Functional Interactions of VirB11 Traffic ATPases with VirB4 and VirD4 Molecular Motors in Type IV Secretion Systems

Abstract

Pilus biogenesis and substrate transport by type IV secretion systems require energy, which is provided by three molecular motors localized at the base of the secretion channel. One of these motors, VirB11, belongs to the superfamily of traffic ATPases, which includes members of the type II secretion system and the type IV pilus and archaeal flagellar assembly apparatus. Here, we report the functional interactions between TrwD, the VirB11 homolog of the conjugative plasmid R388, and TrwK and TrwB, the motors involved in pilus biogenesis and DNA transport, respectively. Although these interactions remained standing upon replacement of the traffic ATPase by a homolog from a phylogenetically related conjugative system, namely, TraG of plasmid pKM101, this homolog could not replace the TrwD function for DNA transfer. This result suggests that VirB11 works as a switch between pilus biogenesis and DNA transport and reinforces a mechanistic model in which VirB11 proteins act as traffic ATPases by regulating both events in type IV secretion systems.

INTRODUCTION

Type IV secretion systems (T4SSs) export proteins and virulence effectors to other bacteria and eukaryotic cells (1–3). They are also responsible for genetic exchange between bacteria during conjugation (4, 5). T4SSs are large macromolecular assemblies formed by 12 different protein subunits, named VirB1 to VirB11 and VirD4, following the Agrobacterium tumefaciens T4SS nomenclature. Three of these proteins (VirD4, VirB4, and VirB11) are hexameric ATPases (6–8) that provide the energy for substrate transport and T4SS biogenesis.

VirB11 proteins are traffic ATPases that belong to a large AAA⁺ secretion protein superfamily, which also includes proteins of type II secretion systems and the type IV pilus biogenesis and archaeal flagellar assembly machineries (9). All of them are soluble, hexameric proteins located at the cytoplasmic side of the inner membrane. The monomer is characterized by the presence of an N-terminal domain (NTD) and a C-terminal domain (CTD), which are connected by a flexible linker that plays a key role in the catalytic cycle (10–12). Comparison of the crystal structures of VirB11 from Brucella suis (13) and its homolog in Helicobacter pylori, HP0525 (10, 14), revealed that this linker is responsible for a large domain swap of the NTD over the CTD without affecting the hexameric assembly (13).

VirB11 was reported to assist VirB4 during pilus biogenesis by dislocating pilin subunits from the inner membrane to the periplasmic space, thus promoting pilus polymerization (15). VirB4 proteins are the largest and most conserved components of T4SSs. TrwK, the VirB4 homolog of plasmid R388, consists of a hexameric double ring with a barrel-shaped structure (16). The atomic structure of the C-terminal domain of the VirB4 homolog in Thermoanaerobacter pseudethanolicus was recently obtained (17), revealing a striking structural similarity with TrwB, the VirD4 homolog of the R388 plasmid. TrwB was extensively characterized as a DNA-dependent ATPase (6, 18), playing an essential role in connecting the relaxosome complex with the secretion channel. The structural similarity between TrwB and well-known molecular motors, such as F₁ ATPase or ring helicases, suggests that TrwB uses the energy released from ATP hydrolysis to pump DNA through its central channel (19).

Interactions between VirB11, VirB4, and VirD4 were mainly inferred from two-hybrid and coimmunoprecipitation assays (20–22). Here, we report direct evidence for interactions of TrwD with TrwK and TrwB proteins, the VirB11, VirB4, and VirD4 homologs in the conjugative plasmid R388. Furthermore, replacement of TrwD by its homolog in plasmid pKM101 (TraG) or Bartonella tribocorum (TrwD) does not affect the interactions with TrwK, whereas replacement by its homolog in RP4 (TrbB) disrupts it. In contrast to TrwD and TraG, which have a large linker domain similar to that of B. suis VirB11, the linker domain of TrbB is short, like that of H. pylori HP0525. These results suggest that this linker domain is crucial in establishing VirB11-VirB4 interactions. Interestingly, TrwK-TraG binding is not sufficient to promote R388 conjugation, which indicates that more specific interactions are required for DNA transfer. Moreover, we have also found that TrwD is able to interact with the other T4SS motor, TrwB, which further supports the idea that VirB11 proteins act as traffic ATPases by regulating pilus morphogenesis and substrate transport steps in type IV secretion systems.

MATERIALS AND METHODS

Cloning of VirB11 homologs.

R388 genes trwD and trwK were amplified by PCR and cloned into pCDF-1b and pET28a expression vectors (Novagen, Madison, WI), by using NcoI-BamHI or NdeI-BamHI restriction sites, respectively. The trwD K203A mutant was generated by QuikChange site-directed mutagenesis (Stratagene) using plasmid pJR01 as the template (data not shown) and cloned into the pCDF-1b vector. The traG and trbB genes were amplified by PCR and cloned into the pET28a and pCDF-1b expression vectors using the pKM101 (23) and RP4 (24) plasmid DNAs as the templates, respectively. The Bartonella tribocorum trwD gene was amplified by PCR and cloned into pET3a and pET28a using pAB2 plasmid DNA as the template (25). All the primers used for the cloning procedures are available from the authors.

Coexpression and purification.

Coexpression of TrwK and VirB11 homologs was achieved by using the pET28a vector combined with the compatible vector pCDF-1b. Protein overexpression was induced into the Escherichia coli C41(DE3) strain (26) by addition of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside). After 6 h of induction, cells were harvested as reported previously (7) and stored at −20°C. Thawed cells were lysed as described previously (6). Lysates were collected by centrifugation, diluted four times in buffer A (50 mM Tris-HCl, pH 7.2, 100 mM NaCl, 70 mM imidazole, 10% glycerol, 0.5 mM phenylmethylsulfonyl fluoride [PMSF]), and applied to a HisTrap HP (1-ml) column (Amersham, GE). Proteins were eluted from the column in a linear gradient of imidazole by using buffer B (50 mM Tris-HCl, pH 7.2, 100 mM NaCl, 500 mM imidazole, 10% glycerol, 0.5 mM PMSF).

Immunodetection of TrwK and TrwD proteins.

Protein samples eluted from the HisTrap column were run in an SDS-polyacrylamide gel, transferred to a nitrocellulose membrane, and incubated with anti-TrwK (27) or anti-TrwD rabbit antiserum (8). Images were obtained after incubation with an IRDye anti-rabbit IgG (goat) antibody conjugate (Thermo Scientific), using an Odyssey scanner (LI-COR Biosciences).

In vivo complementation assays.

Conjugation donor strains were derivatives of Escherichia coli K-12 strain D1210 carrying either plasmid R388 (28) or plasmid pSU4039 (an R388 mutant with a transposon insertion within trwD [29]), plus derivatives of vector plasmids containing different virB11 homologs (Table 1). These strains were mated with recipient strain E. coli DH5α for 1 h at 37°C on the surface of agar plates, as described previously (6). Transconjugants were selected on LB agar plates containing nalidixic acid (20 μg/ml) plus chloramphenicol (25 μg/ml), except in the case of R388, which was selected in nalidixic acid plus trimethoprim (20 μg/ml). Results are represented as frequencies of transconjugants per donor and are the means of three independent experiments.

Table 1.

Conjugation frequencies of R388 and trwD mutant transcomplemented with different VirB11 homologs

Plasmid(s) in donora	VirB11 homolog protein in donor	Conjugation frequency
R388	TrwD	1.3 × 10−1
pSU4039	None	<10−7
pSU4039 + pJR01	TrwD (R388)	7.6 × 10−2
pSU4039 + pJR06	TrwD (R388)	1.2 × 10−2
pSU4039 + pJR27	TraG (pKM101)	<10−7
pSU4039 + pJR24	TrbB (RP4)	<10−7
pSU4039 + pJR33	TrwD (B. tribocorum)	1.1 × 10−2
R388 + pJR27	TrwD + TraG (pKM101)	1.1 × 10−1
R388 + pJR24	TrwD + TrbB (RP4)	2.3 × 10−1
R388 + pJR33	TrwD + TrwD (B. tribocorum)	0.9 × 10−1

^aDonor cells (E. coli K-12 strain D1210) carrying the plasmids shown in the first column were mated with strain DH5α, as described in Materials and Methods.

Molecular modeling.

Atomic models of TrwD (R388), TraG (pKM101), and TrbB (RP4) were generated by molecular threading, using the protein homology and recognition engine Phyre2 (30). The atomic coordinates of B. suis VirB11 (Protein Data Bank [PDB] accession number 2gza) (13) were used as the template for model building of TrwD from plasmid R388 (TrwD_R388) and TraG from plasmid pKM101 (TraG_pKM101). The HP0525 structure (PDB accession number 1nlz) (14) was used as the template for modeling of RP4 carrying TrbB (TrbB_RP4). Hexameric models were built on the basis of the hexameric structures of VirB11 and HP0525, using the UCSF Chimera package (31).

RESULTS

Interactions between TrwD and TrwK.

Interactions between VirB11 and VirB4 of the A. tumefaciens T4SS were inferred from two-hybrid genetic analysis and immunoprecipitation assays (21, 22). Here, we demonstrate protein-protein interactions between the VirB11 and VirB4 homologs in the R388 plasmid (TrwD and TrwK, respectively) by affinity chromatography. TrwD and TrwK, the latter of which was cloned with a histidine tag in the N-terminal domain (His-TrwK), were coexpressed in C41 cells. Interactions were followed by Western blot analysis of complexes eluted from Ni-nitrilotriacetic acid (NTA) affinity columns. Fractions eluted from the column revealed the presence of His-TrwK/TrwD complexes when analyzed with anti-TrwD antibodies (Fig. 1B). These fractions were also positive for anti-TrwK antiserum (data not shown). To corroborate this interaction, TrwD was cloned with a His tag at the N terminus and was coexpressed with TrwK. Interactions were followed in the same manner described above. Fractions eluted from the affinity column and analyzed by Western blotting with antibodies against TrwK revealed the presence of His-TrwD/TrwK complexes (see Fig. 3B), thus confirming the results shown in Fig. 1B. The formation of complexes between TrwK and TrwD was not dependent on ATP hydrolysis by TrwD, as a mutant variant on the Walker A motif (TrwD K203A) was also able to interact with TrwK (Fig. 1C). Correspondingly, mutations on the Walker B motif in TrwK did not affect the interaction with either TrwD or the TrwD K203A mutant (see Fig. 3C and D, respectively). As negative controls, proteins not related to T4SS were coexpressed with either His-TrwK or His-TrwD. No signal for these proteins was observed in the Western blots (data not shown).

Fig 1.

TrwD-TrwK interactions detected by affinity chromatography. Proteins were coexpressed in C41 cells, as described in Materials and Methods. Soluble lysates were loaded into a Ni-NTA column, and proteins were eluted from the column with an imidazole gradient. Samples were analyzed by Western blotting with anti-TrwD rabbit antiserum. (A) control experiment in the absence of His-TrwK protein; (B to F) elution profiles obtained after the coexpression of His-TrwK with different TrwD constructs. Lanes L and W, the sample loaded into the column and the sample collected after washing with binding buffer, respectively. Black triangle, increasing imidazole concentration gradient in which samples were eluted.

Fig 3.

Interactions of TrwK with different VirB11 homologs. Proteins were coexpressed and eluted from the Ni-NTA column as described in the legend to Fig. 1. Samples were analyzed by Western blotting with anti-TrwK rabbit antiserum. The elution profiles obtained after the coexpression of TrwK with different His-VirB11 homologs are shown. (A) Control experiment in the absence of His-TrwD protein. Lanes L and W, the sample loaded into the column and the sample collected after washing with binding buffer, respectively. Black triangle, increasing imidazole concentration gradient in which samples were eluted.

Recently, we have shown that the C-terminal end of TrwK plays a regulatory role on the ATPase activity of the protein, since removal of the last α helix of TrwK induced a large increase in ATP turnover relative to that for wild-type TrwK (27). This autoinhibitory region prevents futile ATP hydrolysis, suggesting that interaction with specific partners of the transport machinery could stimulate TrwK ATPase activity. This α helix, however, is not involved in TrwK-TrwD interaction, since the TrwK mutant (residues 1 to 801) binds TrwD with the same efficiency as the wild-type protein (Fig. 1D). On the other hand, when TrwK was coexpressed with truncated variants of TrwD, including variants with truncation of the CTD [TrwD(1-121)] or the NTD (TrwDΔN164), no interactions were observed (Fig. 1E and F, respectively), suggesting that the presence of the linker region was essential to observe these interactions.

The length of the TrwD linker domain is a key factor in the formation of TrwD-TrwK complexes.

VirB11 proteins are composed by an NTD and a CTD connected by a flexible linker of variable length (10, 13, 14). This linker is responsible for a large domain swap of the NTD over the CTD (13). Sequence comparison of VirB11 homologs revealed a high degree of variability in this linker region (13), being shorter in homologs such as HP0525 from H. pylori than in VirB11 from B. suis. TrwD from R388 is the homolog with the largest linker. Based on these differences, we wondered if this linker region could influence the interactions with TrwK.

For these experiments, we selected four TrwD homologs involved in bacterial conjugation that contain linkers of different lengths (Fig. 2A). The shorter linker corresponded to TrbB from plasmid RP4, which is similar to the linker of H. pylori HP0525. This linker region lacks a long α helix (αC2) that is present in TrwD and B. suis VirB11 but absent in HP0525 and TrbB. Another difference is the presence of a short α helix at the C terminus of TrwD/VirB11 (αJ) that is absent in HP0525/TrbB. The other homologs studied were TrwD from B. tribocorum and TraG from plasmid pKM101, which are phylogenetically closer to TrwD from R388 and which contain both the αC2 and the αJ regions. Computer modeling of the TrwD homologs (Fig. 2B) revealed that the shorter linker in TrbB_RP4 induced the same domain swap as that observed in HP0525 (13). In addition to the domain swap, it is worth mentioning that both the αC2 and the αJ regions present in TrwD and TraG_pKM101 form a lid at the base of the hexamer that is absent in TrbB (shown in green in Fig. 2B).

Fig 2.

Sequence and structural comparison of VirB11/TrwD homologs. (A) Sequences corresponding to representative members of the VirB11 family were aligned with the T-coffee program (37) and represented by the Jalview program (38). Percent sequence identity is depicted by gray intensity. Prediction of the secondary structure of TrwD is indicated by cylinders (α helices) and arrows (β sheets). Additional sequences corresponding to linker B and helices αC2 and αJ (13) are highlighted in green. (B) Structural three-dimensional models of TrwD, TraG, and TrbB were generated by molecular threading using the crystal structures of VirB11 (for TrwD and TraG) and HP0525 (for TrbB) as templates. Hexamers were built by matching the monomeric models on the VirB11 and HP0525 hexamers. Side and top views of the hexamers highlighting only one monomer in each hexamer are shown. The N-terminal domains and C-terminal domains of TrwD, TraG, and TrbB are depicted in pink and magenta, orange and red, and cyan and blue, respectively. Linker B and helices αC2 and αJ are depicted in green, as in panel A.

The ability of the four VirB11 homologs to bind TrwK of plasmid R388 was analyzed by affinity chromatography. In this case, experiments were carried out by cloning the VirB11 homologs with an N-terminal His tag sequence and using antibodies against TrwK to confirm the coelution of the two proteins. TrwD from R388 and its homolog from Bartonella tribocorum share 80% amino acid identity. As expected, TrwD of B. tribocorum binds TrwK as efficiently as TrwD_R388 (Fig. 3E and B, respectively). Conservation with the TraG_pKM101 plasmid is only moderate, as they share 37% identity. However, we found that TraG binds TrwK as efficiently as TrwD (Fig. 3F). In contrast, TrbB_RP4, which shares only 21% sequence identity with TrwD_R388, did not bind TrwK (Fig. 3G). As a control, we conducted the complementary experiment, coexpressing His-TrwK with Trb_RP4, and we did not observe any band corresponding to the RP4 homolog (data not shown).

The TrwD homolog of pKM101 plasmid binds TrwK, but it cannot replace TrwD in bacterial conjugation.

The ability of a VirB11 homolog to replace TrwD in vitro was tested in vivo by complementation assays in R388-mediated bacterial conjugation (Table 1). Without a functional trwD gene, there was no conjugation of plasmid pSU4039 (an R388 mutant with a transposon insertion within trwD). Conjugation was restored independently of the cloning vector used (plasmid pJR01 or pJR06; these plasmids contain the trwD gene cloned into pET28a and pCDF expression vectors, respectively) when pSU4039 was complemented with a plasmid containing a functional trwD gene. Conjugation of pSU4039 could also be restored to wild-type frequencies by transcomplementation with plasmid pJR33 carrying trwD of B. tribocorum, which is in agreement with previous reports (25, 32). However, when plasmid pJR24 coding for TrbB of plasmid RP4 was used as a complementing plasmid, the biological function was not restored. Intriguingly, TraG of plasmid pKM101, which binds to TrwK as efficiently as TrwD of R388 (Fig. 3F), was unable to transcomplement the TrwD mutant. Therefore, although TrwD-TrwK binding is an essential step in conjugation (21), the fact that TraG of pKM101 binds efficiently to TrwK but cannot replace the TrwD function for R388 conjugation implies that TrwD has at least another partner with an essential role in conjugation that requires a more specific interaction.

Conjugation assays were repeated in the presence of the wild-type R388 plasmid. As observed in Table 1, none of the TrwD homologs, not even TraG of plasmid pKM101, exerted a negative dominant effect. This result supports the idea that there are specific regions present in TrwD and absent in TraG which are essential for the interaction with other partners of T4SSs involved in substrate transfer.

Interactions of the coupling protein TrwB with TrwD and TrwK.

Evidence for a VirD4-VirB11 interaction had previously been obtained by immunoprecipitation assays (21). This VirD4-VirB11 interaction, which is independent of other T4SS subunits, is the first step in substrate transfer. Here, we decided to prove the interaction between TrwD and TrwB, the VirB11 and VirD4 homologs of plasmid R388, respectively, by affinity chromatography. His-TrwB (TrwB with a His tag in the N-terminal domain) and TrwD were coexpressed in C41 cells and treated as described above. As in the case of His-TrwK, TrwD was retained in the column in the presence of His-TrwB (Fig. 4A), although the TrwD interaction with TrwB was weaker than that observed in the case of TrwK. The interactions between TrwB and TrwD were not dependent on ATP hydrolysis by TrwD, as the Walker A TrwD K203A mutant bound TrwB with the same efficiency as the wild type (Fig. 4B). As a control, we coexpressed His-TrwB with Trb_RP4, the TrwD homolog without the linker region, and we did not observe any interaction (data not shown).

Fig 4.

Interactions between the three T4SS ATPases detected by affinity chromatography. Proteins were coexpressed and eluted from the Ni-NTA column as described in the legend to Fig. 1. Samples were analyzed by Western blotting with anti-TrwD (A, B, and D), anti-TrwK (C and D), and anti-TrwB (D) rabbit antiserum. Elution profiles obtained after the coexpression of His-TrwB with TrwD or TrwK are shown in panels A and C, respectively. Analysis of the interaction between His-TrwB, TrwK, and TrwD is shown in panel D. Lanes L and W, the sample loaded into the column and the sample collected after washing with binding buffer, respectively. Black triangle, increasing imidazole concentration gradient in which samples were eluted.

Next, we tested the interactions between His-TrwB and TrwK. As shown in Fig. 4C, the intensity of the TrwK band was much higher than that of the TrwD band, suggesting that TrwB interacts with TrwK with a higher affinity than TrwD. This result is in agreement with that from our previous work, in which we showed that TrwK inhibits the DNA-dependent ATPase activity by TrwB (16). It is also in agreement with the result of recent work reporting the interactions between PrgJ and PcfC, the VirB4 and VirD4 homologs in Enterococcus faecalis plasmid pCF10, respectively (33).

Finally, we explored the possibility that the presence of TrwK might affect the interaction between TrwD and TrwB. The three ATPases were coexpressed in C41 cells at the same time. As shown in Fig. 4D, a His-tagged TrwB was able to retain the other two ATPases in the affinity column, suggesting that they could form ternary complexes. Interestingly, the intensity of the TrwD band was again lower than that of the TrwK band. Moreover, the observed intensities were similar to those obtained when each of the ATPases was assayed against His-TrwB separately.

DISCUSSION

VirB11 proteins play an essential role in pilus biogenesis and substrate secretion (15, 34). This dual function suggests that VirB11 acts as a molecular switch between these two distinct events in T4SSs. Bacterial two-hybrid analysis and coimmunoprecipitation assays revealed that VirB11 is able to interact with the other two motors in T4SSs, the VirB4 and VirD4 proteins (20–22). Here, by affinity chromatography, we show direct evidence for protein interactions between the molecular motors of the R388 T4SS. Furthermore, by comparing the interactions between TrwK and different VirB11 homologs, we have characterized the biochemical and structural basis of the interactions between the molecular motors of the T4SSs.

VirB11 proteins belong to the AAA⁺ traffic ATPase superfamily, which also includes ATPases from type II secretion systems and type IV pilus and archaeal flagellar biogenesis systems (9). These proteins form hexameric ring structures in which the N-terminal domain (NTD) likely interacts with the bacterial membrane, whereas the C-terminal domain (CTD), containing a RecA-like domain, is involved in ATP hydrolysis (35). The two domains are connected by a flexible linker of variable length. This linker has been proposed to be a key factor during catalysis, as it is responsible not only for the flexibility of the enzyme but also for a large domain swap of the NTD over the CTD (13). This domain swap is instigated by the presence of an inserted sequence that forms the linker B and αC2 helix, which are present in VirB11 of B. suis and also in related homologs, such as TraG of plasmid pKM101 or TrwD of plasmid R388. TrbB of plasmid RP4, however, lacks this sequence, and it presents an HP0525-like topology, as inferred by computer modeling. Interestingly, the linker B and αC2 helix, although located in the middle of the sequence, form a lid at the base of the hexameric ring, making contacts with the last α helix at the C-terminal domain (αJ), as observed in the crystal structure of B. suis VirB11 (13) (Fig. 2B).

On the basis of these structural studies on VirB11 proteins and the genetic evidence for interactions between VirB11 with the other two T4SS ATPases, VirB4 and VirD4, we decided to test the ability of different VirB11 homologs to interact with these molecular motors. For these experiments, we used the T4SS motors TrwD, TrwK, and TrwB of conjugative plasmid R388 as a model system. TrwD homologs were selected according to the sequence identity and the secondary and tertiary structure predictions, in order to include the most representative members of the VirB11 family of proteins involved in bacterial conjugation. First, we explored the putative interactions between TrwD and TrwK, as there is evidence for a partnership between these proteins during pilus morphogenesis (15). In this work, we demonstrate that the formation of TrwK-TrwD complexes is not dependent on ATP hydrolysis and that the regulatory C-terminal domain of TrwK (27) is not involved in these protein-protein interactions. However, truncated variants of TrwD, including either the NTD or the CTD, were unable to interact with TrwK, suggesting that the linker region between both domains plays an essential role in these interactions.

As mentioned above, the linker region in B. suis VirB11-like proteins is responsible for the domain swap of the NTD over the CTD (13). Accordingly, we decided to investigate putative protein interactions between TrwK and TrwD homologs with different linker regions. The results revealed that only TrbB of RP4, which is the TrwD conjugative homolog with the shortest linker domain, was unable to interact with TrwK, confirming the role of the linker region in these interactions. Therefore, it is tempting to think that the lid domain formed by linker B and the αC2 and αJ helices at the bottom of the TrwD structure is directly involved in TrwK binding. However, alternative scenarios without a direct involvement of this region in the interactions could also be envisaged. The domain swap induced by this linker region might be responsible for the observed differences, as it affects the subunit interface within the hexameric ring (13). Thus, adjacent subunits within the B. suis hexamer are held together by linker-CTD and CTD-CTD interactions, whereas in the HP0525 hexamer, neighboring subunits are held together mainly by NTD-CTD interactions. These differences in the subunit interface dramatically affect the nucleotide binding site. In HP0525, the nucleotide binding site is coordinated between the NTD and the CTD of the same monomer, while in B. suis, the binding site is located between the NTD and the CTD of neighboring subunits. The main consequence of this difference in the nucleotide binding site location is that the ATPase activity by B. suis VirB11 would be dependent on the oligomerization of the protein, while HP0525 would be able to hydrolyze ATP independently of the formation of the hexamer (10). It might be possible that the interaction with TrwK requires the preassembly of the TrwD hexameric ring, and therefore, the results observed with the different VirB11 homologs could be explained by differences in the dynamics of hexamer formation. Interestingly, ATPase activity was not needed for the interactions, as mutations on the Walker motifs of TrwD or TrwK had no effect on the respective interactions. This is corroborated by the fact that the presence of TrwD did not have any effect on the ATPase activity of TrwK, and vice versa (data not shown).

An interesting result is that TraG of pKM101 efficiently binds to TrwK of R388; however, it cannot replace TrwD for R388 transfer. This result indicates a failure of TraG to interact with other partners of the T4SS in plasmid R388 involved in DNA transfer, supporting the dual role already proposed for VirB11 proteins. It is tempting to think that DNA transfer requires more specific interactions that do not allow the replacement of VirB11, not even with phylogenetically related homologs. Involvement of VirB11 in the first events of DNA transfer by VirD4 was initially reported by Cascales and Christie (34). Recent findings support a model whereby early stages in DNA transfer would be triggered by DNA ligand binding to VirD4 and VirB11. This event would stimulate ATP hydrolysis, which, in turn, would induce conformational changes in the T4SS channel (36). In this work, we present direct evidence for protein-protein interactions between TrwD and the coupling protein TrwB. The observed interactions are weaker than those observed between TrwD and TrwK, suggesting that the formation of the complex between TrwD and TrwB is transitory, while that formed between TrwK and TrwD is more stable. Moreover, the fact that the three ATPases could interact, forming a ternary, albeit transitory, complex, reinforces the hypothesis that TrwD is a traffic ATPase that could assist the exchange of TrwK and TrwB at the bottom of the secretion channel, as previously suggested (16).

In summary, the results shown here reinforce the idea of VirB11 proteins as traffic ATPases that modulate pilus biogenesis and substrate transport by interacting with the other molecular motors in T4SSs, VirB4 and VirD4. In pilus morphogenesis, VirB4 would work as a dislocation motor, whereas VirB11 would act as a modulator of the VirB4 pilin dislocase activity. In nucleoprotein/protein transfer, VirB11 would interact with the coupling protein VirD4 by assisting TrwB to proceed with nucleoprotein substrate transport. Interestingly, TrwB shares a high degree of structural similarity with the C-terminal domain of VirB4 proteins. Moreover, there are reports of interactions between VirD4 and VirB4 homologs in R388 and in Enterococcus faecalis (16, 33), suggesting the possibility that both proteins could exchange their positions at the base of the T4SS channel. This role of VirB11 proteins as a molecular switch is depicted in Fig. 5. The model is supported by the results presented in this work and also by previous data, published by Sagulenko et al. (39), demonstrating that some virB11 mutations uncouple substrate translocation from T-pilus production.

Fig 5.

Model of VirB11 proteins acting as a switch between pilus biogenesis and DNA transport. (A) During pilus biogenesis, VirB4/TrwK (yellow) is associated with the inner membrane (IM), at the base of the core complex (gray). VirB11/TrwD (green) interacts with VirB4/TrwK, modulating its pilin dislocase activity. Successive polymerization reactions of pilin monomers would promote pilus growth. (B) In substrate transfer, VirB11/TrwD assists VirD4/TrwB with nucleoprotein complex transfer. VirD4/TrwB (salmon) displaces VirB4 from the base of the channel, as suggested by various reports of interactions between VirD4 and VirB4 homologs in different conjugative systems (16, 33). OM, outer membrane.