The bacterial strains and plasmids used in this study are listed in Table 1. The wild-type (WT) V. fluvialis 85003 and its derivative, as well as other V. fluvialis environmental and clinical isolates, were routinely grown in Luria-Bertani (LB) broth (pH 7.4) containing 170 mM NaCl at 30°C unless specifically indicated. E. coli SM10 λpir and Rosetta (DE3) were routinely cultured at 37°C. Culture media were supplemented with ampicillin (Amp, 100 μg/ml), streptomycin (Sm, 100 μg/ml), rifampicin (Rfp, 50 μg/ml), kanamycin (Km, 50 μg/ml), tetracycline (Tc, 10 μg/ml for E. coli, 2.5 μg/ml for V. fluvialis), chloramphenicol (Cm, 10 μg/ml), or isopropyl-β-D-thiogalactopyranoside (IPTG), as required.

To determine the sequence of the 3' end of VflT6SS2, primers vfu-vasL-up2153177 and vfu-rbsD-dn2158250 were designed according to the corresponding sequence of Vibrio furnissii NCTC11218. Similarly, primer pairs vfuA01956-up/vfuA01959-dn and vfuB01009-up/vfuB01011-dn were designed to amplify the hcp-vgrG alleles in 85003. The polymerase chain reaction (PCR) was performed with TranStart FastPfu Fly DNA polymerase (Transgen Biotech, China) with 85003 genomic DNA used as a template and the products were used for commercial sequencing.

To determine whether T6SS genetic loci are prevalent in isolates of V. fluvialis, PCR was performed with water-boiled templates of isolates to detect the presence of multiple chromosomal fragments specific for each T6SS gene cluster. For VflT6SS1, fragments covering tssC1 (impC)-impD, impD-tssF1 (impG), tssF1 (impG)-tssG1 (impH), tssH1, tssH1 (clpB)-tagH1 (impI), tagH1 (impI)-tssK1 (impJ), and tssL1 (impK)-tssM1 (impL) were respectively amplified by using primer pairs impC/impD-F and impC/impD-R, impD/impG-F, and impD/impG-R, impG/impH-F and impG/impH-R, clpB-F/clpB-R, clpB/impI-F and clpB/impI-R, lip/impJ-F, and lip/impJ-R, and impK/impL-F and impK/impL-R. For VflT6SS2, fragments covering tssC2 (vipB)-tssF2 (vasA), tssG2 (vasB)-tagH2 (vasC), tssJ2 (vasD)-tssK2 (vasE), tssL2 (vasF)-tssH2 (vasG), vasH-vasI, vasI-tssA2 (vasJ), and tssM2 (vasK)-vasL were separately amplified with primer pairs vipB/vasA-F and vipB/vasA-R, vasB/vasC-F and vasB/vasC-R, vasD/vasE-F, and vasD/vasE-R, vasF/vasG-F and vasF/vasG-R, vasH/vasI-F and vasH/vasI -R, vasI/vasJ-F, and vasI/vasJ-R, and vasK/vasL-F and vasK/vasL-R. The oligonucleotide primers used are listed in Table 2.

The in-frame deletion mutants ΔvasH, ΔtssD2a, ΔtssD2b, and ΔtssDI2c were constructed by allelic exchange using the clinical strain 85003 as a WT precursor as described previously (Wu et al., 2015). We used ΔtssD2a as a precursor to construct the double mutant, ΔtssD2ab. Briefly, the chromosomal fragments containing upstream and downstream nucleotide sequences of target genes were respectively amplified using the corresponding primers listed in Table 2 and then were stitched together by overlapping PCR. The 1.79 kb ΔvasH and 1.69 kb ΔtssD_c-tssI2_c fragments were individually cloned at SalI-SacI sites in a pCVD442 suicide plasmid to yield pCVD-VFΔvasH and pCVD-ΔtssDI2_c. The 1.74 kb ΔtssD2_a and 1.72 kb ΔtssD2_b fragments were cloned at BamHI-SmaI sites and NotI-SmaI sites in the suicide plasmid pWM91 to yield pWM-ΔtssD2_a and pWM-ΔtssD2_b, respectively. The resulting recombinant suicide plasmids were conjugated into V. fluvialis from E. coli SM10λpir and transconjugants were selected on LB media containing Amp and Sm. The transconjugants were counter-selected by growing them on media containing 15% sucrose. Sucrose-resistant colonies were tested for Amp sensitivity, and target deletion was identified by PCR and confirmed by DNA sequencing.

The vasH expression plasmid was constructed for complementation test. Primers vasH-For-HindIII and vasH-Rev-NdeI were used to amplify the ORF of vasH with PrimeSTAR® HS DNA Polymerase (TaKaRa, Dalian, China) using WT 85003 genomic DNA as a template. The 1,596 bp PCR product with blunt ends was digested with NdeI and cloned at NdeI and SmaI sites in pSRKTc to generate pSRvasH, which expresses vasH from the lac promoter with the induction of IPTG. The pSRvasH was mobilized into ΔvasH mutant from E. coli SM10λpir by conjugation, and the expression vector pSRKTc without an insert was used as a negative control.

An anti-Hcp polyclonal antibody was raised in rabbits using the following method: First, the hcp gene was amplified by PCR using primer pair VF-hcp-up-NcoI/VF-hcp-dn-XhoI and was cloned into pET30a (+) using NcoI and XhoI restriction enzyme sites. His-tagged Hcp expression was induced by IPTG in E. coli Rosetta (DE3) host cells and was purified by Ni-IDA affinity chromatography (Novagen) under native conditions according to the manufacturer's instructions. The purified Hcp protein was used for immunization of rabbits by Wuhan Abzome Biotechnology, China. The polyclonal antiserum was purified with an antigen affinity purification procedure before use.

Overnight cultures of bacterial strains were diluted to 1:100 in 20 mL of fresh LB containing 170 mM NaCl with shaking, and were incubated to an optical density at 600 nm (OD₆₀₀) of 1.5. In the vasH complementation test, overnight cultures of ΔvasH mutants containing plasmid pSRvasH (ΔvasH/pSRvasH) or pSRKTc (ΔvasH/pSRKTc) were grown in the above LB with tetracycline to an OD₆₀₀ of 0.5. Each culture was then divided in half. One half was induced by the addition of IPTG (final concentration of 0.5 mM), and the other half (uninduced) was used as a control. The cultures were continually incubated until they reached an OD₆₀₀ of 1.5 at 30°C with shaking.

Cells were pelleted at high speed in a tabletop microcentrifuge for 5 min. Equal volumes of supernatants were filtered through 0.22 μm low protein-binding polyvinylidine fluoride syringe filters (Millipore), and then a 1/100 volume of 2% sodium deoxycholate was added, and the samples were incubated for 30 min on ice. Proteins were precipitated with 10% trichloroacetic acid (TCA) for 30 min on ice, pelleted by centrifugation at 12,000 rpm for 20 min at 4°C, and washed twice with ice-cold acetone to remove residual TCA. Protein precipitates were resuspended in 40 μl SDS-PAGE loading buffer (10 mM Tris-HCl, pH 8.0; 50 mM DTT; 1% SDS; 10% glycerol; 0.08% bromophenol blue) and boiled for 10 min. Cell pellets were resuspended in PBS and the cell density was adjusted to 3.5 McFarland standards. Then pellets from 1 ml aliquots of above samples were suspended with 150 μl of SDS-PAGE lysis buffer and boiled for 10 min. Protein samples were separated by SDS-PAGE (12%), transferred onto PVDF membranes, and analyzed by Western blotting using a rabbit anti-Hcp polyclonal antibody and an anti-E. coli CRP antibody (BioLegend, USA). The secondary antibodies used were horseradish peroxidase (HRP)-conjugated goat anti-rabbit and goat anti-mouse antibodies. Proteins were visualized using an Enhanced Chemiluminescence system (TaKaRa, Dalian, China). Each immunoblot experiment was repeated at least two times.

Total RNA extraction and quantitative reverse transcription (qRT)-PCR were performed as described previously (Wu et al., 2015). In brief, strain 85003 was grown to OD600 1.5 under the indicated conditions with agitation. Three independent samples were tested in triplicate. A control mixture using total RNA as a template was added in each reaction to exclude the effects of chromosomal DNA contamination. Relative expression values (R) were calculated using the equation R = 2^{−(ΔCq target − ΔCq reference)}, where Cq is the fractional threshold cycle. We used recA mRNA as a reference. The following primer combinations were used: VF-recA-qPCR-up and VF-recA-qPCR-dn for recA; hcp-qPCR-F-com and hcp-qPCR-R-com for tssD2 (hcp); VF-vasF-qPCR-F and VF-vasF-qPCR-R for tssL2 (vasF); VF-vasH-qPCR-F and VF-vasH-qPCR-R for vasH; VF-vasK-qPCR-F and VF-vasK-qPCR-F for tssM2 (vasK); VF-tssD1-qPCR-F and VF-tssD1-qPCR-R for tssD1; vflB629-qPCR-F and vflB629-qPCR-R for tssM1; vflB643-qPCR-F and vflB643-qPCR-R for tssA1; vflB631-qPCR-F and vflB631-qPCR-R for tssK1 and vflB647-qPCR-F and vflB647-qPCR-R for rhs. Detailed primer sequence information is listed in Table 2.

V. fluvialis predator strains 85003 and ΔvasH, and E. coli prey MG1655, were grown overnight on LB agar containing 170 mM NaCl at 30°C. Bacterial lawns were resuspended and normalized to 1.5 McFarland standards in PBS and mixed at a 9:1 ratio (predator: prey) in triplicates. A total of 10 μl of the mixtures were spotted on LB plates with 340 mM NaCl and incubated at 30°C for 5 h. The colony forming units (CFU) per milliliter of the attacker and prey in the mixtures (0 h) were determined by plating 10-fold serial dilutions on streptomycin- and rifampicin-containing agar plates, respectively. Bacterial spots were harvested from LB plates after 5 h, and the CFU per milliliter of the surviving attacker and prey (5 h) were determined on selective media plates as described above.

In the vasH complementation assay, IPTG was added to LB plates at a final concentration of 1 mM to induce vasH expression. V. fluvialis strain ΔvasH/pSRvasH was grown overnight at 30°C on tetracycline-containing LB agar with 170 mM NaCl in the presence or absence of IPTG. The mixtures with prey were prepared as described above and spotted on LB plates containing 340 mM NaCl with and without IPTG. Strain ΔvasH/pSRKTc was used as a control.

The sequences of VflT6SS1, VflT6SS2, and the three hcp-vgrG homologs from V. fluvialis 85003 were deposited in the NCBI database under accession numbers KY319183, KY319184, KY319185, KY319186, and KY319187, respectively. The genomic sequences of V. furnissii NCTC11218 (accession numbers NC_016602.1 and NC_016628.1), V. fluvialis 33809 (accession numbers CP014034 and CP014035), V. cholerae N16961 (accession numbers NC_002505.1 and NC_002506.1) and V. splendidus LGP32 (accession numbers NC_011753.2 and NC_011744.2) were downloaded from the NCBI database.

The comparative analysis of T6SS sequences from V. fluvialis strains 85003 and 33809, V. cholerae N16961, V. furnissii NCTC11218 and V. splendidus LGP32 were performed using the BLAST software with an e-value of 1e-2, and the alignments of >1 kilobase (kb) were kept. VasH protein sequence alignments were performed with the GENEDOC program.

Our previous draft genome analysis of V. fluvialis 85003 revealed the presence of putative homologs of T6SSs, which clustered and grouped into two genetic modules which are highly homologous to two separate regions on chromosome 2 of V. furnissii (Lu et al., 2014). In this study, we defined the detailed gene contents and organization of the two T6SSs, the flanking regions of the two T6SS clusters, and further identified two other “orphan” hcp-vgrG homologs by PCR with primers designed based on the sequence of V. furnissii NCTC11218.

The cluster homologous to the V. furnissii vfu_B00799~vfu_B00780 was named VflT6SS1, and also shared partial synteny with the genetic region of VS_1337~VS_1318 of V. splendidus. In V. furnissii, three genes, vfu_B01185, vfu_B01189, and vfu_B01191, were predicted to encode T6SS-related subunits ClpB, IcmF-like protein and VgrG, respectively. In V. splendidus, only VS_1326 was annotated to encode ClpB. The VflT6SS1 is a locus of around 24 kb consisting of 19 open reading frames (ORFs) with tight intergenic spaces and the same gene orientation. Homologs of V. splendidus VS_1324 and VS_1325 were not found in either V. furnissii or V. fluvialis. We designated 13 out of the 19 ORFs as VflT6SS1 tssA1-M1 (Type Six Secretion A1-M1) following the proposed nomenclature for T6SS components (Shalom et al., 2007; Figure 1). These showed 82–100% and 27–79% amino acid identity with the corresponding homologs of V. furnissii and V. splendidus, respectively. The VflT6SS1 cluster contains the T6SS hallmark gene, tssD (hcp), tssH (clpV/vasG) encoding an ortholog of the ClpB ATPase, which is considered important for T6SS function, and two additional genes, tssL (icmH/dotU/impK) and tssM (icmF/impL), encoding homologs of T4SS stabilizing proteins (Boyer et al., 2009). Of note, the hcp of VflT6SS1 had only 40% nucleotide identity to the homologs in V. cholerae (VC1415 and VCA0017). To gain insight into the genetic background of VflT6SS1, its flanking regions were examined. The flanking sequence of 85003 VflT6SS1 was identical to that of V. furnissii NCTC 11218. The left junction of the sequences encodes a putative ABC-type transport periplasmic component and a MFS transporter, and the right junction possesses the sensor histidine kinase, BaeS. However, no homologous gene cluster of VflT6SS1 was found in V. fluvialis strain ATCC 33809, a newly whole-genome sequenced V. fluvialis isolate from Bangladesh.

The cluster homologous to the V. furnissii vfu_B01176~vfu_B01191 genomic region was named VflT6SS2. It spans 21.22 kb and contains 16 ORFs which display the same organization and gene orientation as the “core” gene cluster of V. cholerae T6SS (VCA0107~VCA0124). The major characteristics of the VflT6SS2 gene cluster of V. fluvialis strain 85003 and its most closely related orthologs in other species are summarized in Table 3.

VflT6SS2 is located between the usp and rsbD genes in 85003, the same as in V. cholerae and V. furnissii, indicating its conserved location. Apart from the core gene cluster, V. cholerae T6SS has two copies of hcp-vgrG alleles encoded in two small auxiliary clusters (VCA0017~VCA0021 and VC1415~VC1421). In the 85003 draft genome, we only identified one copy of the hcp-vgrG homolog and the counterparts of vfu_B001191 and VCA0124 (vgrG3) from V. furnissii and V. cholerae, respectively, were missing in the originally aligned VflT6SS2 genetic locus. Sequence alignments showed that the two copies of hcp-vgrG modules exist in similar genetic locations in both V. furnissii and V. cholerae. Therefore, we reasoned that undiscovered hcp-vgrG copies likely existed in V. fluvialis 85003. V. fluvialis and V. furnissii are the genetically closest species among Vibrionaceae, with a mean BLASTp identity of 86.8% for the homologous genes (Lu et al., 2014). V. furnissii was originally regarded as an aerogenic biogroup of V. fluvialis and was later confirmed to be a separate species from V. fluvialis (Brenner et al., 1983). Using primer pairs vfu-vasL-up2153177/vfu-rbsD-dn2158250, vfuA01956-up/vfuA01959-dn, and vfuB01009-up/vfuB01011-dn, which were designed on the basis of the flanking sequences of V. furnissii vfu_B01191~vfu_B01192 and the two hcp-vgrG modules (vfu_A01957~vfu_A01958 and vfu_B01010~vfu_B01011, respectively), we amplified the corresponding regions of 85003 with long fragment PCR.

Primer pair vfu-vasL-up2153177/vfu-rbsD-dn2158250 yielded a 5093 bp fragment for V. fluvialis 85003. A sequence analysis of the fragment revealed a 3039 bp ORF which had 82 and 60% nucleotide identity to vfu_B001191 and VCA0123 (vgrG3), respectively. Additionally, a 384 bp ORF was predicted immediately downstream and has 4 bp overlap with the 3,039 bp ORF. This organizational feature was similar to VCA0123 and VCA0124, which were experimentally demonstrated to encode a vgrG3 effector and its immunity effector, TsiV3, in V. cholerae (Dong et al., 2013). Thus, we named the two ORF tssI2 and tsiI2, respectively. A sequence analysis and alignment of the 3,145 and 3,878 bp PCR products of primer pair vfuA01956-up/vfuA01959-up and vfuB01009-up/vfuB01011-dn indicated that each encodes a copy of the hcp-vgrG homolog, which was different from the originally recognized hcp-vgrG allele. In order to distinguish them, we designated the newly- identified alleles as tssD2_a-tssI2_a and tssD2_b-tssI2_b (corresponding to V. cholerae hcp1-vgrG1 and hcp2-vgrG2, respectively), and the original one was named tssD2_c-tssI2_c. The three copies of TssD (Hcp) show high identity (99.42%) in their nucleotide sequences and have 95.35% amino acid identity to the well-characterized Hcp protein of V. cholerae. TssI2_a and TssI2_b display higher homology with up to 88% amino acid identity. TssI2_c shows 60 and 61% identity to TssI2_a and TssI2_b respectively. The corresponding sequences of these three hcp-vgrG alleles were deposited in GenBank under accession numbers KY319185, KY319186, and KY319187, respectively.

In many species, T6SS gene clusters are located on pathogenicity islands or compositionally distinct regions of the genome indicating its horizontal acquisition and transfer. To better understand the gene content, sequence similarity, synteny, and the distribution of T6SS clusters in V. fluvialis, we carried out PCR screening of each T6SS locus in 34 different V. fluvialis isolates from both human and aquatic products from different provinces in China. The genetic loci of tssC1 (impC)-impD, tssE1 (impF)-tssF1 (impG), tssF1 (impG)-tssG1 (impH), tssH1, tssH1 (clpB)-tagH1 (impI), tagH1-tssK1 (impJ), and tssL1 (impK)-tssM1 (impL) within VflT6SS1 were all detected in seven out of 34 tested isolates (21% detection rate). The genetic loci of tssC2 (vipB)-tssF2 (vasA), tssG2 (vasB)-tagH2 (vasC), tssJ2 (vasD)-tssK2 (vasE), tssL2 (vasF)-tssH2 (vasG), vasH-vasI, vasI-tssA2 (vasJ), and tssM2 (vasK)-vasL for VflT6SS2 were detected in 100, 97, 97, 100, 91, 94, and 94% of the isolates, respectively. These results revealed that in general, VflT6SS2 is more prevalent than VflT6SS1 in V. fluvialis. The different detection rates of VflT6SS2 component gene loci may indicate that there is variation in the gene composition of in the T6SS2 cluster or sequence variation from the corresponding primer annealing regions in different isolates. These variations need to be investigated in greater detail in future studies. Since we found that VflT6SS2 is highly prevalent in V. fluvialis strains, we first focused on the functional characterization of VflT6SS2 in V. fluvialis 85003.

The hallmark of a functional T6SS is the presence of Hcp and VgrG in the culture supernatant. Therefore, we examined the Hcp secretion in V. fluvialis cultures. To this end, we first cloned and expressed hcp in pET30a with induction by IPTG. The purified Hcp protein was used to prepare anti-Hcp polyclonal antiserum in rabbits, because there is no commercially-available antibody against this protein.

To determine whether VflT6SS2 was functionally expressed, clinical isolates 85003, VF42, and CICC21612, and environmental isolate VF54, all of which contain the complete VflT6SS2 based on a PCR screen, were routinely cultured in LB with 170 mM NaCl at 30°C. The growth curves of VF42, VF54, and CICC21612 displayed consistent growth trends before entering a stationary growth phase, while 85003 grew slowly relative to these other strains (Figure S1). However, all four strains were in the logarithmic growth phase at an OD600 of 1.5. The cell pellets and culture supernatants of each strain at an OD₆₀₀ of 1.5 were examined by immunoblot analyses using polyclonal anti-Hcp antiserum. As shown in Figure 2, a large amount of Hcp was detected in both the cell pellets and culture supernatants of 85003, CICC21612, and VF54, indicating the functional expression of T6SS2 in these V. fluvialis strains. Of note, the absence of the cytosolic Crp in the supernatants demonstrated that the bacteria were intact, and thus, Hcp was being actively secreted and not released by cell lysis. This control was routinely used in the subsequent Hcp secretion analysis. For strain VF42, Hcp was only detected in the cell pellets, indicating that, though Hcp was synthesized in these bacteria, it was not efficiently secreted into the supernatant. Taken together, our results demonstrated that functional T6SS exists in at least some V. fluvialis isolates.

It's should be noted that the apparent molecular weight of Hcp in the Western blot analysis (around 25 kDa) is quite different from the predicted size according to its ORF sequence (19 kDa). This inconsistency has also been observed in V. cholerae (Ishikawa et al., 2009) and V. alginolyticus (Sheng et al., 2012), and possible modification of two cysteine residues of Hcp at positions 5 and 22 was suggested to be responsible for the difference (Williams et al., 1996).

Following the confirmation of the functional expression of VflT6SS2 in V. fluvialis, we tested the effects of the bacterial growth phase on the expression and secretion of Hcp. Strain 85003 was grown in LB medium containing 170 mM NaCl at 30°C, and the protein levels of Hcp in the cell pellets and culture supernatants were measured at different growth points (i.e., OD₆₀₀ 0.2, 0.5, 1.0, 1.5, 2.0, 2.5, 3.8, and overnight; Figures 3A,B). Hcp expression was detected in the cell pellets at all growth points except overnight, and the expression level at an OD₆₀₀ of 0.2 was obviously lower than that of the other growth points (Figure 3B, lane 1). However, the secretion of Hcp was only detected at OD₆₀₀ values of 1.0, 1.5, and 2.0, and the highest secretion level was observed at an OD₆₀₀ of 1.5 (Figure 3B, lanes 11–13). Our results indicated that, although the Hcp was continuously synthesized at various OD₆₀₀ values (except overnight culture), the active T6SS was only detected during a narrow window of the growth phase. It appears that when the cell density reaches an OD₆₀₀ of 2.5, the secretion of Hcp stops. At the later stationary phase after overnight incubation, when no Hcp was detected, the expression of hcp may have been turned off or Hcp may have been proteolytically degraded.

Like many species of Vibrios, V. fluvialis can survive under different environmental conditions, from the aquatic milieu, such as seas, estuaries, and brackish waters, to the human intestine. We wondered whether the temperature affects the function of VflT6SS2 in V. fluvialis. Therefore, we cultured the 85003 strain at different temperatures (25°, 30°, or 37°C) and compared the expression and secretion of Hcp. Although 85003 grows faster at 37°C than at 25° and 30°C (Figure S2), the OD₆₀₀ of 1.5 lies at the same log growth phase. Hcp was expressed and secreted at both 25°C and 30°C, but not at 37°C (Figure 4A). The secretion of Hcp at 30°C was higher than that at 25°C (Figure 4A, compare lane 4 with lane 5). Our results revealed that VflT6SS2 is active under cool and warm temperature, similar to T6SS2 in V. parahaemolyticus, which is most active at cold (23°C) and warm (30°C) temperatures under low salt conditions (Salomon et al., 2013), suggesting that VflT6SS2 may play a role in the environmental fitness of V. fluvialis.

To explore the molecular mechanism(s) responsible for the undetectable protein level of Hcp at 37°C, we measured the transcription levels of tssD2 (hcp), together with three selected VflT6SS2 core genes [tssL2 (vasF), vasH, and tssM2 (vasK)] by qRT-PCR using cultures grown at 30° and 37°C. As shown in Figure 4B, the tssD2 (hcp) mRNA level was dramatically decreased (more than 150-fold) at 37°C compared to that at 30°C. However, no considerable alterations were observed in the mRNA levels of tssL2 (vasF), vasH, and tssM2 (vasK) between 30° and 37°C, suggesting that the loss of function of VflT6SS2 at 37°C is mainly due to the extremely low transcription level of the tssD2 (hcp) gene at this temperature.

As V. fluvialis is a halophilic species of Vibrio and normally resides in coastal and estuarine environments, we investigated if the salinity influences the expression and secretion of Hcp in V. fluvialis. We cultured strain 85003 at 30°C to an OD₆₀₀ of 1.5 in LB containing increasing concentrations of NaCl. We observed that comparable amounts of Hcp were detected in the cell pellets regardless of the salt concentration (Figure 5A, left panel). Significant amounts of secreted Hcp were readily detected in supernatants from cultures containing 170, 340, 513, or 855 mM NaCl, but not in supernatants from cultures containing 85 mM NaCl (Figure 5A, right panel, compare lane 6–lanes 7–10), although the secretion levels at 513 and 855 mM NaCl appeared to be slightly reduced compared to those at 170 and 340 mM NaCl. These results indicated that Hcp secretion was stimulated by high NaCl concentrations. To obtain more detailed information, we investigated whether salinity can influence the expression of tssD2 and the VflT6SS2 core gene cluster. As shown in Figure 5B, tssD2 and the all three selected VflT6SS2 core genes, tssL2, vasH, and tssM2, were apparently induced when strain 85003 was grown at 30°C in LB containing 340 mM NaCl compared to cultures grown with 85 mM NaCl.

We then determined whether the salinity could also affect VflT6SS2 at 37°C. We therefore cultured the 85003 strain at 37°C in LB containing different concentrations of NaCl and analyzed the expression and secretion of Hcp. We found that steadily increasing amounts of Hcp were detected in pellets from cultures supplemented with 340, 513, or 855 mM NaCl (Figure 5C, lanes 3–5), and the secreted form of Hcp was detected under 513 or 855 mM NaCl conditions (Figure 5C, lanes 9–10). On the basis of these results, we inferred that high NaCl concentrations could induce both the expression and secretion of Hcp in a dose-dependent manner at 37°C. Consistently, increased mRNA levels of tssD2, tssL2, vasH, and tssM2 were confirmed when the 85003 strain was grown at 37°C in LB containing 855 mM NaCl (Figure 5D). These results indicated that, under certain circumstances, such as high salinity, VflT6SS2 would be activated in an infected host. The precise stimuli and signaling mechanism(s) involved in this response should be further investigated in future studies.

In V. cholerae A1552, it was reported that Hcp secretion was induced by osmolarity rather than salinity (Ishikawa et al., 2012). To test whether this is also the case in V. fluvialis, strain 85003 was grown at 30°C in LB supplemented with mannitol, trehalose or sucrose at a concentration of 510 mM, or KCl at a concentration of 255 mM, in addition to 85 mM NaCl to yield an equivalent osmolarity as 340 mM NaCl. Based on the growth curve (Figure 6A), we collected samples from cultures with an OD₆₀₀ of 1.5. Our results demonstrated that although the protein levels of Hcp in pellets were comparable under low and high osmolarity conditions (Figure 6B, left panel), the protein was more efficiently secreted under high osmolarity conditions (Figure 6B, compare lane 7 with lanes 8–12). Furthermore under the same osmolarity conditions, NaCl, mannitol and trehalose seemed to more potently enhance the secretion of Hcp than sucrose and KCl (Figure 6B, compare lanes 8, 9, 11 with lanes 10 and 12). Together, these results indicate that high osmolarity induces Hcp secretion in V. fluvialis, and the osmolarity-dependent secretion of Hcp may be a common characteristic among Vibrio species.

To test if VflT6SS2 activity contributes to interbacterial virulence or competition, we performed an E. coli killing assay. We first generated a vasH in-frame deletion mutant, ΔvasH, which was used as an avirulent control. VasH is considered to be a global transcriptional regulator of T6SS and is required to initiate the transcription of T6SS genes in V. cholerae (Kitaoka et al., 2011). The VasH of V. fluvialis encodes 531 amino acids and shares 73% identity with that of V. cholerae. Of note, the Walker A (GETGTGKE) motif, Walker B (GTLFLDEIG) motif, and the central GAFSGA-loop 1 region of the ATPase domain of VasH are exactly the same in V. cholerae and V. fluvialis (Figure S3). Consistent with the findings in V. cholerae, the deletion of vasH in V. fluvialis resulted in a lack of expression and secretion of Hcp (Figure 7A, compare lane 1 to lane 2, and lane 7 to lane 8), i.e., the naturally functional VflT6SS2 was completely disabled. However, the expression and secretion of Hcp were restored by introducing a complemented plasmid pSRvasH under IPTG induction, but the restoration did not occur if the pSRKTc control plasmid was used or no IPTG induction was employed (Figure 7A). The mixtures of the prey E. coli K-12 MG1655 and V. fluvialis predators were spotted on LB nutrient agar plates containing 340 mM NaCl and incubated at 30°C. The number of surviving MG1655 was determined after 5 h of incubation. The results showed that the survival of MG1655 was significantly reduced in the presence of wild-type V. fluvialis, but not when vasH was deleted (Figure 7B). Re-expression of vasH from pSRvasH recovered the ability of V. fluvialis to compete against the E. coli, although not to the level of the wild type (Figure 7B). The above results collectively indicate that VflT6SS2 provides V. fluvialis with increased competitive fitness.

To determine whether VflT6SS1 is involved in the competitive bacterial killing demonstrated by V. fluvialis, we measured the protein level of TssD1, as well as the mRNA levels of three VflT6SS1 genes located at the beginning of the VflT6SS1 gene cluster (rhs, tssA1, tssD1), and two other genes at its end (tssK1 and tssM1) using qRT-PCR for bacteria cultured under the same conditions as in the killing assay (340 mM NaCl at 30°C). We could not detect TssD1 expression in either the cell pellet or the supernatant by a Western blot analysis (data not shown). qRT-PCR analyses revealed that the genes selected from the VflT6SS1 cluster showed extremely low mRNA levels (0.015, 0.063, 0.016, 0.003, and 0.004 for tssM1, tssD1, tssK1, tssA1, and rhs, respectively) compared to vasH and tssD2 from the VflT6SS2 cluster (0.199 and 2.124 for vasH and tssD2, respectively; Figure 7C). These results suggested that VflT6SS1 is likely inactive, at least under the tested conditions.

Since VflT6SS2 contains three copies of tssD2 (namely tssD2_a, tssD2_b and tssD2_c), we generated individual deletion mutants and a tssD2_ab double-deletion mutant to pinpoint which copies contribute to the function of VflT6SS2. As shown in Figure 8A, deletion of the individual copies (tssD2_a, tssD2_b or tssDI2_c) did not affect the expression or secretion of Hcp, while the combined deletion of tssD2a and tssD2b completely abolished the secretion of Hcp. Consistently, each deletion mutant of tssD2_a, tssD2_b or tssDI2_c showed similar interbacterial virulence to the wild type organism, but the double mutant (ΔtssD2ab) displayed significantly decreased antibacterial virulence (Figure 8B). Based on these results, we speculated that tssD2_a and tssD2_b function redundantly and are required for VflT6SS2 to mediate its antibacterial activity, while tssD2_c is probably dispensable for the formation of VflT6SS2 structural apparatus or the assembly of functional VflT6SS2.

Taken together, these results demonstrate that VflT6SS2 provides an advantage for V. fluvialis in competition with bacterial neighbors, which may contribute to its fitness and pathogenesis.

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.