Introduction

Vibrio parahaemolyticus, a common resident in marine and estuarine environments (Daniels et al. 2000), infects humans and typically results in gastroenteritis (Nair et al. 2007). In addition to a well-known virulence factor of V. parahaemolyticus, thermostable direct hemolysin (TDH) (Honda and Iida 1993), the type III secretion system (T3SS), is considered to be an important virulence factor for delivering effectors into host cells (Park et al. 2004). Genome sequencing of V. parahaemolyticus RIMD2210633 has revealed the presence of two T3SS, T3SS1 and T3SS2 (Makino et al. 2003). Roles of these T3SSs were examined using T3SS1- or T3SS2-deficient V. parahaemolyticus strains, in which the vcrD1 or vcrD2 gene, respectively, was mutated. The phenotypes of these mutants indicated that T3SS1 plays a role in cytotoxicity toward tissue culture cells, whereas T3SS2 is associated with bacterial enterotoxicity in the rabbit ileal loop model (Makino et al. 2003; Park et al. 2004). A T3SS1-deficient V. parahaemolyticus strain was also defective in inducing the death of HEp-2 cells, which occurs via a MAPK-dependent (p38 and ERK1/2) and caspase-independent mechanism (Yang et al. 2011). Expression of T3SS1 genes is induced when V. parahaemolyticus is cultivated in tissue culture medium in which ExsA and ExsD play roles as a positive and negative regulators, respectively (Zhou et al. 2008).

In this study, several polar flagellar proteins were secreted at lower levels in the media grown by the vcrD1 mutant V. parahaemolyticus than the culture media of the wild-type strain. The secretion system for bacterial flagellar components belongs to the T3SS superfamily (Macnab 2004). While the T3SSs responsible for secreting effectors are defined as non-flagellar (NF) T3SSs, they are thought to have evolved from flagellar T3SSs (Abby and Rocha 2012). Interestingly, V. parahaemolyticus has two flagellar systems, i.e., a polar sheathed flagellum and lateral unsheathed flagella (reviewed by McCarter 1999; reviewed by Duan et al. 2013). The polar flagellum functions constitutively and is responsible for bacterial swimming. On the other hand, synthesis of lateral flagella is induced under a specific condition such as growth on solid surfaces (Belas et al. 1986).

Approximately sixty genes are involved in formation of the polar flagellum in V. parahaemolyticus (Kim and McCarter 2000). Most of the polar flagellar genes are found in two locations of the larger chromosome of V. parahaemolyticus with exceptions of several motor genes, motAB, motX, and motY (McCarter 2001). Transcription of these flagellar genes is temporally regulated in a tight coordination with the orders of flagella assembly (Kim and McCarter 2000). These flagellar genes are grouped into three hierarchies according to their temporal orders of transcription during flagella formation. The early genes, flaK and flaLM, are master regulators, which interact with σ54 and play a role in transcription of the middle genes. Interestingly, flaK mutant V. parahaemolyticus was found to retain the swimming motility until the bacteria had an additional defect in formation of lateral flagella, indicating the presence of cross-regulation between these two flagellar systems of V. parahaemolyticus (Kim and McCarter 2004). The middle flagellar genes transcribed by σ54 encode proteins involved in assembly of the hook-basal-body structure, HAP1, HAP3, MotY, some chemotaxis proteins, and an alternative sigma factor, σ28. Sigma factor 28 then transcribes the other set of flagellar genes, the late flagellar genes, which encode additional motor proteins, additional chemotaxis proteins, 5 flagellins, a distal capping protein HAP2, and the anti-σ28 FlgM.

Besides flagellar T3SSs (Kim and McCarter 2000), V. parahaemolyticus possesses two NF T3SSs (T3SS1 and T3SS2) to convey its virulence factors into the host cells (Park et al. 2004). In this study, we further characterized the role of NF-T3SS1 in polar flagella formation by examining the altered expression and secretion of flagellar components in a ΔvcrD1 mutant.

Materials and methods

Bacterial strains, plasmids and culture conditions

The bacterial strains and plasmids used in this study are listed in Table 1. Escherichia coli strains used to prepare plasmid DNA and transfer of the plasmid by conjugation were grown in Luria–Bertani (LB) broth (1 % bacto-tryptone, 0.5 % yeast extract, and 1 % NaCl) or on LB plate containing 1.5 % agar. V. parahaemolyticus RIMD2210633 (ATCCBAA-238; American Type Culture Collection, Manassas, VA, USA) was used as the wild-type strain in this study and was cultured in LB medium supplemented with an additional 2 % NaCl (LBS). Ampicillin was added to the medium at 100 μg/ml to maintain the plasmids in E. coli. Chloramphenicol (2 μg/ml) was used to select V. parahaemolyticus exconjugants. For V. parahaemolyticus carrying pRK415 or pRK415-derived plasmid, tetracycline was added into the medium at a concentration of 3 μg/ml. All medium components were purchased from Difco (Lawrence, KS, USA), and the chemicals and antibiotics were from Sigma (St. Louis, MO, USA).

Table 1 Bacterial strains and plasmids used in this study
Full size table

Preparation of secreted proteins from V. parahaemolyticus

Secreted proteins were prepared from wild-type and ΔvcrD1 V. parahaemolyticus strains cultivated in LBS broth at 37 °C. In addition, secreted proteins were harvested from these strains grown in Dulbecco’s modified eagle medium (DMEM) as an inducing condition for T3SS1 as reported (Zhou et al. 2008). Once bacterial cells grew to an optical density at 600 nm (OD600) of 0.8, they were centrifuged at 8,000 rpm for 15 min at 4 °C and the resultant supernatants were passed through a 0.2-μm pore membrane filter. Proteins were concentrated using a Centricon YM-10 (Millipore, Bedford, MA, USA) at 4 °C. Filtered supernatants were mixed with an equal volume of pre-chilled 10 % trichloroacetic acid (TCA) and incubated on ice for 30 min. After a 30-min centrifugation at 9,000 rpm and 4 °C, the pellets were washed with acetone three times, dried, and resuspended in lysis buffer (50 mM NaH2PO4, 300 mM NaCl, and 10 mM imidazole, pH 8.0). The protein content in the secretome preparations was determined with a Protein Assay kit (Bio-Rad, Hercules, CA, USA).

2D gel electrophoresis of secreted proteins

Secreted proteins (wild-type and ΔvcrD1 mutant V. parahaemolyticus) were subjected to isoelectric focusing with the IPGphor system and Immobiline Drystrip gel strips (Life Sciences, New York, USA). Briefly, 200-μl aliquots of sample protein in rehydration solution were applied to the strips (pH 4–7, 18 cm) and rehydrated for 16 h at room temperature. After rehydration, the proteins were focused for 36 kV/h at 20 °C (100 V for 2 h, 500 V for 1 h, 1,000 V for 1 h, 2,000 V for 1 h, 4,000 V for 1 h, 6,000 V for 2 h, and 8,000 V for 2 h). The strips were then soaked in sodium dodecyl sulfate (SDS) equilibrium buffer containing 6 M urea, 2 % SDS, 5× gel buffer (pH 8.8), 50 % glycerol, and 2.5 % acrylamide for 15 min. The strips were transferred onto 10 % acrylamide gels and subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE). The gels were stained with Coomassie brilliant blue G-250 containing 17 % ammonium sulfate, 3 % phosphoric acid, 34 % methanol, and 0.1 % Coomassie blue G-250 in distilled water.

In-gel trypsin digestion and MALDI-TOF mass spectrophotometry

Protein spots were selected, which were at decreased levels in the secretome of ΔvcrD1 mutant (below to 33 % of the wild type), but present in considerable amounts pertinent to further analysis. Protein spots of interest were excised and washed three times with distilled water. The excised spots were incubated in destaining solution (15 mM potassium ferricyanide, 50 mM sodium thiosulfate) and washed with 25 mM ammonium bicarbonate/50 % acetonitrile until the Coomassie blue G-250 dye disappeared. After drying in a vacuum concentrator, the gel was rehydrated with 5 μl trypsin (12.5 ng/μl in 50 mM NH4HCO3) at 4 °C for 45 min and incubated overnight at 37 °C. The tryptic peptides were extracted with 60 % acetonitrile and 0.1 % trifluoroacetic acid (TFA), and dried with a vacuum concentrator. The resultant peptide mixtures were dissolved in 0.5 % TFA. Using a saturated solution of α-cyano-4-hydroxycinnamic acid in 0.1 % TFA/50 % acetonitrile as a matrix, mass spectrometry was performed on a MALDI-TOF/TOF mass spectrometer (Applied Biosystems, Foster City, CA, USA). The Mascot program (http://www.mtrixsceince.com) was used to search the Swiss-Prot and NCBInr databases.

Preparation of polyclonal antibodies

The DNA encompassing the open reading frame (ORF) of each selected protein (FlgL, FlgM, and OmpV) was amplified using a pair of specific primers listed in Table 2 and then cloned into a pET21b (+) expression vector (Novagen, Darmstadt, Germany). The recombinant candidate protein was overexpressed in E. coli BL21 (DE3) by adding 1 mM isopropyl thio-β-galactosidase and then separated by 12 % SDS-PAGE. The induced protein band was excised and resuspended with phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, and 2 mM KH2PO4, pH 7.4). This protein was then injected intraperitoneally into pathogen-free rats (CrjBgi: CD[S.D.]IGS, 6-week-old, female) as an antigen to produce polyclonal antibodies. After three injections at 2-week intervals, serum was obtained from the immunized rats, and the resultant antibody titers were tested.

Table 2 Primers used in this study
Full size table

Western blot analysis

Bacterial extracts were prepared by sonicating bacterial cells in TNT buffer (10 mM Tris–HCl, pH 8.0, 150 mM NaCl, and 0.05 % Tween 20). Protein in the culture supernatants was concentrated 200-fold by precipitation with TCA or filtration through Amicon/Centricon YM-10 columns (Millipore, Billerica, MA, USA). Cell lysates and concentrated supernatants in sample buffer (50 mM Tris–HCl, pH 6.8, 100 mM dithiothreitol, 2 % SDS, 0.1 % bromophenol blue, and 20 % glycerol) were separated by SDS-PAGE and transferred to nitrocellulose membranes (Millipore). For each extract, two different amounts of proteins were analyzed by Western blot. Membranes were blocked with 5 % skim milk in Tris-buffered saline with Tween 20 (150 mM NaCl, 50 mM Tris–HCl, and 0.1 % Tween 20) and incubated overnight at 4 °C with polyclonal antibodies (1:5,000 dilution). Horseradish peroxidase-conjugated secondary antibodies were used. Immunoreactive bands were visualized using an enhanced chemiluminescence system (Amersham, Buckinghamshire, United Kingdom).

In addition to antibodies against candidate proteins, secretion of FlgE and FlaA was monitored using antibodies specific to homologous proteins from V. vulnificus (Lee et al. 2004; Lee and Kim unpublished result, respectively). Antibodies specific to TDH (Yang et al. 2011) were included in Western blot analyses as a loading control for each secretome.

Construction of deletion mutants of V. parahaemolyticus

For construction of ΔexsA mutant, the downstream region of the exsA gene was amplified from the genomic DNA of V. parahaemolyticus RIMD2210633 with the primers, exsAdownF and exsAdownR (Table 2). The resultant 792-bp DNA fragment was then digested with PstI and SacI, and ligated into pBlueScript (II) SK (+) (Stratagene, La Jolla, CA, USA) to produce pSKexsAD. The upstream region of the exsA gene was amplified using the primers, exsAupF and exsAupR (Table 2). The resultant DNA fragment of 747 bp was treated with ApaI and PstI and ligated into pSKexsAD to yield pSKexsAUD. The 1,539-bp ApaI–SacI DNA fragment of pSKexsAUD was transferred into a suicide vector pDM4 (Miller and Mekalonos 1988), resulting in formation of pDMexsAUD. The plasmid pDMexsAUD in SM10 λpir (Simon et al. 1983) was mobilized to V. parahaemolyticus RIMD2210633 or ΔvcrD1 mutant, and the conjugants were selected by plating the conjugation mixture of E. coli and V. parahaemolyticus on LBS plates supplemented with 2 μg/ml chloramphenicol. A colony with characteristics indicating a double homologous recombination event (resistance to 5 % sucrose and sensitivity to chloramphenicol) was further confirmed by PCR using the primers, exsAupF and exsAdownR, and then named MJ21 or MJ22 for the ΔexsA or ΔexsAΔvcrD1 mutant strain, respectively.

A mutant V. parahaemolyticus losing both FlaK and LafK, key regulators for polar flagellum and lateral flagella, respectively (Kim and McCarter 2004), was constructed and included for motility assays as a non-motile control. For construction of the ΔflaK mutant, the upstream (766-bp) and downstream (647-bp) regions of the flaK gene were amplified using the primer set of flaKupF/flaKupR and flaKdownF/flaKdownR, respectively (Table 2). The XhoI–SpeI DNA fragment of pSKflaKUD was transferred into pDM4 to produce in pDMflaKUD, which was then used to generate ΔflaK mutant as described above. A plasmid (pSKlafKUD) includes the upstream (750 bp) and downstream (640 bp) regions of the lafK gene, which have been amplified by the following primer sets, lafKupF/lafKupR and lafKdownF/lafKdownR (Table 2). The ApaI–SacI DNA fragment of the resultant plasmid was ligated into pDM4 to produce pDMlafKUD, which was used to make the ΔflaKΔlafK mutant, as described above.

Motility assay

Wild-type and mutant V. parahaemolyticus strains (ΔvcrD1, ΔexsA, ΔexsAΔvcrD1, and ΔflaKΔlafK mutants) were freshly grown in LBS at 37 °C with aeration to an OD600 of 0.8, washed, and resuspended in PBS to a final concentration of 106 colony forming units/ml. The cell suspension was inoculated into LBS medium containing 0.3 % agar and incubated for 4 h at 37 °C. Motility of the vcrD1 mutant V. parahaemolyticus carrying the complementation plasmid, pRKvcrD1 (Yang et al. 2011) was also examined in the same manner along with two control strains (wild type carrying pRK415 and vcrD1 mutant carrying pRK415). The degree of bacterial motility was quantified by measuring the diameters (mm) of the motility halos made by these V. parahaemolyticus strains. Data are presented as the means ± standard deviation from three independent experiments.

Transmission and scanning electron microscopy

Wild-type and ΔvcrD1 mutant strains were examined for the presence of a flagellum. Bacterial cells were negatively stained with 2 % uranyl acetate (pH 7.4) on a Formvar carbon-coated grid and observed with a transmission electron microscope (CM-10; Philips, Amsterdam, Netherlands) at 75 kV.

Vibrio parahaemolyticus strains for the complementation study (wild type carrying pRK415, vcrD1 mutant carrying pRK415, and vcrD1 mutant carrying pRKvcrD1) were also examined for the presence of flagellum using a scanning electron microscope (S-3000 N, Hitachi, Tokyo, Japan) at 30 kV. Before examination, the specimens were coated with an electro-conductive layer of gold, which was evaporated by an ion sputter coater (E101, Hitachi).

Fractionation of bacterial cells

Wild-type and ΔvcrD1 V. parahaemolyticus strains were grown to OD600 = 0.8 at 37 °C in LBS broth. Through centrifugation, the culture supernatant was separated from the bacterial cells. The harvested cells were divided into cytoplasmic and periplasmic fractions as described (Manoil and Beckwith 1986). Harvested cells were resuspended in a spheroplast buffer (100 mM Tris–HCl, pH 8.0, 0.5 mM EDTA, 0.5 mM sucrose, 1 mM MgCl2, and 20 μg/ml phenylmethylsulfonyl fluoride) and subjected to centrifugation (3 min, 13,000 rpm, 4 °C). The pellets were warmed, resuspended in ice-cold water, and treated with 20 mM MgCl2. The osmotic shocked-cells were centrifuged (5 min, 13,000 rpm, 4 °C). The supernatants were saved as periplasmic fractions, and the pellets were further resuspended in PBS with 1 % Triton X-100. The cells in the pellet were lysed by sonication. After centrifugation, the resultant supernatant was saved as the cytoplasmic fraction.

The culture supernatant, periplasmic fraction, membrane fraction, and cytoplasmic extract of each strain were analyzed by Western blot using anti-FlaA antibodies or anti-FlgM antibodies. Antibodies against glucose-specific enzyme IIAGlc (Lee and Lee unpublished result) and anti-OmpV antibodies were used as a loading control for cytoplasmic and membrane fractions, respectively. Anti-TDH antibodies were used as a loading control for the secretome.

Quantitation of flagellar mRNAs

The degree of expression of seven flagellar mRNAs was evaluated in wild-type and ΔvcrD1 mutant V. parahaemolyticus by real-time PCR (RT-PCR). V. parahaemolyticus was cultured in LBS broth at 37 °C with shaking. At OD600 = 0.8, the bacteria cells were centrifuged at 8,000 rpm for 5 min at 4 °C. Total RNAs were isolated from harvested cells using TRizol (BioRad) and then treated with the RNase-free DNase (Qiagen, Hilden, Germany). After amount of RNA was measured using NanoDrop ND-1000 (Thermo Fisher Scientific, Waltham, MA, USA), cDNA was synthesized from 4 μg of RNA using PrimeScript RT reagent kit (TaKaRa, Shiga, Japan) as directed by the manufacturer. The cDNAs were analyzed by quantitative RT-PCR (qRT-PCR) on a Light Cycler 480 II real-time PCR system (Roche Applied Science, Mannheim, Germany) using Light Cycler 480 DNA SYBR green I master kit (Roche Applied Science). The qRT-PCR was carried out in a 96-well plate. The amount of each transcript was estimated using the specific pair of primers (Table 2). Each experiment was performed in triplicate for each sample and repeated three times with independent RNA samples. The gap gene of V. parahaemolyticus was used as an endogenous control to normalize each transcript in wild-type and ΔvcrD1 mutant cDNAs. Data analysis was based on the relative quantification method by determining the crossing point (C p) value using the Light Cycler 480 II real-time PCR system software program (Roche Applied Science, version LSC480 1.5.0.39).

Statistical analysis

Data are presented as the means ± standard deviation from three independent experiments. Statistical analyses were performed using Student’s t test (SYSTAT, SigmaPlot version 11; Systat Software Inc. Chicago, IL, USA) to evaluate the statistical significance of the results. Differences were considered significant when P < 0.05. Data with P < 0.01 are indicated with two asterisks, whereas data with P values between 0.01 and 0.05 are indicated with a single asterisk.

Results

Identification of extracellular proteins via T3SS1 in culture supernatants of V. parahaemolyticus

To identify the proteins secreted via the T3SS1 apparatus, 2D-gel electrophoresis (2DGE) was performed with bacterial culture supernatants of the wild-type and the ΔvcrD1 mutant V. parahaemolyticus strains, and their proteomic patterns were compared. Culture supernatants were prepared from bacteria grown in two different media, LBS (Fig. 1a) and DMEM (Fig. 1b). Among proteins that were present at lower levels in the mutant supernatant (at least threefold less), protein spots in amount pertinent to further analysis were selected. Nineteen protein spots were identified and are listed in Table 3 in case of LBS-grown cells. Fifteen protein spots were chosen for further analysis in the supernatants derived from the DMEM-grown bacteria (Table 4).

Fig. 1
figure 1

Comparative analysis of secreted proteins using 2D gel electrophoresis of culture supernatants of wild-type and ΔvcrD1 V. parahaemolyticus strains. Bacteria were incubated in LBS broth (a) or DMEM (b) at 37 °C. Secreted proteins were analyzed by 2D-gel electrophoresis, including isoelectric focusing (pH 4–8) and SDS-PAGE. The resultant gels were stained with Coomassie brilliant blue G-250. Numbered proteins were identified by MALDI-TOF mass spectrometry. Identity of these proteins is listed in Tables 3 and 4 for LBS-secretome and DMEM-secretome, respectively

Full size image
Table 3 Proteins with diminished secretion in the vcrD1 mutant compared with wild type (cultivated in LBS)
Full size table
Table 4 Proteins with diminished secretion in the vcrD1 mutant compared with wild type (cultivated in DMEM)
Full size table

In case of secretome prepared from bacterial cells grown in LBS, a majority of proteins secreted less in the ΔvcrD1 mutant were associated with parts of the polar flagellum: FliK, FlaH, FlgK, and FlgL, which are in the flagellar hook; FlgD, which is a cap protein of a flagellar hook; and FlaA, FlaD, and FlaG, which are flagellins. This result suggests that VcrD1 plays a significant role in the assembly and development of a polar flagellum of V. parahaemolyticus. Earlier studies reported that T3SS is evolutionarily related to the secretion system for bacterial flagella, flagellar T3SS, through sharing structural and functional features (Jorges 2008). The other proteins are enzymes, including a chitodextrinase, chitinases, and an alkaline serine protease. Two outermembrane proteins, OmpA and OmpV, were also secreted at lower levels in the ΔvcrD1 mutant. Finally, six hypothetical proteins without any known function were present at lower levels in the secretome of the mutant.

Comparative proteomic analysis on the secretomes derived from V. parahaemolyticus grown in DMEM also demonstrated that secretion of the polar flagellin FlaA protein was affected in ΔvcrD1 mutant (Table 4). In addition, an outermembrane protein OmpA and chitinase were found at lower levels in secretome of ΔvcrD1 mutant than that in wild type. Unlike secretomes derived from LBS-grown cells, several metabolic enzymes for amino acids were secreted less from the ΔvcrD1 mutant than wild type. Interestingly, some transport proteins were found at lower levels in the mutant secretome.

Confirmation of lower secretion of the flagellar proteins from the ΔvcrD1 mutant by Western blot

Two of the eight flagellar proteins identified by 2DGE (Tables 3, 4), FlaA and FlgL, were confirmed for their low level in LBS-secretome of the ΔvcrD1 mutant using their specific polyclonal antibodies. A 40-kDa immunoreactive band was present at a lower level in secretome of the vcrD1 mutant than in the wild type (Fig. 2a). The ΔvcrD1 mutant secretome also contained less FlgL than did the wild-type secretome (Fig. 2b). We also examined secretion of an additional flagellar protein, FlgE, in the culture supernatants of wild type and the ΔvcrD1 mutant by Western blot using antibodies against V. vulnficus FlgE (Lee et al. 2004; Fig. 2c). Similar to FlaA and FlgL, the vcrD1 mutant secreted less FlgE than did the wild type.

Fig. 2
figure 2

Western blot analysis of LBS-secretomes of wild-type and ΔvcrD1 mutant strains using polyclonal antibodies against four selected flagellar proteins. Two different amounts of proteins (5–15 μg) were used for Western blot analysis. Western blots using anti-FlaA antibodies (a), anti-FlgL antibodies (b), anti-FlgE antibodies (c), anti-FlgM antibodies (d), and anti-TDH antibodies as a loading control (e). Lanes 1 and 3, wild-type secretomes; lanes 2 and 4, ΔvcrD1 secretomes

Full size image

Lower levels of three flagellar proteins in the vcrD1 mutant suggest that overall expression of flagellar proteins is decreased in the ΔvcrD1 mutant. In the next experiment, secretion of FlgM was examined because it is an anti-σ28 controlling the temporal expression of flagellar genes, especially late genes (Kutsukake and Iino 1994). Thus, accumulation of FlgM in the cytoplasm of the vcrD1 mutant affects the expression of downstream flagellar genes. However, Western blot analysis of LBS-culture supernatants demonstrated that less FlgM was present in the culture supernatant of the ΔvcrD1 mutant (Fig. 2d). The TDH levels in wild-type and vcrD1 mutant secretomes were evaluated by Western blot with anti-TDH antibodies as a loading control.

Both wild-type and mutant secretomes were prepared from V. parahaemolyticus incubated in DMEM and also analyzed by Western blot using these anti-flagellar protein antibodies (Fig. 3). All of the four flagellar proteins, FlaA, FlgE, FlgL, and FlgM, were present at lower levels in the mutant secretome than wild-type secretome indicating that secretion of flagellar proteins was also affected by the vcrD1 mutation in DMEM like the LBS-cultivated cells.

Fig. 3
figure 3

Western blot analysis of DMEM-secretomes of wild-type and ΔvcrD1 mutant strains using polyclonal antibodies against four selected flagellar proteins. Two different amounts of proteins (4 and 8 μg) were used for Western blot analysis using anti-FlaA antibodies, anti-FlgE antibodies, anti-FlgL antibodies, anti-FlgM antibodies, and anti-TDH antibodies. Lanes 1 and 3, wild-type secretomes; lanes 2 and 4, ΔvcrD1 secretomes. Nonspecific protein is indicated with an asterisk

Full size image

Role of vcrD1 in bacterial flagella/motility of V. parahaemolyticus

Inoculation of the bacterial suspensions into LBS medium containing 0.3 % agar showed that the ΔvcrD1 mutant strain did not have any motility, whereas wild-type V. parahaemolyticus had a distinct motile phenotype with a large diffuse spreading halo (Fig. 4a, b). Reduced diameter of the motility halo of the ΔvcrD1 mutant was not caused from its decreased growth as shown in the similar growth curves of wild type and vcrD1 mutant (Fig. S1).

Fig. 4
figure 4

Bacterial motility and polar flagellum of wild-type and ΔvcrD1 mutant strains of V. parahaemolyticus. a Swimming motility of various V. parahaemolyticus strains (wild type, ΔvcrD1 mutant, ΔexsA mutant, and ΔvcrD1ΔexsA mutant) was measured in LBS/0.3 % agar plates for 4 h. A double ΔlafKΔflaK mutant was also included in this motility assay as a non-motile control. b Quantification of bacterial motility by measuring the diameters (mm) of the motility halos made by these V. parahaemolyticus strains. Data are presented as the means ± standard deviation from three independent experiments. Statistical analyses were performed using Student’s t test to evaluate the statistical significance of the results. Differences were considered significant when P < 0.05. Data with P < 0.01 are indicated with two asterisks. c Transmission electron micrographs of wild-type and ΔvcrD1 mutant strains negatively stained with uranyl acetate. The bar represents 2 μm

Full size image

The next experiment was to examine a possibility whether the loss of VcrD1 protein causes defects in various secretion system including flagellar components due to toxic accumulation of unexported T3SS1 components. An additional T3SS1 mutant with a defect in the master T3SS1 regulator, ExsA, was constructed and examined for its motility in combination with the ΔvcrD1 mutant (Fig. 4a, b). The ΔexsA mutant showed intact motility similar with the wild type. Interestingly, ΔvcrD1ΔexsA double mutant demonstrated non-motile phenotype as did the ΔvcrD1 mutant. A mutant V. parahaemolyticus with defects in LafK and FlaK was also included in the motility assays as a non-motile control. No significant difference in diameter of motility halos was detected among the ΔvcrD1, ΔvcrD1ΔexsA, and ΔflaKΔlafK strains when they were incubated up to 7 h in the motility plates (Fig. S2).

Transmission electron micrographs clearly showed that wild-type V. parahaemolyticus retained a polar flagellum but that the ΔvcrD1 mutant had completely lost its flagellum (Fig. 4c). Less than 1 % of the ΔvcrD1 mutant had a flagellum.

To confirm that loss of bacterial motility in the ΔvcrD1 mutant is derived from absence of the vcrD1 gene, a complementation strain, ΔvcrD1 mutant carrying pRKvcrD1, was examined for its motility along with two control strains, wild type with pRK415 and ΔvcrD1 strain with pRK415 (Fig. 5a, b). As expected, wild-type V. parahaemolyticus with pRK415 had motility, whereas ΔvcrD1 mutant carrying pRK415 did not show any motility. The ΔvcrD1 mutant with pRKvcrD1 had a comparable motility with wild type carrying pRK415.

Fig. 5
figure 5

Complementation of bacterial motility with an addition of wild-type vcrD1 gene. a Swimming motility of various V. parahaemolyticus strains (wild-type carrying pRK415, ΔvcrD1 mutant strain carrying pRK415, and ΔvcrD1 mutant strain carrying pRKvcrD1) was measured in LBS/3 % agar plates for 4 h. b Quantification of bacterial motility by measuring the diameters (mm) of the motility halos made by these V. parahaemolyticus strains. Data are presented as the means ± standard deviation from three independent experiments. Statistical analyses were performed using Student’s t test to evaluate the statistical significance of the results. Differences were considered significant when P < 0.05. Data with P < 0.01 are indicated with two asterisks

Full size image

Secretomes of these V. parahaemolyticus strains were analyzed by Western blot using anti-flagellar proteins, FlaA, FlgL, FlgE, and FlgM (Fig. S3A). Unexpectedly, FlaA protein was present at a similar level in these strains (wild type carrying pRK415, ΔvcrD1 mutant with pRK415, and ΔvcrD1 mutant with pRKvcrD1). The rest three proteins were found less in the secretome of the complemented strain, ΔvcrD1 mutant with pRKvcrD1. Scanning electron micrographs of these strains revealed that wild type carrying pRK415 and ΔvcrD1 mutant with pRKvcrD1 had a polar flagellum, whereas ΔvcrD1 mutant with pRK415 did not have (Fig. S3B).

Determination of the expression level of flagellar components in wild-type and ΔvcrD1 mutant V. parahaemolyticus strains

Since four flagellar proteins were present at lower levels in culture supernatant of the ΔvcrD1 mutant, we examined the cellular fractions containing these proteins in ΔvcrD1 mutant V. parahaemolyticus. Specifically, we determined whether FlaA or FlgM in the cytoplasm of the ΔvcrD1 mutant was not successfully secreted through the outer membrane. Extracts of wild-type and ΔvcrD1 V. parahaemolyticus strains were divided into four fractions (cytoplasmic, periplasmic, membrane, and culture supernatant) that were analyzed by Western blot using anti-FlaA or anti-FlgM antibodies (Fig. 6a, b, respectively). FlaA was detected in all four fractions from the wild-type strain, but was barely detected in any fraction from the ΔvcrD1 strain, indicating that decreased secretion of FlaA in the mutant resulted from decreased FlaA synthesis.

Fig. 6
figure 6

Fractionation of wild-type and ΔvcrD1 mutant V. parahaemolyticus into cytoplasmic, periplasmic, membrane and secretome proteins, and Western blot analysis of these fractions using anti-FlaA antibodies (a) or anti-FlgM antibodies (b). Two different amounts of each fraction (5–15 μg) were used for Western blot analysis. Lanes 1 and 3, cytoplasmic proteins of wild type; lanes 2 and 4, cytoplasmic proteins of ΔvcrD1 mutant; lanes 5 and 7, periplasmic proteins of wild type; lanes 6 and 8, periplasmic proteins of ΔvcrD1 mutant; lanes 9 and 11, membrane proteins of wild type; lanes 10 and 12, membrane proteins of ΔvcrD1 mutant; lanes 13 and 15, secretomes of wild type; lanes 14 and 16, secretomes of ΔvcrD1 mutant. Intracellular level of glucose-specific enzyme IIAGlc was monitored as a control for the cytoplasmic proteins, whereas OmpV serves as a loading control for membrane proteins. TDH is a loading control for secretome proteins. Nonspecific proteins are indicated with asterisks, whereas specific proteins (FlaA, FlgM, glucose-specific enzyme IIAGlc, OmpV, and TDH) are denoted with arrows

Full size image

FlgM was found in all of the four fractions of wild-type cells. The amount of FlgM in the cytoplasmic and secreted fractions from the ΔvcrD1 mutant was decreased, whereas it did not show any difference in the periplasmic and membrane fractions of the mutant compared with those of wild type. Western blot analysis of these fractions indicated that IIAGlc was found in the cytoplasm, whereas TDH was only detected in the secreted proteins. In all of the fractions, OmpV was found, but mainly presented in the membrane fractions. Decreased levels of both FlgM and FlaA in the cytoplasmic fraction of the ΔvcrD1 mutant raised a possibility that transcription of flagellar genes is decreased in this mutant.

Determination of transcript level of flagellar genes in wild-type and ΔvcrD1 mutant V. parahaemolyticus

Using primers designed to amplify specific flagellar genes (Table 2), the levels of flagellar gene transcripts were monitored in the wild-type and ΔvcrD1 mutant strains (Fig. 7). The gap gene encoding glyceraldehyde 3-phosphate dehydrogenase was constitutively expressed. Therefore, all transcript levels were normalized to the amount of gap transcript in the same RNA sample. Expression of two early flagellar genes, flaK and flaL, was not decreased in the ΔvcrD1 mutant. Rather that, flaK transcription was significantly increased by the vcrD1 mutation. Transcripts of the middle (flgE, and fliF) and late (fliD, and flaA) flagellar genes were decreased significantly in the ΔvcrD1 mutant (from 28 to 3 % of wild-type). In addition, expression of the flgM gene was also reduced in the ΔvcrD1 mutant, which is an anti-σ factor and one of middle/late genes. This result demonstrated that deficiency of vcrD1 resulted in decreased transcription of flagellar genes, especially middle and late flagellar genes.

Fig. 7
figure 7

Quantitative analysis of transcripts of seven flagellar genes. The wild-type and ΔvcrD1 mutant V. parahaemolyticus cultured in LBS broth up to OD600 = 0.8, were harvested to isolate RNAs using TRizol (BioRad), which were treated with the RNase-free DNase (Qiagen). Four micrograms of RNAs were used to make cDNAs using PrimeScript RT reagent kit (TaKaRa). cDNA was analyzed by qRT-PCR on a Light Cycler 480 II real-time PCR system (Roche Applied Science) using Light Cycler 480 DNA SYBR green I master kit. The amount of each transcript was estimated using the specific pair of primers (Table 2). The experiments were performed in triplicate for each sample. The gap gene of V. parahaemolyticus was used as an endogenous control to normalize each transcript in wild-type and ΔvcrD1 mutant cDNAs. Data analysis was based on the relative quantification method by determining the crossing point (C p) value using the Light Cycler 480 II real-time PCR system software program (Roche Applied Science, version LSC480 1.5.0.39). Data are presented as the means ± standard deviation from three independent experiments. Statistical analyses were performed using Student’s t test to evaluate the statistical significance of the results. Differences were considered significant when P < 0.05. Data with P < 0.01 are indicated with two asterisks, whereas data with P values between 0.01 and 0.05 are indicated with a single asterisk

Full size image

Discussion

Vibrio parahaemolyticus has two distinct NF-T3SSs, T3SS1 and T3SS2 (Park et al. 2004). VcrD protein is postulated to be a conserved member of the NF-T3SS secretion apparatus (Buttner 2012). Knocking out this conserved inner membrane proteins of these NF-T3SSs, VcrD1 or VcrD2, resulted in significant attenuation in bacterial cytotoxicity and enterotoxicity, respectively (Park et al. 2004; Yang et al. 2011). This microorganism has dual flagellar systems: (1) a single polar flagellum that is continuously produced and (2) numerous lateral flagella that are synthesized under certain conditions, such as surface conditions (McCarter 1999). Further characterization of the ΔvcrD1 mutant V. parahaemolyticus showed decreased secretion of several flagellar components (Figs. 1, 2, 3; Tables 3, 4) and raised a possibility that VcrD1, a component of NF-T3SS1, also plays a role in formation of the polar flagellum.

As expected, the ΔvcrD1 mutant did not have motility and the polar flagellum (Fig. 4). Absence of a polar flagellum on the ΔvcrD1 mutant of V. parahaemolyticus indicates that VcrD1-containing T3SS1 system might function in flagellum secretion. Non-motile phenotype of the double ΔexsAΔvcrD1 mutant clearly indicated that the exsA mutation could not alleviate the defect caused by the vcrD1 mutation (Fig. 4a, b). This data excluded a possibility that deficiency of VcrD1 causes malfunction of general secretion systems including the flagellar secretion. V parahaemolyticus has 60 potential flagellar genes, and their roles in flagellum assembly and transcriptional organization have been reported (Kim and McCarter 2000). Therefore, further experiments are needed to reveal how VcrD1 is involved in flagellum formation. For example, physical association of VcrD1 with other flagellar components is an interesting question needed to be verified.

Motility was restored when the ΔvcrD1 mutant was complemented with the wild-type vcrD1 gene (Fig. 5a, b). In contrast with the bacterial motility, flagellar components were found in the secretomes of all the strains (wild type carrying pRK415, mutant carrying pRK415, and the complemented vcrD1 mutant). A hypothesis was made to explain this discrepancy between motility and secretion of flagellar proteins that V. parahaemolyticus carrying pRK415 or pRKvcrD1 may have dramatic change in its membrane topology resulting in secretion of these proteins, but as inappropriate forms to assemble flagella. That is, the ΔvcrD1 mutant carrying pRK415 may secrete flagellar components, but fail to assemble the flagellum in a correct form as demonstrated in Fig. S3B.

Lower cytoplasmic level of FlaA protein in the ΔvcrD1 mutant suggests that decreased FlaA secretion could be caused by decreased expression of FlaA (Fig. 6a). Flagellum morphogenesis is a complex cascade of events, which requires coordinate expression 60 genes encoding structural subunits, regulatory proteins, and chemo-sensor machinery. These genes have been categorized into three groups in relation to their temporal expression during assembly of the polar flagellum: early, middle, and late genes (Aldridge and Hughes 2002).

Early genes encode FlaK and FlaLM, transcriptional regulators, required for expression of the middle genes. The middle genes transcribed by σ54 include ORFs encoding assembly proteins, σ28, hook proteins, and basal body proteins. Sigma 54 also transcribes the late genes encoding flagellins, motor proteins, and the anti-σ factor, FlgM (McCarter 2001). FlgM is an important regulatory factor for temporal coordination of flagella gene expression, which blocks premature expression of late filament genes by binding to FliA. Secretion of FlgM through a complete hook-basal-body results in expression of the late filament genes (Ohnishi et al. 1992). Therefore, we examined whether decreased expression of flagellar proteins is caused by the altered level of cytoplasmic FlgM (Fig. 6b). Western blot analysis indicated that FlgM expression is also decreased in the ΔvcrD1 mutant V. parahaemolyticus. This result suggests that the altered FlaA level is not due to increased amount of cytoplasmic FlgM caused by unsuccessful FlgM secretion.

Quantitative measurement of flagellar transcripts demonstrated that the ΔvcrD1 mutation affected early genes and middle/late genes differently (Fig. 7). Failed secretion of flagellar components has been reported to result in abortion of transcription of downstream flagellar genes (Chilcott and Hughes 2000). Malfunctioning flagellar secretion caused by VcrD1 deficiency does not interfere with the early steps of flagellum assembly, but prevents the transcriptional cascade in the middle and late stages of flagellum morphogenesis.

This observation suggests that the role of T3SS1 in cytotoxicity of V. parahaemolyticus should be re-evaluated. The loss of polar flagellum/motility in the ΔvcrD1 mutant (Fig. 4) may be a factor responsible for the attenuated cytotoxicity of the mutant. In V. vulnificus, flagellum/motility has been reported as an important virulence factor (Lee et al. 2004). So far, three effector proteins delivered by T3SS1 have been identified: VopQ (VepA) (Burdette et al. 2009), VopS (Yarbrough et al. 2009), and VPA0450 (Broberg et al. 2010). To define the exact role of T3SS1 as an NF-T3SS, additional mutants that are defective in secretion of these effectors but retain polar flagellum/motility should be analyzed. Another NF-T3SS of V. parahaemolyticus, T3SS2, has been shown to be involved in delivering several effectors, such as VopA/P, VopC, VopV, VopT, and VopL (Zhang and Orth 2013; Ham and Orth 2012). A T3SS2-deficient V. parahaemolyticus was constructed by deleting the vcrD2 gene that has an amino acid sequence 25 % identical to VcrD1 (Park et al. 2004; Lee and Park unpublished result). This ΔvcrD2 mutant demonstrated intact motility (Noh and Park unpublished result), suggesting that T3SS1 and T3SS2 of V. parahaemolyticus function differentially with respect to flagellum biogenesis.

Concluding remarks

Absence of the VcrD1 protein, which has been considered to be a NF-T3SS component, resulted in a loss of bacterial flagellum/motility in V. parahaemolyticus due to decreased secretion of flagellar components and decreased expression of middle and late flagellar genes. The regulatory mechanism for VcrD1 involvement in flagella formation needs to be elucidated in future studies.