Analyzing the role of CagV, a VirB8 homolog of the type IV secretion system of Helicobacter pylori

The type IV secretion system of Helicobacter pylori (Cag‐T4SS) is composed of ~ 27 components including a VirB8 homolog, CagV. We have characterized CagV and reported that it is an inner membrane protein and, like VirB8, forms a homodimer. Its stability is not dependent on the other Cag components and the absence of cagV affects the stability of only CagI, a protein involved in pilus formation. CagV is not required for the stability and localization of outer membrane subcomplex proteins, but interacts with them through CagX. It also interacts with the inner membrane‐associated components, CagF and CagZ, and is required for the surface localization of CagA. The results of this study might help in deciphering the mechanistic contributions of CagV in the Cag‐T4SS biogenesis and function.

The type IV secretion system of Helicobacter pylori (Cag-T4SS) is composed of~27 components including a VirB8 homolog, CagV. We have characterized CagV and reported that it is an inner membrane protein and, like VirB8, forms a homodimer. Its stability is not dependent on the other Cag components and the absence of cagV affects the stability of only CagI, a protein involved in pilus formation. CagV is not required for the stability and localization of outer membrane subcomplex proteins, but interacts with them through CagX. It also interacts with the inner membrane-associated components, CagF and CagZ, and is required for the surface localization of CagA. The results of this study might help in deciphering the mechanistic contributions of CagV in the Cag-T4SS biogenesis and function.
Type I strains of Helicobacter pylori (H. pylori) harboring type IV secretion system (Cag-T4SS) are responsible for the more severe forms of gastric diseases [1][2][3][4]. Cag-T4SS is composed of~27 components and encoded by a~37-kb genomic region called cag-Pathogenicity Island (cag-PAI) [5]. H. pylori uses this specialized secretion system to translocate the substrate CagA into the host epithelial cells to modulate the host signaling pathways and to induce proinflammatory response [6-9].
The most studied prototypical secretion system, the VirB/D4 T4SS of Agrobacterium tumefaciens (A. tumefaciens) is composed of 11 VirB proteins (VirB1-VirB11) and a coupling protein VirD4 [10,11]. Several Cag proteins have sequence similarity to the VirB/D4 components and are considered as homologs, like CagY as VirB10; CagX as VirB9; CagT as VirB7; CagV as VirB8 etc. [12,13]. Additionally, Cag-T4SS encodes a number of unique components like Cagd, CagZ, CagU, and CagM etc. [12][13][14][15]. VirB proteins that are involved in the assembly of the secretion system could be subdivided into three groups [16]. Cag-T4SS is also presumed to have similar type of organization of the Cag components as shown in a number of hypothetical models [12,13,16,17]. However, a recent report has shown that the architecture of the core Cag-T4SS complex is significantly different from that of other published T4SS core complexes [18]. Very little is, however, known about the architecture of the system and the molecular mechanism of substrate translocation across the bacterial membranes.
In the present study, we have characterized the VirB8 homolog CagV which shares weak sequence homology with its counterpart in other known T4SS. However, on the basis of topological analysis in E. coli, it is predicted to be a VirB8 homolog, an essential component of the VirB/D4 secretion system in A. tumefaciens [19,20]. In H. pylori, mutational analysis revealed that CagV is essential for CagA translocation and IL-8 induction in the host cells [5,21].
In A. tumefaciens, VirB8 is a bitopic membrane protein and is required for the stability of several other VirB proteins underlining its importance in the biogenesis of the secretion complex [22,23]. Hence, VirB8 is suggested to be an assembly factor that is connected to the outer membrane core complex via VirB9 and VirB10 [24,25]. It is also connected to the cytoplasmic core complex via VirB4 and to the periplasmic protein VirB1 for breaking down peptidoglycan layer to facilitate the T4SS assembly across the bacterial cell envelope [19,[26][27][28]. In order to understand the molecular basis of native CagV function in the Cag-T4SS assembly and substrate transportation to the bacterial surface, we have investigated its subcellular localization, its interactions with cag-PAI components and compared the findings with its counterpart VirB8 of A. tumefaciens.

Bacterial strains and growth conditions
All the bacterial strains used in this study are listed in Table 1. Wild-type H. pylori and mutant strains were grown and maintained as described earlier [14]. E. coli cultures were also grown as described earlier [14].
Cloning, construction of mutant strains, and expression of genes All primers used in this study are mentioned in Table 2. pET-28a was used to express CagV (His tag) and pMal-c2X was used to express CagV and CagF with maltosebinding protein (MBP) tag. Proteins were expressed in E. coli strain BL-21 (DE3) at 25°C in LB medium with 0.1 mM IPTG (isopropyl-b-D-1-thiogalactopyranoside) to induce the expression of proteins in the culture medium. All cag null mutant strains were constructed in H. pylori 26695, as described previously except cagV [5,14,29]. To generate the nonpolar cagV null mutant strain, mutator DNA was synthesized by overlapping PCR using a cagV upstream fragment containing the first three codons of cagV and the upstream region of cat cassette (using primers DV1 and DV3), the downstream region of cagV containing the termination codon, and the downstream region of cat cassette (using primers DV2 and DV4) and amplified cat cassette (using primers Fcat and Rcat). Finally a linear mutator PCR product was synthesized using primers DV1 and DV2 and phusion polymerase. Next, the resulting linear PCR product was introduced into H. pylori 26695 by natural transformation following a published protocol and selected on chloramphenicol BHI (Brain Heart Infusion) agar plate [14]. Complementation of the wildtype cagV was performed as described earlier [14]. Briefly, the cagV (under cagA promoter)-containing plasmid was transformed into DcagV and expressed in the presence of appropriate antibiotics and tested for CagV expression. Antibody generation, SDS/PAGE, and western blotting Polyclonal antibodies against Cag proteins such as Cagd, CagH, CagL, CagE, CagY, CagZ, CagX, CagT, CagM, CagI, and CagF were raised by immunization of rabbit and mice in the laboratory as described previously [14,29]. Briefly, to generate polyclonal antibody against CagV, a full-length cagV was cloned in pET-28a vector and overexpressed in E. coli strain BL-21 (DE3). CagV protein band was excised from SDS/PAGE, used for immunization, specificity and titer of rabbit anti-CagV (a-CagVr) and mice anti-CagV antibody (a-CagVm) were tested by western blotting. Anti-CagA, anti-OMP (Outer Membrane Protein), and anti-HSP (Heat Shock Protein) antibodies were from Santa Cruz, CA, USA (Cat. No. Sc-25766, Sc-69935 and Sc-57779, respectively). Anti-MBP antibody was from NEB, USA (Cat. No. E8032L). SDS/PAGE and western blotting were performed as described previously [14,29].

Ethics statement
The present study was approved by the Institutional Animal Ethics Committee-of Jawaharlal Nehru University (Code no: 23/2007 and 22/2012). The animals (Balb/c mice female or New Zealand white rabbit female) were maintained at Central Animal Facility of the Jawaharlal Nehru University as approved by the Institutional Animal Ethics Committee. After experimental procedures were finished, the animals were maintained until their natural death and every effort was made to minimize their suffering.

Cell fractionation
Cellular fractionation of the wild-type H. pylori cells was performed as described earlier [14]. Briefly, bacterial cell extract was centrifuged at 148 000 g in SW-55 rotor (Beckman coulter ultracentrifuge, Irving, TX, USA) at 4°C for 1 h, the supernatant was recovered as cytoplasmic/periplasmic fraction (C/P) and the resulting pellet was recovered as total membrane (TM). Volume of C/P was measured and TM was resuspended in the same volume of PBS. Equal volume of each sample was mixed with 2X-SDS sample buffer, separated in SDS/PAGE, and subjected to western blotting using an appropriate antibody.

Purification of MBP tagged proteins and pulldown assay
For purification of recombinant proteins (MBP-tagged CagV and CagF), E. coli cell pellets containing expressed Table 2. Oligonucleotides used in the present study.

Homology modeling and molecular dynamics simulation of CagV monomer and dimer
The sequence of CagV corresponding to the crystallized periplasmic domains of A. tumefaciens-VirB8 (92-237 aa) and B. suis-VirB8 (97-234 aa) was identified. A multiple template-based model of the identified region was generated using Modeller (v9.14) [33,34]. The highest resolution crystal structures of A. tumefaciens-VirB8 (PDB ID: 2CC3) and B. suis-VirB8 (PDB ID: 4AKZ) were taken as templates [35,36]. Structure with lowest discrete optimized potential energy was used to obtain a more stable state for molecular dynamics (MD) simulation [37]. MD was performed using GROMACS (v5.1) and GROMOS96 54a7 force field [37][38][39][40]. Cubical box filled with the simple point charged water models was used for solvation, under periodic boundary conditions with padding distance of 1.5 nm, keeping protein at center. Few water molecules were replaced by required number of counter ions (Na + /Cl À ) needed to neutralize the system at pH 7.0. A maximum of 10 000 steps of steepest descent was performed to minimize the energy of the system. Solvent and ionic molecules were equilibrated around the protein under NVT and NPT conditions simultaneously (for 200 ps each), to maintain 300 K temperature and 1 bar pressure. For this step a positional restraint was applied on the protein. Finally, a 70-ns long unrestrained production MD was run with time step of 2 fs at 300 K and 1 bar pressure. Energy and the co-ordinates were saved every 10 ps (at every 5000 step) for further analysis. The final step co-ordinates of CagV obtained after simulation was superimposed on both the chains of A. tumefaciens-VirB8 (PDB ID: 2CC3, Chain A and B), one by one and transformed co-ordinates were saved and combined to generate the dimer of CagV. Chain Identifier A and B were assigned to the chains. Initial interchain clashes were removed manually by slightly shifting the chains apart and introducing a slight rotational shift in UCSF chimera [41]. Co-ordinates of the dimer were then subjected to a 50-ns MD simulation following similar procedure applied to the monomer. The conserved interchain contacts at the dimer interface throughout the simulation were calculated using MDCONS package (v2.0). It considers a contact by a cut-off distance of 5 A between heavy atoms of two residues of different chains at the interface [42].
The buried solvent accessible surface area throughout simulation at the dimer interface was calculated by subtracting the total solvent accessible surface area (SASA) of the CagV dimer from the sum of total SASA of each individual monomeric units of the dimer. The buried hydrophilic and hydrophobic surface areas were also calculated using the same approach. GROMACS tool gmx sasa was used to calculate SASA and the residues having charge between À0.2 and 0.2 were treated as hydrophobic, while others as hydrophilic. The plot of interaction across the interface of final CagV dimer co-ordinate for A. tumefaciens-VirB8 and B. suis-VirB8 was generated using DIMPLOT (v.1.4.5) [43]. The binding free energies of complexes were calculated using DCOMPLEX package (http://sparks-lab.org/czhang/com plex.html) [44]. Other trajectory analysis and visualization was carried out using inbuilt gromacs tools, UCSF CHIMERA (v-1.10.2, (http://www.cgl.ucsf.edu/chimera, and VMD v-1.9.2, http://www.ks.uiuc.edu/Research/vmd/) [41,45].

Cellular localization of CagV
Following in silico analysis of the transmembrane domain of CagV and expression of the corresponding gene in E. coli, Buhrdorf et al. predicted that cagV might be a functional virB8 analog [20]. Later, it was accepted as a VirB8 homolog. However, not much is known about the protein. In order to understand the location of native CagV, we performed cell fractionation of the wild-type H. pylori 26695 into soluble cytoplasmic/periplasmic (C/P) and total membrane fractions (TM). The fractions were separated in SDS/PAGE, western blotted using appropriate antibodies and quantitated (Figs 1A,B and S1). As shown in Fig. 1A, like inner membraneassociated CagZ, CagV was also found exclusively in the membrane fraction [46][47][48]. Subsequently, IFM analysis of wild-type H. pylori under permeabilized (P) and nonpermeabilized (NP) conditions supported its localization inside the cell (Fig. 1C). These results, thus, suggest that CagV is not a bacterial surface exposed protein rather; it resides inside the bacterium and is associated with membrane. To visualize exact localization of the protein we performed TEM and detected CagV-specific signals at the inner membrane suggesting its inner membrane association (Fig. 1D,E).

Stability of CagI is partially affected in the absence of CagV
To understand the role of CagV in the stability of Cag components, we tested the levels of individual Cag proteins in the absence of cagV in DcagV by western blot analysis. As shown in Fig. 2A,B, all the tested structural components of Cag-T4SS such as CagT, Cagd, CagY, CagX, CagM, and CagZ along with CagE, chaperone CagF, and substrate CagA were found stable. However, the stability (level) of CagI, the protein involved in the Cag-T4SS pilus formation was found to be partially affected ( Fig. 2A,B). Stability of no other known protein involved in pilus formation, like CagH and CagL (VirB5 homolog) were affected. Complementation of the cagV function in DcagV/cagV supported the result.
In the case of the prototypical T4SS of A. tumefaciens, stabilities of several VirB proteins are reported to be affected in the absence of VirB8. For example, the stability of VirB3 and VirB6 are strongly reduced in its absence, whereas modest effect is observed in the cases of VirB1, VirB4, VirB5, VirB7, and VirB11 [22]. Among these, VirB3 and VirB5 are involved in the T pilus biogenesis [49]. However, when the stability (level) of CagV was tested in the absence of other Cag components, no apparent change in CagV stability was observed (Fig. 2C,D). It is worth mentioning that in A. tumefaciens, however, VirB8 stability is dependent on VirB4, an energy providing protein [50].

CagV-independent stability and surface localization of CagT and CagX
In the prototypical VirB/D4 T4SS and others, VirB8 protein is reported to be an essential assembly factor of the secretion systems [19,[50][51][52]. We, thus, asked the question whether CagV like VirB8 affects the stability and localization of the key outer membrane subcomplex proteins of the Cag-T4SS, such as CagT and CagX. As mentioned above, CagV does not affect the stability of the Cag proteins except CagI. We, therefore, tested the localization of these proteins in the wild-type H. pylori and isogenic DcagV by IFM and TEM. As shown in Figs 3 and 4, like in the wild-type H. pylori cells, CagT-and CagX-specific signals were also observed in the isogenic DcagV. Taken together, these results thus suggest that not only the stability but also the native location of the proteins are independent of CagV. This is in contrary to the VirB/D4 T4SS, where stability, polar localization, and assembly of the components are dependent on VirB8 [22].

CagV-dependent surface localization of CagA
Recent studies have shown that the Cag-T4SS substrate CagA is a cell surface-exposed protein [14, 53,54]. To investigate the role of CagV in the translocation of CagA to the bacterial surface we determined the localization of CagA in the absence of CagV in ΔcagV. We performed IFM on the wild-type H. pylori and ΔcagV cells under permeabilized (P) and nonpermeabilized (NP) conditions. As shown in Fig. S2, unlike in wild-type H. pylori, the CagA-specific signal was not detected in the DcagV under NP condition. However, when IFM was performed under permeabilized condition, the CagA-specific signal was detected in the DcagV cells. These results thus suggest that CagV may be required for the bacterial surface localization of CagA. To further validate the above data, we complemented the cagV function in the DcagV, performed IFM, and observed that CagA again reappeared on the bacterial cell surface like in wild-type H. pylori. We further performed TEM on the wild-type H. pylori, isogenic DcagV and DcagV/ cagV strains. In agreement with the IFM results, CagA-specific signals were detected in the cytoplasmic face of the inner membrane along the perimeter in the DcagV (Fig. 5A,B). However, under a similar condition CagA-specific signals were detected associated with the outer membrane and on the bacterial cell surface in the wild-type and cagV complemented DcagV/ cagV strains (Fig. 5C,E). CagA-specific signals were, however, absent in DcagA strain (used as a negative control; Fig. 5D). Taken together, these results suggest that CagV is essential for CagA translocation across the bacterial membranes.

Comparative sequence analysis and homology modeling of CagV monomer
Comparative sequence analysis of VirB8 homologs has shown that despite having low sequence identity, several of the amino acid residues are conserved across the family members and participate in the interaction either with itself or with other VirB proteins or DNA substrate [35,55]. We, therefore, tested the formation of CagV dimer in SDS/PAGE followed by western blotting. However, we failed to observe any dimer. This result thus prompted us to perform in silico homology modeling of CagV using the crystal structure data of A. tumefaciens-VirB8 and B. suis-VirB8. Studies on the B. suis-VirB8 periplasmic domain identified amino acids required for protein-protein interactions and homodimerization [56]. Similarly, in A. tumefaciens-VirB8 periplasmic domain, V97 and A100 were reported essential for homodimerization [22]. Comparative sequence and structural analysis showed that CagV shares 17% and 19% identity with A. tumefaciens-VirB8 and B. suis-VirB8, respectively (Fig. 6). We observed a strictly conserved region 210-LxxNPxGxxV-219 which is equivalent to 210-RxxNPxGxxV-219 of A. tumefaciens-VirB8 and 217-RxxNPxGxxV-227 of B. suis-VirB8. Other conserved residues are shown on the Fig. 6.
We generated a homology model of CagV periplasmic domain (87-236 aa) by using A. tumefaciens-VirB8 and B. suis-VirB8 as templates (Fig. 7A). It shared 18% average identity with periplasmic domains of the templates. The modeled structure was further subjected to a 70-ns-long molecular dynamics simulation to obtain its stable conformation. Convergence was observed in root mean square deviation (RMSD) value, radius of gyration, and number of intrachain hydrogen bonds, 15 ns onwards (Fig. 7B). Looking at the residue wise RMSF curve, it is clear that fluctuations in amino acid positions during the last 20 ns of the run has decreased as compared to the first 20 ns and 20-50 ns (Fig. 7B). These observations point that the structure of CagV has attained a conformation with high compactness and less fluctuations which is comparatively more stable than the modeled structure. The co-ordinates of the final stable structure were saved for further analysis. It has an RMSD of 1.147 A and 1.118 A with A. tumefaciens-VirB8 and B. suis-VirB8 and consists of four a-helices (a1, a2, a3, and a4) juxtaposed by antiparallel b-sheets (b1, b2 and b3) (Fig. 7A,C). Unlike A. tumefaciens-VirB8 and B. suis-VirB8, b-sheets are relatively smaller in size. One b strand and one a-helix corresponding to b1 and a4 of A. tumefaciens-VirB8 have found to be transformed into coils (Fig. 7C). Eight of the conserved amino acids are exposed on the surface and are accessible to solvent (Fig. 7D,E).

MD simulation of CagV homodimer, comparison of dimer interfaces, and in vitro formation of CagV dimer
The CagV dimer was modeled by the structure superimposition of its monomer on two chains of A. tumefaciens-VirB8 (Fig. 7F). In a 50-ns simulation, we observed a little deviation in RMSD value of individual chains with reference to their starting structure (Fig. 8A). Radius of gyration of individual chain shows some fluctuation in the first 15 ns but converges afterward and starts fluctuating around 1.55-1.6 nm which is very close to the observed Rg value of 1.6 nm, after convergence in the case of monomer simulation (Figs 8A and 7B). Amino acid RMSF value of each individual chain of dimer throughout the simulation is almost similar to the observed RMSF in the last 20-ns simulation of the monomers (Figs 8A and 7B). After convergence of both, monomer and individual chains of dimer form comparable numbers of intrachain hydrogen bonds (Figs 8A and 7B). These comparative studies suggest that dimerization of CagV seems to introduce very little or no change in the compactness, stability, intrachain hydrogen bond-forming pattern of each individual units. The plot of distance between the center of mass (COM) of individual chain shows a sharp decrease of around 1 nm between 10 and 20 ns. It is followed by another small decrease around 28-30 ns. For the rest of the period, distance is converged and fluctuate around 2.25 nm (Fig. 8B). This shows that from the reference structure, the individual chain has moved close to each other, getting stabilized at around 2.25 nm after 30 ns. Based on these observations, we confirm that an optimized stable structure of CagV dimer is obtained.
A comparative study of the dimer interfaces of A. tumefaciens-VirB8 and B. suis-VirB8 showed that the dimer interface buries over 1700 A 2 accessible surface area at their interface [35,55]. We calculated the buried accessible surface area at the interface. The total buried, hydrophilic, and hydrophobic surface area was found to be 2033.313 A 2 , 1140.546 A 2 , and 892.471 A 2 (Fig. 8C). The total buried surface of the CagV dimer is slightly more than the reported areas for the A. tumefaciens-VirB8 and B. suis-VirB8 proteins. The binding energy between respective chains of the A. tumefaciens-VirB8, B. suis-VirB8, and CagV was found to be À11.212 kcalÁmol À1 , À15.910 kcalÁmol À1 , and À11.180 kcalÁmol À1 , respectively. The contacts between the residues at the dimer interface from 31 ns onward were calculated for each frame and plotted on the basis of frequency to identify conserved contacts (Fig. 8D). At a consensus cutoff of 90%, we found 22 conserved contact pairs between both the chains (Table 3). Contact forming residues Y108 and V144 are strictly and A92 and L104 are moderately conserved (Fig. 6 and Table 4). I211 and I212 are part of the conserved motif 210-LxxNPxGxxV-219 which is proposed to position the a-helix to support dimerization [35]. At the dimer interface of A. tumefaciens-VirB8; B. suis-VirB8; and CagV 1, 6, and 6 hydrogen bonds; 25, 19, and 28 hydrophobic contacts; and 78, 80, and 54 non-bonded interactions were found, respectively (Fig. 9A-C). As essential amino acids of A. tumefaciens-VirB8 V97, P214, and B. suis-VirB8 M102 are involved in forming hydrophobic contacts. B. suis-VirB8 Y105 is involved in hydrogen bonding and hydrophobic contacts with itself (Fig. 9B,C), similarly, aligned residues of CagV to these reported essential residues form similar type of interactions (Figs 6 and 9A). Moderately conserved CagV V93, aligned to A. tumefaciens-VirB8 V97 and B. suis-VirB8 M108 forms hydrophobic contacts. CagV S96, aligned to B. suis-VirB8 Y105 forms hydrogen bond and hydrophobic contacts. CagV P214 forms hydrophobic contacts at the dimer interface (Figs 6 and 9A). Thus, the comparative study suggests that the dimer interface of CagV is almost similar to that of the other two homologs and might form homodimer under a physiological condition. With these notions we performed chemical cross-linking of wild-type H. pylori extract using homo bifunctional DSP and analyzed the product in SDS/PAGE followed by western blotting using anti-CagV antibody. As shown in Fig. 9D, CagV forms a dimer like VirB8 homologs.

CagV has multiple interacting partners
It is well documented that VirB8 interacts with a number of VirB proteins and helps in the assembly of the VirB/D4 transporter [28,51,[57][58][59][60][61]. Since CagV is a predicted homolog of the VirB8, we searched for its interacting partners including CagA by IP using anti-CagV antibody. IP was performed using detergentsolubilized wild-type H. pylori, DcagV, and DcagV/ cagV cell extracts by anti-CagV antibody. The immunoprecipitated samples were separated in SDS/ PAGE and western blotted using indicated antibodies. As shown in Fig. 10A, CagE, CagM, CagT, CagX, CagY, and Cagd were found coimmunoprecipitated along with CagV by anti-CagV antibody from the wild-type H. pylori and DcagV/cagV extracts but not CagF, CagZ, and CagA. No protein was, however, precipitated where the DcagV extract was used. These results thus suggest that the interaction between CagV and the outer membrane subcomplex proteins (Cagd, CagY, CagX, CagT, and CagM) and the inner membrane-associated energy-providing component CagE (VirB4 homolog). Since CagF and CagZ are also inner membrane-associated proteins like CagE and CagV, we hypothesized that these proteins could also interact with each other. The failure to detect precipitation of these proteins using anti-CagV antibody could be due to the fact that proteins in a large complex such as T4SS interact with each other with different affinity and during antigen antibody interaction in IP, conformation of the complex might change leading to the dissociation of weakly associated proteins. However, other possibilities cannot be ruled out. It is worth mentioning that interaction between recombinant CagV and GST-CagZ has been reported earlier [62]. Therefore, to test the notion, we employed an alternative approach to detect the interaction between these proteins. In this approach, MBP pull-down experiments were performed using purified, naturally soluble recombinant MBP-CagV and MBP-CagF as baits, and detergent-solubilized H. pylori extract as prey, separately (Fig. S3) [47]. As shown in Fig. 10B, MBP-CagV pulled down CagE, CagM, CagT, CagX, CagY, Cagd, and CagZ from the wild-type H. pylori extract reproducibly, supporting the IP data. Similarly, MBP-CagF was able to pull-down CagV efficiently along with CagA and CagZ from the wild-type H. pylori extract (Fig. 10C). It is worth mentioning that the interaction of CagF with CagA has been reported earlier [46,47]. As expected, no protein was pulled down by MBP alone. In order to identify direct interacting partners of CagV, IP was performed using extracts prepared from different Cag mutant strains by anti-CagV antibody. The precipitated proteins were separated in SDS/PAGE and western blotted using indicated antibodies. As shown in Fig. S4, unlike in the cases of DcagT, DcagM, and Dcagd mutant strains, in the absence of CagX (DcagX), anti-CagV antibody failed to coimmunoprecipitate the outer membrane subcomplex proteins suggesting that the interaction between CagV and the outer membrane subcomplex proteins might be through CagX. Previously we have also reported interaction of CagE with the outer membrane-associated core complex through CagV and CagX following IP using anti-CagE antibody [14]. Taking together these results showed that CagV has multiple interacting partners including CagX-, CagT-, and Cag-T4SS-specific proteins Cagd, CagM, CagF, and CagZ.  Table 4. Conserved residues of CagV. Strictly conserved residues (conserved in all three homologs). Moderately conserved residues (conserved in any two homologs).

Discussion
In this study, we have characterized one essential Cag-T4SS component, CagV, that shares weak sequence similarity with its counterpart VirB8 of A. tumefaciens. However, based on its adaptation of the VirB8 like membrane topology in E. coli, it is considered as a homolog [20]. To decipher its role in the Cag-T4SS biogenesis and substrate transport to the bacterial cell surface, we have performed biochemical, genetic, microscopic, and bioinformatics analysis of the protein in the context of multifunctional VirB8. Agrobacterium tumefaciens-VirB8 plays a key role in the VirB/VirD4 T4SS assembly, substrate transfer, and pilus biogenesis [19]. VirB8 and its homologs are bitopic inner membrane proteins, form homodimers, and interact with multiple VirB components [56][57][58][59][60][61]. Here, we have investigated its subcellular localization and shown that like A. tumefaciens-VirB8 CagV is also an inner membrane protein ( Fig. 1A-E).
Since our initial attempt to identify native CagV dimer failed, we were prompted to perform a comparative sequence analysis of CagV, along with the two of its homologs A. tumefaciens-VirB8 and B. suis-VirB8 and identified the conserved residues (Fig. 6A). Based on the reported crystal structure data of A. tumefaciens-VirB8 and B. suis-VirB8, a homology model of the periplasmic domain of CagV was generated and found to have similar structure and fold (Fig. 7A). Through comparative MD simulation, we predicted the homodimer interface of the CagV and which was found to be similar to that of the templates. Comparative study of the CagV homodimer interface indicates a high probability of homodimerization of CagV. Based on this information we cross-linked CagV protein and observed self association (Fig. 9D). Dimerization of a protein provides large interface for multiple protein interactions and stability [22]. Agrobacterium tumefaciens-VirB8 has multiple interacting partners like VirB1, VirB4, VirB5, VirB9 etc. [22,35,50,59,61]. Likewise, CagV interacts with the Cag-T4SS outer and inner membrane proteins ( Fig. 10A-C). However, the interaction between CagV and the outer membrane associated subcomplex proteins is through CagX (Fig. S4). Thus, it is the CagV which bridges the inner membrane with the outer membrane subcomplex. This result reconfirms our earlier report of the interaction of CagE with the outer membrane core complex through CagV [14]. It is worth mentioning that the interactions between CagV and Cagd, CagZ, CagT, CagM, and CagN have been shown by the yeast two hybrid (Y2H) experiments and some of the interactions have been confirmed by pulldown experiment [17,62].
Agrobacterium tumefaciens-VirB8 is known to direct the polar localization and stability of several VirB proteins and also stabilizes the secretion system [19,22,23]. However, cellular localization of CagX and CagT on the bacterial cell surface is found to be independent of CagV (Figs 3A-D and 4). CagV is also not required for the stability of several essential Cag-T4SS proteins ( Fig. 2A,B). It only partially affects the stability of CagI, a protein involved in the Cag-T4SS pilus formation ( Fig. 2A,B). CagI interacts with CagH and CagL and is required for the Cag-T4SS pilus formation [29,63]. We have tested interaction of CagV with CagI but they do not interact. These results thus indicate that CagV is not involved directly in stabilizing CagI but might be required to retain the intact Cag-T4SS structure which in turn affects CagI.
Agrobacterium tumefaciens-VirB8 interacts with the VirD2 and the single-stranded T-DNA substrates, which in turn passage the DNA-protein complex from the cytoplasm to the periplasm [59,64,65]. Substrate transfer to the A. tumefaciens-VirB8 requires the energizing ATPase VirD4, VirB4, and VirB11 together with the outer membrane core components VirB7, VirB9 and VirB10 in the periplasm [58]. Surprisingly, however, unlike A. tumefaciens-VirB8, CagV does not interact with the substrate. Nonetheless, Cag-T4SS substrate CagA interacts with the inner membraneassociated chaperone CagF and coupling protein Cagb, respectively [46,48]. Earlier, Jurik et al. demonstrated that CagZ interacts with Cagb [48]. CagV on the other hand interacts with CagF, CagE, and CagZ. CagE has also been reported to interact with Cagb [14]. Gene knockout study has shown that CagF is essential for the translocation of CagA into the host cells [5]. Hence, it is proposed that CagF interacts with CagA at the cytoplasmic membrane which in turn recruits CagA into the active secretion system at the cytoplasmic face of the bacterium [46][47][48]. However, how does CagF assist in CagA recruitment into the secretion system and/or what are the other interacting partner(s) of CagF is not known. Recently, Frick-Cheng et al. have shown the interactions between CagF and the outer membrane subcomplex proteins [18]. However, they were unable to detect interaction between CagF, CagZ, and CagV. This could be due to the dissociation of weakly interacting partners of CagF during IP reaction. But other possibilities cannot be ruled out. Nonetheless, following Y2H screening, the interaction between CagF and CagY has been reported [17,62]. Busler et al. also reported an interaction between recombinant CagV and GST-CagZ [62]. Pinto-Santini and Salama also demonstrated interactions between Cagd and a number of Cag-T4SS components like CagT, Cagb, CagD, Cagc, CagA, CagF, and CagY etc. following immuno-purification of Cag proteins using anti-Cagd antibody and subsequent mass spectrometry [66]. We have earlier shown the interaction between CagV and the energizing component CagE that is essential for CagA translocation [14]. We, therefore, suggest that once Cag-T4SS is activated, energizing component CagE might modulate CagV at the cytoplasmic membrane, which in turn assists CagF to recruit CagA into the secretion system. This notion is supported by the experimental data that in the absence of CagV, CagA translocation across the bacterial membrane is stalled and CagA cannot reach to the cell surface. However, when cagV function is complemented, CagA regains ability to cross the membrane barrier (Figs 5 and S2). Other interpretations also cannot be ruled out, like in the absence of CagV, functional Cag-T4SS is not formed. Hence, although CagV is considered to be a VirB8 homolog, it lacks a few prominent functional attributes of its counterparts. The present study, therefore, may aid in explaining the structural and functional diversity of Cag-T4SS from its counterparts.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Fig. S1. Western blots showing specificity of polyclonal anti-CagV antibody.