Multiple sequence alignments were performed by the use of Clustal W2. Three-dimensional (3D) structure was predicated and modeled using Swiss-model. Structural alignment was generated using TM-align servers from Zhang's lab at University of Michigan. Phylogenetic analysis was performed using MEGA version 5 after multiple alignment of the data via CLUSTAL_X. Distances were obtained using options according to Kimura's two-parameter model and clustering was performed by using the neighbor-joining method. The topology of the neighbor-joining phylogenetic tree was evaluated by using bootstrap resampling with 1000 replications.

Plasmids pDM4 and pMMB207 were used for gene deletion and complementation experiments, respectively. E. coli SY327 λpir was used for conjugation (Miller and Mekalanos, 1988). Clinical V. parahaemolyticus strains were obtained from hospitals in Hong Kong. Other strains and plasmids used in this study were listed in Table 1. V. parahaemolyticus was cultured in LB medium supplemented with 2.5% sodium chloride (NaCl) at 37°C. Chloramphenicol (25 μg/ml for E. coli and 5 μg/ml for V. parahaemolyticus), kanamycin (50 μg/ml for E. coli), and 1 mM Isopropyl β-D-1-Thiogalactopyranoside (IPTG) were supplied if necessary. Zinc depletion was carried out using specific zinc chelator, N,N,N′,N′-Tetrakis (2-pyridylmethyl) ethylenediamine (TPEN, Sigma) dissolved in ethanol.

The vpa1307 gene was deleted from V. parahaemolyticus strain VP3218 by homologous recombination using the methods described previously (Liverman et al., 2007; Zhou et al., 2010). Briefly, the upstream and downstream sequences of vpa1307 gene were amplified using primers vpa1307-1F/vpa1307-1R and vpa1307-2F/vpa1307-2R, respectively (Table 2). These two fragments were used as templates for the second round of PCR using primers vpa1307-1F/vpa1307-2R. The purified overlapping PCR product was digested and cloned into the same digested suicide vector, pDM4. E. coli SY327 λ pir carrying the recombinant plasmid, the helper plasmid pPK2013, and V. parahaemolyticus strain VP3218 were mixed (5:5:1, v/v/v), spun down and resuspended in 100 μ l LB broth, poured onto a filter on LB agar plate, and incubated overnight. The bacteria on the filter were resuspended, spread on Thiosulfate-citrate-bile salts-sucrose agar (TCBS) containing 5 μg/ml chloramphenicol to select transconjugants. Randomly selected transconjugants were cultured on LB agar in the presence of 5% sucrose and subjected to repeated serial passages. The knockout mutant, Δ vpa1307 was obtained.

To construct the complementary strain, the vpa1307 gene with additional ribosome-binding site was amplified using primers vpa1307com-F and vpa1307com-R (Zhou et al., 2010) (Table 2). PCR product was digested and cloned into the same digested pMMB207 to create pMMB207:vpa1307. This recombinant plasmid was transformed into E. coli SY327 λ pir and then conjugated into Δvpa1307 with the presence of helper plasmid pPK2013 carrying E. coli SY327 λ pir. Transconjugants were selected on TCBS containing 5 μg/ml chloramphenicol and the strain Δ vpa1307::pvpa1307 was obtained.

Site-directed mutagenesis was generated using GENEART® Site-Directed Mutagenesis kit (Invitrogen Co., NY, USA) with primer pairs H69A-F/H69A-R and H148A-F/H148A-R, (Table 2). Plasmid pMMB207:vpa1307 was used as template. Successful mutations were confirmed by sequencing. Δ vpa1307::pvpa1307 (H⁶⁹A, H¹⁴⁸A) was obtained by the use of the method described above.

Thirty-five micrometers TPEN was added to wild type (WT) log-phase V. parahaemolyticus culture. After induction for 30 min, 1 ml culture was collected and used to extract RNA using Trizol (Invitrogen) following the manufacturer's instructions. DNA was removed from the sample with DNase (Turbo DNase, Ambion) according to the manufacturer's instructions. 0.5 μg RNA was used as template using Superscript one-step RT-PCR system (Invitrogen). No TPEN culture was used as negative control. Primers rtvpa1307-F/rtvpa1307-R and rt16S-F/rt16S-R were used, respectively (Table 2). Primers vpa1307-F and vpa1307-R (Table 2) were used for screening the distribution of vpa1307 in V. parahaemolyticus clinical isolates by PCR approach. The tdh and trh genes were also screened by PCR.

For growth assay, overnight V. parahaemolyticus culture was diluted in LB broth and grown to the exponential growth phase (OD₆₀₀ ≈ 0.6–0.7). The cells were diluted 1:100 into fresh LB broth with or without 35 μ M TPEN, respectively and grown at 37°C with shaking (250 rpm). OD600 was monitored at specific time points. A similar procedure was used in relative growth assay, except that OD₆₀₀ was only monitored at 6 h. 1 mM IPTG plus 5μg/ml chloramphenicol was added when culturing the complementary strains. Relative growth rate was calculated as culture grown with TPEN to that grown without 35 μ M TPEN.

HeLa cells were washed five times with PBS to completely wash the serum-off before bacteria was added and incubated in DMEM (without serum and antibiotics). Overnight V. parahaemolyticus strains were diluted 100-fold using fresh LB broth and grown at 37°C for 4 h. Cultures were then collected, washed, resuspended in DMEM (without serum) and used to infect HeLa cells at a multiplicity of infection (MOI) of ~50 cfu per cell. Supernatants were collected at specific time points and the amounts of LDH released were determined using CytoTox 96 Non-Radioactive Cytotoxicity kit (Promega) following the manufacturer's instructions. Percentage of cytotoxicity was calculated using formula: (test LDH release—spontaneous release)/maximal release. Test LDH release represents the LDH release after infection with different V. parahaemolyticus strains; spontaneous release represents the baseline cell LDH release without infecting with any bacteria, whereas maximal release represents the release of LDH when cells were lysed using lysis solution from the kit.

V. parahaemolyticus strains (10⁸ CFU) were intraperitoneally injected into 6- to 10-week-old C57BL/6 mice (n = 10) as previously described (Hiyoshi et al., 2010; Pineyro et al., 2010; Whitaker et al., 2012) and mice that were alive were measured at the indicated time points. Three independent replicated experiments were performed. The animal experiments were conducted in the National Institute for Communicable Disease Control and Prevention, Chinese Center for Disease Control and Prevention (CDC) following the guidelines and policies approved by the Chinese CDC.

Since PAI is important for the virulence, we focused on genes related to Vp-PAI (Dobrindt et al., 2004). After a close examination of the Vp-PAI region from tdh-positive V. parahaemolyticus RIMD2210633, we found a hypothetical gene, vpa1307 that is localized upstream of the Vp-PAI. A similar vpa1307 was also identified upstream of Vp-PAI of a trh-positive V. parahaemolyticus TH3996 (Figure 1).

BLAST analysis showed that Vpa1307 shares 23% amino acid sequence identity to the zinc binding protein from V. cholerae O1 biovar EI Tor strain N16961. In addition, Vpa1307 possesses three conserved histidine residues, H⁶⁹, H¹⁴⁸, and H²⁰² that are the hallmark of ZnuA family of proteins (Figure 2). It was shown that residues H⁶⁹, H¹⁴⁸, and H²⁰² are critical for zinc binding and activity (Banerjee et al., 2003; Li and Jogl, 2007; Loisel et al., 2008; Yatsunyk et al., 2008; Ilari et al., 2011). The 3D structure of Vpa1307 was modeled and aligned with the crystal structure of ZnuA from Synechocystis sp. PCC 6803, even through the similarity between these two proteins is only 24%. The TM score, an algorithm to calculate the structural similarity of two protein models, is 0.97, which strongly suggests that Vpa1307 is likely to be a member of ZnuA family (Figure 3).

To explore the evolution history of Vpa1307, a phylogenetic tree was constructed (Figure 4). The neighbor-joining phylogenetic tree showed that Vpa1307 together with its four homologs from other Vibrio spp. fell within the lineage of ZnuA family and formed a distinct cluster within members of ZnuA from other genera. Intriguingly, the phylogenetic analysis also showed that Vpa1307 was excluded from the Vibrionaceae clade of ZnuA, suggesting an exogenous origin of Vpa1307 and representing a novel subfamily of ZnuA. The data suggested that vpa1307 is very likely acquired by V. parahaemolyticus through HGT.

Given that vpa1307 group genes were exogenously acquired by Vibrio spp., the prevalence of this gene among V. parahaemolyticus strains was evaluated. Our data showed that the vpa1307 gene was detectable in 8 out of 20 (40%) of the tdh-positive strains but not in tdh- and trh-negative strains, suggesting the exogenous origin of vpa1307.

To test the contribution of Vpa1307 to V. parahaemolyticus growth, both WT V. parahaemolyticus VP3218 clinical strains and the vpa1307 deletion mutant, Δ vpa1307, were grown in normal and zinc depletion conditions. All test strains showed similar growth rate in normal medium, while the growth of Δvpa1307 was inhibited by ~70% in the medium containing 35 μ M TPEN, a zinc chelating agent, compared to growth in normal medium. However, the growth of WT was only slightly inhibited in the medium containing 35 μ M TPEN (Figure 5). This indicated that vpa1307 contributes to the growth of V. parahaemolyticus under the zinc limitation condition. The data prompt us to examine the expression status of vpa1307 in V. parahaemolyticus. It was shown that vpa1307 only expressed under zinc depletion condition (35 μ M TPEN added) (Figure 6), but not in normal medium. The expression regulation feature of vpa1307 was consistent with the contribution of vpa1307 to V. parahaemolyticus growth under zinc depletion condition.

To further confirm that Vpa1307 is a homolog of ZnuA, the contribution of three conserved histidine residues to the activity of Vpa1307 was tested. An elegant design helps us to achieve this goal. First, a complementary construct was made by incorporating a vpa1307 into plasmid pMMB207 and designated as pvpa1307. Second, two histidine residues, H⁶⁹ and H¹⁴⁸ that may be essential for Vpa1307 function, were mutated to alanine to obtain pvpa1307 (H⁶⁹A, H¹⁴⁸A). Mutations of two of the three histidines are enough to inactivate a ZnuA function (Li and Jogl, 2007; Loisel et al., 2008; Ilari et al., 2011). These two constructs were then used to complement the loss of function by vpa1307 deletion mutant, Δvpa1307. As showed in Figure 7, compared to WT V. parahaemolyticus VP3218, vpa1307 deletion mutant, Δ vpa1307, showed about 25% of growth rate. When complemented with vap1307, Δ vpa1307::pvpa1307, the growth rate of V. parahaemolyticus increased to ~75% compared to the WT strain. However, when complemented with the vpa1307 double histidine mutant Δ vpa1307::pvpa1307 (H⁶⁹A, H¹⁴⁸A), the growth rate of V. parahaemolyticus remained at the same level as Δ vpa1307 (25%) suggesting that the mutations, H⁶⁹A, H¹⁴⁸A, completely abolished the activity of Vpa1307. Furthermore, RT-PCR assay has confirmed that the expression level of vpa1307 in WT V. parahemolyticus VP3218 strain, Δ vpa1307::pvpa1307, and Δ vpa1307::pvpa1307 (H⁶⁹A, H¹⁴⁸A) were similar suggesting that the loss of function of vpa1307 (H⁶⁹A, H¹⁴⁸A) was due to the mutation of conserved histidine residues (data not shown). These data further confirmed that Vpa1307 is a member of ZnuA.

Since ZnuA contributed to host cell infection in B. abortus, M. catarrhalis, and S. enterica (Yang et al., 2006; Ammendola et al., 2007; Murphy et al., 2013), we further tested whether vpa1307 gene contributed to the virulence of V. parahaemolyticus. HeLa cells that were maintained in serum-free DMEM were infected with Δ vpa1307, Δ vpa1307::pvpa1307 and WT strains, respectively. Similar to no infection cell control, strain Δ vpa1307 did not cause notable cell rounding and detachment, while strain Δ vpa1307::pvpa1307 showed more cell rounding than infected with WT strain of V. parahaemolyticus and all cells were detached after longer incubation (Data not shown). The cytotoxicity of V. parahaemolyticus was also determined by measuring the amount of LDH released from damaged cells. WT V. parahaemolyticus caused about 70% of LDH release, whereas Δ vpa1307 strain caused ~20% of LDH release after 4 h infection (Figure 8). The complementation of vpa1307, Δ vpa1307::pvpa1307, regained its toxicity back to 60% of LDH release, a level of toxicity similar to that of WT strain. These data indicated that Vpa1307 contributes to the cytotoxicity of V. parahaemolyticus strain VP3218 in HeLa cells. It also suggested that acquisition of zinc from cells is required for the successful infection of V. parahaemolyticus.

To further evaluate the role of Vpa1307 in the pathogenesis of V. parahaemolyticus, mouse infection model was employed. As shown in Figure 9, during the early infection period (within 6h), both WT V. parahaemolyticus and vpa1307 deletion mutant, Δ vpa1307, showed similar effect on mice and caused about 10% death. However, the mortality of Δ vpa1307 was dramatically delayed and attenuated when compared with WT strain at the later infection period. At post 24 h of infection, mortality rate of WT stain reached 80%, while that of Δ vpa1307 was about 30~50%. This indicated that the Vpa1307 contributes to the pathogenesis of V. parahaemolyticus at certain extent.

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.