The bacterial strains, plasmids, and oligonucleotides used in this study are summarized in Tables 1, 2. All V. parahaemolyticus strains used in this study were derived from the HZ strain, a clinical isolate from the Zhejiang Provincial Center for Disease Control and Prevention, Zhejiang, China (Yu et al., 2015). Escherichia coli strains DH5α, BL21, and CC118λpir were used for general manipulation of plasmids, prokaryotic expression of proteins, and mobilization of plasmids into V. parahaemolyticus, respectively. The bacterial strains were grown at 37°C in Luria-Bertani (LB) broth with 1% NaCl (Sambrook et al., 1989) (for E. coli) or LB-NaCl (for V. parahaemolyticus) which is LB broth supplemented with 3% NaCl containing appropriate antibiotics. Plasmids for overexpressing VpDsbAs or VpDsbB or other proteins in V. parahaemolyticus were constructed by cloning the PCR-amplified coding regions into pBAD24Cm (Guzman et al., 1995) and introduced into V. parahaemolyticus by conjugation with the helper strain of CC118λpir (Yu et al., 2015). Plasmids for overexpressing VpDsbAs in E. coli were constructed by cloning the coding regions into pacyc177 which has been modified to obtain an arabinose operon (Chang and Cohen, 1978) and introduced into E. coli strain by electroporation (Sambrook et al., 1989).

In-frame deletion strains used in this study were described in previous publications (Yu et al., 2015; Jiang et al., 2018). In-frame deletions of VpdsbA1 (vp3054), VpdsbA2 (vpa1271), VpdsbB (vp2073) or EcDsbA were constructed by cloning the regions flanking the target genes into suicide vector pDS132 (Philippe et al., 2004) containing a sacB counter selectable marker (Metcalf et al., 1996). The resulting plasmids were introduced into V. parahaemolyticus or E. coli by conjugation (Boyer, 1966; Yu et al., 2015) and deletion mutants were selected for double homologous recombination events (Kaniga et al., 1991).

Zebrafish virulence assays were designed based on a previous publication (Paranjpye et al., 2013). All animal experiments were carried out in strict accordance with the animal protocols that were approved by the Institutional Animal Care and Use Committee of Zhejiang A&F University (Permit Number: ZJAFU/IACUC_2011-10-25-02). Care and feeding of zebrafish followed established protocols¹. Zebrafish were obtained from a commercial supplier between 5 and 6 months old and were raised at our animal facility at Zhejiang A&F University for at least 2 weeks before challenge experiments following previously published protocols (Paranjpye et al., 2013). For challenge experiments, zebrafish were injected intraperitoneally using a repeater dispenser (Hamilton #83700) and a 33 gauge needle (Hamilton 1750LTSN, 33/0.3759/PT4) following previously published protocols (Lefebvre et al., 2009). Ten zebrafish per group were inoculated with each strain at a concentration of 10⁷ cfu 10 μl⁻¹ of the inoculum and observed every 4 or 8 h for 48 h. Control fish injected with 10 μl PBS were included with each experiment. Experiments were repeated at least once. Tris-buffered tricaine at a concentration of 320 μg ml⁻¹ was used to kill the fish on completion of the experiment. All aquaria and water were disinfected with 20% sodium hypochlorite after each experiment. Filters from air pumps were sterilized by autoclaving.

Vibrio parahaemolyticus VpdsbA1, VpdsbA2 or VPA1314 genes without the signal peptide sequence were amplified from the genomic DNA which was purified by QIAamp DNA Mini Kit (Qiagen) from bacterial culture. The PCR products without the signal peptide sequence were inserted into a modified pET-28a (Novagen, Inc.) vector encoding an N-terminal His₆ tag. E. coli BL21(DE3)/pLysS (Karimova et al., 1998) cells were transformed with the plasmid containing the target gene and transformed cells were used for protein expression by autoinduction (Thompson et al., 1994). Proteins were expressed and purified on nickel columns according to the manufacturer’s instructions (Invitrogen).

Ten milligrams of each protein, VpDsbA1, VpDsbA2, and VPA1314 (TDH) purified as described above were used as antigens, and then sent to GenScript Inc. for polyclonal antibodies preparation by immunizing rabbits. Antibody effectiveness was detected by Western blotting of each specific purified protein a week after the fourth immunization.

The protein disulfide reductase activity of VpDsbA was measured in vitro using the insulin-reduction assay in the presence of dithiothreitol (DTT) (Holmgren, 1979). Each DsbA protein (10 μM) was mixed with buffer consisting of 100 mM sodium phosphate, 2 mM ethylenediaminetetraacetic acid (EDTA), 0.33 mM DTT. Insulin (170 μM) was added immediately before measurements were made. Insulin reduction by DTT was monitored spectrophotometrically at 650 nm.

VpDsbA redox potential assays were performed essentially as described by Walden et al. (2012). VpDsbA (2 μM) was incubated in fully degassed buffer consisting of 100 mM sodium phosphate, 1 mM EDTA pH 7.0 containing 1 mM oxidized glutathione (GSSG; Sigma–Aldrich, United States) and a range of reduced glutathione (GSH) concentrations (0.1–2.0 mM) for 24 h at room temperature. After incubation, the reactions were stopped with 10% trichloroacetic acid (TCA) and the precipitated protein pellets were collected by centrifugation at 16,000 × g for 30 min at 4°C. The pellets were washed with 100% ice-cold acetone and dissolved in a buffer consisting of 50 mM Tris-HCl pH 7.0, 1% sodium dodecyl sulfate (SDS), 10 mM 4-acetamide-4′-maleimidylstilbene-2,2′-disulfonate (AMS) to label the free thiols. Separation of the reduced and oxidized forms was performed on 12% SDS-polyacrylamide gels under denaturing conditions. Gels were stained with Coomassie Brilliant Blue and scanned. The relative intensity of the reduced and oxidized forms was analyzed using ImageJ². The fraction of the reduced protein was plotted against the ratio [GSH]²/[GSSG]. The equilibrium constant K_eq was calculated using the equation:

where R is the fraction of reduced protein at equilibrium. The standard redox potential was calculated using the Nernst equation,

where E^0′_GSH/GSSG is the standard potential of −240 mV (Gilbert, 1995), R is the universal gas constant 8.314 J K⁻¹mol⁻¹, T is the absolute temperature in Kelvin, n is the number of electrons transferred, F is the Faraday constant 9.648 × 10⁴ C mol⁻¹ and K_eq is the equilibrium constant.

Overnight cultures of V. parahaemolyticus WT or dsbA mutants were subcultured at a dilution of 1:100 in LB-NaCl medium and incubated without shaking at 37°C for 4 h. Total RNA was purified from bacterial cultures using TRIzol reagent (Invitrogen), DNase digestion and RNA reverse transcription was performed by using the PrimeScript RT reagent Kit with gDNA Eraser (TaKara). Quantitative real-time qPCR was performed in 20 μl reaction mixtures containing SYBR quantitative PCR mix (Toyobo) to measure the transcriptional levels of genes of interest using the Mx3000P PCR detection system (Agilent) with primers specific for tested genes. The 16s rRNA gene were used as internal controls in all reactions.

Cadmium resistance was performed to quantify the relative disulfide oxidase activity of the strains in vivo as described in Xue et al. (2016). Briefly, strains were grown overnight in LB and diluted 1:100 into fresh LB media with appropriate antibiotics. Strains were grown to mid-logarithmic phase at 37°C and serially diluted into phosphate buffer salt (PBS). A 5 μl aliquot of each dilution was plated onto LB plates with a cadmium gradient. Cells were grown at 37°C overnight. All spot titers were performed at least in triplicate.

Affinity and thermodynamics of binding between taurocholate (TC) and VpDsbA proteins were assessed by isothermal titration calorimetry (ITC) using a VP-ITC instrument (MicroCal^TM, GE Healthcare) according to the protocols as described in Xue et al. (2016). Briefly, the sample cell was loaded with 1.5 ml of purified protein at 200 μM concentration in PBS. The syringe was filled with TC in the same buffer as that was used to dilute the proteins at a concentration of 4 mM. Titrations were conducted at 25°C using 25 consecutive injections of 10 μl each delayed by 300 s with a stirring speed of 307 rpm. As a control for background noise, titration of TC into a solution containing the buffer only was performed. The association constant (K_a = 1/K_d), free energy (ΔG), and enthalpy change (ΔH) and entropy change (ΔS) were calculated by fitting the data to a single-site binding model using the MicroCal Origin software (Origin 7.0 SR4 version7.0552β). Parameters reported include the mean ± SD across three replicates. The calculated c-value for these measurements is 12.

Motility assay was performed essentially on soft LB (for E. coli strains) or LB-NaCl (for V. parahaemolyticus strains) agar (0.25%) as described (Paxman et al., 2009). E. coli or V. parahaemolyticus strains were grown overnight on LB or LB-NaCl agar and a single colony of each strain was inserted into soft agar by tooth pick and incubated at 37°C for 8–12 h. Motility was assessed by examining migration of bacteria through agar from the center toward the periphery of the colony. The diameter of the motility circle was analyzed using ImageJ (see text footnote 2).

VpDsbA oxidized by VpDsbB in vitro was performed as previously described (Xue et al., 2016). V. parahaemolyticus strain ΔdsbA harboring a plasmid encoding VpDsbB under the control of an arabinose inducible promoter was cultured at 37°C with 200 rpm shaking in LB-NaCl medium until OD₆₀₀≈0.5. 0.2% of arabinose was added and bacteria were kept cultured in the same conditions for another 16 h. Membranes were prepared according to Kovach et al. (1995). Purified DsbA was reduced by incubation in 50 mM DTT for 10 min at 4°C. DTT was then removed by gel filtration on PD-10 Sephadex columns (GE). Reduced DsbA was stored at −80°C in the presence of 0.1 mM EDTA, pH 8.0. VpDsbB membrane suspension (1 mg/ml) was mixed with 10 mM ubiquinone 1 (UQ1) with or without different concentrations of TC as indicated in PBS. Reactions were started right after 2 mM of reduced VpDsbA was added. TCA (10%) was added at different time points to stop the reaction and protein was precipitated at 4°C overnight. Precipitated proteins were treated with AMS as described. Negative controls were the membrane of V. parahaemolyticus ΔdsbA/dsbB.

Caco-2 cells (Sigma-Aldrich Inc.) were maintained in Dulbecco’s Modified Eagle Medium (DMEM, Invitrogen) supplemented with 10% fetal bovine serum (FBS, Invitrogen) at 37°C in 5% CO₂. Adhesion assays were carried out as previously described (Liu and Chen, 2015). Briefly, Caco-2 cells were infected with freshly prepared V. parahaemolyticus cultures of tested strains at a multiplicity of infection (MOI) of 10. As a control, bacterial cells were added to the empty wells of the cell culture plates and incubated for the same time as the binding experiment to determine the final total number of V. parahaemolyticus for the experiment. After 1 h infection, cells were washed three times with PBS and then lysed with 1% Triton X-100 at 37°C for 10 min. The cell lysates and control bacteria were serially diluted and plated on LB-NaCl agar. Attachment rate was calculated by dividing bound bacteria to the total bacterial load.

Kanagawa hemolytic (KH) hemolytic activity was determined as described in Chun et al. (1975). Briefly, Wagatsuma blood agar (WBA) (Chun et al., 1975) contained 10% washed rabbit erythrocytes was prepared freshly before each assay. Bacteria from freshly prepared V. parahaemolyticus cultures of test strains were pelleted by centrifugation and washed with PBS. ∼10⁸ CFU in 10 μl bacterial suspension were dropped on WBA plates, and the result was observed after incubation for 24 h at 37°C. Well-defined, clear hemolysis around the bacterial growth was recorded and the hemolysis around area was analyzed using ImageJ (see text footnote 2).

The cytotoxic assays were performed as described previously (Hiyoshi et al., 2010). HeLa cells (Thermo Fisher Scientific Inc.) were seeded in 96-well plates. The overnight cultured strains were sub-cultured at the ratio 1:50 to the fresh LB-NaCl broth medium and grow at 37°C to OD₆₀₀≈0.5. Bacteria were pelleted by centrifugation and washed with PBS, then suspended in two volumes of unsupplemented DMEM (DMEM without FBS). Before infection, HeLa cells were also washed with unsupplemented DMEM. Infection was performed at an MOI of 10. After infection, the release of lactate dehydrogenase (LDH) into the medium was quantified at each indicated time point with a CytoTox96 kit (Promega, Madison, WI, United States) used according to the manufacturer’s instructions.

All data are presented as the mean ± SD of three determinations in each experimental condition. Statistical significance was determined using one-way ANOVA. P < 0.05 was considered statistically significant.

VP3054 (NP_799433.1) and VPA1271 (NP_800781.1) are predicted to be DsbA homologs in V. parahaemolyticus. The amino acid sequences of VP3054 and VPA1271 were compared and thioredoxin-fold molecular features were observed (Figure 1A). Both VP3054 and VPA1271 show about 40% amino acid sequence identity to E. coli DsbA (EcDsbA), Klebsiella pneumonia DsbA (KpDsbA) or Salmonella enterica DsbA (SeDsbA), and the identities are even higher when compared to V. cholerae DsbA from (VcDsbA). VP3054 shows 79% identity to VcDsbA while VPA1271 is 59% identical (Figure 1A). Both VP3054 and VPA1271 encode a DsbA-like domain, including a conserved CXXC active-site motif (CPHC in this case) and a putative cisPro motif (Ren et al., 2009) (Figure 1A). Comparison of loop sequences on the catalytic face of these DsbA proteins revealed that both the two DsbAs from V. parahaemolyticus should belong to Class-Ia DsbA (McMahon et al., 2014; Totsika et al., 2018). To differentiate these two proteins, we named VP3054 as VpDsbA1 and VPA1271 as VpDsbA2.

We found that both VpdsbA1 and VpdsbA2 are expressed in V. parahaemolyticus (Figure 1B), and mutating one gene did not appreciably affect the expression of the other gene. In a zebrafish infection model, ΔVpdsbA1/2 had significantly reduced virulence while both ΔVpdsbA1 and ΔVpdsbA2 were as virulent as the wild type strain (Figure 1C). Each mutant grew similarly to the wild-type strain in broth culture (Supplementary Figure S1). These results suggest that both VpDsbA1 and VpDsbA2 are functional in V. parahaemolyticus and they may function redundantly, because a virulence defect was only seen in the double mutant.

To investigate the enzyme activity of VpDsbA1 and VpDsbA2, we performed a standard reductase activity assay using folded insulin as the substrate and DTT as the electron donor (Holmgren, 1979). Insulin comprises A and B chains which are linked by two disulfide bonds, and the reduction of the disulfide bonds leads to B-chain precipitation, which can be monitored as an increase in absorbance at 650 nm. B chain precipitation and a consequent increase in the OD₆₅₀ was measurable after 22 min with EcDsbA (Figure 2A). B chain precipitation was observed after 15 min with VpDsbA1 and after 22 min with VpDsbA2. Insulin reduction was observed after 10 min with VcDsbA.

To better characterize the redox properties of VpDsbA1 and VpDsbA2, we determined the standard redox potential of these two proteins relative to the redox potential of GSH/GSSG (−240 mV) (Gilbert, 1995). By calculating the fraction of reduced VpDsbA at different concentrations of [GSH]²/[GSSG], the equilibrium constants were determined (Figure 2B,C). The K_eq for VpDsbA1 was 8.4+0.06 × 10⁻⁵ M and VpDsbA2 was 1.2+0.03 × 10⁻⁴ M. The corresponding redox potentials of VpDsbA1 and VpDsbA2 are −118 mV and −123 mV rat pH 7.0. Thus, VpDsbA1 is relatively oxidizing, with a redox-potential value similar to that of the thioredoxin-fold proteins, VcDsbA (E^0′ = −116 mV) (Walden et al., 2012) and the redox-potential value of VpDsbA2 is similar to that of VcDsbA (E^0′ = −122 mV) (Huber-Wunderlich and Glockshuber, 1998).

It was previously reported that E. coli ΔdsbA showed cadmium sensitivity due to the high affinity of Cd²⁺ for protein free thiols (Vallee and Ulmer, 1972). We tested V. parahaemolyticus dsbA mutant cadmium sensitivity as an indicator of oxidase capacity. V. parahaemolyticus WT and mutants were grown in the presence of increasing concentrations of cadmium (0–1.25 mM). The wild type strain of V. parahaemolyticus grew on concentrations of up to 1 mM cadmium, but the ΔVpdsbA1, ΔVpdsbA2, and ΔVpdsbA1/2 mutants were cadmium-sensitive (Figure 3A and Table 3). These results suggest that both VpDsbA1 and VpDsbA2 possess oxidase activity.

E. coli ΔdsbA or ΔdsbB were reported to be defective in motility because they fail to assemble flagella due to a lack of disulfide bonds in the P-ring protein (FlgI) (Dailey and Berg, 1993). However, we did not observe the same motility defect in V. parahaemolyticus dsb mutants (Figure 3B,C). To test if VpDsbA1 and VpDsbA2 can rescue this defect in E. coli ΔdsbA, we introduced VpdsbA1 or VpdsbA2 on plasmids with a low copy number replication origin under the control of the P_BAD promoter in E. coli ΔdsbA. Both VpDsbA1 and VpDsbA2 were able to restore the motility of E. coli ΔdsbA (Figure 3B,C).

When V. parahaemolyticus enters into the human body through the consumption of contaminated water or uncooked food, a set of virulence determinants controlled by a regulatory network are produced in response to the chemical signals present in the small intestine where it normally causes disease (Rivera-Cancel and Orth, 2017). V. parahaemolyticus uses bile salts to regulate virulence factor expression (Gotoh et al., 2010; Li et al., 2016). Bile salts also induce virulence gene expression in V. cholerae by interfering with the redox reaction of DsbA proteins (Xue et al., 2016). We first investigated the interaction between the bile salt TC and VpDsbA1 and VpDsbA2 by using ITC. We found that both VpDsbA1 and VpDsbA2 can bind TC, with a K_D of 131 ± 7 μM and 164 ± 11 μM, respectively (Figure 4A). We also tested TrxA from E. coli binding to TC by ITC, but EcTrxA and TC did not show any specific interaction (Supplementary Figure S2) which indicated that the binding of VpDsbA1 or VpDsbA2 with TC is a specific interaction event.

Insulin reduction mediated by either VpDsbA1 or VpDsbA2 was inhibited by TC (Figure 4B). The cadmium sensitivity assay showed that V. parahaemolyticus was more sensitive to cadmium in the presence, than in the absence of TC (Figure 3A and Table 3), indicating that TC might reduce the efficiency of disulfide bond formation catalyzed by VpDsbAs in the periplasm of V. parahaemolyticus. DsbA reductase activity correlates with DsbA reoxidation by DsbB (Ren et al., 2009). To test if VpDsbA activity inhibited by TC in vivo is because TC interferes with the reaction between VpDsbA and VpDsbB, we studied the oxidization of VpDsbA by VpDsbB in vitro. The oxidization of both VpDsbA1 and VpDsbA2 by the membrane protein VpDsbB was inhibited by TC in vitro (Figure 4C).

Vibrio parahaemolyticus has evolved several regulatory networks that control the production of a wide range of virulence factors which enables them to cause disease in the host, including the thermostable direct hemolysin (tdh) and TDH related hemolysin (trh), adhesin, and secreted effectors (Zhang and Orth, 2013; Liu and Chen, 2015). We found that V. parahaemolyticusΔVpdsbA1/2 exhibits reduced virulence in the zebrafish infection model (Figure 1C). To understand why, we first investigated the effect of VpDsbAs on host cell adhesion (Liu and Chen, 2015). Deleting either VpdsbA1 or VpdsbA2 decreased V. parahaemolyticus attachment to Caco-2 epithelial cells (Figure 5A). Several adhesion factors present at the surface of V. parahaemolyticus have been reported to be important in host cell binding, including MAM7 (Multivalent Adhesion Molecule 7) (Krachler et al., 2011), MSHA pilus (O’Boyle et al., 2013), PilA (Shime-Hattori et al., 2006), and VpadF (Liu and Chen, 2015). To determine whether VpDsbAs are involved in maintaining the stability of these adhesion factors, we quantified the abundance of MAM7 (VP1611), MshA (VP2697), PilA (VP2523), and VpadF (VP1767). The abundance of VpadF, but not MAM7 was significantly reduced in the ΔVpdsbA1/2 mutant but not in ΔVpdsbA1 or ΔVpdsbA2 single knock-out mutant (Figure 5B) which indicates that either one of these two VpDsbA proteins is necessary for VpadF stability. The VpDsbAs had no effect on the transcription levels of all these adhesion factors (Figure 5B and Supplementary Figure S3). These results indicate that the stability of VpadF depends on a functional DsbA protein present in the periplasm of V. parahaemolyticus which plays an important role in V. parahaemolyticus adhesion to epithelia cells.

Vibrio parahaemolyticus strains isolated from clinical samples are able to lyse human erythrocytes when plated on a high-salt media called Wagatsuma agar, a process termed the Kanagawa (KP) test (Nishibuchi and Kaper, 1995). To determine whether VpDsbAs contribute to the β-hemolytic activity of V. parahaemolyticus, we performed the KP test with WT and the dsbA mutants. We found that the β-hemolytic activity significantly decreased when both VpdsbA genes are deleted (Figure 5C,D). TDH is the major toxin that contributes to the β-hemolytic activity of V. parahaemolyticus (Honda et al., 1988; Nishibuchi et al., 1992). The mRNA level of TDH was not affected by VpDsbAs (Figure 5E,F). However, the amount of TDH protein decreased significantly in the ΔVpdsbA1 and ΔVpdsbA1/2 mutants (Figure 5E,F), suggesting that VpDsbA1 plays a major role in maintaining TDH stability.

Vibrio parahaemolyticus harbors two type III secretion systems (T3SS) encoded in chromosomes 1 (T3SS1) and 2 (T3SS2) (Makino et al., 2003). T3SS1 is responsible for cytotoxicity and T3SS2 is primarily involved in enterotoxicity, as well as in cytotoxic activity against some specific cell lines (Hiyoshi et al., 2010). To test whether VpDsbAs affect cytotoxicity of V. parahaemolyticus against HeLa cells, HeLa cell lysis was measured by monitoring the release of LDH after infection with V. parahaemolyticus tdhAs and T3SS deletion mutant strains in the presence or absence of VpDsbAs. VtrA is the master regulator of T3SS2 in V. parahaemolyticus, so T3SS2 will stop working with vtrA deletion (Kodama et al., 2010); and T3SS1 will lose function when the inner membrane protein VP1696 is deleted from the genome (Park et al., 2004). Consistent with previous reports (Park et al., 2004; Hiyoshi et al., 2010), T3SS1 of V. parahaemolyticus works dominantly in cytotoxicity of HeLa cells (Figure 5G). After 6 h infection, HeLa cells were nearly completely lysed by V. parahaemolyticus T3SS2 mutant strains containing a functional T3SS1, but strains without VpDsbAs showed ∼40% less lysis (Figure 5G). This suggests that VpDsbAs play an important role in epithelial cell infection through T3SS1. By assessing the transcription level and the secretion level of some major effectors and the activators of T3SS1, we found that VpDsbAs do not have a significant effect on the transcription of genes involved in T3SS1 activity (Supplementary Figure S4). However, the secretion efficiency of VPA0450 decreased in dsbA mutant compared with that in WT strain, while VP1683 was secreted more efficiently in the mutant strain under the conditions that we tested (Supplementary Figure S5). We did not figure out what’s the mechanism that VpDsbAs work differently in the secretion of these two effectors, yet, this strongly suggests that VpDsbAs affect T3SS1 functions at the post-translational level.