Interaction between the co-inherited TraG coupling protein and the TraJ membrane-associated protein of the H-plasmid conjugative DNA transfer system resembles chromosomal DNA translocases

Bacterial conjugation is a mechanism of horizontal gene transfer, which has both evolutionary and medical implications. Plasmids can represent a major contribution to overall genetic content, and conjugation is the principal route for the dissemination of antibiotic resistance determinants (Mazel & Davies, 1999). The conjugative pilus initiates plasmid transfer from the donor cell by recruitment of recipient cells. A membrane-spanning protein complex called the mating pair formation (Mpf) complex, which constitutes the greater part of the conjugative type IV secretion system (T4SS), enables the production of conjugative pili, and facilitates the transfer of plasmid DNA from donor to recipient bacteria (for reviews, see Cascales & Christie, 2003, 2004; Lawley et al., 2003a). Plasmid DNA is processed to a transfer intermediate in the cytoplasm by the relaxosome, a DNAprotein complex containing the relaxase and accessory proteins (Furste et al., 1989). The T4SS also includes the coupling protein at the cytoplasmic membrane, and by serving as the link between the relaxosome and Mpf complex, the transferring DNA strand contacts the cell envelope and then proceeds onwards to the recipient cell.

The incompatibility group HI1 plasmid (IncHI1) R27 is a large (180 kb) self-transmissible resistance plasmid (Sherburne et al., 2000). There has been some confusion over the years as to the origin of R27, but Datta & Olarte (1974) have clearly distinguished it as a plasmid characteristic of Salmonella enterica serovar Typhi. However, R27 and a related plasmid Tp117 have been isolated from Salmonella typhimurium strain 1M1407 (Lawn et al., 1967; Smith et al., 1973). The proteins required for R27 transfer are located within two separate transfer regions, Tra1 and Tra2 (Lawley et al., 2002, 2003b). Encoded within Tra1 are the relaxosome, the coupling protein and part of the Mpf complex, and the remainder of the Mpf is encoded in Tra2. Interactions between the relaxosome and coupling protein have been identified in the IncW, IncP and IncF plasmid families (Disque-Kochem & Dreiseikelmann, 1997; Llosa et al., 2003; Schroder et al., 2002). The coupling protein then interacts with the Mpf/T4SS at the cytoplasmic membrane through the TraB/VirB10 protein (Gilmour et al., 2003; Llosa et al., 2003). The R27 coupling protein is termed TraG, and this family of proteins has been structurally characterized. The hexameric ring structure of the IncW plasmid-coupling protein TrwB resembles the tertiary structure of a number of P-loop ATPases, such as the F1 ATPase, suggesting that this complex is actively involved in DNA translocation (Gomis-Ruth et al., 2001).

Three other essential genes encoded in the Tra1 region (traI, traH and traJ) are not required for the production of the H-conjugative pilus, and consequently are not components of the Mpf complex (Lawley et al., 2002). The TraI protein shares homology with the relaxase family of proteins, whereas TraH contains a DNA-binding motif and an N-terminal coiled-coil domain (Lawley et al., 2002). TraJ homologues from a variety of genomic sources have been identified (Gilmour et al., 2004), and an alignment of these homologues has indicated that the predicted four transmembrane (TM) domains are well-conserved among this family of proteins. A conserved module of traItraGtraJ has been identified in each of the genomic sources that encode a TraJ homologue (Gilmour et al., 2004). Few of the TraJ-encoding genomes encode R27-like Mpf genes, suggesting that the module of traIGJ genes can be inherited independently from other transfer determinants.

The objective of the present study was to identify additional R27-encoded interaction partners for the coupling protein TraG. Here, we identified an interaction between TraJ and TraG, and propose that because of co-localization at the cytoplasmic membrane, binary heterologous interaction, and co-inheritance with conserved domain organization of known DNA translocases, TraJ is a protein which associates with the coupling protein TraG and is involved in DNA translocation.

Bacterial strains and growth conditions.
Escherichia coli strains used in this study are listed in Table 1. All strains were grown in LuriaBertani (LB) broth, Lennox formulation (Difco) at 37 °C with shaking at 200 r.p.m., unless otherwise stated. Antibiotics used in this broth were as follows: ampicillin (100 µg ml¹), kanamycin (50 µg ml¹), tetracycline (10 µg ml¹), rifampicin (50 µg ml¹), nalidixic acid (20 µg ml¹) and trimethoprim (25 µg ml¹). For the bacterial two-hybrid (BTH) assay, conjugative transfer proteins were fused to catalytic domain fragments of adenylate cyclase from Bordetella pertussis by cloning into the BTH vectors pKT25 and pUT18C (Karimova et al., 1998). Plasmids encoding fused proteins were co-transformed into competent BTH101 cells, processed and analysed as previously described (Gilmour et al., 2003). Liquid conjugative matings and complementation experiments with a derepressed derivative of R27, which contains an insertion in the htdA gene that causes a derepressed phenotype, were performed as described previously (Lawley et al., 2002; Taylor & Levine, 1980). Complementation experiments with the F plasmid were performed similarly, with the exception that donor and recipient cells were grown at 37 °C, and mating experiments were performed at 37 °C for 16 h. For solid matings, conjugative transfer of RP4 and R388 plasmids was performed as described by Bradley et al. (1980).

Table 1. Bacterial strains and plasmids used in this study Abbreviations: Nalr, nalidixic acid resistance; Rifr, rifampicin resistance; Tcr, tetracycline resistance; Kanr, kanamycin resistance; Ampr, ampicillin resistance; Strr, streptomycin resistance; Cmr, chloramphenicol resistance; Tpr, trimethoprim resistance.

DNA manipulation and cloning.
Standard recombinant DNA methods were used, as described by Sambrook et al. (1989). PCR products were amplified with terminal restriction enzymes added within the primer sequences (Table 2). Amplified DNA was initially cloned into pGEM-T (Promega), digested with appropriate enzymes, and subcloned into the expression vector pMS119EH/pMS119HE or the adenylate cyclase fusion vectors pKT25 and pUT18C. Restriction enzyme endonucleases were used according to the manufacturer's specifications, and digested products were resolved on agarose gels.

Table 2. Primers used in this study

Production of coupling protein homologues.
A His₆ tag was incorporated into the primer design for coupling protein amplification, such that each protein contained a C-terminal histidine tag for detection of the recombinant proteins. E. coli strain DH5α, containing pMS119EH encoding each tagged coupling protein, was grown at 28 °C to OD₆₀₀ 0.50.7. The bacteria were induced for 1 h at 28 °C with IPTG at a final concentration of 0.4 mM. Whole-cell lysates were resolved on 10 % SDS-PAGE, and following transfer to nitrocellulose membranes, a primary mouse anti-His₆ mAb (Invitrogen) and a secondary goat anti-mousehorseradish peroxidase (HRP) antibody conjugate were used to probe for protein synthesis.

Immunoprecipitation.
Immunoprecipitation of R27 transfer proteins was performed as described by Couturier et al. (2006). Briefly, E. coli DH5α cells containing the plasmid pUT18C-TraJ were co-transformed with pBAD33-TraG_FLAG or pBAD33-TrhB_FLAG for the interaction studies. Cells were induced at mid-logarithmic growth phase (OD₆₀₀ 0.50.7) with 0.4 mM IPTG and 0.2 % arabinose for 1 h prior to cell lysis. Cells were harvested by centrifugation and resuspended in a threefold volume of lysis buffer [150 mM PBS, 7.15 % (w/v) sucrose, 1 % Nonidet P-40, 1 mM PMSF, 1/66 vol. 15 µg ml¹ lysozyme). The cells were frozen and thawed three times, and lysed cells were removed by centrifugation at 14 000 r.p.m. for 10 min. The supernatant was removed and anti-FLAG M2 agarose (Sigma) beads were added. The beads had been incubated at 4 °C overnight with rotation in a 5 % BSA solution to avoid non-specific precipitation. The supernatant and beads were incubated at 4 °C with rotation for 2 h. The supernatant was removed after centrifugation at 14 000 r.p.m. for 30 s, and the beads were washed three times with lysis buffer. The immunoprecipitate fraction was collected by resuspending the beads in 50 µl 1xLaemmli sample buffer (5 % β-mercaptoethanol). Upon resolution of the cleared lysate and immunoprecipitation by 10 % SDS-PAGE, proteins were transferred to nitrocellulose and probed with the anti-adenylate cyclase mAb 3D1 (List Biological Laboratories). This mAb is specific for the distal portion of the adenylate cyclase catalytic domain (aa 377399) (Lee et al., 1999).

Membrane preparations.
Overnight culture (20 ml) of E. coli DH5α (pUT18C-TraJ) with ampicillin was used to inoculate 500 ml LBLennox broth. Cells were grown to mid-exponential phase at 28 °C and 225 r.p.m., and induced with 0.4 mM IPTG for 20 min. Bacterial cells were harvested by centrifugation (7000 g, 7 min), resuspended in 10 ml buffer A [10 mM Tris/HCl, pH 7.6, 5 mM MgCl₂, one Complete Protease Inhibitor Cocktail tablet (Roche), 1 mM PMSF, 100 U micrococcal nuclease, 1 mg RNase]. One milligram of lysozyme was added to cells for 30 min on ice, and cells were lysed using sonication for 3 min (Fisher model 300 sonicator). Unlysed cells were removed by centrifugation (twice at 6800 g for 10 min), and this preparation was called the whole-cell lysate. The cytoplasmic/periplasmic fraction was obtained by ultracentrifugation (100 000 g for 2 h at 4 °C) and the supernatant fraction was transferred to a new tube. The resulting pellet containing the membrane fraction was resuspended in 1 ml buffer B [55 % sucrose (w/v), 10 mM Tris/HCl, pH 7.6, 5 mM EDTA].

Immunodetection of proteins.
Protein preparations were boiled for 10 min and separated using 10 % SDS-PAGE, transferred to nitrocellulose membranes, and blocked overnight in 1 % (w/v) skimmed milk and 0.1 % Tween 20 in PBS. Membranes were probed with the anti-adenylate cyclase mAb 3D1, anti-DnaK mouse antibody (Stressgen) or anti-CpxA rabbit antibody (Raivio et al., 1999) for 1 h at room temperature, washed, and the secondary anti-mouse IgG or anti-rabbit IgG (Sigma) was applied. Detection was performed using the ECL kit (Amersham Biosciences) according to the manufacturer's instructions.

Phylogenetic analysis.
Protein sequences were aligned using CLUSTAL_W (Gonnet matrix, gap penalty=10, extension penalty=0.2) (Thompson et al., 1994), and phylogenetic trees were constructed using the unweighted pair group method with arithmetic mean (UPMGA) algorithm of MEGA3 (Kumar et al., 2004). To test the statistical significance of the tree, 500 bootstrap samples were generated from the alignment data.

Web-based computer programs.
PSI-BLAST (), ScanProsite (), TMHMM (), CLUSTAL_W () and EMBOSS pairwise local alignment (Blosum40 matrix, gap penalty=10, extension penalty=0.5; ) were used.

Although it has been difficult to identify interacting partners within the membrane-associated components of the conjugative apparatus because of the extracytoplasmic locations of the interacting proteins, BTH studies with the IncH plasmid R27 have previously demonstrated an interaction between the coupling protein TraG and VirB10 homologue TrhB (Gilmour et al., 2003). Elegant studies have identified similar interactions between TrwB (coupling protein) and TrwE (VirB10 homologue) from the IncW plasmid R388, as well as VirB10 homologues in IncN and IncX plasmids (Llosa et al., 2003). In this study, BTH technology was used to investigate TraG and potential interactions with non-Mpf proteins TraJ, TraH and TraI, located contiguously in the Tra1 region of R27 (Lawley et al., 2002).

In vivo detection of the R27 TraG and TraJ interaction
The R27 relaxase gene traI, non-Mpf gene traH and non-Mpf gene traJ (encoding a putative membrane-associated transfer protein) were each cloned into the BTH vector pUT18C. The R27 coupling protein gene traG was cloned in the appropriate reading frame of the pKT25 vector, and the resulting gene product TraG_AC25 was a fusion between the 78 kDa coupling protein and the C-terminal 25 kDa region of the adenylate cyclase catalytic domain. Screening for pairwise interactions using the BTH system did not identify an interaction that involved R27 relaxase TraI or TraH. However, co-production of R27 TraJ_AC18 with TraG_AC25 allowed for functional complementation of the B. pertussis adenylate cyclase domains, which were initially screened by plating transformed BTH101 cells onto media containing X-Gal. More than 99 % of the colonies containing TraJ_AC18 with TraG_AC25 constructs were blue within 48 h of plating. The BTH positive control (vectors containing leucine zippers) showed similar blue colonies, whereas the BTH101 cells containing empty BTH vectors were white. A standard Miller assay was used to quantify β-galactosidase activity (Fig. 1). The in vivo interaction between TraJ and TraG was found to produce comparable levels of β-galactosidase activity to those of the BTH leucine zipper-positive control (Fig. 1).

(23K): Fig. 1. In vivo BTH interactions between the R27 coupling protein TraG and R27 transfer protein TraJ. Conjugative proteins were fused with adenylate cyclase domains and co-produced in BTH101. The BTH controls were BTH101 cells containing pUT18C and pKT25 (negative control), and pUT18C-leucine zipper with pKT25-leucine zipper (positive control). Liquid cultures of transformed BTH101 cells were analysed for β-galactosidase activity using a Miller assay (Miller, 1972). Results of a representative experiment are shown.

Both TraG and TraJ contain multiple, well-conserved predicted TM domains (Gilmour et al., 2004), and to ensure that this interaction was not a non-specific interaction of membrane-spanning domains caused by overexpression of R27 transfer proteins, the R27 virB10 homologue trhB was cloned into the BTH vector pUT18C. The VirB10 family of proteins are cytoplasmic membrane-associated conjugative proteins that have been shown to interact with coupling proteins (Gilmour et al., 2003; Llosa et al., 2003). When TraJ_AC18 was co-produced with TrhB_AC25 in BTH101, >99 % of the colonies were white after 48 h incubation on media containing X-Gal. The inability of TrhB to interact with TraJ implies that the productive TraGTraJ interaction was not simply due to predicted co-localization of transfer proteins at the cytoplasmic membrane.

Immunoprecipitation of a TraG and TraJ complex
To confirm the interaction between the R27 coupling protein and TraJ, we purified R27-tagged protein complexes using an immunoprecipitation technique. The C termini of TraG and TrhB were fused with a FLAG octapeptide Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys tag. Cell lysates were prepared after co-production of TraJ_AC18 and TraG_FLAG, and in the presence of Sepharose beads conjugated with anti-FLAG antibody, the TraJ_AC18 construct was precipitated by TraG_FLAG (Fig. 2). The mAb 3D1, which recognizes an epitope in the 18 kDa adenylate cyclase fragment of B. pertussis (Lee et al., 1999), was used to detect precipitated TraJ_AC18. When TraG_FLAG was omitted from the immunoprecipitation experiment, no TraJ_AC18 was recovered by the anti-FLAG bead matrix (Fig. 2), ensuring that TraJ was not bound non-specifically by a component of the immunoprecipitation reaction. Additionally, to ensure that the TraG and TraJ interaction was not due to overproduction of two membrane-associated proteins, TrhB was included in an immunoprecipitation experiment with TraJ. Although TraJ_AC18 was soluble and detectable in the cleared lysate prior to immunoprecipitation, TrhB_FLAG did not precipitate TraJ_AC18 (Fig. 2). Furthermore, the inability of TraJ to interact with TrhB confirmed that the interaction of TraJ_AC18 and TraG_FLAG was not due to non-specific interaction between the 18 kDa adenylate cyclase domain and TraG_FLAG. The precipitation of a TraGTraJ complex confirmed the BTH data, which showed that the R27 coupling protein interacted with the non-Mpf protein TraJ.

(26K): Fig. 2. Co-immunoprecipitation of an IncH R27 TraJTraG complex. The cellular lysates (CL) of E. coli DH5α cells expressing TraJAC18 and TraGFLAG were mixed with M2 anti-FLAG affinity gel beads, washed, and resolved with 10 % SDS-PAGE. After transfer to nitrocellulose, the samples were probed with the anti-adenylate cyclase mAb 3D1, which detected TraJ in the immunoprecipitate samples (IP). As a control for non-specific precipitation of TraJAC18, cellular lysates containing TraJAC18 and TrhBFLAG, or TraJAC18 alone were mixed with M2 anti-FLAG affinity gel beads, and resolved as described above.

TraJ localizes to the cellular membrane
Sequence alignments of R27-encoded TraJ and TraJ homologues have identified that this group of proteins are similar in size (∼220 aa), and contain four conserved hydrophobic TM regions (Gilmour et al., 2004). To determine if the predicted TM regions correlated with localization of TraJ to the cellular membrane, a fractionation study was performed on E. coli containing pUT18C-TraJ. Separation of the cytoplasmic membrane and outer membranes (M fractions) from the soluble cytoplasmic and periplasmic fractions (S fractions), and subsequent Western blot analysis revealed that TraJ_AC18 associated exclusively with the membrane fraction (Fig. 3A).

(25K): Fig. 3. TraJAC18 localizes to the cellular membrane. Bacteria containing pUT18C-TraJ were fractionated into a cytoplasmic and periplasmic soluble fraction (S), and a total membrane fraction (M). Whole-cell lysate (L) samples were also run on SDS-PAGE, and transferred to nitrocellulose membranes. Membranes were individually probed with (A) anti-AC 3D1 to detect TraJAC18, (B) anti-DnaK and (C) anti-CpxA. There was evidence of non-specific banding with the application of the polyclonal anti-CpxA antibody.

To ensure the efficacy of the fractionation procedure, the M and S fractions were probed with antibodies specific for cytoplasmic and membrane-associated proteins. DnaK is a well-characterized cytoplasmic heat shock protein that is essential for λ replication (Liberek et al., 1988), and an anti-DnaK antibody identified this protein exclusively in the S fraction (Fig. 3B). CpxA is a component of the Cpx pathway involved in the envelope stress response (see Raivio, 2005, for a review), and anti-CpxA antibody detected this protein exclusively in the M fraction (Fig. 3C).

Immunofluorescent analysis of the R27 coupling protein TraG has previously demonstrated a similar membrane localization (Gunton et al., 2005). Fluorescently labelled R27 coupling protein was visualized as discrete foci in the periphery of the cell. The discovery that TraJ is membrane bound and able to interact with the membrane-associated R27 coupling protein indicates that TraJ may act as an interacting partner for the coupling protein during conjugation.

Non-cognate coupling protein interactions with the IncH Mpf
Complementation of conjugative plasmids containing coupling protein mutations by non-cognate coupling proteins has indicated that the Mpfcoupling protein interaction is non-specific (Cabezon et al., 1994; Hamilton et al., 2000; Llosa et al., 2003). Furthermore, recent studies on the coupling proteins and VirB10 homologues from the IncW R388 plasmid, IncN plasmid pKM101 and IncX plasmid R6K have demonstrated that each coupling protein is able to interact with cognate and heterologous VirB10 proteins (Llosa et al., 2003). To expand upon these studies, the coupling proteins from the conjugative plasmids R27 (IncH), RP4 (IncP), F (IncF) and R388 (IncW) were cloned into the pUT18C BTH vector. When each coupling protein was co-produced in BTH101 cells with the R27 Mpf construct TrhB_AC25, there were successful interactions between each coupling protein and the VirB10-homologue TrhB (Fig. 4A). The values obtained in the BTH studies of the interactions between the coupling proteins and TrhB were similar to those produced by the BTH leucine-zipper positive control. This seemed remarkable, given the low level of overall similarity shared between members of the coupling protein family. The coupling proteins from the RP4, R388 and F plasmids had similar levels of sequence identity with TraG^R27: TraG^RP4, 19.6 % identity over an alignment length of 749 aa; TraD^F, 19.5 % identity over an alignment length of 845 aa; and TrwB^R388, 20.1 % identity over an alignment length of 750 aa.

(21K): Fig. 4. BTH data for R27 transfer protein TraJ or the Mpf protein TrhB produced with the coupling protein from IncH (R27), IncF (F plasmid), IncP (RP4) and IncW (R388) plasmids. The BTH plasmid pKT25 was used to produce IncH R27adenylate cyclase fusion proteins TraJAC25 and TrhBAC25. The coupling protein homologues were cloned into pUT18C, resulting in the expression of TraGH, TraDF, TraGP and TrwBW. Protein interaction was determined by assaying for the β-galactosidase activity produced by co-expression of (A) TrhBH or (B) TraJH with TraGH, TraDF, TraGP and TrwBW. The BTH controls were BTH101 cells containing pUT18C and pKT25 (negative control), and pUT18C-leucine zipper with pKT25-leucine zipper (positive control). Results of a representative experiment are shown.

As the interface between coupling protein and Mpf is proposed to occur at the cytoplasmic membrane, an interaction was also investigated between the membrane-associated protein TraJ and the non-cognate coupling proteins TraG^RP4, TraD^F and TrwB^R388. These BTH studies should have indicated if the lack of specificity in the coupling proteinTrhB interaction extended to the coupling proteins and TraJ, although a limitation of the BTH analysis is that the adenylate cyclase fusions may create steric interference that prevents interactions. The only in vivo interaction that was detected using the BTH technology was between the R27 coupling protein and the R27 TraJ protein (Fig. 4B). These data suggest that the coupling proteinTraJ interaction is specific to conjugative systems that encode traJ.

Because the coupling proteins TraG^RP4, TraD^F and TrwB^R388 did not interact with TraJ in a BTH screen, and because TraJ is essential for conjugation (Lawley et al., 2002), it is unlikely that these non-cognate coupling proteins would be functionally interchangeable with the R27 coupling protein TraG^R27 in context with the remainder of the R27 conjugative apparatus. To investigate this, E. coli DY330R cells containing an R27 traG insertional mutant were transformed with the expression vector pMS119 expressing C-terminal His-tagged TraG^R27, TraG^RP4, TraD^F and TrwB^R388. Immunoblot analysis confirmed expression of each coupling protein (data not shown), and RP4, R388 and F plasmids encoding coupling protein mutations were successfully complemented by the constructs expressing their cognate coupling proteins (data not shown). The TraG coupling protein from R27 was the only protein that restored conjugative transfer of the R27 traG mutant.

Functional co-inheritance of TraJ and TraG homologues
Bioinformatic analyses of the IncHI2 plasmid R478 have revealed that traG and traJ homologues are encoded by a number of genomic sources, including plasmids, conjugative elements and chromosomes (Gilmour et al., 2004). Similar genomic analyses were repeated, and adjacent determinants encoding proteins similar to TraG and TraJ were observed in 38 gamma- and beta-proteobacteria strains, including many members of the order Burkholderiales (Table 3). The newly observed TraJ homologues were all of a similar size (201252 aa), and typically encoded four or five predicted TM domains. Phylogenetic analysis of TraJ homologues encoded within this co-inherited module revealed that the majority of the plasmid-encoded determinants clustered on a different node from the chromosomal determinants (Fig. 5). Not all of the presented sequences were fully annotated, so the genomic location (chromosome versus plasmid) was uncertain in some instances, although the majority of TraJ homologues encoded from a chromosomal source had an additional ∼30 aa at the N terminus of the protein, and the traJ and traG coding sequences overlapped (Table 3). The protein sequences similar to TraG were also examined, and a tree with very similar topology to that of TraJ was observed, except that IncP, IncF and IncW coupling proteins, and Ti plasmid-coupling plasmid orthologue VirD4 were included, and these were divergent from the IncH TraG-like sequences (Fig. 6). This comparison of coupling protein homologues indicates that the IncP, IncW and IncF, and Ti plasmid sequences are distinct from the IncH-type coupling proteins, and the requirement for TraJ as an interacting partner may partly explain the divergence of the IncH-type coupling proteins.

Table 3. Distribution of the traJtraG module in gamma- and beta-Proteobacteria The criteria for selection were gene products that were similar to R27-encoded TraG and TraJ (by protein sequence identity by BLAST, similar size in amino acids and similar number of predicted TM domains), and were also encoded by adjacent or nearly adjacent coding sequences.

Table 3. The distance score scale bar is shown at the bottom.

Table 3. The distance score scale bar is shown at the bottom.

To ascertain if an interaction between the coupling protein and TraJ was a common feature within this module of traGtraJ genes, we selected representative TraJ homologues from different nodes in the TraJ comparison. TraJ homologues encoded on R27, R478, R391 and Pseudomonas fluorescens chromosomal DNA were cloned into the BTH vector pKT25. The coupling protein homologues encoded on each respective genomic source were cloned into the BTH vector pUT18C. The in vivo BTH interaction data demonstrated that the cognate TraJTraG interaction from each distinct genomic source was conserved (Fig. 7). The diversity of this TraJTraG interaction was further characterized by using the BTH technology to determine if non-cognate TraJ homologues interacted with each coupling protein homologue. As expected, the highly related IncH plasmids R27 and R478 encoded TraJ homologues that were interchangeable in interactions with their coupling proteins (Fig. 6). The R391 plasmid-encoded TraJ and TraG homologues were not able to interact with the respective proteins from the other three genomic sources. Intriguingly, the coupling protein homologue from the P. fluorescens genome showed a strong interaction with TraJ from R27. In contrast, there was no evidence of an interaction with the opposite configuration, in which TraG from R27 was tested with the P. fluorescens TraJ homologue.

(18K): Fig. 7. BTH data for TraJ and TraG homologues. The traJ homologous genes from R27, R478, R391 and P. fluorescens (Pfl) were cloned into the BTH plasmid pKT25. The traG homologous genes from the same genomic sources were cloned into pUT18C, and co-transformed into BTH101 in a pairwise combination with each pKT25TraJ construct. The BTH controls were BTH101 cells containing pUT18C and pKT25 (negative control), and pUT18C-leucine zipper with pKT25-leucine zipper (positive control). Liquid cultures of BTH101 containing the BTH vectors were analysed for β-galactosidase activity using a Miller assay (Miller, 1972). Results of a representative experiment are shown.

Architectural evidence that TraJ is a functional part of the IncH-type coupling protein complex
During the search of GenBank for homologues of TraJ, we identified that TraJ has sequence identity to the FtsK/SpoIIIE family of proteins. The sequence and structural similarity of coupling proteins to FtsK/SpoIIIE-type proteins is already known (Errington et al., 2001; Wu et al., 1995). These DNA translocase ATPases are essential for coordinating chromosome segregation with cell division (FtsK; Weiss, 2004), or for mediating DNA partitioning during the process of sporulation (SpoIIIE; Wu & Errington, 1994). The similarity to conjugative coupling proteins is located primarily within the region surrounding the Walker A and B boxes, which are essential for nucleoside triphosphate binding and hydrolysis (Walker et al., 1982). At the N terminus of the DNA translocases, R27-encoded TraJ is 27 % identical over 110 aa, and has 20.1 % identity over an alignment length of 249 aa to Streptococcus pneumoniae SpoIIIE (GenBank accession no. NP_345365). This region of the DNA translocases contains between four and five TM domains that target SpoIIIE and FtsK to the site of cell division (Bath et al., 2000; Sharp & Pogliano, 2003; Wang & Lutkenhaus, 1998). This coincides with the well-conserved TM domains of TraJ, and an alignment of R27-encoded TraJ and TraG to FtsK of Shewanella oneidensis illustrates the domains with which the conjugative transfer proteins share homology (TraJ, 21.4 % identity over an alignment length of 243 aa; TraG, 21.1 % identity over an alignment length of 913 aa) (Fig. 8).

(8K): Fig. 8. Schematic representation of the homology shared by R27-encoded TraG and TraJ proteins with an FtsK DNA translocase, Sh. oneidensis MR-1 (NP_717901). Arrows indicate the region of FtsK with which the R27 transfer proteins share the greatest level of homology. The level of this homology is represented as a percentage identity with the alignment length represented in parentheses, as determined by EMBOSS pairwise local alignment. TM regions are indicated by hatched regions, as predicted by TMHMM. Walker A boxes determined by ScanProsite are on the left and Walker B boxes with the motif (R/K-X(7-8)-h(4)-D) (Walker et al., 1982) are on the right. A scale bar for amino acids is indicated.

The co-inheritance of the traGtraJ genes (and interacting gene products) within this module across multiple genetic elements suggests that both plasmid and chromosomal sequences share a common ancestor, but the cellular function of the chromosomal sequences has not been determined. Notably, many of these sequences are present in genomes that do not encode IncH-type Mpf components, indicating that traGtraJ forms a module that is inherited independently from the Mpf genes, and that the chromosomal determinants may be involved in functions other than conjugation. The majority of conjugative DNA transfer systems do not encode a TraJ homologue; however, in genetic elements producing both TraG and TraJ, the latter protein may play an essential role in the function of the coupling protein. We propose that the domains comprising TraG and TraJ cumulatively resemble the domain architecture of the FtsK/SpoIIIE DNA translocase proteins. This study confirms that TraJ and TraG interact in vivo, and it is possible that these proteins function together to transport plasmid DNA molecules through the cytoplasmic membrane. We thank Daniel Ladant, Institut Pasteur, Paris, for providing all of the BTH vectors and strains, and we thank Eric Lanka, Max-Planck-Institut für Molekulare Genetik, Berlin, Laura Frost, University of Alberta, and Fernando de la Cruz, Universidad de Cantabria, for generously providing pDB127, pOX38 : : TraD411 and pSU1456, respectively. We thank Tracy Raivio, University of Alberta, for the generous gift of the CpxA antibody. We are grateful to Bart Hazes, Mark Peppler, Joanne Simala-Grant and Marc Couturier for revision of the manuscript. This work was supported by grant MOP6200 to D. E. T. from the Canadian Institutes for Health Research (CIHR). J. E. G. and M. W. G. are recipients of a training award from the Alberta Heritage Foundation for Medical Research (AHFMR), and T. D. L. is a recipient of a training award from both AHFMR and CIHR. D. E. T. is a Senior Investigator with AHFMR.

Edited by: C. W. Chen