|Home | About | Journals | Submit | Contact Us | Français|
Transfer of conjugative plasmids requires relaxases, proteins that cleave one plasmid strand sequence specifically. The F plasmid relaxase TraI (1,756 amino acids) is also a highly processive DNA helicase. The TraI relaxase activity is located within the N-terminal ~300 amino acids, while helicase motifs are located in the region comprising positions 990 to 1450. For efficient F transfer, the two activities must be physically linked. The two TraI activities are likely used in different stages of transfer; how the protein regulates the transition between activities is unknown. We examined TraI helicase single-stranded DNA (ssDNA) recognition to complement previous explorations of relaxase ssDNA binding. Here, we show that TraI helicase-associated ssDNA binding is independent of and located N-terminal to all helicase motifs. The helicase-associated site binds ssDNA oligonucleotides with nM-range equilibrium dissociation constants and some sequence specificity. Significantly, we observe an apparent strong negative cooperativity in ssDNA binding between relaxase and helicase-associated sites. We examined three TraI variants having 31-amino-acid insertions in or near the helicase-associated ssDNA binding site. B. A. Traxler and colleagues (J. Bacteriol. 188:6346-6353) showed that under certain conditions, these variants are released from a form of negative regulation, allowing them to facilitate transfer more efficiently than wild-type TraI. We find that these variants display both moderately reduced affinity for ssDNA by their helicase-associated binding sites and a significant reduction in the apparent negative cooperativity of binding, relative to wild-type TraI. These results suggest that the apparent negative cooperativity of binding to the two ssDNA binding sites of TraI serves a major regulatory function in F transfer.
Transfer of conjugative plasmids between bacteria contributes to genome diversification and acquisition of new traits. Conjugative plasmids encode most proteins required for transfer of one plasmid strand from the donor to the recipient cell (reviewed in references 11, 24, and 43). In preparation for transfer, a complex of proteins assembles at the plasmid origin of transfer (oriT). Within this complex, called the relaxosome, a plasmid-encoded relaxase or nickase binds and cleaves one plasmid strand at a specific oriT site (nic). As part of the cleavage reaction, the relaxase forms a covalent linkage between an active-site tyrosyl hydroxyl oxygen and a single-stranded DNA (ssDNA) phosphate, yielding a 3′ ssDNA hydroxyl (19, 30). Upon initiation of transfer, the plasmid strands are separated, and the cut strand is transported into the recipient. The relaxase is likely transferred into the recipient (12, 31) while still physically attached to plasmid DNA. The transferred relaxase may then join the ends of the ssDNA plasmid copy in the final step of plasmid transfer. Complementary strand synthesis in the donor and the recipient generates a double-stranded plasmid that is competent for further transfer. Successful conjugation requires effective temporal regulation, yet the mechanisms governing this regulation are poorly understood.
The F plasmid oriT is ~500 bp long and includes multiple binding sites for integration host factor (IHF), TraY, and TraM and a single site for TraI, the F relaxase (11). IHF, TraY, and TraM, participants in the relaxosome, bind double-stranded DNA to facilitate the action of TraI, perhaps by creating or stabilizing the ssDNA conformation around nic required for TraI recognition. The F TraI minimal high-affinity binding site includes ~15 nucleotides around nic (39), and throughout the text, we refer to oligonucleotides that contain the TraI wild-type (wt) or variant binding site as oriT oligonucleotides. F TraI is 192 kDa (42), and in addition to its relaxase activity, TraI has a 5′-to-3′ helicase activity (4). These activities must be physically joined to allow efficient plasmid transfer (29), yet how the two activities are coordinated is a mystery. The relaxase region of F TraI has been defined as the N-terminal ~300 amino acids (aa) (6, 40). Conserved helicase motifs, including those associated with an ATPase, lie between amino acids 990 and 1450. The C-terminal region (positions 1450 to 1756) plays an important role in bacterial conjugation, possibly involving protein-protein interactions with TraM (32) and/or inner membrane protein TraD (28).
The 70-kDa central region of TraI that lies between the relaxase and helicase domains has been implicated in two functions. Haft and colleagues described TraI variants with 31-amino-acid insertions in this TraI region that facilitated plasmid transfer with greater efficiency than that afforded by the wild-type protein when these proteins are expressed at high levels (16). On the basis of this observation, the authors proposed that the region participated in a negative regulation of transfer. Matson and Ragonese demonstrated that this central region is required for TraI helicase function, likely due to participation in ssDNA recognition essential for the helicase activity (28). We wondered whether the proposed regulatory and ssDNA binding roles of the central region are linked and whether this region might help modulate TraI helicase and relaxase activities. Our objectives in this study were to confirm the role of the central region in ssDNA recognition, to assess the affinity and specificity of the ssDNA recognition by the central region, and to determine whether the relaxase and central domain ssDNA binding sites demonstrate cooperativity in binding. Our work yielded two significant and surprising results. First, the binding site within the TraI central region binds ssDNA with high affinity and significant sequence specificity, both unusual characteristics for a helicase. Second, the central region and relaxase ssDNA binding sites show an apparent strong negative cooperativity of binding, possibly explaining the role of the central region as a negative regulator and providing clues about how the timing of conjugative transfer might be regulated.
The oligonucleotides used in this study were purchased from Integrated DNA Technologies (Coralville, IA). Fluorophore-labeled oligonucleotides were high-performance liquid chromatography (HPLC) purified by Integrated DNA Technologies, while other oligonucleotides were desalted but not further purified. The sequences of all fluorophore-labeled oligonucleotides used in this study are listed in Table Table1.1. Expression plasmids pHP2 (44), for R1 TraI, and pAR45 (41), for R1 TraIΔN308, an R1 TraI fragment lacking the relaxase domain (the N-terminal 308 amino acids), were kindly provided by Ellen Zechner (University of Graz, Austria). Expression vectors for TraI i31 insertion variants (16) were graciously provided by Beth Traxler (University of Washington School of Medicine). Enzymes NdeI and EcoRI were purchased from New England Biolabs (Ipswich, MA).
Expression constructs for TraI fragments were engineered from the pET24a-traI plasmid (40) by PCR amplifying the plasmid, excluding a region encoding either an N-terminal (TraIΔN constructs) or a C-terminal (TraIΔC constructs) region. The primers for the TraIΔN and TraIΔC constructs contained NdeI and EcoRI sites, respectively. Primer sequences are available upon request. The PCR products were digested with NdeI or EcoRI and their ends ligated. The construct sequences were verified by DNA sequencing.
For F TraI expression and purification, Escherichia coli strain BL21(DE3)/pET24a-traI (40) was grown at 37°C in 1 liter of LB medium containing 30 μg/ml of kanamycin to an A600 of ~0.6. Protein expression was induced by adding 0.75 mM isopropyl-1-thio-β-d-galactopyranoside, and cultures were incubated overnight at 20°C. Cells were harvested by centrifugation (5,000 × g for 15 min at 4°C) and resuspended in 35 ml of buffer I (10% glycerol, 100 mM NaCl, 50 mM Tris-HCl [pH 7.5], 1 mM EDTA). Phenylmethylsulfonyl fluoride was added to give a final concentration of 50 μM. Cells were lysed by sonication, and lysate was centrifuged at 4°C for 30 min at 16,000 × g. All chromatography steps were performed using a Gradifrac system (GE Healthcare Life Sciences, Piscataway, NJ). The supernatant was applied to a 5-ml HiTrap heparin column (GE Healthcare Life Sciences, Piscataway, NJ) equilibrated in buffer I. A gradient from 100 mM to 1 M NaCl in buffer I was applied to the column. Elution of proteins was monitored by following absorption at 280 nm, and composition of column fractions was assessed using 7.5% polyacrylamide Tris-Tricine gels (36) and subsequent Coomassie staining. TraI eluted in a peak at approximately 0.4 M NaCl. Peak fractions were collected and diluted with a solution of 10% glycerol, 50 mM Tris-HCl (pH 7.5), and 1 mM EDTA to yield a sample with a final NaCl concentration of 0.1 M. The sample was loaded on a 5-ml HiTrap Q column (GE Healthcare Life Sciences, Piscataway, NJ) equilibrated in buffer I. A gradient from 100 mM to 1 M NaCl in buffer I was applied and protein eluted in a peak at 250 mM NaCl. Peak fractions were collected, and ammonium sulfate was added to give a final concentration of 0.8 M. Protein solution was loaded on a 5-ml HiTrap phenyl column (GE Healthcare Life Sciences, Piscataway, NJ) equilibrated in buffer II [10% glycerol, 0.8 M (NH3)2SO4, 50 mM Tris-HCl (pH 7.5), 1 mM EDTA]. A gradient from 0.8 M to 0 M ammonium sulfate in buffer I was applied and TraI eluted at 0.3 M (NH3)2SO4. Peak fractions were collected and loaded on a 5-ml HiTrap Blue column (GE Healthcare Life Sciences, Piscataway, NJ) equilibrated in buffer I. A gradient from 100 mM to 4 M NaCl was applied. TraI eluted in a broad peak between 2 M and 4 M NaCl. Fractions containing TraI were extensively dialyzed against buffer III (10% glycerol, 20 mM Tris-HCl [pH 7.5], 100 mM NaCl) and concentrated using Amicon Ultra-4 or Centricon Plus-20 centrifugal filter devices (Millipore) to approximately 20 mg/ml. TraI concentration was estimated by absorbance at 278 nm using an extinction coefficient of 135,800 M−1·cm−1, calculated using ProtParam (14) as implemented at http://www.expasy.org/. The TraI yield was approximately 10 mg/liter of culture, with the TraI protein approximately 95% pure, judging from Coomassie-stained Tris-Tricine gels.
F TraI fragments were expressed and purified using essentially the same protocols. R1 TraI was expressed and purified as described previously (8).
The affinities of proteins for ssDNA were measured by following the fluorescence emission intensity and anisotropy changes of a 3′-carboxytetramethyl-rhodamine (TAMRA)-labeled oligonucleotide upon protein binding. Data were collected at 25°C with an ATF-105 automatic titrating fluorometer (Aviv Biomedical, Inc., Lakewood, NJ). Prior to measurement, fluorescently labeled oligonucleotides were diluted to 4 nM in binding buffer (100 mM NaCl, 20 mM Tris-HCl [pH 7.5], 1 mM EDTA). TraI or a TraI fragment in binding buffer was titrated into the oligonucleotide solution. The fluorophore was excited at 520 nm, and emission data were collected at 580 nm. Each data point was averaged over 15 s, and the equilibration time was 75 s. Fluorescence emission intensity data were volume corrected using the equation
where Icorr is the corrected fluorescence emission intensity, I is observed fluorescence emission intensity, Vtot is the total volume, and V is the initial volume.
Fluorescence data were fit to a single-site binding model using KaleidaGraph (Synergy Software) and equation 2:
where D is the concentration of labeled ssDNA, x is the protein concentration, A is amplitude, and A × B is the lower baseline.
Binding stoichiometry was measured using a method similar to the affinity assay. Protein in binding buffer was titrated into 300 nM 3′-TAMRA-labeled ssDNA oligonucleotide in binding buffer. Each data point was averaged over 15 s, and the equilibration time was 75 s. Concentration changes due to volume changes during the experiment were corrected using an equation analogous to equation 1.
The fluorescence emission intensity and anisotropy changes of 4 nM 3′-TAMRA-labeled oligonucleotide were followed upon protein binding at 25°C during a competition assay. In the competition assay, 50 or 300 nM unlabeled ssDNA, which competes with the fluorescently labeled ssDNA for TraI binding, was included in the oligonucleotide solution. TraI in binding buffer was titrated into the oligonucleotide solution. The concentration changes resulting from increased volume during the experiment were small and were corrected using an equation analogous to equation 1. The oligonucleotide sequences used in the competition assays are given in Table Table11.
TraI from F and related plasmids is a bifunctional protein containing both helicase and relaxase activities (1-4, 9, 13, 15, 22, 37, 42). We demonstrated previously that the F TraI relaxase domain recognizes its ssDNA oriT binding site with a subnanomolar KD (equilibrium dissociation constant) and a high level of sequence specificity (17, 18, 39). A separate TraI ssDNA recognition function, associated with the helicase, is less well characterized. We set out to examine the ssDNA binding by the helicase of TraI from the F plasmid. In parallel, we examined TraI from the closely related R1 plasmid. The TraI proteins of R1 and F each are composed of 1,756 amino acids (aa), bind identical oriT sites, and share >97% sequence identity. Even though these two proteins show relatively few sequence differences, we examined both in hopes that we could identify functional differences that we could then assign to individual amino acid differences. We had considerable success with a similar approach when comparing F and R100 relaxase domains (17, 39). The differences in recognition that we observed between F and R1 helicases, however, were minor, and therefore, we will consider the two proteins and their fragments as essentially interchangeable for most purposes. In support of this contention, when complementing an F traI deletion mutant, we found that F and R1 traI under similar promoters restored plasmid transfer efficiency to similar levels (L. Dostál and J. F. Schildbach, unpublished results).
We measured affinities by following the changes in fluorescence intensity and anisotropy of 3′-TAMRA-labeled oligonucleotides upon binding. Binding of fluorophore-labeled oligonucleotides to proteins can be detected as changes in fluorescence anisotropy, caused by the increased mass and slowed rotational diffusion of the protein-DNA complex relative to the level for the oligonucleotide alone and, for some fluorophores, as a difference in fluorescence emission intensity caused by a change in fluorophore environment upon protein binding (5). We used the fluorophore TAMRA, which has an emission intensity that can be significantly influenced by the fluorophore's environment, and this in turn can influence the signal change caused by binding of the protein to the oligonucleotide. Therefore, we calculated many of the affinities by fitting a single-site binding model to the anisotropy data and, independently, to the intensity data. We report both numbers. The sequences of the fluorophore-labeled oligonucleotides are listed in Table Table11.
We initially examined binding of TraI to oriT oligonucleotides that are bound with high affinity by the relaxase domain. Binding curves for both F and R1 TraI could be fit with single-site binding models, yielding subnanomolar apparent KD values for 17-base and 22-base wild-type F oriT oligonucleotides (Table (Table2)2) (26, 28). As anticipated, these results were consistent with the idea that F and R1 TraI molecules contain a single high-affinity ssDNA binding site for the oriT oligonucleotide with an affinity similar to that observed for the relaxase domain binding site.
To assess the binding to the ssDNA site associated with the helicase, we measured TraI affinity for 17- and 22-base G144′C variant oriT oligonucleotides. The single G144′C base substitution, relative to the wild-type F oriT oligonucleotide, increases the KD for the F relaxase domain over 1,000-fold and dramatically reduces the mobilization of a plasmid containing the variant oriT (39). Both F and R1 TraI bind the variant oligonucleotides with nM-range KD values, strongly suggesting that regions C-terminal to the relaxase bind ssDNA oligonucleotides with high affinity (Table (Table2).2). The high affinity of the helicase-associated binding site, coupled with a specificity different from that observed for the relaxase-associated site, should allow us to measure binding to the former site in the intact protein with little interference from the latter site. Consistent with this observation, both F and R1 TraI also bind an oligonucleotide containing the oriT sequence of the closely related R100 plasmid, even though this sequence, which contains two base substitutions relative to the F sequence, is bound poorly by the F relaxase domain (TraI36) (17, 39). As would be expected for a nonspecific binding interaction, F and R1 TraI bind to the 22-base R100 oriT and F G144′C oriT oligonucleotides with similar affinities and with higher affinities than that observed for the 17-base F G144′C oligonucleotide.
We also examined the binding of F and R1 TraI to two 30-base oligonucleotides, one with a sequence taken from χ174 (30-base PhiX) and the second taken from the pET24 vector (30-base HelAct). Neither binds detectably to the relaxase domain (not shown). The 30-base HelAct oligonucleotide was bound by F and R1 TraI with a subnanomolar KD and a higher affinity than those observed for the 17-base and 22-base G144′C oligonucleotides. In contrast, the 30-base PhiX sequence was bound by F (~20 nM KD) and R1 (~3 nM KD) TraI approximately as well as the 17-base F G144′C oligonucleotide and worse than the 22-base R100 and F G144′C oligonucleotides. The lower affinity for the 30-base oligonucleotide than for the 22-base oligonucleotides indicates that there is a degree of sequence specificity in the ssDNA binding activity of TraI. Additional observations support this assessment, with a 60-base poly(dT) oligonucleotide competing poorly for binding to the helicase-associated ssDNA binding site relative to other oligonucleotides, including 2X-G144′C oriT, a 44-base oligonucleotide consisting of a repeat of the 22-base G144′C oriT sequence (not shown).
As a further examination of the ssDNA sequence binding specificity of TraI, we measured binging of TraI to a fluorescently labeled 22-base R100 oriT oligonucleotide in a competition experiment. We assessed the binding of the labeled oligonucleotide alone in the presence of an unlabeled R100 competing oligonucleotide or in the presence of an unlabeled oligonucleotide of the same length and base composition as the R100 oligonucleotide but with a differing sequence (R100 RI, R100 RII, and R100 RIII) (Table (Table11 and Fig. Fig.1).1). None of these oligonucleotides has a propensity to form a secondary structure, as assessed using the program mfold (45). The R100 RII oligonucleotide causes the greatest shift in the binding curves, indicating that it competes most effectively with the fluorescently labeled 22-base R100 oriT oligonucleotide for the TraI ssDNA binding site. R100 RI and R100 RIII bound with slightly lower affinity than RII, and all bound better than the 22-base R100 oriT oligonucleotide. While the differing effects caused by the unlabeled oligonucleotides indicated sequence specificity in the recognition by the helicase-associated binding site, the relatively small range of the effects suggested that the affinity differences are modest, and the affinities for these oligonucleotides all fall within the nM range.
Having demonstrated that TraI contains two distinct ssDNA binding activities, we set out to localize the helicase-associated binding site. Byrd and colleagues reported previously that an F TraI fragment containing the residues C-terminal to residue 309 possessed high-affinity ssDNA binding (nM range KD) and significant helicase activity but that a C-terminal fragment starting at amino acid 348 had significant albeit reduced ssDNA affinity but no helicase activity (6). We characterized the ssDNA binding of R1 TraI and an R1 TraI fragment that starts at residue 309 (R1 TraIΔN308) (Table (Table2).2). R1 TraIΔN308 binds ssDNA with an affinity reduced approximately 10-fold relative to the level for R1 TraI (Table (Table2),2), indicating that amino acids N-terminal to residue 309 affect TraI ssDNA binding. The R1 TraIΔN308 fragment, however, displays wild-type helicase activity (41).
We also examined the binding of an F TraI fragment that starts at residue 331 (F TraIΔN330), which showed significantly reduced affinity for ssDNA (KD, ~100 nM), relative to the intact protein. The reduced binding affinity of the TraIΔN330 variant did not appear to be an indirect effect resulting from the unfolding of the protein, because the protein purified readily, did not show a dramatically increased susceptibility to proteolysis during purification, and did not precipitate at high (20 mg/ml) concentrations. Byrd and colleagues also reported that a TraI fragment starting at amino acid 348 was soluble and could be purified (6). The helicase-associated ATP hydrolysis by TraIΔN330 is minimal (L. Dostál and J. F. Schildbach, unpublished results), but that is likely the result of the significantly reduced binding affinity of the protein for ssDNA.
The sum of these observations strongly suggests that the high-affinity helicase-associated ssDNA binding activity of TraI requires residues located between amino acids 309 and 331. There is no interpretable electron density for this region in the crystal structures of F TraI36 (10, 23). This observation could mean that this segment is flexible, providing contact residues but stably folding only upon ssDNA binding. Alternatively, this region may be folded, but only in the presence of the region C-terminal to position 330.
To further define the TraI helicase-associated ssDNA binding activity, we engineered multiple F TraI fragments based on trypsin digestion experiments and results from the program PONDR (Predictor of Naturally Disordered Regions) (33; data not shown). Figure Figure22 depicts a schematic of the fragments used in this study. Consistent with the results described above, no fragment starting at amino acid 331 demonstrated high affinity (with approximately nM-range KD values) for any ssDNA oligonucleotide tested. Fragments F TraIΔN973 and F TraI 974-1520, which both contain all helicase motifs, had no detectable affinity for ssDNA. Fragments F TraIΔC1520, F TraIΔC973, and F TraIΔC822 (Table (Table2)2) bound wt F oriT, G144′C oriT, and R100 oriT oligonucleotides with affinities similar to those observed for F and R1 TraI. These results indicate that (i) both relaxase and helicase-associated ssDNA binding sites are intact in these fragments, (ii) the helicase-associated ssDNA binding site does not extend C-terminal to amino acid 822, and (iii) the helicase-associated ssDNA binding does not require any helicase motifs.
To gain insight into the organization of TraI regions whose structures are as yet undetermined, we submitted the amino acid sequences of these regions, in segments, to the Phyre web server (http://www.sbg.bio.ic.ac.uk/~phyre/) (20). The TraI region comprising positions 830 to 1473, which contains helicase motifs, yielded a highly significant (E value = 5.5 × 10−27) match with aa 4 to 606 from E. coli RecD from the RecBCD structure (Protein Data Bank identification code [PDB ID] 1W36) (38). E. coli RecD, like TraI, contains a 5′-to-3′ helicase activity (7). Surprisingly, the TraI region comprising positions 303 to 844 also yielded a significant (E value = 9.5 × 10−23) match with E. coli RecD (aa 22 to 608). These observations are consistent with the idea that a single RecD-like domain within a TraI precursor was duplicated, with the N-terminal domain subsequently specializing for ssDNA binding and losing the helicase motifs and the C-terminal region losing ssDNA binding while retaining functional helicase motifs.
To place our results for the TraI helicase-associated ssDNA binding site in a structural context, we examined a structure of the E. coli RecBCD complex with bound DNA (PDB ID 3K70) (34) as well as structures of Deinococcus radiodurans RecD2 complexed with DNA, with (PDB ID 3GPL) (35) or without (PDB ID 3GP8) (35) a bound ATP analogue. In all three structures, DNA docks into a shallow cleft in RecD and is contacted by amino acids from multiple discontiguous regions of the protein. There is, however, no contribution to the RecD binding site from amino acids corresponding to the region comprising positions 309 to 330 of TraI. These results suggest that while the segments of RecD and TraI that comprise the helicase-associated ssDNA binding sites (within the first 822 N-terminal amino acids) may have similar folds, they likely differ significantly in the details of binding.
Examination of the structure of Deinococcus radiodurans RecD2 complexed with ssDNA identified a short loop region, termed the “pin,” that forms a β-hairpin in close proximity to the bound ssDNA (34). Deletion of the pin renders the RecD2 helicase inactive, but the protein retains wild-type DNA-dependent ATPase activity. The Phyre results suggest that the pin motif is present in the C-terminal RecD-like fold (positions 830 to 1473) of TraI but is absent in the N-terminal domain (positions 303 to 844).
The two TraI high-affinity binding sites for ssDNA are each capable of binding to the F 22-base oriT oligonucleotide with a nanomolar or subnanomolar KD. To confirm the stoichiometry of binding, we titrated F TraI into a solution of 300 nM fluorophore-labeled F 22-base oriT oligonucleotide. To our surprise, we found that anisotropy and intensity increase linearly until the TraI concentration reaches ~300 nM, where binding is saturated (Fig. (Fig.3,3, squares), consistent with a 1:1 TraI/ssDNA ratio. Equivalent results were obtained with TraI36 and R1 TraIΔN308 (Fig. (Fig.3,3, triangles and circles, respectively), which each contain only one ssDNA binding site.
These results demonstrate that binding of ssDNA to the relaxase domain is not independent of binding of ssDNA to the helicase-associated ssDNA binding site. This phenomenon resembles negative cooperativity, in which binding of a ligand to one binding site decreases the affinity of the ligand for a second binding site. Binding of a 22-base fragment of F oriT to the relaxase binding site interferes with binding of ssDNA to the helicase-associated site, and vice versa. Because the phenomenon described here involves distinct instead of identical oligonucleotides and binding sites, we refer to it as apparent negative cooperativity of binding. To further examine this effect, we performed binding assays in which F TraI or TraI36 is titrated into a solution of 50 nM unlabeled ssDNA plus 4 nM 3′-TAMRA-labeled 22-base F oriT oligonucleotide. Figure Figure4A4A shows the results from the relaxase domain TraI36. The data for TraI36 show little effect by any unlabeled oligonucleotide other than the 22-base F oriT oligonucleotide, consistent with the idea that TraI36 possesses a single, sequence-specific site. Data from TraI are presented in Fig. Fig.4B.4B. When either unlabeled 22-base F oriT or unlabeled 44-base 2X-G144′C oriT was used as the competing oligonucleotide, the binding of the labeled oligonucleotide to TraI was similarly and dramatically reduced. Therefore, an oligonucleotide that is capable of binding only the helicase-associated ssDNA binding site with high affinity is as effective a competitor for the labeled oligonucleotide as is an unlabeled oligonucleotide that, like the labeled oligonucleotide, can bind to both TraI sites with high affinity. The unlabeled 60-base poly(dT) oligonucleotide has a considerably less pronounced effect, consistent with its lower affinity for the helicase-associated ssDNA binding site. Therefore, we conclude that there is an apparent negative cooperativity between the relaxase ssDNA binding site and the helicase-associated ssDNA binding site.
TraIi369, TraIi593, and TraIi681 are TraI variants generated by Haft and colleagues (16) that have 31-aa insertions within regions of TraI essential for ssDNA binding. TraIi369, TraIi593, and TraIi681 yield higher transfer efficiencies than wild-type TraI when provided in trans from the pTrc99a plasmid to complement an F traI deletion. The observed increase in F plasmid transfer facilitated by TraIi369, TraIi593, and TraIi681 was interpreted as a release from negative regulation.
We measured the affinities of TraIi369, TraIi593, and TraIi681 for ssDNA and compared the resulting affinities with that of TraI (Table (Table3).3). All three variants showed significant affinity reductions for the G144′C, R100, and HelAct oligonucleotides, indicating that the i31 insertions are negatively affecting ssDNA binding. We also tested for the apparent negative cooperativity of binding by examining the effect of excess unlabeled 2X-G144′C oriT oligonucleotide on the binding of the labeled 22-base oriT oligonucleotide (Fig. (Fig.5).5). The TraI36 competition curve shows no obvious effect of the presence of the 2X-G144′C oligonucleotide on the binding of the relaxase domain. The R1 TraIΔN308 competition curve is shifted well to the right, reflecting a higher affinity of R1 TraIΔN308 for 2X-G144′C than for the 22-base oriT oligonucleotide. The curve for F TraI lies between curves for F TraI36 and R1 TraIΔN308. The curves for the TraI i31 variants show dramatic shifts in the curves relative to TraI, demonstrating that binding of ssDNA to the relaxase domains of TraIi369, TraIi593, and TraIi681 is less affected by an oligonucleotide that can bind to the helicase-associated ssDNA binding site. Thus, the i31 variants demonstrate a less pronounced apparent negative cooperativity, at least under the conditions tested.
Using oligonucleotides selected based upon our earlier examination of the binding specificities of the F and R100 TraI proteins (17, 39), here we confirm that F TraI contains two ssDNA binding sites, both of which demonstrate sequence specificity and, importantly, show an apparent negative cooperativity of binding. The first site is associated with the relaxase, located in the N-terminal 300 amino acids, while the second, associated with the helicase, is located within first 822 amino acids. Both are capable of binding ssDNA oligonucleotides with nM or sub-nM range KD values, and both show sequence specificity. We present evidence that suggests that the region comprising positions 309 to 330 of F TraI is essential to the helicase-associated binding site. One implication of the role of the region comprising positions 309 to 330 in helicase-associated binding is that the relaxase domain and at least a portion of the helicase-associated sites are physically close to each other.
While there is sequence specificity evident in both sites, the helicase-associated site appears the less specific of the two. As shown by the dramatically different affinities of the F relaxase domain for the F wild-type oriT, F G144′C oriT, and R100 oriT oligonucleotides (17, 39; this paper) as well as other oligonucleotides (18, 39), a single base substitution can result in a nearly 10,000-fold difference in relaxase domain binding affinity. In contrast, as shown by the results for R1 TraIΔN308 (Table (Table2),2), the helicase-associated site of TraI binds these oligonucleotides with similar high affinities. Furthermore, although scrambling the R100 oligonucleotide sequence definitely affects the affinity for the helicase-associated binding site, all R100-derivative oligonucleotides were bound relatively well (Fig. (Fig.1).1). In other evidence for a moderate sequence specificity, a 60-base poly(dT) oligonucleotide competed poorly for binding to the helicase-associated binding site relative to several shorter oligonucleotides; data for one of these shorter oligonucleotides, 2X-G144′C oriT, are shown in Fig. Fig.4B.4B. The sum of results from the binding assays strongly suggests a significant sequence specificity for the helicase-associated site, but the degree of sequence specificity does not rival that of the relaxase site.
Binding experiments using high concentrations of fluorescently labeled oligonucleotide clearly demonstrate an apparent negative cooperativity between the two binding sites (Fig. (Fig.3).3). At the 300 nM concentration used, if both binding sites were accessible, 150 nM TraI should be able to bind 300 nM oligonucleotide. Instead, 300 nM TraI is required for the binding of the oligonucleotide. The 1:1 binding model is surprising because it indicates that the apparent negative cooperativity is extremely strong and allows for only one site of TraI to be occupied. This does explain, however, our failure to detect simultaneous occupation of both sites by use of fluorescence resonance energy transfer or fluorescence correlation spectroscopy (L. Dostál and J. F. Schildbach, unpublished observations). In addition, it explains why the TraI binding data shown in Table Table22 could be fit with a single binding site model even though the protein had two high-affinity ssDNA binding sites. Because the relaxase binding site is located near the portion comprising positions 309 to 330 of the helicase-associated site in three-dimensional space, binding at one site may influence binding at the second through occlusion of the second site. Alternatives, including an allosteric model in which conformational changes propagate from one site to the other site upon binding of ssDNA, are also possible.
The apparent negative cooperativity in the binding to the two TraI ssDNA binding sites could play an essential role in the regulation of TraI ssDNA activities. In cells containing conjugative plasmids (21), covalent complexes of plasmid and protein, subsequently shown to include relaxases, are established and maintained even in the absence of recipients. On the basis of the greater recognition of sequences 5′ to nic (23, 39) and on the sequestration of the 3′ hydroxyl at nic while in the complex (27), we imagine that any TraI molecule participating in the relaxosome would have its relaxase ssDNA binding site occupied and would thus have an effectively inactive helicase. Once the relaxosome is disrupted and, potentially, the ssDNA on the cut strand 5′ to nic is released from the relaxase binding site, the helicase-associated site might not only be competent to bind ssDNA but, because of its proximity to the relaxase binding site, would also be located near the ssDNA at nic. This could facilitate the loading of the helicase onto plasmid DNA in preparation for strand separation and transfer.
Sut and colleagues recently reported evidence suggesting that R1 TraI possesses a regulatory mechanism that influences the efficiency of its helicase activity (41). Using heteroduplex oligonucleotide substrates that contain a 60-bp noncomplementary region in the center to allow for helicase loading, they showed that TraI less effectively separates substrates that have 30 or more bases 5′ to nic within the noncomplementary region than a substrate that has only 15 bases 5′ to nic within this region. In contrast, R1 TraIΔN308 separates all substrates equally well. The substrate that is separated most effectively by TraI is also the substrate with the shortest single-stranded region 5′ to nic and may offer the least access to the TraI relaxase domain. If so, Sut and colleagues may be seeing another manifestation of the apparent negative cooperativity that we report here. If the relaxase domain cannot efficiently bind a heteroduplex substrate, the relaxase domain will not exert influence over substrate binding by the helicase, thus rendering the helicase more effective.
We also examined a trio of TraI i31 insertion variants isolated and characterized by the Traxler laboratory (16). These variants are a product of a transposon mutagenesis process that yields in-frame 31-aa insertions within cloned genes (25, 26). The three variants were among several generated that, when complementing a traI deletion from a high-copy-number plasmid, actually facilitated plasmid transfer at a greater efficiency than complementation by the wild-type traI gene. All three variants (i369, i593, and i681, where the number indicates the site of the insertion in the amino acid sequence) had moderately reduced affinity for ssDNA relative to wild-type TraI (Table (Table3).3). In addition, the insertion variants all demonstrated reduced levels of apparent negative cooperativity in binding of ssDNA to their relaxase and helicase-associated sites (Fig. (Fig.5).5). These two observations are probably linked, with the change in apparent negative cooperativity being an indirect result of a reduced affinity for a competing oligonucleotide. It is intriguing, however, that all the insertion variants tested exhibit this characteristic. Haft and colleagues concluded that the enhanced transfer efficiencies observed when a traI deletion was complemented in trans with these insertion variants were due to a release of these proteins from a negative regulation (16). If these insertion variants are indeed demonstrating a loss of negative cooperativity in binding to its two ssDNA binding sites, this loss of negative cooperativity could represent the observed release from negative regulation. On the other hand, the reduced ssDNA affinity and the increased efficiency of transfer seen with the i31 variants may be coincidental. Certainly, observing increased transfer rates from variants that presumably would perform one of their tasks less efficiently (for example, the maximal helicase-associated ATP hydrolysis rate of variant TraIi369 is reduced to one-third of that of TraI [L. Dostál and J. F. Schildbach, unpublished results]) is counterintuitive. Much is still unknown about the molecular processes involved in conjugation and the activities of TraI in conjugation. Placing our results within the larger context of TraI function and F transfer will require additional study.
We thank Beth Traxler for providing TraIi31 expression vectors and Ellen Zechner for providing R1 TraI and R1 TraIΔN308 expression vectors. We also thank Chris Larkin, Sarah Williams, and Nils Walter for helpful discussions and Nate Wright for comments on the manuscript.
This work was supported by National Institutes of Health grant GM61017 to J.F.S.
Published ahead of print on 30 April 2010.