, we have generated a number of chimeric SSBs by swapping different regions of EcoSSB and MtuSSB and analyzed them for their in vitro and, in vivo properties using a sensitive assay system designed in this study.

Generation of chimeric SSBs
Chimeric constructs were designed based on the three dimensional structures of EcoSSB and MtuSSB ( Figure 1). Details of generation of the chimeric constructs are provided in the supporting material (Methods S1), and shown schematically is Figure 2. In our earlier study [35], we generated a chimeric MtuEcoSSB which has been renamed as mb1-b6 SSB (Figure 2, iii), possessing the Nterminal region (amino acids 1-130; the initiating methionine is numbered as 1) from MtuSSB and the C-terminal region (131-178) from EcoSSB. The crystal structure of MtuSSB [32] revealed a novel hook like structure formed by the presence of the b6 strand, an element absent from EcoSSB (Figure 1, i; Figure 2, i and ii). Hence, the mb1-b5 SSB containing b1 to b5 strands (first 111 amino acids) from MtuSSB and the remainder of the sequence from EcoSSB (Figure 2, iv) was also designed.
In other constructs, various secondary structural elements in the N-terminal domain of EcoSSB were replaced with the corresponding regions of MtuSSB ( Figure 2). The mb4-b5 SSB, contained the first 73 amino acids comprising b1, b2, b3 strands and the a-helix from EcoSSB, and amino acids 74 to 111 comprising b4, b45 1 , b45 2 and b5 strands from MtuSSB, followed by amino acids 112 to the end of the protein from EcoSSB (Figure 2, v). In the mb1 SSB ( Figure 2, vii) amino acids 6-11in the b1 strand of EcoSSB were substituted with the corresponding MtuSSB sequence (Table 1, S3).
The above constructs possessed substitutions of EcoSSB regions involved in subunit-subunit interactions. Hence, we generated mb19b2 SSB wherein amino acids 21 to 45 comprising b19 and b2 strands positioned in the exterior of the tetramer (Figure 1, ii), were exchanged with the corresponding sequences from MtuSSB (Figure 2, viii). The mb19b2 ESWR SSB was generated from mb19b2 SSB by replacing four amino acids of the MtuSSB origin at positions 39 to 42 (PRIY, in the b2 strand) with the EcoSSB specific sequence, ESWR (Figure 2, ix). And, a clone with deletion of C-terminal domain of EcoSSB, DC SSB ( Figure 2, vi), was identified serendipitously during sequence analysis of the generated constructs.

Oligomerization status of chimeric SSBs
Analysis of the purified SSBs using native PAGE is shown in Figure 3A. On such a gel, EcoSSB and MtuSSB were shown to migrate as homotetramers [31,35]. The migration of many chimeras was comparable to EcoSSB or MtuSSB suggesting their homotetrameric nature. However, we observed diffuse migration of mb1 and mb19b2 SSBs, suggesting alteration(s) in their oligomerization/folding properties. Interestingly, introduction of EcoSSB specific 'ESWR' sequence in mb19b2 ESWR SSB in place of MtuSSB specific 'PRIY' sequence (in mb19b2 SSB), restored its mobility as a tetramer ( Figure 3A, lanes 8 and 9). To further analyze the oligomerization status of the chimeric SSBs, we performed gel filtration chromatography wherein EcoSSB eluted as tetramer ( Figure 3C, i). The oligomeric nature of other SSBs was determined from a standard plot of Ve/Vo versus log molecular weight ( Figure 3B). Consistent with the diffuse mobility of mb1 SSB in native PAGE ( Figure 3A), it eluted in the void volume suggesting alteration in its oligomerization/folding properties ( Figure 3C, ii). On the other hand, while a fraction of the mb19b2 SSB eluted as tetramer, its elution continued beyond the tetramer peak suggesting poor tetramerization ( Figure 3C, iii). However, as expected from the native gel analysis, introduction of ESWR sequence in mb19b2 SSB (in place of PRIY sequence) restored its oligomeric status as a tetramer, and it eluted same as EcoSSB ( Figure 3C, iv). Other chimeras eluted as tetramers ( Figure  S1). As expected, the elution profile of DC SSB suggested it to be tetramer but smaller in molecular weight.

DNA binding activity of chimeric SSBs
Using electrophoretic mobility shift assays (EMSA) [31,35] we observed that all SSBs formed protein-DNA complexes ( Figures 4A-C). However, the complex formation with mb1 SSB was extremely poor and detectable only at the highest concentration of the protein ( Figure 4C, lanes 1-4). Also, in such assays, EcoSSB and MtuSSB have been shown to bind longer DNA oligomers in two forms corresponding to SSB 35 and SSB 56/65 binding modes of EcoSSB [31]. As a 79mer DNA was used, at higher molar ratios of SSB to DNA, two major complexes were seen with all except mb1 SSB which showed poor binding, and the DC SSB showed additional complexes (compare lanes 3 and 4, with 2; 7 and 8 with 6; 11 and 12 with 10 in Figures 4A-C, respectively). However, at low SSB to DNA ratios, a single complex of mobility corresponding to SSB 56/65 was seen (compare lanes 2 with 1; 6 with 5; and 10 with 9 in Figures 4A and 4B). Highly compromised binding of mb1 SSB ( Figure 4C, lanes 1-4) is consistent with its altered oligomerization/folding properties ( Figure 3). However, the DNA binding ability of mb19b2 SSB ( Figure 4C, lanes 5 to 8) which showed weak tetramerization ( Figure 3C) appeared not as compromised. The nature of the complexes seen with DC SSB, was not investigated. However, the presence of multiple bands ( Figure 4B, lane 12) may indicate that the C-terminal domain may contribute to remodeling DNA binding predominantly in SSB 35 and SSB 56/65 modes.

Susceptibility of chimeric SSBs to chymotrypsin digestion
Chymotrypsin which cleaves at the carboxyl side of the aromatic amino acids (F, Y and W) has 14 cleavage sites in EcoSSB. While the sites within the N-terminal domain are protected by a well formed structure, the sites within the Cterminal domain are sensitive to cleavage. As shown earlier [18], the sites at 136 and 148 positions are particularly prone to cleavage, and at early time points in the reaction, two bands corresponding to ,14 kD and ,15 kD are seen. Upon DNA binding the conformational changes in the C-terminal domain make the site at position 136 more accessible and a single product corresponding to ,14 kD is seen [18]. We used this assay to probe for conformational changes in the chimeric SSBs upon ssDNA binding. Chymotrypsin cleavage pattern of the free and DNA bound EcoSSB ( Figure 5A) was the same as reported [18]. The digestion of MtuSSB also resulted in two products migrating as a doublet; and the presence of ssDNA resulted in a single band corresponding to the lower band of the doublet ( Figure 5B). The cleavage patterns of the chimeric SSBs followed the same trend, resulting in relative accumulation of the smaller product (bands marked with arrowheads) upon DNA binding (Figures 5C-E and I). The DC SSB lacking C terminal domain, as expected, did not result in change in the digestion pattern ( Figure 5F). The mb1 SSB, defective in oligomerization/folding, was more sensitive to digestion. And, consistent with its poor DNA binding ( Figure 4C) it did not show a relative accumulation of the smaller sized product upon DNA binding ( Figure 5G). While mb19b2 SSB ( Figure 5H) was somewhat more sensitive to chymotrypsin than the other constructs ( Figures 5A-E), substitution of the PRIY sequence in its b2 strand with ESWR (in mb19b2 ESWR SSB) rescued it from its protease sensitivity ( Figure 5I), as it did its oligomerization and DNA binding properties ( Figure 3A, lanes 8 and 9; 3C, iii and iv; and Figure 4C, lanes 5-12). These observations lend further support to the observations (Figures 3 and 4) that mb1 and mb19b2 SSBs suffer from structural alterations.

Fluorescence titration of SSBs
Fluorescence reverse titrations have been used to determine the binding site sizes of SSBs [19][20][21][22]27]. We performed such experiments to determine the kinetic parameters for DNA binding to various SSBs using poly (dT) in 50 and 200 mM NaCl. Except for mb1 SSB which revealed altered structure, highly compro-mised DNA binding, and high sensitivity to chymotrypsin (Figures 3, 4, 5), other SSBs resulted in similar fluorescence quenching ( Figure S2). We then processed [36][37][38][39][40] these data to estimate values of maximal fluorescence quenching (Q max ), binding site size (n), binding constant (K obs ) and co-operativity (v) for each of the chimeras except for mb1 SSB ( Table 2). The values of Q max , n and v of EcoSSB, MtuSSB and the other SSBs were comparable. As reported previously [19], we observed that the binding site size of EcoSSB increased from 50 to 68 when NaCl concentration was changed from 50 mM to 200 mM, respectively. On the contrary, MtuSSB did not show a similar increase in binding site size upon increase in salt concentration. Binding site size of MtuSSB was observed to be 72 in 50 mM NaCl and it changed to 76 in 200 mM NaCl. The chimeric SSBs (except mb19b2 ESWR SSB) exhibited comparable binding site sizes in the presence of 200 mM NaCl. At 50 mM salt, ssDNA binding with mb19b2, mb19b2 ESWR , mb1-b6 SSBs showed binding site sizes comparable to EcoSSB. The binding site sizes of mb1-b5, mb4-b5 and DC

In vivo complementation analysis of the chimeric SSBs
To further characterize the chimeric SSBs, it was of interest to determine if they complemented E. coli for the in vivo function of SSB. We first used the 'plasmid bumping' method [35,41; and Methods S2] where the test ssb construct (in a ColE1 ori plasmid, Amp R ) was introduced in a Dssb (ssb::kan) strain (RDP317, Kan R ) of E. coli [42] harboring a wild-type ssb gene on another ColE1 ori plasmid (pRPZ150, Tet R ), and the transformants cultured for multiple rounds in the presence of Amp and Kan. As both plasmids possess ColE1 ori, under the growth conditions, the strain would lose the original Tet R plasmid if the test ssb plasmid (Amp R ) substituted for the essential function of EcoSSB [35,41] giving rise to Amp R Tet S population. Using this assay, we observed that besides the positive control of EcoSSB, only mb19b2 ESWR SSB resulted in 'bumping' of the original plasmid (Table S1). However, it may be that in this assay a weakly complementing SSB construct does not 'bump' the original plasmid due to fitness disadvantage. Hence, a new assay wherein pHYDEcoSSB (Cam R ) construct sustained the E coli RDP317-1 strain (Kan R ) was developed. As the replication of pHYDEcoSSB is dependent on the presence of IPTG, its withdrawal from the growth medium results in the loss of the plasmid and failure of the strain growth unless sustained by the test SSB.
The SSB constructs were subcloned into a ColE1 ori (Amp R ) plasmid wherein their expression was inducible by arabinose (the pBAD series of constructs, Table 1) and introduced into the RDP317-1 strain (Kan R ) harboring pHYDEcoSSB (Cam R ). Transformants were selected on LB agar containing Kan, Amp and 0.02% arabinose ( Figure 6A). Six of the constructs (EcoSSB, MtuSSB, and the mb1-b6, mb1-b5, mb19b2 and mb19b2 ESWR SSBs) yielded transformants, indicating that they functioned in E. coli. The remaining three constructs (DC, mb1 and mb4-b5 SSBs) did not yield any transformants. When checked for their expression, all chimeric SSBs showed expression in E. coli TG1 (Figures 6B and 6C). The transformants for DC, mb1 and mb4-b5 SSBs constructs could be obtained in the presence of IPTG (to allow replication of pHYDEcoSSB). Unfortunately, these transformants (unlike the ones obtained with EcoSSB, MtuSSB, and mb1-b5, mb1-b6, mb19b2, mb19b2 ESWR SSB constructs) retained the support plasmid (Cam R ) even after their subsequent growth in the absence of IPTG. The reasons for this observation are unclear.
When the transformants obtained with EcoSSB, MtuSSB, and mb1-b5, mb1-b6, mb19b2, and mb19b2 ESWR SSB constructs, were streaked on a fresh plate, except for the mb1-b5 construct, the other constructs supported efficient growth upon induction of SSB expression by arabinose ( Figure 6D). Interestingly, when grown in liquid medium, even the transformant harboring mb1-b5 SSB reached saturation. However, total viable count determination revealed that viability of cells harboring mb1-b5 SSB was severely compromised ( Figure 6E). The growth curve experiment (Figure 7) revealed that the mb19b2 ESWR SSB supported E. coli growth nearly as well as EcoSSB (Figure 7, panel ii). The growth in the presence of mb19b2 SSB and MtuSSB was weak. Importantly, mb1-b6 SSB wherein the C-terminal domain of MtuSSB was replaced with that from EcoSSB, supported better growth (Figure 7, panel ii). As MtuSSB and mb1-b6 SSB are expressed to similar levels ( Figure 6C), this observation is consistent with the importance of interactions of the various cellular proteins with the C-terminal of the homologous EcoSSB [13][14][15][16]. In the absence of induction of SSB expression, the basal level expression of only the wild-type EcoSSB and the mb19b2 ESWR SSB resulted in some visible growth (Figure 7, panel i). Furthermore, we observed that with the increase in the concentration of the inducer, the lag phases in the cases of less efficiently functioning SSBs (MtuSSB, mb1-b6, mb19b2 and mb1-b5) decreased suggesting a dose dependent complementation of an E. coli Dssb strain by these SSBs.
Further analysis using fluorescent microscopy revealed that the less efficiently functioning SSBs resulted in an elongated cell/ filamentation phenotype of E. coli (Figures 8B-E). In fact, the mb1-b5 SSB caused a notable filamentation phenotype with increased number of nucleoids per cell, as revealed by the DAPI staining ( Figure 8D). Importantly, the morphology of E. coli cells harboring mb19b2 ESWR SSB was very similar to those harboring wild-type EcoSSB (Figures 8F and 8A).

Discussion
We used the crystal structure data of EcoSSB [12] and MtuSSB [32] to design a series of chimeric SSBs, which complement E. coli Dssb strain with variable efficiencies. One of the constructs, mb19b2 ESWR SSB complements the strain as well as EcoSSB. And, while the MtuSSB, mb1-b6 SSB and mb19b2 complement the strain weakly, they show a limited improvement in rescuing it in a dose dependent manner as the inducer concentration is increased (Figure 7). However, under the same conditions, the rescue offered by the mb1-b5 SSB is poor. The mb4-b5, mb1 and DC SSBs do not rescue the strain for its growth highlighting the intricacies and significance of the specificity of inter-subunit interactions for a fully functional SSB. Furthermore, as the mb1-b5 and mb4-b5 SSBs are proficient in tetramerization and DNA binding, our observations suggest that the importance of inter-subunit interactions is not limited to merely provide these functions. The nature of these interactions may be crucial in allowing conformational changes ('cross-talk') between various regions of SSB, necessary for the in vivo function of SSBs. For example, in mb1-b5 SSB, presence of EcoSSB sequences downstream to mb1-b5 resulted in a change in the mode of DNA binding. However, an additional presence of MtuSSB sequences (b6) in mb1-b6 SSB resulted in a mode of binding comparable to EcoSSB and also resulted in better growth.
Moreover, the tip of the L 45 loop in EcoSSB ( Figure 1A, panel ii) undergoes a conformational change of ,2 Å upon DNA binding [43]. The L 45 loop at the tetramer-tetramer interface is predicted to be important for the SSB 35 mode of DNA binding [12,43] and is thus important in cooperativity of SSB binding to DNA. We should say that while our fluorescence reverse titrations (Table 2)   do not reveal significant differences in the cooperativity of DNA binding by the SSBs, small differences, undetectable in this assay may be significant in vivo. The mb4-b5 SSB possesses the L 45 loop region from MtuSSB, and while it retains the oligomeric status and DNA binding ability, it may be compromised for in vivo cooperativity. Recent computational analysis has indeed suggested that the movement of L 45 loop in EcoSSB, MtuSSB, and Streptomyces coelicolor SSB is different [34]. However, as the mb1-b6 SSB construct complemented E. coli for the essential function of SSB, albeit less efficiently, our observations suggest that the L 45 loop movements could be influenced by the context of the neighboring sequence. This may also be a reason why the mb1-b5 SSB lacking the MtuSSB specific region downstream of the b5 strand, is unable to offer a significant rescue of the Dssb strain of E. coli for its growth. Further studies would be required to understand the contributions of specific interactions of the L 45 loop with the neighboring sequences.
Furthermore, as revealed by the native-PAGE, gel filtration chromatography and chymotrypsin digestion analyses, the mb1 SSB wherein the b1 strand was from MtuSSB, was destabilized at least in its quaternary structure and highly compromised for DNA binding, suggesting that the b1 strand is involved in specific interactions not compensated for by the heterologous sequences from MtuSSB. Recently, the b1 strand of EcoSSB was shown to be involved in direct hydrogen bonding in monomer-monomer interactions; whereas the same region in MtuSSB establishes water mediated hydrogen bonds [34]. Replacement of b19 and b2 strands of EcoSSB with those from MtuSSB (in mb19b2 SSB) also resulted in structural alterations (Figure 3). However, unlike the mb1 SSB, the mb19b2 SSB is able to bind DNA and resist complete digestion by chymotrypsin ( Figures 4C; 5G and 5H). Interestingly, micromanipulation of this construct by introduction of EcoSSB specific 'ESWR' sequence (in place of the MtuSSB specific 'PRIY') important for oligomerization of SSB [34],   converted the new chimera, mb19b2 ESWR SSB, into a more efficient protein (Figures 3, 5, 7 and 8).
In EcoSSB, W40 and W54 are important for DNA binding. In MtuSSB, these residues are replaced by I39 and F54. The model of MtuSSB-ssDNA reveals that the absence of W54 in MtuSSB is compensated for by W60. Also, there are ten basic residues in MtuSSB as opposed to six in EcoSSB. Additional ionic residues are predicted to compensate for the absence of W40 of EcoSSB. Even though the DNA binding properties of EcoSSB, MtuSSB, and various nonfunctional chimeras are similar, the precise mode of DNA binding in them may be different due to alteration of the residues crucial in determining the mode of DNA interactions [32]. And, as is evident from the elongated cell/filamentation phenotypes of MtuSSB, mb1-b6, mb19b2 and mb1-b5 SSBs, even the minor deficiencies in the varied DNA transaction activities of SSB may be significant from the in vivo perspective.
It had been reported that overexpression of SSB in E. coli results in elongated cell phenotype [44], which appeared unlikely due to a marginal overexpression (1.2 to 1.5 fold) of sfi gene product known to cause inhibition of cell division. In our studies, the level of expression of mb1-b5 SSB is the same as those of MtuSSB or mb1-b6 SSB ( Figure 6C). However, among these while the mb1-b5 SSB caused a filamentation phenotype, the other two (MtuSSB and mb1-b6 SSB) resulted in a milder phenotype of elongated cells. Also, the level of expression of EcoSSB, and mb19b2 ESWR SSBs (readily detected by commassie blue staining of SDS-PAGE, Figure 6B) is much higher than that of MtuSSB, mb1-b6, or mb1-b5 SSBs (immunoblotting was needed for their clear visualization, Figure 6C). However, neither the EcoSSB nor the mb19b2 ESWR SSB result in either an elongated cell or filamentation phenotypes. Taken together, these observations suggest that, at least in our study, the elongated cell/filamentation phenotype is not due to overexpression of SSB, but rather due to inefficient function of SSB. In fact, in a more recent report [45], it was observed that when the SSB levels were decreased, it resulted in a filamentation phenotype in E. coli. Importantly, further studies using mb1-b5 SSB may prove useful in understanding the mechanism of filamentation phenotype in E. coli.
Finally, a recent study on DraSSB having only two C-terminal tails, showed that it complemented E. coli for its essential function of EcoSSB in the 'plasmid bumping' assay. Hence, it was somewhat surprising that using the same assay, both in our earlier study [35] as well as the present study, we failed to see complementation of E. coli Dssb strain by MtuSSB or some of the chimeric SSBs. Likewise, despite the fact that the DNA binding domain of HsmtSSB shared similarity with the corresponding domain of EcoSSB, a chimera wherein the DNA binding domain of EcoSSB was replaced with the corresponding domain of HsmtSSB [46,47] failed to function in E. coli [48]. At least, in the case of MtuSSB and mb1-b6 SSB (and the other constructs), it is now clear that the conditions used for the 'plasmid bumping' assay did not overcome the fitness disadvantage for the E. coli Dssb strains to sustain exclusively on these SSBs (as opposed to those harboring both the EcoSSB and the test SSB). Importantly, the new assay developed in this study, overcomes the fitness disadvantage of a weakly functioning SSB by selective blocking of the replication of the parent plasmid by (i. e. by withdrawal of IPTG needed for the replication of pHYDE-coSSB). In fact, this assay allowed us to detect in vivo functioning of even the mb1-b5 SSB, wherein the total viable counts, at saturation, were about three orders of magnitude lower than the strain harboring EcoSSB. Also, this assay has the advantage of not requiring multiple sub-culturing to bump out the original EcoSSB construct. Thus, we believe that the assay developed in this study may be better suited to detect activities of the SSB constructs that offer weak complementation.

Cloning, overexpression and purification of SSBs and their analysis on native gels
To generate chimeric SSBs, EcoSSB sequences were substituted with the corresponding sequences from MtuSSB (Table 1, S3, and Methods S1). SSB open reading frames were also subcloned into pET11D, pUC18R or pBAD/His B (Invitrogen) from the respective pTrc99C constructs using standard methods [49]. The pET11D based expression constructs for mb1-b6, mb1-b5, mb4-b5, mb1 and DC SSBs were introduced into E coli BL21(DE3). The pTrc99c based expression constructs for EcoSSB, MtuSSB, mb19b2 SSB and mb19b2 ESWR SSB; and the pUC18R based construct for mb1 SSB were introduced into E. coli TG1. Cultures (1.2 L) were grown to OD 600 of ,0.5 to 0.6 at 37uC under shaking, supplemented with 0.5 mM isopropyl-b-D-galactopyranoside (IPTG) and the growth continued further for 4 h. Cells were harvested and processed [31] to obtain pure SSB preparations, estimated by Bradford's method using BSA as standard, and stored in 50 mM Tris.HCl, pH 8.0, 0.1 mM Na 2 EDTA, 500 mM NaCl and 10% glycerol. Analysis of the proteins on the native polyacrylamide gels (native-PAGE) was as described before [31].

Gel Filtration analysis of SSB proteins
Oligomeric status of various SSB proteins were determined by gel filtration chromatography. Proteins were chromatographed on Superose TM 6HR 10/30 column (bed volume ,24 ml) attached to an AKTA basic FPLC (GE Healthcare Lifesciences). The column was equilibrated with buffer containing 20 mM Tris.HCl pH 8.0, 500 mM NaCl and 0.1 mM Na 2 EDTA. The flow rate was maintained at 0.3 ml 21 and elution profile was monitored by absorbance at 280 nm. The void volume (Vo) was determined by blue dextran and the column was calibrated using following standard molecular size markers: thyroglobulin (670 kDa), EcoSSB (76 kDa), chicken globulin (44 kDa), equine myoglobin (17 kDa), vitamin B12 (1.3 kDa). Various amounts of SSB proteins (10-200 mg) were loaded on to the column. The Vo of the column was found to be 7.5 ml. The elution volumes (Ve) of marker proteins and SSB proteins were determined and the oligomeric status of SSB proteins was determined from the plot of Ve/Vo versus log of molecular size markers.

Digestions of SSBs with chymotrypsin
Reactions (15 ml) with ,2 mg of SSBs were set up [18] in buffer consisting of 10 mM Tris.HCl, pH 8.1, 0.3 M NaCl, 1 mM CaCl 2 and 5% glycerol (v/v), incubated for 30 min on ice in the absence or presence of 2 mg sheared and heat denatured DNA. The DNA was prepared by digestion of ,100 mg E. coli MG1655 genomic DNA with MspI (50 U) followed by chloroform-phenol extraction and ethanol precipitation. The digested DNA was sonicated at pulse rate of 2 s (on/off) for 1 min, heat denaturation (90uC for 30 min) and chilled on ice. The reactions were initiated by adding ,100 ng chymotrypsin (Amresco) for various times at 37uC, stopped by adding 16 SDS sample loading dye and heating at 90uC for 5 min, and analyzed on SDS-PAGE (17.5%) followed by coomassie brilliant blue staining.
Estimation of binding constant (K obs mM 21 ), maximal fluorescence quenching (Q max ), binding site size (n) and co-operativity (v) These parameters were estimated by fitting a non-linear least squares isotherm onto the data-points obtained by reverse fluorescence titration experiments. We used the binding curve fitting protocol [36] and estimated the above parameters by performing least squares minimization using the 'Solver' tool addin in Microsoft Excel [37]. The equations applied to obtain initial values for K obs , Q max and v are as follows: where L b i ð Þ = protein bound to DNA, L t i ð Þ = total protein concentration, Q obs i ð Þ = observed fluorescence quenching, where v i ð Þ = moles of bound ligand per mole of total lattice residues, D t exp i ð Þ = i th experimental total nucleic acid concentration.
These values are substituted into the McGhee-von Hippel model (McGhee & von Hippel, 1974), as modified by Lohman and Mascotti (1992) to include both noncooperative and cooperative binding, to obtain the concentration of free protein, L f i ð Þ: n is the site size (i.e. the number of bases occluded by binding), v is the cooperativity parameter, and K obs is the intrinsic binding constant observed at the specified pH and salt concentrations. A new value for L b i ð Þ is calculated from the total protein concentration, and the corresponding value for the total DNA concentration, D t exp i ð Þ, is calculated using the definition of the binding density D tcalc i ð Þ is compared to the experimental value D t exp i ð Þ, and v i ð Þ is iteratively incremented until the difference between the calculated and experimental D t values is acceptably small (typically less than 0.01% error). The value of Q obs i ð Þ which corresponds to the final D tcalc i ð Þ [i.e. D t exp i ð Þ], is calculated by rearrangement of eq 1: Thus, Q obs i ð Þ has been calculated for a given value of D t exp i ð Þ. The function requires four parameters, K obs , n, v, and Q max , which were optimized by nonlinear regression (Bevington & Robinson, 1992).
We report the parameters that yielded the minimum value for the sum of the squared differences between the newly calculated Q calc i ð Þ and the actual Q obs i ð Þ. The function requires four parameters, K obs , n, v, and Q max , which were optimized by nonlinear regression [40]. The parameters reported in Table 2 yielded the minimum value for the sum of the squared differences between the newly calculated Q obs (i) and the actual Q obs (i).

Complementation analysis
The pBAD based expression constructs were introduced into E. coli RDP317-1 harboring pHYDEcoSSB (ColE1 ori, Cam R ) whose replication is dependent on the presence of IPTG, and the transformants selected on LB agar containing Kan, Amp and 0.02% arabinose (or Kan, Amp and 0.5 mM IPTG, as control). The isolated colonies were grown in 2 ml LB containing Kan, Amp and 0.02% arabinose to late stationary phase and streaked on LB agar containing Kan and Amp with various concentration of arabinose.

Growth curve analysis
Five independent colonies were inoculated in LB containing Kan, Amp and 0.02% arabinose to obtain late stationary phase cultures; and inoculated at 0.1% level in LB containing Kan, Amp and arabinose (as indicated) in the honeycomb microtitre plates. The growth was recorded at 600 nm using Bioscreen C growth reader (OY growth, Finland) at 37uC on hourly basis. Average values (6SEM) for three isolates were plotted.

Microscopic studies
Fresh transformants of E. coli Dssb strain harboring SSB constructs (pBAD series) were grown to log phase (7-9 h in 2 ml LB containing 0.02% arabinose). Bacterial cells were collected by centrifugation at 5,000 rpm for 5 min, washed with PBS (20 mM sodium phosphate, pH 7.2 containing 0.8% NaCl), suspended in 500 ml of 4% paraformaldehyde solution in 0.1 M sodium phosphate buffer (pH 7.2), and incubated at 4uC for ,4 h. The fixed cells were collected by centrifugation at 5000 rpm for 5 min and resuspended in 666 diluted PBS. The wells of the multi-well slide were coated with 10 ml of 0.1% (w/v) poly-L-lysine (Sigma-Aldrich) for 10 min. Poly-L-lysine was removed and 10 ml of fixed bacterial cells (appropriately diluted) were kept on the wells for 15 min, washed first with PBS and then with 666 diluted PBS. The bacterial cells were stained with 0.25 mg ml 21 solution of 49, 6-diamidino-2-phenylindole (DAPI) in 666 diluted PBS for 5 min in dark, washed with PBS followed by 666 diluted PBS, and visualized in fluorescence microscope (ZEISS, Axio Imager) with 1006 objective lens.

Supporting Information
Methods S1 Generation of chimeric constructs of SSB.