Bacterial strains and plasmids are listed in Table S1. Bacteria were routinely cultured at 37°C on LB agar (Miller) or LB broth (Lennox). Antibiotic selection was performed using ampicillin (Amp) at 100 μg/ml unless otherwise stated. Isopropyl-β-D-thiogalactopyranoside (IPTG) was used at a final concentration of 0.2 mM where appropriate. Eukaryotic tissue culture was performed at 37°C in the presence of 5% CO₂. HeLa cells were cultured in Dulbecco's Modified Eagle Medium (DMEM, Gibco) supplemented with 10% heat-inactivated fetal calf serum (FCS), 1% L-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin.

To create N-terminal glutathione S-transferase (GST)-fusion proteins, the full length genes encoding BipC (B. pseudomallei 10276) and SipC (S. enterica serovar Typhimurium 4/74) were amplified from genomic DNA, cloned into the vector pGEX-4T-1, propagated in E. coli XL1 Blue (Agilent) and then confirmed by Sanger sequencing (Source Bioscience). pGEX-BimA₄₈₋₃₈₄ (Stevens M. P. et al., 2005) was used as a positive control in relevant assays. pGEX, pGEX-BipC, -SipC, and -BimA₄₈₋₃₈₄ were transformed into E. coli Rosetta Bl21 (Novagen). An overnight culture of E. coli Rosetta BL21 containing pGEX, pGEX-BipC, -SipC, or -BimA₄₈₋₃₈₄ was used to inoculate LB containing Amp followed by incubation for 3 h with shaking at 37°C. IPTG was added followed by incubation overnight at 16°C. The bacteria were collected by centrifugation and lyzed with ice-cold lysis buffer [1% Octylthioglucoside, 200 mM NaCl, 1 mM EDTA, 20 mM Tris Cl, 1 mM DTT, 0.1% Triton X-100, 10 μl Lysonase Bioprocessing Reagent (Merck)]. The insoluble fraction was removed by centrifuging at 10,000 RCF. In the case of GST-SipC, the recombinant protein was insoluble after induction. Insoluble GST-SipC was purified from inclusion bodies and refolded using the Novagen protein refolding kit according to the manufacturer's instructions.

For GST, GST-BimA, and GST-BipC, glutathione (GSH)-linked Sepharose beads (GE Healthcare) were added to the bacterial lysate and incubated at 4°C for 1 h with agitation. The beads were collected by centrifugation, washed extensively with PBS followed by Re-suspension in an equal volume of PBS. When necessary, GST-fusion proteins were eluted in elution buffer (50 mM reduced GSH, 50 mM Tris, pH 8.0). The solution was dialyzed three times in PBS for a minimum of 2 h each to remove the detergents and salts. The dialyzed protein was concentrated using a 10 kDa cut-off spin column and protein concentrations determined using a Direct Detect (Merck Millipore).

Murine spleens^* were homogenized in 1x polymerization buffer with protease inhibitors (10 mM Tris—pH 7.5, 1 mM ATP, 2 mM MgCl₂, 50 mM KCl, 0.5 mM EGTA, 1 mM PMSF, 1 μg/ml pepstatin A, 10 μg/ml leupeptin, 1 μg/ml aprotinin). Lysates were clarified by ultracentrifugation at 100,000 RCF. To perform pull downs, an aliquot of murine splenic lysate was diluted 1:3 with 1x polymerization buffer and CaCl₂ was added to a final concentration of 100 μM. The GST or GST-fusion proteins of interest (-BimA₄₈₋₃₈₄, -BipC, and -SipC) bound to GSH-linked Sepharose beads were added to the lysate and incubated for 1 h at room temperature with agitation. The Sepharose beads were then washed five times with ice-cold PBS. Samples were subjected to SDS-PAGE and visualized by Coomassie blue staining (Thermo Fisher). For immunoblotting, proteins were transferred to PVDF membranes and visualized using goat α-actin primary antibody (Santa Cruz Biotechnology) followed by rabbit α-goat HRP-linked secondary antibody (Invitrogen). The antibodies were detected using ECL (Thermo Fisher) followed by film exposure.

^*Tissues were obtained from wild-type mice in accordance with the UK Animals (Scientific Procedures) Act 1986, following local ethical review of protocols.

Confocal microscopy was used to investigate the ability of GST-fusion proteins to directly bind actin as described in (Stevens J. M. et al., 2005; Stevens M. P. et al., 2005). To the center of a glass microscope slide, 10 μl of 1x polymerization buffer or PBS containing 1 μM rhodamine labeled actin was added. To this, 1 μl of the GST-fusion proteins of interest (GST, GST-BimA₄₈₋₃₈₄, or GST-BipC) bound to GSH-linked Sepharose beads was added. The slide was immediately visualized using a Zeiss LSM 710 Scanning Laser Confocal Microscope using an excitation of 535 nm and an emission of 585 nm, and data collected using Zen 2011 software (Zeiss). Images were further processed using Image J (https://imagej.nih.gov/ij/).

Mg-actin was prepared by incubating 4 μg/μl rhodamine-labeled actin (Cytoskeleton) in the presence of 50 μM MgCl₂, and 125 μM EGTA for 10 min at room temperature. To the mixture, 10x polymerization buffer was added to a final concentration of 1x and incubated for 2 h at room temperature to polymerize the G-actin into F-actin. Sedimentation reactions were prepared in G-Mg buffer (5 mM Tris-HCl pH8, 0.2 mM MgCl₂, 0.2 mM ATP, 0.5 mM DTT) by adding 2.3 μM of F-actin and 1 μM of GST-fusion protein. The reaction was incubated at room temperature for 30 min in the dark before being centrifuged at 100,000 RCF for 1 h at 20°C to sediment the F-actin. The soluble fraction (monomeric G-actin fraction) was separated from the pellet (filamentous F-actin) and the fractions were re-suspended in 2x Laemmli buffer, separated by SDS-PAGE and visualized using Silver staining (Pierce Scientific) according to the manufacturer's protocol. Densitometry was performed using Image Studio Lite software (LI-COR Biosciences).

A pyrene actin polymerization assay was performed essentially as previously described (Stevens et al., 2003). Samples were excited with a wavelength of 365 nm and emission was collected at a wavelength of 407 nm. Data was collected every 30 s for up to 1 h using a FLUOstar Omega plate reader (BMG Labtech). Monomeric actin was used in polymerization assays at a final concentration of 2–3 μM and GST or GST-fusion proteins were used at a final concentration of 1 μM. Each well contained polymerization buffer at a final concentration of 1x. The assay was performed in three independent replicates with three technical measurements per replicate. Data were plotted in Excel. The rates of polymerization were calculated as the rise in fluorescence units per second during the linear phase of polymerization, between 200 and 800 s. The mean rates of polymerization for each protein were analyzed by pairwise Student t-test using R software (https://www.r-project.org/) and P-values of ≤ 0.05 were taken as significant.

To assess the ability of BipC to stabilize actin filaments, an actin depolymerization assay was performed essentially as previously described (McGhie et al., 2001). Mg-actin was prepared by incubating 1 mg/ml pyrene-labeled actin with 125 μM EGTA and 50 μM MgCl₂ for 10 min at room temperature. The Mg-actin was then polymerized by adding 10x polymerization buffer to a final concentration of 1x, followed by incubation at room temperature in the dark for 2–4 h. The depolymerization assay was set up in a 96 well black opaque plate with each well containing 2.3 μM F-actin, plus or minus 2.5 μM of the GST-fusion proteins. Data was collected every 30 s for up to 1 h using a FLUOstar Omega plate reader (BMG Labtech), with samples excited at 365 nm and emission data collected at 407 nm. The assay was performed in three independent replicates with two measurements per replicate. Data was plotted in Excel. Relative fluorescence intensity was calculated by dividing all of the sample measurements by its value at time = 0 s. The rates of depolymerization were calculated as the decrease in relative fluorescence units per second during the linear phase of depolymerization, between 100 and 600 s. The mean rates of depolymerization for each protein were analyzed by pairwise Student t-test using R software (https://www.r-project.org/) and P-values of ≤ 0.05 were taken as significant.

In order to investigate the effects of ectopically expressed BipC in eukaryotic cells, HeLa cells were transfected with the constitutive eukaryotic expression vector pRK5-myc-BipC, which encodes BipC with an in-frame N-terminal c-Myc tag. The plasmid pEGFP (Clontech) which encodes enhanced green fluorescent protein (EGFP) was used as a control. HeLa cell transfections were performed using Lipofectamine 2000 (Invitrogen) according to the manufacturers protocol, using 4.0 μg of plasmid and 10 μl of Lipofectamine 2000 per well. The transfected HeLa cells were incubated for 48 h before being fixed in 4% paraformaldehyde in PBS. The cells were washed twice with PBS and permeabilized using PBS containing 0.5% Triton X-100 for 15 min. Cells were washed twice with PBS and then blocked for 30 min with blocking buffer (PBS containing 0.5% bovine serum albumin w/v, 0.02% NaN₃ w/v). In order to detect the Myc-BipC protein, cover slips were incubated with rabbit α-c-Myc primary antibody (0.5 μg/ml, Santa Cruz) at 37°C for 1 h. The cover slips were washed three times with PBS followed by incubation with goat α-rabbit⁴⁸⁸ secondary antibody (10 μg/ml, Molecular Probes) for 1 h. The cover slips were washed three times with PBS and incubated with phalloidin⁵⁶⁸ (10 μg/ml, Invitrogen) for 15 min. The cover slips were washed six times with PBS, mounted and visualized using a Zeiss LSM 710 Scanning Laser Confocal Microscope using Zen 2011 software (Zeiss). Images were processed using Image J software (https://imagej.nih.gov/ij/).

BipC (annotated as BPSS1531 in the reference K96243 genome) is encoded within the Bsa T3SS locus and is very highly conserved in pathogenic B. pseudomallei and B. mallei (100% identity), and in the closely related avirulent B. thailandensis, though to a lesser extent (86% identity). The B. thailandensis BipC protein contains numerous single amino acid differences as well as two minor deletions in comparison to the BipC proteins of the other two species (Figure S1). BipC shares homology with SipC (58% coverage and 24% identity at the amino acid level) and IpaC (58% coverage, 22% identity) from Salmonella spp. and S. flexneri, respectively (Figure S2). We previously identified BipC in the secretome of B. pseudomallei using a quantitative proteomics approach (Vander Broek et al., 2015). Due to the similarity in sequence between BipC and SipC/IpaC, we hypothesized that BipC would be an actin-binding protein and may possess the ability to bundle F-actin. In the following experiments, we have directly compared the activities of BipC with Salmonella SipC or the B. pseudomallei BimA protein as positive controls. We have previously characterized BimA, which is required for actin-based motility of B. pseudomallei, as an actin monomer binding and actin polymerizing factor (Stevens M. P. et al., 2005; Sitthidet et al., 2011). Both SipC and IpaC interact with and nucleate actin polymerization in vitro (Hayward and Koronakis, 1999; Terry et al., 2008), facilitating bacterial invasion of host cells (Mounier et al., 2009; Myeni and Zhou, 2010). In order to investigate whether BipC associates with cellular actin, pull-down experiments were performed using a GST-BipC fusion protein and murine splenic lysates under actin polymerizing conditions. In a similar manner to the control SipC and BimA proteins, Figure 1A shows that BipC can interact with actin in a murine splenic lysate. Because pull-downs may identify interactions that are indirect through large protein complexes (reviewed in Kool et al., 2011), further work was required to determine if BipC has the ability to bind actin directly. To test this, GSH-linked Sepharose beads coated with GST fusion proteins were incubated with rhodamine-labeled actin, under either polymerizing or non-polymerizing conditions (PBS), followed by immediate visualization using a confocal microscope. A protein that is able to directly bind actin creates a high concentration of the rhodamine-labeled actin at the surface of the Sepharose bead, which appears as a red “halo” or ring around the clear bead in microscope images. The GST-BimA₄₈₋₃₈₄ coated beads directly bound actin in both polymerization buffer and PBS as indicated by a distinct red halo (Figure 1B), and in agreement with previous studies by our laboratory (Stevens J. M. et al., 2005; Stevens M. P. et al., 2005; Sitthidet et al., 2011). The GST-BipC coated beads also directly bound actin in both polymerization buffer and PBS, but appeared to bind actin more efficiently under polymerizing conditions (Figure 1B). At the time of submission of this manuscript for peer review we believed this to be the first evidence that BipC is an actin-binding protein. However, we have since become aware of a similar study by Kang et al. (2016), which also demonstrates that B. pseudomallei BipC is an actin-binding protein.

As BipC appeared to bind actin more efficiently under polymerizing conditions, we were interested in testing whether it preferentially binds to F-actin or G-actin using a quantitative sedimentation assay as we have previously described (Stevens M. P. et al., 2005). Purified GST and GST-fusion proteins (GST-BipC, GST-SipC, and GST-BimA₄₈₋₃₈₄) were pre-incubated with polymerized actin (F-actin). Ultra-centrifugation was then used to sediment the F-actin along with any interacting GST fusion protein. Under these conditions the monomeric G-actin, along with any G-actin interacting proteins or with non-interacting proteins (such as GST), remain in the supernatant. The supernatants and pellets were then subjected to SDS-PAGE and visualized by silver staining. Densitometry was used to quantify the levels of protein present in each sample. Three independent replicates of the assay were performed to ensure reproducibility. As a control, the assay was also performed in the absence of actin to demonstrate that the GST proteins would not sediment on their own (Figure S3). As expected the GST protein was primarily present in the supernatant fraction as it does not possess actin binding activity (Figure 1C). GST-BimA₄₈₋₃₈₄ was also most abundant in the supernatant fraction due to its intrinsic G-actin binding activity (Figure 1C), and in concordance with our previous published observation (Stevens M. P. et al., 2005). GST-SipC associated predominantly with the pellet fraction due to its preferential binding to F-actin, which is in agreement with data published by Hayward and Koronakis (1999) (Figure 1C). Similarly, GST-BipC also co-sedimented with the F-actin pellet in this assay, indicating that it too preferentially binds F-actin under these conditions (Figure 1C).

We have recently investigated the interaction of B. pseudomallei BimA with cellular proteins using a yeast two-hybrid assay (Jitprasutwit et al., 2016). Whilst we have been able to confirm that BimA is an actin-interacting protein in this assay (Jitprasutwit et al., 2016), we were unable to detect a direct interaction between BipC and actin by yeast two-hybrid assay (Figure S4). In our assay, the BipC protein was fused to the GAL4 DNA-binding domain and actin was fused to the GAL4 activation domain. It is possible that the fusion of BipC to the GAL4 DNA-binding domain affected the tertiary structure of the protein and disrupted the interaction. It is also a possibility that the actin-GAL4 fusion protein is unable to polymerize into F-actin, which we have demonstrated that BipC predominantly associates with (Figure 1C). In support of this hypothesis, we have been unable to find examples of the use of the yeast two-hybrid system to dissect interactions of F-actin binding proteins with actin in the literature.

Previous work has demonstrated that Salmonella SipC not only has the ability to bind and bundle F-actin, but also to polymerize actin (Hayward and Koronakis, 1999). To determine if BipC is also able to polymerize actin, a polymerization assay was performed with pyrene-labeled actin, which emits fluorescence over time as monomers assemble into actin filaments (Stevens M. P. et al., 2005). GST, GST-SipC, and GST-BipC were incubated with pyrene actin in polymerization buffer and their rates of actin polymerization compared by measuring fluorescence intensity over time. SipC showed a statistically significant increased rate of polymerization (average rate of increase in fluorescence intensity = 14.66 ± 5.01 AU/s) when compared to the GST control which mimics the rate of intrinsic actin polymerization in the absence of a polymerizing factor (average rate of increase in fluorescence intensity = 7.45 ± 3.65 AU/s) (Figure 2A). BipC also showed a statistically significant increase in the rate of actin polymerization (average rate of increase in fluorescence intensity = 9.88 ± 3.87 AU/s) when compared to the GST control indicating that this protein has an intrinsic actin polymerizing activity (Figure 2A).

With the establishment that BipC binds and polymerizes actin in vitro, we wanted to investigate what effect BipC would have when expressed ectopically in eukaryotic host cells. Full length BipC was transiently expressed with an in-frame N-terminal c-Myc tag in HeLa cells. At 48 h post-transfection, the cells were fixed and permeabilized, probed using phalloidin (to label F-actin) and rabbit α-Myc antibodies to label BipC, and imaged using a confocal microscope. In cells expressing BipC, the BipC protein localized around the periphery of the cells (Figure 2B). The cells expressing BipC were morphologically distinct from untransfected cells, displaying multiple pseudopodia-like structures, indicating that BipC modulates actin dynamics within host cells (Figure 2B).

In addition to stimulating the polymerization of actin in vitro, Salmonella SipC also stabilizes F-actin polymers, preventing their dissociation in solution. The actin polymerizing and F-actin bundling activities of SipC are enhanced by another T3SS effector protein, SipA (McGhie et al., 2001). Interestingly, there is no direct homolog of SipA encoded by the B. pseudomallei genome. To investigate whether BipC has the ability to stabilize F-actin polymers in its own right, an actin depolymerization assay was performed. The inverse of the polymerization assay, the depolymerization assay measures a reduction in the rate of relative fluorescence over time as pyrene-F-actin filaments disassemble into actin monomers (G-actin). GST, GST-SipC, or GST-BipC were incubated with pre-polymerized pyrene-actin under depolymerizing conditions and fluorescence intensity was measured over time. As expected, the GST negative control showed similar levels of depolymerization to that of the actin-only control since it lacks any actin-binding or F-actin stabilizing activities (Figure 2C). Under the same assay conditions F-actin incubated with SipC was significantly stabilized and showed significantly slower rates of depolymerisation, as previously demonstrated by McGhie et al. (2001). Surprisingly we did not need to include recombinant SipA in our assay, although addition of SipA may well have further enhanced the F-actin stabilizing activity of SipC. In contrast the GST-BipC protein showed no significant F-actin stabilizing activity in this assay, demonstrating that this protein, unlike its homolog SipC, does not have intrinsic F-actin bundling activity (Figure 2C). Since submission of this manuscript for peer review, a study by Kang et al. (2016) suggested that a His-tagged BipC protein exhibited F-actin stabilizing activity in a similar assay. However, it is notable that much higher concentrations of BipC protein were required in these experiments than used in our study, or that of McGhie et al. (2001).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.